21 October 2011 | $10

©2011PerkinElmer,Inc.400227_01.AlltrademarksorregisteredtrademarksarethepropertyofPerkinElmer,Inc.and/oritssubsidiaries.

When attrition rates, regulations and recall concerns are rising, getting better information faster

is critical. PerkinElmer’smultimode detection instruments, imaging and analysis systems, versatile

assay platforms and informatics solutions enable you to identify and characterize viable drug

candidates sooner. And alongwith ourmultivendor lab services, we help you succeed faster.

See us in action at www.perkinelmer.com/biomedstories1

GeneArt® gene synthesis, build your

next breakthrough

©2011 Life Technologies Corporation. All rights reserved. The trademarksmentioned herein are the property of Life Technologies Corporation or their

respective owners. For research use only. Not intended for any animal or human therapeutic or diagnostic use. CO18747

Go to www.invitrogen.com/geneart

Experience the new generation of gene synthesis and assembly tools

GeneArt® solutions give you predictability, adaptability, and confidence. Whether

increasing protein expression rates as much as 100X or assembling genetic constructs

with unprecedented precision, GeneArt® gene synthesis and assembly tools enable you

to break through your most complex challenges.

GE Healthcare

Life Sciences

For every success there’s a successor. It’s called

progress. So while we knew we had something

special in Amersham™ ECL Plus, our Western

blotting team was quietly working on the next

generation of detection reagent. The result: a new

substrate that operates with superior levels of

sensitivity, signal intensity and stability than even

its famous predecessor, making it an excellent

choice for CCD imagers.Welcome to ECL Prime.

Amersham and ECL are trademarks of GE Healthcare companies.

© 2010 General Electric Company – All rights reserved. First published November 2010.

GE Healthcare Bio-Sciences AB, Björkgatan 30, 751 84 Uppsala, Sweden.

Find out more about Amersham ECL Prime at

www.gelifesciences.com/eclprime

Comparison of Amersham ECL Prime and Amersham ECL Plus

detection of ß-catenin in NIH 3T3 whole cell lysates using different

dilutions of rabbit anti-ß-catenin and HRP-conjugated anti-rabbit IgG.

Thank you ECL™Plus,

you’ve been great.

Signal intensity sustained at higher antibody dilutions

Ab dilution Amersham ECL Prime Amersham ECL Plus

Primary Secondary 10 μg 156 ng 10 μg 156 ng

1:3000 1:300 000

1:10 000 1:500 000

CONTENTS Volume 334 Issue 6054

pages 315 & 389

page 304

EDITORIAL

289 The Cost of Doing NothingKathy Caldwell

NEWS OF THE WEEK

294 A roundup of the week’s top stories

NEWS & ANALYSIS

298 Vaccine Trial Meets Modest Expectations, Buoys Hopes

299 Drug-Screening Program Looking for a Home

300 Navy Dives Into Program Offering Cash for Good Scores

301 China Looks to Purge Academia of ‘Trash Journals’

302 Pre-Clovis Mastodon Hunters Make a Point>> Report p. 351

303 In Northern Aral Sea, Rebound Comes With a Big Catch

NEWS FOCUS

304 The Sterile Neutrino: Fertile Concept or Dead End?>> Science Podcast

307 Social Science for Pennies

308 Open-Source Ecology Takes Root Across the World

LETTERS

310 Editorial Expression of ConcernB. Alberts

Martial Arts Research: Prudent SkepticismJ. M. Strayhorn and J. C. Strayhorn

Martial Arts Research: Weak EvidenceJ. Mercer

ResponseA. Diamond and K. Lee

311 CORRECTIONS AND CLARIFICATIONS

311 TECHNICAL COMMENT ABSTRACTS

BOOKS ET AL.

312 Sybil ExposedD. Nathan, reviewed by B. Harris

EDUCATION FORUM

313 Rethink Summer Student ResearchF. A. Carrero-Martínez

PERSPECTIVES

315 The Strength of Electrical SynapsesS. Hestrin>> Report p. 389

316 Watery DisksR. Akeson>> Report p. 338

317 Antenna-Guided LightN. Engheta>> Research Article p. 333

318 Eddies Masquerade as Planetary WavesD. J. McGillicuddy Jr.>> Research Article p. 328

320 Up Close with Membrane Lipid-Protein ComplexesJ. Whitelegge>> Report p. 380

321 Every Bit CountsP. J. Thomas>> Report p. 354

REVIEW

323 Globalization, Land Use, and the Invasion of West Nile VirusA. M. Kilpatrick>> Science Podcast

CONTENTS continued >>

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 281

COVER

Illustration of a generalized form of Snell’s Law that fully accounts for light interaction with subwavelength structured materials, which can produce phase jumps at the interface between media. To illustrate this effect, origami ribbons are folded into tapered cylinders (phase jumps at interfaces), and the lines (light rays) form angles that depend on the degree of taper. Initially diverging rays converge after passing through two interfaces. See page 333.

Image: Nanfang Yu/Harvard University; Yi Tan and Jinhua Tan

DEPARTMENTS

285 This Week in Science290 Editors’ Choice292 Science Staff395 New Products396 Science Careers

CONTENTS

pages 320 & 380

page 373

page 328

RESEARCH ARTICLES

328 The Infl uence of Nonlinear Mesoscale Eddies on Near-Surface Oceanic ChlorophyllD. B. Chelton et al.Large ocean eddies are the cause of some sea-surface height and chlorophyll anomalies previously ascribed to Rossby waves.>> Perspective p. 318

333 Light Propagation with Phase Discontinuities: Generalized Laws of Refl ection and RefractionN. Yu et al.Light propagation can be controlled with plasmonic interfaces that introduce abrupt phase shifts along the optical path.>> Perspective p. 317

REPORTS

338 Detection of the Water Reservoir in a Forming Planetary SystemM. R. Hogerheijde et al.The detection of cold water vapor in a nearby planet-forming disk suggests that water ice exists in its outer regions. >> Perspective p. 316

340 Supramolecular Linear Heterojunction Composed of Graphite-Like Semiconducting Nanotubular SegmentsW. Zhang et al. A supramolecular self-assembly approach is used to make a nanotubular heterojunction.

343 Dynamics of the Reaction of Methane with Chlorine Atom on an Accurate Potential Energy SurfaceG. Czakó and J. M. BowmanTheory helps explain the counterintuitive impacts of vibrational excitation in a widely studied reaction.

347 800,000 Years of Abrupt Climate VariabilityS. Barker et al.Greenland climate variability for the past 800,000 years was inferred from the Antarctic ice-core temperature record.

351 Pre-Clovis Mastodon Hunting 13,800 Years Ago at the Manis Site, WashingtonM. R. Waters et al.Further dating of the Manis site shows that people were hunting mastodons in North America by 14,000 years ago.>> News story p. 302; Science Podcast

354 Information Transduction Capacity of Noisy Biochemical Signaling NetworksR. Cheong et al.Noise limits information transfer through a single signaling pathway in a single cell to just one bit.>> Perspective p. 321

358 ER Tubules Mark Sites of Mitochondrial DivisionJ. R. Friedman et al.Mitochondrial division occurs at positions where endoplasmic reticulum tubules contact mitochondria and mediate constriction.

362 Antimicrobial Peptides Keep Insect Endosymbionts Under ControlF. H. Login et al.A beetle species synthesizes an antimicrobial peptide to constrain a bacterial symbiont in specialized organs.

366 Stochastic Pulse Regulation in Bacterial Stress ResponseJ. C. W. Locke et al.Energy stress activates an alternative sigma factor in stochastic pulses and modulates pulse frequency to control activity.

369 Transgenerational Epigenetic Instability Is a Source of Novel Methylation VariantsR. J. Schmitz et al.Spontaneous methylation rates that may affect phenotype in the plant Arabidopsis are higher than the mutation rate.

373 Computation-Guided Backbone Grafting of a Discontinuous Motif onto a Protein ScaffoldM. L. Azoitei et al.A two-segment HIV epitope grafted into a scaffold protein maintains high affi nity for a broadly neutralizing antibody.

376 Antagonists Induce a Conformational Change in cIAP1 That Promotes AutoubiquitinationE. C. Dueber et al.Antagonist binding to an apoptosis inhibitor releases inhibition by promoting dimerization required for autoubiquitination.

380 Mass Spectrometry of Intact V-Type ATPases Reveals Bound Lipids and the Effects of Nucleotide BindingM. Zhou et al.The effect of lipids and nucleotides on the soluble head domain and membrane base domain is examined in an intact adenosine triphosphatase.>> Perspective p. 320

385 Cerebellum Shapes Hippocampal Spatial CodeC. Rochefort et al.Cerebellar protein kinase C–dependent mechanisms process self-motion information needed for spatial representation and accurate navigation.

389 Activity-Dependent Long-Term Depression of Electrical SynapsesJ. S. Haas et al.Paired bursting in coupled neurons depresses electrical synapses while their asymmetry increases after unidirectional use.>> Perspective p. 315

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org 282

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 283

CONTENTS

SCIENCEXPRESSwww.sciencexpress.org

Structural Dynamics of a Catalytic Monolayer Probed by Ultrafast 2D IR Vibrational EchoesD. E. Rosenfeld et al.A method to track fast vibrational motion in solution has been extended to catalytically important solid/liquid interfaces.10.1126/science.1211350

Wolbachia Enhance Drosophila Stem Cell Proliferation and Target the Germline Stem Cell NicheE. M. Fast et al.A bacterial endosymbiont up-regulates mitosis of Drosophila germline stem cells and blocks programmed cell death.10.1126/science.1209609

Endocannabinoid Hydrolysis Generates Brain Prostaglandins That Promote Neuroinfl ammationD. K. Nomura et al.A new tissue-specifi c pathway for the synthesis of proinfl ammatory prostaglandins is described. 10.1126/science.1209200

Recent Synchronous Radiation of a Living Fossil N. S. Nagalingum et al.Despite their ancient origin, the majority ofextant cycad species radiated within the past 10 million years.10.1126/science.1209926

Polarization of PAR Proteins by Advective Triggering of a Pattern-Forming System N. W. Goehring et al.Patterning of Caenorhabditis elegans zygotes involves passive as well as active mechanisms.10.1126/science.1208619

TECHNICALCOMMENTS

Comment on “How Cats Lap: Water Uptake by Felis catus”M. NauenbergFull text at www.sciencemag.org/cgi/content/full/334/6054/311-b

Response to Comment on “How Cats Lap: Water Uptake by Felis catus”R. Stocker et al.Full text at www.sciencemag.org/cgi/content/full/334/6054/311-c

SCIENCENOWwww.sciencenow.org Highlights From Our Daily News Coverage

Cute TV Chimps May Harm Their Wild BrethrenSeeing dressed-up chimpanzees doesn’t make viewers care more.http://scim.ag/chimpads

Seaweed With a Deadly Touch Toxins on the surfaces of common South Pacifi c algae are deadly to corals.http://scim.ag/CSlamphora

Winged Robots Hint at the Origins of FlightWhat were primitive wings used for before bird ancestors could fl y?http://scim.ag/robot-fl ight

SCIENCEONLINE

SCIENCE (ISSN 0036-8075) is published weekly on Friday, except the last week in December, by the American Association for the Advancement of Science, 1200 New York Avenue, NW, Washington, DC 20005. Periodicals Mail postage (publication No. 484460) paid at Washington, DC, and additional mailing offi ces. Copyright © 2011 by the American Association for the Advancement of Science. The title SCIENCE is a registered trademark of the AAAS. Domestic individual membership and subscription (51 issues): $149 ($74 allocated to subscription). Domestic institutional subscription (51 issues): $990; Foreign postage extra: Mexico, Caribbean (surface mail) $55; other countries (air assist delivery) $85. First class, airmail, student, and emeritus rates on request. Canadian rates with GST available upon request, GST #1254 88122. Publications Mail Agreement Number 1069624. Printed in the U.S.A.

Change of address: Allow 4 weeks, giving old and new addresses and 8-digit account number. Postmaster: Send change of address to AAAS, P.O. Box 96178, Washington, DC 20090–6178. Single-copy sales: $10.00 current issue, $15.00 back issue prepaid includes surface postage; bulk rates on request. Authorization to photocopy material for internal or personal use under circumstances not falling within the fair use provisions of the Copyright Act is granted by AAAS to libraries and other users registered with the Copyright Clearance Center (CCC) Transactional Reporting Service, provided that $25.00 per article is paid directly to CCC, 222 Rosewood Drive, Danvers, MA 01923. The identifi cation code for Science is 0036-8075. Science is indexed in the Reader’s Guide to Periodical Literature and in several specialized indexes.

SCIENCESIGNALING www.sciencesignaling.org The Signal Transduction Knowledge Environment18 October issue: http://scim.ag/ss101811

RESEARCH ARTICLE: Agrobacterium Counteracts Host-Induced Degradation of Its Effector F-Box ProteinS. Magori and V. CitovskyA plant pathogen prevents degradation of its key virulence factor in infected host cells.

REVIEW: Structural Basis for Activation and Inhibition of Class I Phosphoinositide 3-KinasesO. Vadas et al.The regulatory interactions between PI3Ks and their binding partners could be exploited for therapies.

JOURNAL CLUB: α-Synuclein Promotes Neuroprotection Through NF-κB–Mediated Transcriptional Regulation of Protein Kinase CδR. Aoki and Y. R. Li α-Synuclein may protect dopaminergic neurons from apoptosis by reducing the abundance of a kinase involved in cell death.

FUNDING SOURCESCheck out the latest update in opportunities for funding signaling research and training.

SCIENCETRANSLATIONAL MEDICINEwww.sciencetranslationalmedicine.org Integrating Medicine and Science19 October issue: http://scim.ag/stm101911

FOCUS: Containing the Contagion— Treating the Virus That Inspired the FilmB. Lee

RESEARCH ARTICLE: A Neutralizing Human Monoclonal Antibody Protects African Green Monkeys from Hendra Virus Challenge K. N. Bossart et al.A neutralizing human monoclonal antibody can fully protect nonhuman primates from disease after a lethal Hendra virus challenge.

PERSPECTIVE: Hedgehog Rushes to the Rescue of the Developing CerebellumO. Baud and P. Gressens

RESEARCH ARTICLE: A Small-Molecule Smoothened Agonist Prevents Glucocorticoid-Induced Neonatal Cerebellar InjuryV. M. Heine et al.

RESEARCH ARTICLE: Preterm Cerebellar Growth Impairment After Postnatal Exposure to GlucocorticoidsE. W. Y. Tam et al.A small molecule that targets the sonic hedgehog pathway protects against cerebellar injury caused by glucocorticoids, which may be given to preterm infants for treating lung disease.

SCIENCECAREERSwww.sciencecareers.org/career_magazine Free Career Resources for Scientists

Tooling Up: Views on an Interview, Part 1D. Jensen Scott Jackson’s interview at ABC technologies ranges from mundane to terrifying. http://scim.ag/TU_Views_1

Perspective: An Auxiliary (Research) Engine D. AlbertTo attain your goals, it is sometimes necessary to employ indirect means.http://scim.ag/Aux_Engine

Content Collection: Careers at Nonprofi tsE. PainScientists interested in a social or humanitarian cause can fi nd diverse research or alternative careers within nonprofi t organizations.http://scim.ag/CC_Nonprofi ts

SCIENCEPODCASTwww.sciencemag.org/multimedia/podcastFree Weekly Show

On the 21 October Science Podcast: mastodon hunting before Clovis, globalization and West Nile virus, the search for the “sterile neutrino,” and more.

SCIENCEINSIDERnews.sciencemag.org/scienceinsiderScience Policy News and Analysis

Built forperformance.DNA Ligases fromNew England Biolabs.

Why compromise any aspect of performance in your

cloning experiments? DNA ligases from NEB set the

industry standard for function and purity. Continuous

product development and extensive quality controls

ensure success in your ligation reactions.

DNA ligases from NEB – building a strong foundation

for your cloning experiments.

SPECIAL OFFERPurchase NEB 5-alpha Competent Cellswith NEB’s Quick Ligation Kit andreceive the item of lesser value for FREE

For more information, visit

www.neb.com/DNAligases SilverXpress™ is a trademark of Invitrogen.

Monarch caterpillar

© Maryellen Baker/Botanica

Equivalent amounts of protein were loaded and analyzed by

Coomassie gel and silver stained using SilverXpress™. Marker M

is NEB’s Broad Range Protein Marker (NEB #P7702).

Experience extreme purity with

NEB’s T4 DNA Ligase

NEB NEBLot Lot

M 101 102 A B C D

Supplier

285

EDITED BY STELLA HURTLEY

CR

ED

ITS

(TO

P TO

BO

TTO

M):

A. H

ED

DI A

ND

S. B

ALM

AN

D/B

F2I L

AB

OR

ATO

RY; Z

HA

NG

ET

AL.

Continued on page 287

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011

Virus InvasionWest Nile virus is spread through mosquitoes to birds, wildlife, and humans and has established itself at an astonishingly rapid rate since it was introduced to North America in 1999. How did the West Nile virus establish itself so successfully to the detriment of human and wildlife popula-tions? Kilpatrick (p. 323) reviews the scenarios and dynamics that point to the key bird hosts and the relative predilections of the associated mosquito vectors to feed on a variety of animals, including humans.

Making WavesPatterns of ocean chlorophyll variability from satellite observations have been attributed to oce-anic Rossby waves—slow-moving features with wavelengths of hundreds of kilometers but with sea surface heights of only centimeters—that take months or years to cross ocean basins from the west to the east. Chelton et al. (p. 328, pub-lished online 15 September; see the Perspective by McGillicuddy) report that the cause of these chlorophyll anomalies has been misidentifi ed. Analysis of 10 years of remotely observed sea sur-face height fi elds and concurrent observations of upper-ocean chlorophyll concentrations suggests that the dominant mechanism controlling the development of these anomalies is the horizontal advection of chlorophyll-rich surface water caused by the rotational motions of eddies.

Nano-Heterojunction Self-Assembly

Nanoscale materials can now be syn-thesized by a wide range of methods, including self-assembly techniques.

The junction regions between dissimilar materials often have unusual and desirable properties. Zhang et al. (p. 340) were able to extend the self-assembly toolbox to make heterojunctions of semiconducting nanotubular segments. The con-joined segments could transport electrical charge and also increase the lifetime of photogenerated charge carriers.

Controlling Light The behavior of light as it propagates through a material and from one material to another is very well understood in terms of classical optics. Yu et al. (p. 333, published online 1 September; see the cover; see the Perspective by Engheta) now dem-onstrate a powerful new method to control light propagation, based on introducing abrupt phase shifts along the optical path. These phase disconti-

nuities are constructed using plasmonic interfaces that consist of an optically thin two-dimensional matrix of optical antennas with subwavelength separation. The fl exibility of the technique should prove useful for developing a wide variety of small-footprint planar optical components.

Whence the Water Vapor?Water vapor has been detected in the inner regions of planet-forming disks—where terrestrial planets are created. Using the Heterodyne Instrument for the Far-Infrared on board the Herschel Space Observatory, Hogerheijde et al. (p. 338; see the Perspective by Akeson) now report the detection of water vapor over the full extent of the disk around the young star TW Hydrae. In the outer regions of this planet-forming disk, water vapor could only originate from icy grains. Thus, the result suggests the presence of a large reservoir of water ice in the region where comets and giant planets form.

When Cl and CH4 CollideThe simplest class of two-body chemical reaction is the formation of a diatomic molecule. Two atoms come together and, generally speaking, the only variable is their relative velocity. Things get considerably more complicated if you add another atom to the mix and consider its reaction with a preformed diatomic. Now, there are rela-

tive spatial orientations, and the diatomic might be vibrating or rotating. Nonetheless, over the past half-century or so, chemists have developed a fi rm grasp of how these reactions work in detail. The next frontier will be to understand how an atom reacts with a polyatomic, which has many different ways of vibrating. Czakó and Bowman (p. 343) simulated the reactivity of methane with a chlorine atom, providing a theoretical basis for a multitude of pivotal experiments on this system.

Polar Connections The climate records extracted from ice cores re-covered from the Greenland Ice Sheet are detailed but relatively short in duration—around 120,000 years. Ice cores from Antarctica, on the other hand, have lower temporal resolution but extend back more than 800,000 years. In order to infer how Greenland’s climate may have varied over a longer interval, Barker et al. (p. 347, published online 8 September) used the Antarctic tempera-ture record, data from Chinese speleothems, and the concept of the bipolar seesaw to produce a well-dated reconstruction of inferred Greenland temperature variability. Abrupt shifts in Northern Hemisphere climate appear to have occurred throughout the Late Pleistocene, and glacial terminations may have been linked to oscillations of the bipolar seesaw.

Control and Cooperation How do hosts regulate internal symbionts to prevent them from taking over their bodies without compromising the advantages of the relationship? Login et al. (p. 362) explored the balance between host innate immune responses and bacterial replication of an endosymbiont in weevils, an important beetle pest of wheat. A single peptide, coleoptericin-A (ColA), synthesized by the beetle constrained the bacteria within bacteriocytes and blocked bacterial replication. When the weevil’s ColA expression was silenced, the bacteria were able to replicate normally, escape the bacteriocyte, and spread throughout the insect.

Building upon your voice and our vision,

Roche scientists have developed new

reagents that position reliability and

comprehensive efficiency across

common and hard-to-transfect cell lines.

Created to meet the needs of researchers who require

high efficiency, ease of use and reproducibility,

X-tremeGENE reagents provide more viable cells with

more physiologically relevant results. Our newest

transfection reagents deliver results-driven performance,

allowing you to focus your efforts on discovery, and

helping you to shape the future. With Roche, you can

have complete confidence in the technology behind you.

Transfect with confidence.

For life science research only.

Not for use in diagnostic procedures.

X-TREMEGENE is a trademark of Roche.

© 2011 Roche Diagnostics. All rights reserved.

Roche Diagnostics Corporation

Roche Applied Science

Indianapolis, Indiana

Introducing X-tremeGENE DNA Transfection Reagents

Request your free sample today at

x-tremegene.roche.com

This Week in ScienceC

RE

DIT

: FR

IED

MA

N E

T A

L.

Hunting MastodonsA mastodon skeleton containing an embedded projectile tip was discovered in the late 1970s near Manis, Washington. It was initially dated as about 14,000 years ago but the age, and whether the bone containing the projectile was directly associated with the rest of the skeleton, has been questioned. Waters et al. (p. 351) provide new dates on the fossils that confi rm an age of about 14,000 years ago. The data, together with genetic analyses, show that the skeletal elements are related and also that the projectile was fashioned from a mastodon bone. This age predates the Clovis culture in North America and, along with other sites, shows early exploitation of megafauna.

Tumor Necrosis Factor ResponseEngineers use information theory to analyze how noise infl uences information transfer, for example, in telephone systems. Cheong et al. (p. 354, published online 15 September; see the Perspective by Thomas) have now applied such analysis in biological experiments by monitoring the response of thousands of single mouse fi broblasts to stimulation with various doses of tumor necrosis factor (TNF). Signal transmission was surprisingly noisy, meaning that the cells could only really differ-entiate whether TNF was present or not. Such limitations of a single signaling pathway appear to be overcome by the cooperation of multiple signaling pathways in networks, or by groups of cells collectively averaging their response to the same signal.

Mitochondrial Division Mitochondrial division regulates both the shape and the distribution of the mitochondrial network, which is important in maintaining cellular health. Friedman et al. (p. 358, pub-lished online 1 September; see the 14 October Perspective by Rambold and Lippincott-

Schwartz) demonstrate that mitochondria in both yeast and mammalian cells are constricted and divide at positions where they form stable contact sites with the endoplasmic reticulum.

Not in the GenesThe mechanism, distribution, and function of DNA methylation in plant genomes have been char-acterized, but the stability of DNA methylation over multiple generations and the rate of change are less well understood. Schmitz et al. (p. 369, published online 15 September) determined the methylation status of several previously sequenced Arabidopsis lines, including three ancestral and fi ve descendant lines separated by 30 generations. The frequency of DNA methylation changes was 5 orders of magnitude higher than genetic changes. Also, unlike genetic changes that were mostly random, DNA methylation changes occurred in hotspots.

Improving the FitDesigning proteins for specifi c functions often relies on grafting functional groups onto existing protein scaffolds. Success has been limited because backbone remodeling, which might allow more complex grafting, has been computationally challenging. Azoitei et al. (p. 373) integrated com-putational design and directed evolution to enable the manipulation of protein backbone structure required for transplantation of the backbone and side chains of discontinous functional motifs. They grafted a two-segment HIV gp120 epitope that is targeted by the cross-neutralizing antibody b12, onto an unrelated scaffold. The fi nal design showed high affi nity and specifi city for b12 and the complex structurally mimicked the gp120-b12 interaction.

Navigating with the CerebellumTo navigate in space, animals use two strategies: landmark-assisted or map-navigation, which requires self-motion cues. Although the cerebellum is known to assist in the coordination of self-motion informa-tion, its role in spatial navigation is unclear. Rochefort et al. (p. 385) examined whether impairment of the cerebellum affects the spatial code in the hippocampus, using transgenic mice with a selective disruption of protein kinase C–dependent plasticity at parallel fi ber–Purkinje cells synapses. Although landmark-assisted navigation was robust, when the mice had to rely on motion-generated cues—for example, in the dark—navigation suffered.

Continued from page 285

membercentral.aaas.org

105things you didn’t

(and 3 you probably

shouldn’t) know

about some of

your most

respected

colleagues.One more data point on why

you should spend more time

at membercentral.aaas.org.

There you can enjoy a feast

of blogs, videos, webinars,

discounts, and downloads

created by and for the most

insatiable brains around.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011

Sign Up At:www.sciencemag.org/webinar

REGISTER NOW!

Few, if any, proteins carry out their function in isolation. More likely is that they interact with one or more other proteins, during

which conformational changes and even physical modifi cations such as phosphorylation may occur. The ability to detect both

the interaction and modifi cation of proteins is critical to elucidating function and dysfunction in cells and tissues, leading

to a deeper understanding of normal as well as diseased states. This webinar will focus on current methodologies available

to study protein-protein interactions, as well as looking at new technologies for characterizing protein binding events in a

noninvasive manner that more closely resembles a cell’s natural environment, producing more physiologically relevant data.

During the webinar, our expert panelists will:

• Explain how protein structure and function can be affected by

posttranslational modifi cations and interactions with diverse partners

• Discuss current state-of-the-art methods and technologies for

characterizing protein-protein interactions and modifi cations

• Describe how multiplexing can be used to distinguish similar

biomolecular events, such as multiphosphorylated molecules or

multiple interacting partners

• Answer your questions live and in real time!

Webinar sponsored by

Brought to you by the

Science/AAAS Custom

Publishing Offi ce

Participating Experts

Anton Simeonov, Ph.D.National Institutes of Health

Bethesda, MD

James Bradner, M.D.Dana-Farber Cancer InstituteBoston, MA

Mathieu Arcand, Ph.D.PerkinElmerMontréal, Canada

Detecting Native Protein Interaction and PhosphorylationNew Approaches and Novel Technologies

WEBINAR November 9, 201112 noon Eastern, 9 a.m. Pacifi c, 5 p.m. GMT

The Cost of Doing Nothing AS THE U.S. CONGRESS GRAPPLES FOR SOLUTIONS TO THE ECONOMIC CRISIS, IT IS CRITICAL to recognize that rebuilding and modernizing infrastructure will be a key driver of economic growth. Recent reports issued by the American Society of Civil Engineers (ASCE),* the Urban Land Institute, and Building America’s Future‡ have described the deterioration of the nation’s energy, water, and transportation infrastructure. This week, ASCE convenes its annual meeting, gathering international scientists, engineers, policy-makers, and educa-tors to share sustainable solutions across a broad spectrum of concerns in the natural and built environment. Key among the discussions will be the role of scientists and engineers in developing effective public policy, helping to produce an infrastructure that incorporates new materials, technologies, and strategies to improve environmental and social well-being.

When the ASCE issued its 2009 Report Card for America’s Infrastructure, it gave the cumulative grade of “D” to the condition and performance of 15 of the country’s infrastructure systems.* Among the worst were roads and drinking water. The United States not only loses about seven bil-lion gallons of clean drinking water every day due to leaking water systems, but pipe failures and resulting fl oods have collapsed roads, destroyed homes, and endangered people. It would require an esti-mated $2.2 trillion over 5 years to raise the grade for all 15 infra-structure systems to an acceptable level. Sadly, the situation has not changed since the report was published. Earlier this year, the ASCE’s report Failure to Act, The Economic Impact of Current Investment Trends in Surface Transportation Infrastructure§ determined that the defi cient surface transportation infrastructure alone will cost U.S. businesses an added $430 billion (cumulative to 2020) in transpor-tation costs. By 2020, it is projected that exports will be $28 billion lower, 70,000 jobs will be lost, households will lose more than $7000 in personal income, and the country’s gross domestic product will take a hit of $897 billion. Businesses will need to divert increasing portions of income to pay for transportation delays, wasting money that could instead be invested in innovation. Nearly all sectors will suffer, but those associated with technology and innovation would probably be the hardest hit.

To meet the many infrastructure challenges, more fi nancing is needed. Now that the American Jobs Act has failed to pass Congress, there is discussion of breaking the bill into pieces that should be easier to pass. The proposed act includes $50 billion to modernize road, rail, and air transportation systems, and it would establish a National Infrastructure Bank to leverage public and private capital toward these endeavors. This level of priority and invest-ment is needed, or the United States will continue its downward slide. Indeed, this year’s report from the Urban Land Institute warns that the United States has fallen behind Brazil, China, and India in bolstering transportation, water, and sewage infrastructure.

Promoting a more sustainable and resilient infrastructure must also be part of this con-versation. Improved design and construction standards to withstand extreme conditions will require further R&D. Climate change and environmental preservation also require innova-tive infrastructure designs. Research is needed to determine the best ways to expand power generation and transmission. And the growing demands for information technology mean that underground utilities must be carefully planned.

Infrastructure investments provide an opportunity to improve the economy in the short term by creating jobs, while also driving the long-term growth needed to compete in the global mar-ketplace. Although repairing and modernizing the country’s infrastructure may seem daunting in lean times, the cost of doing nothing will be exponentially greater.

10.1126/science.1214039

– Kathy Caldwell

289

EDITORIALC

RE

DIT

S: (L

EFT

) ASC

E; (

RIG

HT)

FO

TOSE

AR

CH

Kathy Caldwell is president of the American Society of Civil Engineers. E-mail: kcaldwell@asce.org.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011

*www.asce.org/reportcard. www.uli.org/~/media/Documents/ResearchAndPublications/Reports/Infrastructure/Infrastructure2011.ashx. ‡www.bafuture.org/report. §www.asce.org/uploadedFiles/Infrastructure/Report_Card/ASCE-FailureToActFinal.pdf.

290

EDITED BY KRISTEN MUELLER AND JAKE YESTON

EDITORS’CHOICE

CR

ED

ITS

(TO

P TO

BO

TTO

M):

JAM

ES

KIN

G-H

OLM

ES/

SCIE

NC

E P

HO

TO L

IBR

ARY

; FO

TOSE

AR

CH

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org

O C E A N S C I E N C E

What Keeps the Storms Away?

The number of hurricanes that develop over the Atlantic Ocean each year and the number that make landfall in North and Central America are two distinct quantities. The difference between the two has great practical consequence, as hurricanes that remain offshore cause few deaths and do little damage to human prop-erty or infrastructure. What controls hurricane tracks, then? Wang et al. looked at sea surface temperature data from 1970 to 2009 and found that the size and location of the Atlantic warm pool help to steer hurricanes by infl uencing both where over the Atlantic they form and to what extent ensuing atmospheric circulation patterns push the storms away from the eastern seaboard of the United States. When the Atlantic warm pool is large, storms form more to the east, further from potential landfall, and the winds along their paths blow more strongly toward the northeast, also reducing the chance that the storms ultimately reach a vulnerable coast. Although these are not the only factors that control hurricane tracks, consideration of the sea surface temperature fi elds of the North Atlantic Ocean may help improve forecasts of potential hurricane dangers. — HJS

Geophys. Res. Lett. 38, L19702 (2011).

P L A N E T A R Y S C I E N C E

Why No Clay Up North?

The surface of Mars can be divided into two major regions: the northern lowlands and the southern highlands. The lowlands, covering around 1/3 of the planet, are thought to have once been the site of an ancient, great northern

ocean. However, this hypothesis is at odds with the record

of the presence of clay deposits. These

sediments, whose formation requires the presence of liquid water, are widespread in the southern highlands but very

scarcely distributed in the northern low-

lands. Using a climate model, Fairén et al. deter-

mined the surface temperatures on early Mars, assuming a southern super-continent and a northern ocean. The model tem-peratures imply that the northern ocean would

H Y D R O L O G Y

Rolling Down the River

All rivers naturally move loads of sediments, from coarse sand grains rolling along river-beds to tiny clay and silt particles carried in suspension. When sediment load gets too high though, either naturally or from hu-man activities, biodiversity suffers and water quality deteriorates. Identifying the sources of increased sediment loads, which can vary with such regional factors as land use and precipita-

have had to be a glacial ocean similar to the seas in Earth’s polar regions. Calculations of the rate of clay formation at subzero temperatures support the lack of clays in the northern low-lands, because their formation would have been inhibited at those cold temperatures. Moreover, the presence of glaciers surrounding the north-ern ocean would have limited the transport of continental sediments into the ocean, as is the case in the Arctic and Antarctic coastal regions of Earth. — MJC

Nat. Geosci. 4, 667 (2011).

E C O N O M I C S

Resource InvestmentAs Newton famously noted, researchers stand on the shoulders of giants, building on accu-mulated knowledge. But the mere production of knowledge does not ensure its use by others; societal benefi t depends also on mechanisms for storing and accessing knowledge. Research-ers have sought to understand how different institutions and policies can promote knowledge use and impact. To explore impacts of institutional resources in the life sciences, Furman and Stern studied the American Type Culture Collection (ATCC). Among the world’s largest bio re-source centers, ATCC maintains and distributes a vast collection of cell lines and microbiology cultures. Because each specimen deposited in ATCC is accompanied by an initial character-ization in a journal article, bibliometric analyses of article citations provided tools to assess impacts of ATCC. Besides comparing articles that did and did not link to ATCC specimens, the authors also analyzed the timing of citation “boosts,” because the deposition of specimens at ATCC often did not occur until sometime after the initial journal article describing the speci-men. The ATCC-deposit citation boost ranged from 57 to 135%, was higher for articles in less prestigious journals, and was concentrated on follow-up research into more complex subject matter. A rough approximation of “cost per citation” suggested that funders might consider increased investments in ensuring access to existing research rather than focusing so much on new research. — BW

Am. Econ. Rev. 101, 1933 (2011).

EDITORS’CHOICEC

RE

DIT

: NIK

ITA

S A

ND

LE

CU

IT

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011

tion, is critical for implementing remediation strategies. As a case study for determining the mechanism of sediment transport on the scale of several watersheds, Belmont et al. integrated a number of data sources—including geochemi-cal tracers, hydrologic fi eld measurements, and remote sensing—from tributaries or lakes in Minnesota along North America’s largest river, the Mississippi. Over the past 150 years, not only has the amount of fi ne-grained sediment increased by a factor of 10, but the sources have also changed. Historically, upland erosion of soil contributed most of the sediment load; however, the new data suggest that up to 70% of the sediment comes from the erosion of riverbanks and ravines themselves. The shift is probably a function of both natural and anthropogenic activity, including increased precipitation and extensive modifi cation of drainage networks for agricultural purposes. — NW

Environ. Sci. Technol. 45, 10.1021/es2019109 (2011).

M I C R O B I O L O G Y

Rapid Transport

E-cadherin is a species-specifi c receptor for the foodborne pathogen Listeria monocytogenes but it is located out of reach beneath the tight junctions formed between gut epithelial cells. Does the dynamic nature of the intestinal epi-thelium, which is being remodeled constantly, with cells being shed and mucus secreted, allow for E-cadherin to be accessed? Taking a step back from molecu-lar studies of pathogen cell invasion, Nikitas et al. watched how Listeria invades the body, using humanized mice and two-photon and confocal microscopy. They found that E-cadherin is not perpetually out of sight but becomes exposed to the intestinal lumen when cells are extruded and cell junctions are disrupted by contracting goblet cells or folds in the villi. Once inside the cell, the bacteria have no need for any other virulence factors, neither listeriolysin-O nor ActA (which polymerizes the cell’s actin). All that is required is the bacterial surface protein InlA for rapid apico-basal translocation mediated by the cell’s microtubules and exocytosis into the lamina propria. Thirty minutes after invasion, and Listeria had entered the spleen undetected by immune surveillance and a systemic infection was established. — CA

J. Exp.Med. 208, 10.1084/jem.20110560 (2011).

P S Y C H O L O G Y

That’s Not Yours!

Much ink has been spilled in arguments about what it is that children have learned when they begin to grasp the possibility that other people’s beliefs can differ from their own. But what do children comprehend of other people’s rights, such as the ownership of property? Rossano et al. describe experimental results indicating that 3-year-old children exhibit a more sophisticated understanding of the rights conferred by ownership—in this instance, the disposal of a cap or scarf—than 2-year-olds. Children of both ages complained when their own hat was thrown away by a puppet, and they did not protest when the puppet threw away his own article of clothing; the key distinction was that older kids registered a normative objection when the puppet discarded a hat belonging to a third party (the experimenter). In their introduction to an edited collection, Friedman and Ross enumerate the reasons why research on the developmental origins of ownership will yield fi ndings of interest. — GJC

Cognition 121, 219 (2011); New Dir. Child Ado-lesc. Dev. 132, 1 (2011).

C E L L S I G N A L I N G

Enlightening the Load

If you have driven a car with a manual transmis-sion, you are aware that the response to the throttle is quite different when the drive train is

connected to a “load” (when the clutch is engaged and the engine drives the wheels) than when it is not (when the engine spins freely with the clutch disengaged). Jiang et al. explored whether a similar concept of “load” applies to biochemical signaling systems; that is, whether the dynamic properties of a signaling mecha-nism were altered in the presence or absence of substrate molecules that are targets of the system.

Combined experiments and mathematical modeling showed that the presence of substrate could alter the response time of the system, increasing it when one of the enzymes in the signaling system was operating at a maximal rate (saturated) but decreasing it when theenzymes were operating in a linear manner. The authors discuss how such effects of downstream targets on the responsiveness of signaling systems might be used to design appropriate responses when modifying biological systems or designing synthetic ones. — LBR

Sci. Sig. 4, ra67 (2011).

AAAS is here –

bringing educationalinfrastructure to thedeveloping world.

AAAS is helping the Rwandan

government rebuild its educa-

tional infrastructure as a way

to help drive economic growth

and development. By providing

materials such as the Project

2061 Atlas of Science Literacy,

lesson plans from Science

NetLinks, and access to Science

digital libraries, AAAS is helping

the people of Rwanda work

toward a future built around

science and technology. As a

AAAS member your dues

support these efforts. If you’re

not yet a AAAS member, join us.

Together we can make a difference.

To learn more, visit

aaas.org/plusyou/rwanda

292 21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org

www.sciencemag.org

SENIOR EDITORIAL BOARDA. Paul Alivisatos, Lawrence Berkeley Nat'l. LaboratoryCori Bargmann, The Rockefeller Univ.Ernst Fehr, Univ. of ZurichRichard Losick, Harvard Univ.Michael S. Turner, University of Chicago

BOARD OF REVIEWING EDITORSAdriano Aguzzi, Univ. Hospital ZürichTakuzo Aida, Univ. of TokyoSonia Altizer, Univ. of GeorgiaSebastian Amigorena, Institut CurieAngelika Amon, MITKathryn Anderson, Memorial Sloan-Kettering Cancer CenterSiv G. E. Andersson, Uppsala Univ.Peter Andolfatto, Princeton Univ.Meinrat O. Andreae, Max Planck Inst., MainzJohn A. Bargh, Yale Univ.Ben Barres, Stanford Medical SchoolJordi Bascompte, Estación Biológica de Doñana, CSICFacundo Batista, London Research Inst.Ray H. Baughman, Univ. of Texas, DallasDavid Baum, Univ. of WisconsinYasmine Belkaid, NIAID, NIH Philip Benfey, Duke Univ. Stephen J. Benkovic, Penn State Univ.Gregory C. Beroza, Stanford Univ.Peer Bork, EMBLBernard Bourdon, Ecole Normale Superieure de Lyon Ian Boyd, Univ. of St. AndrewsPaul M. Brakefi eld, Univ. of Cambridge Christian Büchel, Universitätsklinikum Hamburg-EppendorfJoseph A. Burns, Cornell Univ. William P. Butz, Population Reference BureauGyorgy Buzsaki, Rutgers Univ.Mats Carlsson, Univ. of Oslo Mildred Cho, Stanford Univ. David Clapham, Children’s Hospital, Boston David Clary, Univ. of Oxford J. M. Claverie, CNRS, Marseille Jonathan D. Cohen, Princeton Univ.Robert Cook-Deegan, Duke Univ. Alan Cowman, Walter & Eliza Hall Inst. Robert H. Crabtree, Yale Univ.Wolfgang Cramer, Medit. Inst. for Ecology & PaleoecologyF. Fleming Crim, Univ. of Wisconsin Jeff L. Dangl, Univ. of North Carolina

Tom Daniel, Univ. of WashingtonStanislas Dehaene, Collège de FranceEmmanouil T. Dermitzakis, Univ. of Geneva Medical SchoolRobert Desimone, MITClaude Desplan, New York Univ.Ap Dijksterhuis, Radboud Univ. of NijmegenDennis Discher, Univ. of Pennsylvania Jennifer A. Doudna, Univ. of California, BerkeleyJulian Downward, Cancer Research UK Bruce Dunn, Univ. of California, Los Angeles Christopher Dye, WHODavid Ehrhardt, Carnegie Inst. of WashingtonMichael B. Elowitz, Calif. Inst. of TechnologyTim Elston, Univ. of North Carolina at Chapel Hill Gerhard Ertl, Fritz-Haber-Institut, Berlin Barry Everitt, Univ. of Cambridge Paul G. Falkowski, Rutgers Univ. Ernst Fehr, Univ. of ZurichTom Fenchel, Univ. of Copenhagen Alain Fischer, INSERM Wulfram Gerstner, EPFL LausanneKarl-Heinz Glassmeier, TU BraunschweigDiane Griffi n, Johns Hopkins Bloomberg School of Public HealthElizabeth Grove, Univ. of ChicagoTaekjip Ha, Univ. of Illinois at Urbana-ChampaignChristian Haass, Ludwig Maximilians Univ.Steven Hahn, Fred Hutchinson Cancer Research CenterGregory J. Hannon, Cold Spring Harbor Lab.Martin Heimann, Max Planck Inst., JenaIsaac Held, NOAA James A. Hendler, Rensselaer Polytechnic Inst.Janet G. Hering, Swiss Fed. Inst. of Aquatic Science & TechnologyRay Hilborn, Univ. of WashingtonMichael E. Himmel, National Renewable Energy Lab.Kai-Uwe Hinrichs, Univ. of BremenKei Hirose, Tokyo Inst. of TechnologyDavid Hodell, Univ. of CambridgeOve Hoegh-Guldberg, Univ. of QueenslandDavid Holden, Imperial CollegeLora Hooper, UT Southwestern Medical Ctr at DallasJeffrey A. Hubbell, EPFL LausanneSteven Jacobsen, Univ. of California, Los AngelesKai Johnsson, EPFL LausannePeter Jonas, Universität FreiburgWilliam Kaelin, Dana-Farber Cancer Inst.Barbara B. Kahn, Harvard Medical School

Daniel Kahne, Harvard Univ.Bernhard Keimer, Max Planck Inst., StuttgartJoel Kingsolver, Univ. of North Carolina at Chapel Hill Robert Kingston, Harvard Medical SchoolAlberto R. Kornblihtt, Univ. of Buenos AiresLeonid Kruglyak, Princeton Univ.Mitchell A. Lazar, Univ. of PennsylvaniaDavid Lazer, Harvard Univ. Virginia Lee, Univ. of PennsylvaniaOttoline Leyser, Cambridge Univ.Olle Lindvall, Univ. Hospital, LundMarcia C. Linn, Univ. of California, BerkeleyJohn Lis, Cornell Univ.Jianguo Liu, Michigan State Univ.Richard Losick, Harvard Univ.Jonathan Losos, Harvard Univ. Ke Lu, Chinese Acad. of SciencesLaura Machesky, CRUK Beatson Inst. for Cancer ResearchAndrew P. MacKenzie, Univ. of St Andrews Anne Magurran, Univ. of St AndrewsOscar Marin, CSIC & Univ. Miguel HernándezCharles Marshall, Univ. of California, BerkeleyMartin M. Matzuk, Baylor College of MedicineGraham Medley, Univ. of WarwickYasushi Miyashita, Univ. of TokyoRichard Morris, Univ. of EdinburghEdvard Moser, Norwegian Univ. of Science and TechnologySean Munro, MRC Lab. of Molecular BiologyThomas Murray, The Hastings CenterNaoto Nagaosa, Univ. of Tokyo James Nelson, Stanford Univ. School of Med. Timothy W. Nilsen, Case Western Reserve Univ. Pär Nordlund, Karolinska Inst.Helga Nowotny, European Research Advisory BoardLuke O'Neill, Trinity College, DublinStuart H. Orkin, Dana-Farber Cancer Inst.Christine Ortiz, MITElinor Ostrom, Indiana Univ.Andrew Oswald, Univ. of WarwickJane Parker, Max-Planck Inst. of Plant Breeding ResearchDonald R. Paul, Univ. of Texas at AustinP. David Pearson, Univ. of California, BerkeleyReginald M. Penner, Univ. of California, IrvineJohn H. J. Petrini, Memorial Sloan-Kettering Cancer CenterSimon Phillpot, Univ. of Florida Philippe Poulin, CNRS Colin Renfrew, Univ. of CambridgeTrevor Robbins, Univ. of Cambridge

Barbara A. Romanowicz, Univ. of California, BerkeleyJens Rostrup-Nielsen, Haldor TopsoeEdward M. Rubin, Lawrence Berkeley National LabMike Ryan, Univ. of Texas, AustinShimon Sakaguchi, Kyoto Univ.Miquel Salmeron, Lawrence Berkeley National LabJürgen Sandkühler, Medical Univ. of ViennaRandy Seeley, Univ. of CincinnatiVladimir Shalaev, Purdue Univ.Joseph Silk, Univ. of OxfordDenis Simon, Univ. of OregonAlison Smith, John Innes Centre Davor Solter, Inst. of Medical Biology, SingaporeJohn Speakman, Univ. of AberdeenAllan C. Spradling, Carnegie Institution of WashingtonJonathan Sprent, Garvan Inst. of Medical ResearchElsbeth Stern, ETH ZürichIra Tabas, Columbia Univ.Yoshiko Takahashi, Nara Inst. of Science and TechnologyJohn Thomas, Duke Univ.Herbert Virgin, Washington Univ. Bert Vogelstein, Johns Hopkins Univ.Cynthia Volkert, Univ. of GottingenBruce D. Walker, Harvard Medical SchoolDouglas Wallace, Leibniz Inst. of Marine SciencesIan Walmsley, Univ. of OxfordDavid A. Wardle, Swedish Univ. of Agric SciencesDavid Waxman, Fudan Univ.Detlef Weigel, Max Planck Inst., TübingenJonathan Weissman, Univ. of California, San FranciscoSue Wessler, Univ. of California, RiversideIan A. Wilson, The Scripps Res. Inst.Timothy D. Wilson, Univ. of Virginia Jan Zaanen, Leiden Univ.Kenneth Zaret, Univ. of Penn. School of MedicineMayana Zatz, University of Sao PaoloJonathan Zehr, Univ. of California, Santa CruzHuda Zoghbi, Baylor College of Medicine Maria Zuber, MIT

BOOK REVIEW BOARDJohn Aldrich, Duke Univ.David Bloom, Harvard Univ.Angela Creager, Princeton Univ.Richard Shweder, Univ. of ChicagoEd Wasserman, DuPontLewis Wolpert, Univ. College London

1200 New York Avenue, NW Washington, DC 20005

Editorial: 202-326-6550, FAX 202-289-7562News: 202-326-6581, FAX 202-371-9227Bateman House, 82-88 Hills Road

Cambridge, UK CB2 1LQ+44 (0) 1223 326500, FAX +44 (0) 1223 326501

SUBSCRIPTION SERVICES For change of address, missing issues, new orders and renewals, and payment questions: 866-434-AAAS (2227) or 202-326-6417, FAX 202-842-1065. Mailing addresses: AAAS, P.O. Box 96178, Washington, DC 20090-6178 or AAAS Member Services, 1200 New York Avenue, NW, Washington, DC 20005

INSTITUTIONAL SITE LICENSES please call 202-326-6755 for any questions or informationREPRINTS: Author Inquiries 800-635-7181 Commercial Inquiries 803-359-4578

PERMISSIONS 202-326-7074, FAX 202-682-0816MEMBER BENEFITS AAAS/Barnes&Noble.com bookstore www.aaas.org/bn; AAAS Online Store www.apisource.com/aaas/ code MKB6; AAAS Travels: Betchart Expeditions 800-252-4910; Apple Store www.apple.com/eppstore/aaas; Bank of America MasterCard 1-800-833-6262priority code FAA3YU; Cold Spring Harbor Laboratory Press Publications www.cshlpress.com/affiliates/aaas.htm; GEICO Auto Insurance www.geico.com/landingpage/go51.htm?logo=17624; Hertz 800-654-2200 CDP#343457; Office Depot https://bsd.offi cedepot.com/portalLogin.do; Seabury & Smith Life Insurance 800-424-9883; Subaru VIP Program 202-326-6417; VIP Moving Services www.vipmayfl ower.com/domestic/index.html; Other Benefi ts: AAAS Member Services 202-326-6417 or www.aaasmember.org.science_editors@aaas.org (for general editorial queries)science_letters@aaas.org (for queries about letters)science_reviews@aaas.org (for returning manuscript reviews)science_bookrevs@aaas.org (for book review queries)

Published by the American Association for the Advancement of Science (AAAS), Science serves its readers as a forum for the presentation and discussion of important issues related to the advancement of science, including the presentation of minority or conflicting points of view,rather than by publishing only material on which a consensus has been reached. Accordingly, all articles published in Science—including editorials, news and comment, and book reviews—are signed and refl ect the individual views of the authors and not offi cial points of view adopted by AAAS or the institutions with which the authors are affi liated.

AAAS was founded in 1848 and incorporated in 1874. Its mission is to advance science, engineering, and innovation throughout the world for the benefi t of all people. The goals of the association are to: enhance communication among scientists, engineers, and the public; promote and defend the integrity of science and its use; strengthen support for the science and technology enterprise; provide a voice for science on societal issues; promote the responsible use of science in public policy; strengthen and diversify the science and technology workforce; foster education in science and technology for everyone; increase public engagement with science and technology; and advance international cooperation in science.

INFORMATION FOR AUTHORS

See pages 784 and 785 of the 11 February 2011 issue or access www.sciencemag.org/about/authors

EDITOR-IN-CHIEF Bruce Alberts EXECUTIVE EDITOR NEWS EDITOR

Monica M. Bradford Colin NormanMANAGING EDITOR, RESEARCH JOURNALS Katrina L. KelnerDEPUTY EDITORS R. Brooks Hanson, Barbara R. Jasny, Andrew M. SugdenEDITORIAL SENIOR EDITORS/COMMENTARY Lisa D. Chong, Brad Wible; SENIOR EDITORS Gilbert J. Chin, Pamela J. Hines, Paula A. Kiberstis (Boston), Marc S. Lavine (Toronto), Beverly A. Purnell, L. Bryan Ray, Guy Riddihough, H. Jesse Smith, Phillip D. Szuromi (Tennessee), Valda Vinson, Jake S. Yeston, Laura M. Zahn (San Diego); ASSOCIATE EDITORS Kristen L. Mueller, Jelena Stajic, Sacha Vignieri, Nicholas S. Wigginton; BOOK REVIEW EDITOR Sherman J. Suter; ASSOCIATE LETTERS EDITOR Jennifer Sills; EDITORIAL MANAGER Cara Tate; SENIOR COPY EDITORS Jeffrey E. Cook, Cynthia Howe, Harry Jach, Lauren Kmec, Barbara P. Ordway, Trista Wagoner; COPY EDITOR Chris Filiatreau; SENIOR EDITORIAL COORDINATORS Carolyn Kyle, Beverly Shields; EDITORIAL COORDINATORS Joi S. Granger, Anita Wynn; PUBLICATIONS ASSISTANTS Ramatoulaye Diop, Aneera Dobbins, Jeffrey Hearn, Michael Hicks, Lisa Johnson, Scott Miller, Jerry Richardson, Teresa R. Sakon, Brian White; EDITORIAL ASSISTANT Patricia M. Moore; EXECUTIVE EDITORIAL ASSISTANT Yolanda O'Bannon (San Francisco); EXECUTIVE ASSISTANT Alison Crawford; ADMINISTRATIVE SUPPORT Maryrose Madrid; EDITORIAL FELLOW Melissa R. McCartney EDITORIAL DIRECTOR, WEB AND NEW MEDIA Stewart Wills; RESEARCH

ASSOCIATE Corinna Cohn; WEB DEVELOPMENT MANAGER Martyn Green;WEB DEVELOPER Andrew Whitesell; INTERNS Andrew Green, Kerry KleinNEWS DEPUTY NEWS EDITORS Robert Coontz, David Grimm (Online), Eliot Marshall, Jeffrey Mervis, Leslie Roberts, John Travis; CONTRIBUTING EDITORS Elizabeth Culotta, Polly Shulman; NEWS WRITERS Yudhijit Bhattacharjee, Adrian Cho, Jennifer Couzin-Frankel, Carolyn Gramling, Jocelyn Kaiser, Richard A. Kerr, David Malakoff, Greg Miller, Elizabeth Pennisi, Robert F. Service (Pacific NW), Erik Stokstad; WEB DEVELOPER Daniel Berger; INTERN Daniel Strain; CONTRIBUTING CORRESPONDENTS Jon Cohen (San Diego, CA), Daniel Ferber, Ann Gibbons, Sam Kean, Andrew Lawler, Mitch Leslie, Charles C. Mann, Virginia Morell, Gary Taubes; COPY EDITORS Linda B. Felaco, Melvin Gatling, Melissa Raimondi; ADMINISTRATIVE SUPPORT Scherraine Mack; BUREAUS San Diego, CA: 760-942-3252, FAX 760-942-4979; Pacifi c Northwest: 503-963-1940PRODUCTION DIRECTOR Wendy K. Shank; ASSISTANT MANAGER Rebecca Doshi; SENIOR SPECIALISTS Steve Forrester, Chris Redwood, Anthony Rosen; PREFLIGHT DIRECTOR David M. Tompkins; MANAGER Marcus Spiegler; SPECIALISTS Jason Hillman, Tara KellyART DIRECTOR Yael Fitzpatrick; ASSOCIATE ART DIRECTOR Laura Creveling; SENIOR ILLUSTRATORS Chris Bickel, Katharine Sutliff; ILLUSTRATOR Yana Hammond; SENIOR ART ASSOCIATES Holly Bishop, Preston Huey, Nayomi Kevitiyagala, Matthew Twombly; ART ASSOCIATE Kay Engman; PHOTO EDITOR Leslie Blizard

SCIENCE INTERNATIONAL

EUROPE (science@science-int.co.uk) EDITORIAL: INTERNATIONAL MANAGING EDITOR Andrew M. Sugden; SENIOR EDITOR/COMMENTARY Julia Fahrenkamp-Uppenbrink; SENIOR EDITORS Caroline Ash, Stella M. Hurtley, Ian S. Osborne, Peter Stern; ASSOCIATE EDITOR Maria Cruz; LOCUM EDITOR Helen Pickersgill; EDITORIAL SUPPORT Samantha Hogg, Alice Whaley; ADMINISTRATIVE SUPPORT John Cannell, Janet Clements, Louise Hartwell; NEWS: DEPUTY NEWS EDITOR, U.K. Daniel Clery; NEWSWRITER Sara Reardon; CONTRIBUTING

EDITOR, EUROPE Martin Enserink; CONTRIBUTING CORRESPONDENTS Michael Balter (Paris), John Bohannon (Vienna), Gretchen Vogel (Berlin)ASIA Japan Offi ce: Asca Corporation, Tomoko Furusawa, Rustic Bldg. 7F, 77 Tenjin-cho, Shinjuku-ku, Tokyo 162-0808, Japan; +81 3 6802 4616, FAX +81 3 6802 4615, inquiry@sciencemag.jp; ASIA NEWS EDITOR Richard Stone (Beijing: rstone@aaas.org); CONTRIBUTING CORRESPONDENTS Dennis Normile [Japan: +81 (0) 3 3391 0630, FAX +81 (0) 3 5936 3531; dnormile@gol.com]; Hao Xin [China: cindyhao@gmail.com]; Mara Hvistendahl [China: mara@marahvistendahl.com]; Pallava Bagla [South Asia: +91 (0) 11 2271 2896; pbagla@vsnl.com]

EXECUTIVE PUBLISHER Alan I. LeshnerPUBLISHER Beth Rosner

FULFILLMENT SYSTEMS AND OPERATIONS (membership@aaas.org); CUSTOMER SERVICE SUPERVISOR Pat Butler; SPECIALISTS LaVonda Crawford, Vicki Linton, April Marshall; DATA ENTRY SPECIALISTS Shirlene Hall, Tarrika Hill, William Jones

BUSINESS OPERATIONS AND ADMINISTRATION DIRECTOR Deborah Rivera-Wienhold; BUSINESS SYSTEMS AND FINANCIAL ANALYSIS DIRECTOR Randy Yi; MANAGER, FULFILLMENT SYSTEMS Frits Buningh; SYSTEMS ANALYST NicoleMehmedovich; MANAGER, BUSINESS ANALYSIS Eric Knott; MANAGER,BUSINESS OPERATIONS Jessica Tierney; BUSINESS ANALYSTS Priti Pamnani,Celeste Troxler; Christine Wehrli; RIGHTS AND PERMISSIONS: ADMINISTRATOR Emilie David; ASSOCIATE Elizabeth Sandler; MARKETING DIRECTOR Ian King; MARKETING MANAGERS Allison Pritchard, Alison Chandler, Julianne Wielga, Samantha Smith; MARKETING ASSOCIATES Aimee Aponte, Mary Ellen Crowley; SENIOR MARKETING EXECUTIVE Jennifer Reeves; DIRECTOR, SITE LICENSING Tom Ryan; DIRECTOR, CORPORATE RELATIONS Eileen Bernadette Moran; SENIOR PUBLISHER RELATIONS SPECIALIST Kiki Forsythe; PUBLISHER RELATIONS MANAGER Catherine Holland; PUBLISHER RELATIONS, EASTERN REGION Phillip Smith; PUBLISHER RELATIONS, WESTERN REGION Ryan Rexroth; CUSTOMER RELATIONS MANAGER Iquo Edim; MARKETING MAN-AGER Christina Schlecht; MARKETING ASSOCIATE Chad Johnson; ELEC-TRONIC MEDIA: DIRECTOR Lizabeth Harman; ASSISTANT MANAGER Lisa Stanford; SENIOR PRODUCTION SPECIALIST Ryan Atkins; PRODUCTION SPE-CIALISTS Antoinette Hodal, Nichele Johnston, Kimberly Oster; DIREC-TOR, WEB AND NEW MEDIA Will Collins; PROJECT MANAGER Trista Snyder; SENIOR PRODUCTION SPECIALIST Christopher Coleman; COMPUTER SPECIALISTS Walter Jones, Kai ZhangADVERTISING DIRECTOR, WORLDWIDE AD SALES Bill Moran COMMERCIAL EDITOR Sean Sanders: 202-326-6430ASSISTANT COMMERCIAL EDITOR Tianna Hicklin 202-326-6463

PRODUCT (sc ience_advert is ing@aaas.org) ; MIDWEST Rick Bongiovanni: 330-405-7080, FAX 330-405-7081; EAST COAST/E. CANADA Laurie Faraday: 508-747-9395, FAX 617-507-8189; WEST COAST/W. CANADA Lynne Stickrod: 415-931-9782, FAX 415-520-6940; UK/EUROPE/ASIA Roger Goncalves: TEL/FAX +41 43 243 1358; JAPAN ASCA Corporation, Makiko Hara: +81 (0) 3 6802 4616, FAX +81 (0) 3 6802 4615; ads@sciencemag.jp; CHINA/TAIWAN Ruolei Wu: +86 1367 1015 294 rwu@aaas.org WORLDWIDE ASSOCIATE DIRECTOR OF SCIENCE CAREERS Tracy Holmes: +44 (0) 1223 326525, FAX +44 (0) 1223 326532CLASSIFIED (advertise@sciencecareers.org); U.S.: MIDWEST/WEST COAST/SOUTH CENTRAL/CANADA Tina Burks: 202-326-6577; EAST COAST/INDUSTRY Elizabeth Early: 202-326-6578; SALES ADMINISTRATOR Marci Gallun; EUROPE/ROW SALES Simone Jux, Alex Palmer; SALES ASSIS-TANT Lisa Patterson; JAPAN Makiko Hara +81 (0) 90 9853-9982, FAX +81 (0)3-6369-4491; careerads@sciencemag.jp; CHINA/TAIWAN Ruolei Wu: +86 1367 1015 294 rwu@aaas.org; ADVERTISING SUPPORT MANAGER Karen Foote: 202-326-6740; ADVERTISING PRODUCTION OPERATIONS MANAGER Deborah Tompkins; SENIOR PRODUCTION SPECIALIST/GRAPHIC DESIGNER Amy Hardcastle; PRODUCTION SPECIALIST Yuse Lajiminmuhip; SENIOR TRAFFIC ASSOCIATE Christine Hall; SALES COORDINATOR Shirley Young AAAS BOARD OF DIRECTORS RETIRING PRESIDENT, CHAIR Alice Huang; PRESIDENT Nina Fedoroff; PRESIDENT-ELECT William Press; TREASURER David E. Shaw; CHIEF EXECUTIVE OFFICER Alan I. Leshner; BOARD Nancy Knowlton, Stephen Mayo, Raymond Orbach, Julia M. Phillips, Sue V. Rosser, David D. Sabatini, Inder Verma, Thomas A. Woolsey

I help higher educationprofessionals with their

RETIREMENT

SAVINGS

every day.

I’m

READY

TO HELP

you, too.

Keep in mind that investing involves risk. The value of your investment will fluctuate over time and you may gain or lose money.Although consultations are one on one, guidance provided by Fidelity is educational in nature, is not individualized, and is not intended to serve asthe primary or sole basis for your investment or tax-planning decisions.

Fidelity Brokerage Services LLC, Member NYSE, SIPC, 900 Salem Street, Smithfield, RI 02917. © 2011 FMR LLC. All rights reserved. 586605.1.0

866.715.6111

Fidelity.com/meetfidelity

I’m dedicated to helping higher education professionals feel confident

about retirement. It’s what I love to do — and I’m here for you, too.

� ���� �� ��� �� �� � ��� ������� ���� �� �� ����� ��� ��� �

� ���� �� ��� �� �� ��� ����� ������ ��� ����� ������� ��� ��� �����

� ���� �� ��� �� ������� ��� ��� ���� �� �� ���� �� �����

As a leading retirement plan provider for colleges and universities

across the country, we’re with you at every step. Start with a free,

no-obligation consultation with a Fidelity Investment Professional

— even if you aren’t working with us today.

Diane Bolden

Fidelity Investment Education Consultant

Meet us today.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org294

NEWS OF THE WEEK

CR

ED

ITS

(TO

P TO

BO

TTO

M):

RE

UTE

RS;

CN

ES

Sriharikota on the Bay of Bengal. The $125 million mission will study the

dynamics of cloud formation over the trop-ics, and how climate change could be affect-ing the monsoon. As it circles Earth near the equator, Megha Tropiques will revisit the same regions more than a dozen times each day, simultaneously measuring water vapor, clouds, precipitation, and radiation. These multiple measurements, mission scientists say, will give unique insights into the minu-tiae of how clouds are born and die during the monsoon.

India and France have agreed to make scientifi c data from the satellite freely avail-able—a welcome prospect for atmospheric scientists, says Christian Kummerow at Colorado State University, Fort Collins. “It is a very exciting mission and we do look forward to receiving the data from its instruments.”

Indian Ocean 3

Tsunami Warning System Passes Critical TestTwenty-three Indian Ocean nations came together on 12 October to test a new warn-ing communications network that might save lives the next time the region is pummeled by a tsunami. The $100 million Indian Ocean Tsunami Warning and Mitigation System performed well, though not fl awlessly, dur-ing a simulation modeled on the devastating tsunami of 26 December 2004 that killed over 230,000 people in 14 countries. Sev-eral countries also conducted dry runs of their own emergency response plans. India,

Kenya, and Malaysia conducted evacuation drills. An actual warning will depend on rap-idly analyzing data from numerous seismic stations, instrumented buoys, and sea-fl oor pressure sensors deployed over the past 6 years, and spreading the word through the networks tested last week.

The Intergovernmental Oceanographic Commission, a part of the United Nations Educational, Scientifi c and Cultural Orga-nization, which coordinated development of the system, declared the test a success. Australia, India, and Indonesia will now take responsibility for issuing warnings to the region. The Japan Meteorological Agency

and the United States’ Pacifi c Tsunami Warning Center have issued regional

warnings since 2005.

Moscow 4

Russian Scientists Rally to Protest Funding FreezeHundreds of researchers, many in lab coats, rallied in Moscow’s Pushkin Square 13 October to protest a funding freeze at Russia’s two grant organizations and on procurement regulations that they call major obstacles to research. The rally was organized by the trade union of the Russian Academy of Sciences (RAS) and the Young Scientists Council, together with associa-tions of Moscow State University students, and young scientists.

The Russian government recently froze the budgets for Russia’s two funding agen-cies, the Russian Foundation for Basic Research (RFBR) and the Russian Founda-tion for Humanities (RFH), leaving only $200 million for both agencies. Protestors urged the government to restore the old rule, under which RFBR received 6% of the overall budget for civilian science and RFH 1%. The protestors also demanded radical reform of laws governing public procure-ment, which severely limit grantees’ freedom

London 1

Royal Society: Plan Ahead For Nuclear Power Despite projections of low nuclear power growth in Europe and the United States, a renaissance of nuclear power construc-tion in China, Southeast Asia, and Russia is likely, Britain’s Royal Society notes in a report released 12 October. As a result, the report says, governments and international bodies need to develop long-term policies to account not only for safety but also for security, proliferation risk, and fuel cycle management.

“Spent fuel can no longer be an after-thought and governments worldwide need to face up to this issue,” Roger Cashmore, head of the U.K. Atomic Energy Authority and chair of the Royal Society working group that drafted the report, said in a statement.

The panel recommends that countries place their civil nuclear programs under international safeguards run by the Inter-national Atomic Energy Agency (IAEA), so that spent fuel cannot be diverted for weap-ons use. Countries that already have nuclear weapons should separate their civil and military nuclear programs. It also suggests setting up a World Nuclear Forum, made up of CEOs and government leaders, to discuss nuclear developments and responsibilities.http://scim.ag/nuclearUK

Sriharikota, India 2

Monsoon Satellite Promises Data DelugeThe Indo-French satellite Megha Tropiques, tasked with helping scientists understand the water and energy balance that controls monsoons, launched success-fully 12 October from the Indian space port

14

5

6

2

3

AROUND THE WORLD

Deluged. Damage in Phuket, Thailand, following the 2004 tsunami.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 295

NEWSC

RE

DIT

: © N

AN

CY

KE

DE

RSH

A

Making the Invisible Visible

For her colorful microscopy images that reveal the once-hidden secrets of cells, biochem-ist Nancy Kedersha has won the Lennart Nilsson Award for scientifi c and medical photog-raphy, presented annually in honor of Swedish photographer Lennart Nilsson. The award comes with a cash prize of SEK 100,000 (about $15,500). Kedersha will receive the award at a ceremony 8 November at the Berwald Hall in Stockholm, Sweden.

“Nancy Kedersha´s colour images open our eyes to the smallest components of life,” the award selection panel stated in a press release 14 October. “With the aid of a confocal microscope, she has turned biological data into an artistic experience.”

Kedersha is a researcher at Harvard Medical School and director of the confocal microscopy core at Brigham and Women’s Hospital in Boston. While working in the lab of biochemist and cell biologist Leonard Rome at the University of California, Los Angeles, in the 1980s, Kedersha developed a technique to stain and photograph cells to reveal their inner workings. Using this technique, she co-discovered a mysterious organelle called a vault that exists in everything from humans to slime molds. She has continued to develop techniques to identify different cell functions, distinguish healthy cells from cancerous ones, and observe cells dividing.

THEY SAID IT

“ If I’m going to take money from a citizen to put into education then I’m going to take that money to create jobs. So I want that money to go to degrees where people can get jobs in this state. Is it a vital interest of the state to have more anthropologists? I don’t think so.”

—Florida Governor Rick Scott (R) to the Sarasota Herald-Tribune on 10 October.

to spend the money as they see fi t. “This rally is a warning,” says Evgeny

Onishchenko of the RAS Institute of Phys-ics, one of the organizers. “We want to make it clear that if nothing is done to meet our demands, there will be much more serious rallies of researchers all over the country.”

Klong Luang, Pathum Thani, Thailand 5

Thai Floods Spare Research ParkFlood waters creeping toward Thailand’s big-gest research park forced the evacuation of dozens of public and private labs last week. The Thailand Science Park, 30 kilometers north of Bangkok, is home to 2700 employ-ees working in four national research insti-tutes under the National Science and Tech-nology Development Agency as well as in the labs of 60 private companies.

The campus closed 13 October; although the science park was still dry on Monday, it remained closed through 19 October due to high water in surrounding areas. The fl ooding, the worst in 50 years, has already claimed more than 300 lives and, according to a Businessweek report, caused over $5.1 billion in damage.

Washington, D.C. 6

House Panel Lays Out Spending PreferencesA climate-science satellite, some tech-nology commercialization efforts, and a chemical risk assessment program are all among the federal R&D programs that Republican leaders of the House of Repre-sentatives Committee on Science, Space, and Technology would cut to rein in the U.S. budget defi cit. The ideas, which also include protecting the core budgets of the National Science Foundation (NSF) and the Department of Energy’s (DOE’s) Offi ce of Science, were highlighted in an unusu-ally detailed 14-page letter that the law-makers sent on 14 October to Congress’s bipartisan Joint Select Committee on Defi -cit Reduction, which must devise a plan to trim at least $1.2 trillion from the defi cit over 10 years.

In general, the Republican lawmakers took a back-to-basics approach, arguing for protecting traditional science pro-grams while trimming many newer efforts championed by the Obama Administra-tion. They took an especially dim view of climate-related research; taxpayers could save $149 million over 5 years,

for instance, by axing NASA’s Orbiting Carbon Observatory-2, designed to map greenhouse gas emissions. They would phase out DOE’s Advanced Research Proj-ects Agency–Energy (ARPA-E), saving $180 million. All told, they proposed cuts totaling $1.5 billion.

The panel’s ranking Democrat, mean-while, penned a less specifi c plea for spar-ing the knife and fattening the federal purse. When it comes to funding science, it is “critically important” for the committee to “include serious revenue enhancements in its set of recommendations,” wrote Representa-tive Eddie Bernice Johnson (D–TX). Neither letter is likely to have a major impact on the defi cit committee, which faces a 23 Novem-ber deadline for delivering its plan.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org296

NEWS OF THE WEEK

entry, “Microstructure-Property rela-tionships in Ti2448 components pro-duced by Selective Laser Melting,” Miller, 32, fl ies in silvery spandex and a cape as he dances with women representing titanium’s alpha and beta crystalline forms. He receives $1000 and a trip to Belgium to be crowned the winner on 22 November at TEDxBrussels.

Category winners include “X-ray Crystal Structure of Human Protein Phosphatase,” by FoSheng Hsu of

Cornell University (Chemistry); “Smell-Mediated Response to Relatedness of Potential Mates,” by Cedric Tan of the Uni-versity of Oxford in the United Kingdom (Biology); and “A Study of Social Inter-activity Using Pigeon Courtship,” by Emma Ware of Queen’s University in Canada (Social Science). Videos of this year’s 55 Ph.D. dances are at www.gonzolabs.org/dance.

CR

ED

ITS

(TO

P TO

BO

TTO

M):

JOE

L M

ILLE

R; ©

MIN

ER

ALI

EN

TAG

E M

ÜN

CH

EN

NEWSMAKERS

CSI: Amphorae?

What did amphorae, the ubiquitous ceramic jugs of the ancient Mediterranean maritime world, actually contain? To fi nd out, mari-time archaeologist Brendan Foley of the Woods Hole Oceanographic Institution in Massachusetts and colleagues turned to a CSI-like method: swiping the insides of the amphorae with a swab. They swabbed nine 5th to 3rd century B.C.E. amphorae for DNA and compared it with snippets of DNA from various plants.

They identifi ed DNA from a range of commodities, including olive oil, olives, and wine, as well as traces of DNA from oregano, thyme, mint, and juniper, the team reports in a study published online this month in the Journal of Archaeological Science. Eight of the nine amphorae bore DNA from a complex mixture of foods, suggesting that amphorae were reused, Foley says.

Not everyone is convinced that the tech-nique works, however. It is “remarkable” that an amphora “should release endog-enous DNA by simply swabbing the sur-face,” Oliver Craig of the University of York in the United Kingdom said via e-mail. Craig, who specializes in recovering DNA and other molecules from ancient artifacts, says he would need to see more control tests to be convinced. http://scim.ag/CSIamphora

Nearly Intact Dino Fossil Found in Germany

An exceptionally well-preserved baby dinosaur, with traces of skin and protofeathers, will be the main attraction at a fossil and gem show next week in Munich. The juvenile theropod, which lived between 145 million and 150 million years ago and was probably less than a year old when it died, is 98% intact. That makes it the most nearly complete dinosaur ever found in Europe, says Oliver Rauhut, curator at the Bavarian State Collection for Palaeontology and Geology in Munich, who led the fi rst examinations of the fossil.

The fossil’s hairlike protofeathers may help researchers understand how and when feath-ers evolved. The roughly 70-cm specimen, not yet fully classifi ed or named, was unearthed near the Bavarian town of Kelheim and has been registered as a German cultural artifact, which means that it can’t leave the country. The mineral show will display the fossil for 4 days starting 27 October; a spokesperson for the show has said the unnamed owner plans to lend the fossil to a museum.

FINDINGS

Dances With TitaniumFifty-fi ve scientists around the world moon-lighted as choreographers for Science’s fourth annual “Dance Your Ph.D.” contest—and the results are in. This year’s winning dances were based on protein x-ray crystal-lography, fruit-fl y sex, and pigeon courtship, each scooping $500 prizes.

The grand winner, announced 20 Octo-ber, is Joel Miller, a biomedical engineer at the University of Western Australia in Perth. Miller’s dance, which won the phys-ics category, depicts his work with lasers to create titanium alloys strong and fl ex-ible enough for long-lasting hip replace-

ments. “We didn’t have a video camera,” says Miller. So he and his friends converted 2200 still photographs of the dance into stop-motion animation. In his winning

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 297

NEWSC

RE

DIT

S (T

OP

TO B

OTT

OM

): C

OU

RTE

SY J

ON

ATH

ON

KE

ATS

; MU

SEU

M O

F LO

ND

ON

AR

CH

AE

OLO

GY;

HO

ME

R S

YK

ES

AR

CH

IVE

/ALA

MY

BY THE NUMBERS1 billion tons Amount of extra food that that could be grown on agricultural lands now devoted to animal feed and biofuel produc-tion, according to an analysis in this week’s Nature.

200,000 amps Maximum power generated by a new airplane industry–sponsored lab at Cardiff University that studies the effects of lightning on materials. An average lightning strike generates 10,000 to 30,000 amps.

Random Sample

The Story Is Dead. Long Live the Story.Artist and self-styled experimental philosopher Jonathon Keats is hoping to persuade the art world to join scientists in the Copernican Revolution—nearly 5 centuries late. In 1543, Nicolaus Copernicus made the humbling observation that the Earth revolves around the sun. Modern physicists often cite the “Copernican principle” that, as nature’s rules are the same every-where, the human viewpoint isn’t unique.

But the art world, Keats says, is still stubbornly Ptolemaic, in that it emphasizes the “exceptionalism” of humans and centers on stories about ourselves. So, in “The First Copernican Art Manifesto,” an exhibit that opened Thursday at the Modernism gallery in San Fran-cisco, California, Keats will feature art that refl ects banal, average truths about the universe.

The pieces don’t assume a human audience or viewpoint—and they don’t aim to appeal to us, either. One canvas is painted a bland tan, the average color of the starlight of all stars measured by astronomers. Hydrogen gas released from glassware suspended above otherwise empty pedestals assumes a form invisible to human eyes. A quarter of the notes in a once-orderly Bach composition are rearranged—refl ecting the increasing entropy of the universe since its tidy, pre–big bang singularity.

Although not for humans, the exhibition is aimed at a particular demographic, in a way. “Were the aliens to land and see our show, they wouldn’t say, ‘Now I understand humanity,’ ” Keats says. “They’d say, ‘Now I have a better understanding of the universe.’ ” The exhibit runs through the end of November.

Cute TV Chimps May Harm Wild BrethrenSome entertainment industry moguls claim that chimpanzees dressed in clothes and clowning around fosters sympathy for the species. But a study published 12 October in PLoS ONE suggests the opposite: Peo-ple who watch such shows or ads decide chimpanzees are abundant in the wild and don’t need further protection.

Evolutionary anthropologist Brian Hare at Duke University in Durham, North Caro-lina, and colleagues asked 165 people to answer a questionnaire about the status of chimpanzees in the wild after watching tele-vision ads for products such as toothpaste and soft drinks. Mixed in with the ads was one of three short fi lms about chimpanzees. One showed Jane Goodall urging for their protection; another showed footage of chim-panzees in the wild; and the third showed chimpanzees “acting” in ads.

The results suggested absolutely “no support for the familiarity hypothesis,”

”

t -

Hare says. More than 35% of those who watched the humor-ous ads thought individuals should have the right to own a chimpanzee as a pet, compared with only 10% of those who watched the two other fi lms. Those who watched the enter-tainment chimps were also least likely to donate to a conserva-tion charity.http://scim.ag/chimpads

Black Death Spawned Modern Plague These skeletons—excavated in the 1980s from a 14th century graveyard in London—belonged to six of the estimated 30 million people who died from the Black Death, the plague epidemic that swept Europe between 1347 and 1351. Research-ers used teeth from the same graveyard—home to 2500 plague victims—to recon-struct 99% of the genome of Yersinia pestis, the bacterium that causes plague. An analysis of that microbial DNA pub-lished online 12 October in Nature sug-

gests that Y. pestis strains currently circu-lating around the world are all descendents of the medieval strain believed to have killed 30% to 60% of Europe’s population. The 14th century genome closely resem-bled those of modern strains and did not have any obvious unique mutations that might explain its unprecedented virulence. Other factors—such as the population’s susceptibility or the ecology of rodents and fl eas, which help spread the disease—were probably responsible for the medi-eval calamity, the team concludes.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org 298

NEWS & ANALYSIS

CR

ED

IT: J

OSE

PH O

KA

NG

A/R

EU

TER

S

Sometimes just meeting expectations is a major achievement. Initial, eagerly awaited results from the world’s first large-scale trial of a malaria vaccine, carried out at 11 sites in seven African countries, show that it reduced episodes of the disease by about half in babies and toddlers. That confi rms the effi cacy seen in earlier, much smaller trials of the experimental vaccine, so far the only one to show signifi cant benefi t against malaria in real-world settings.

The new fi ndings keep the candidate on track to become the fi rst licensed vaccine against the disease, says Christopher Plowe, a malaria vaccine expert at the University of Maryland School of Medicine in Baltimore, who is not involved in the trial. Even a par-tially effective vaccine, used in combination with other tools like bed nets, could curtail malaria’s massive death toll signifi cantly, experts say. But the vaccine will be expen-sive by developing-world standards, and its cost-effectiveness is yet to be determined.

“I am thrilled,” says Joe Cohen, one of the vaccine’s original developers and leader of the malaria vaccine project at GlaxoSmithKline (GSK) Biologicals in Rixensart, Belgium. The fact that the huge trial confirms results from smaller pre-decessors is “fabulous,” he says. Robert Newman, head of the World Health Orga-nization’s (WHO’s) malaria program, agrees. “The results are in line with what

we expected. But one fears they won’t hold up, so ‘in line’ is very encouraging.”

The fi rst round of results was published online on 18 October by The New Eng-land Journal of Medicine; they were also announced by Bill Gates—who called them “phenomenal”—at a Seattle meeting hosted by the Bill & Melinda Gates Foundation, which has given more than $200 million to support trials of the vaccine. The data show that in 6000 children aged 5 to 17 months, three doses of the vaccine cut the risk of any episode of malaria by 56% and the risk of severe disease by 47%. That’s far from the 90% effi cacy that most vaccines against viral and bacterial disease achieve. But the Plas-modium falciparum parasite, with its multi-ple life stages, is a much more diffi cult target, and no one expected a fi rst-generation vac-cine to be more than partially effective. “This vaccine will not be a magic bullet against what is a very, very diffi cult disease,” Cohen says. “It is one weapon to be added to an arse-nal of other interventions.”

The vaccine, called RTS,S, was developed in 1987 by researchers working for a prede-cessor to GSK Biologicals. It contains an engineered protein that combines a protein fragment from P. falciparum and a protein from the hepatitis B virus that helps trig-ger a strong immune response. The vaccine is designed to block the parasite’s ability to infect the liver and mature there.

After early human trials in 1997 showed promising results—protecting six of seven adult volunteers—GSK entered a public-private partnership with the PATH Malaria Vaccine Initiative (MVI) to further develop the vaccine. The fi rst fi eld trials in 2000 children in Mozambique, launched in 2003, showed that the vaccine lowered the risk of devel-oping malaria symptoms by 30%, with no severe side effects (Science, 22 October 2004, p. 587). Since then, phase II trials in Mozam-bique, Kenya, and Tanzania have consistently shown that the vaccine can cut the number of malaria episodes by between 35% and 53% (Science, 12 December 2008, p. 1622).

The phase III trial—the f inal test—enrolled more than 15,000 babies aged 6 to 12 weeks and toddlers between 5 and 17 months across sub-Saharan Africa. All were scheduled to receive three doses, each 1 month apart; a subgroup will receive a booster dose 18 months later. The results announced this week are for the toddlers and cover the 12 months after their first shot. (Infants were enrolled slightly later, so results from that group won’t be available until the end of 2012.) Children who missed one or two of the doses were almost as well protected as those who received all three shots, the researchers report.

The vaccine also looks fairly safe. Chil-dren who received the vaccine had a slightly higher rate of seizures than those who received the control injection, a rabies vac-cine. But the independent safety board that keeps watch over the trial has not raised any concerns, says MVI Director Christian Loucq. Children enrolled in the trial had a very low risk of dying from malaria—even if they received the control injections—mainly because clinics put in place proce-dures to detect and treat cases as soon as possible. There were only 10 malaria deaths in the fi rst 2 years of the study, Cohen says.

In a separate analysis, the researchers looked at the rate of severe malaria to date in all 15,460 children enrolled in the study; they found that the vaccine reduced the rate of severe, life-threatening disease by 35%. That hints that effectiveness might be lower in the babies, but Loucq and Cohen caution that the number is very preliminary.

The babies received their doses at the same time they receive the standard infant vaccinations recommended by WHO. Add-ing the malaria vaccine to existing vaccina-

Vaccine Trial Meets Modest Expectations, Buoys Hopes

M A L A R I A

Promising jab. A baby receives a dose of the experi-mental malaria vaccine at a trial site in Kilifi , Kenya.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 299

NEWSC

RE

DIT

S (T

OP

TO B

OTT

OM

): TH

IER

RY R

OG

E/R

EU

TER

S; M

LP/N

IH

tion schedules would be the most practical approach if the vaccine is to be widely used, and smaller-scale trials have suggested that it is effective and safe when given along with the other shots.

The trial will continue until 2014 and will follow the children until 30 months after their third dose. Once the full results are unblinded, researchers will know more about how long protection lasts and will also be able to com-pare the vaccine’s performance at different study sites, which have different rates and seasonal patterns of malaria trans mission. That could help governments and public health experts decide where RTS,S might have the most impact. The complex vaccine is expensive to make, and its cost-effective-ness is a major issue, says Scott Filler of the Global Fund to Fight AIDS, Tuberculosis and Malaria. “The key question is going to be cost,” he says, given the limited funds avail-able for fi ghting malaria.

GSK, which has invested more than $300 million in RTS,S to date, has pledged to keep the price as low as possible—just manufacturing costs plus a small return to be reinvested in development of second-

generation malaria vaccines or vaccines against other neglected tropical diseases. Even so, a full vaccine course is likely to cost more than other prevention methods, such as insecticide-treated bed nets, which also offer partial protection. Many countries have already rolled out massive net distribu-tion projects, and 75% of the children in the trial slept under a net, Cohen says; the results show that the vaccine can provide an extra layer of protection on top of the nets, he says.

WHO is expected to take all such issues into account when it drafts policy recom-mendations for use of the vaccine after the trial’s fi nal results come in. The vaccine has “incredible potential” to reduce suffering, Filler says, but deciding how and where to use it will take much more work. “These are going to be incredibly challenging questions for which we—the community as a whole—don’t have answers yet.”

–GRETCHEN VOGEL AND LESLIE ROBERTS

A $70-million-a-year program launched 7 years ago at the National Institutes of Health (NIH) to help academic research-ers move into industry-style drug discovery may soon be forced to scale back sharply. NIH Director Francis Collins has been one of its biggest champions. But the NIH Molecular Libraries, according to plan, must be weaned starting next year from the NIH director’s office Common Fund and fi nd support at other NIH institutes. In a time of tight budgets, nobody wants it.

The fate of the Molecular Libraries pro-gram became “an extremely sensitive politi-cal issue” earlier this year when NIH realized it would not be easy to fi nd a new home for the program, said one NIH offi cial speaking on background. It illustrates the “complexity of moving projects out of the Common Fund,” says the fund’s overseer, James Anderson, director of NIH’s Division of Program Coor-dination, Planning, and Strategic Initiatives. “Obviously” the process would be easier if NIH’s budget were growing, he says.

The Molecular Libraries began as a large piece of the NIH Roadmap, a set of cross-cutting initiatives announced in 2003 by then–NIH Director Elias Zerhouni. (The Roadmap later became the Common Fund.) As described by Collins, who was then the genome institute director, academic

researchers would submit protein or cell assays to a set of academic screening cen-ters, which would be paid by NIH to look for biological interactions with thousands of chemicals. Chemists would then refi ne these “hits” into research “probes,” some of which could become drug candidates.

Although some industry scientists were skeptical of government-led drug research, NIH scaled up the program from a pilot phase in 2008 (Science, 8 August 2008, p. 764). The program now includes four large screening centers—one part of NIH’s in-house research program, the rest extramural—and five supporting centers. They have produced more than 240 probes, one of which is now in clinical trials as a mul-tiple sclerosis (MS) drug. Hugh Rosen, direc-tor of a center at Scripps Research Institute in San Diego, California, said papers in top jour-nals—including those on the MS drug—show that the program “has transformed academic chemical discovery and target validation.”

The centers were told, however, that their Common Fund money would end in 2014 and that they would have to fi nd other sponsors. Only the intramural center has found a poten-tial home so far: It will move to NIH’s planned National Center for Advancing Translational Sciences, if Congress approves.

The extramural centers aren’t so lucky.

Their Common Fund awards will drop by 33% next June and another 33% in June 2013. Other institutes are “reluctant” to pick them up because they

need the money for research grants, accord-ing to Molecular Libraries program director Carson Loomis. Loomis says NIH intends to replace some of the declining money with new grants to users of the extramural cen-ters focused on drug discovery. Still, he says, “there will be some belt-tightening.”

John Reed, head of the Sanford-Burnham Medical Research Institute screening center in San Diego, which receives about $16 mil-lion a year from the Common Fund, says his center has so far attracted only modest fund-ing from drug companies. He expressed frus-tration with the Common Fund process. “NIH has put a huge investment into [the Molecular Libraries], and it’s running very well,” he says. “If there’s not a long-term commitment to keep it available to the academic community, why did we make this hundreds of millions of dollars investment?” –JOCELYN KAISER

Drug-Screening Program Looking for a HomeN AT I O N A L I N ST I T U T E S O F H E A LT H

2ath

Screen test. NIH’s Molecular Libraries has yielded scores of probes.

Long-term investment. Joe Cohen of GSK Biologicals has been working on the RTS,S vaccine for 24 years.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org 300

NEWS&ANALYSIS

SOU

RC

E: N

ATI

ON

AL

MA

TH A

ND

SC

IEN

CE

INIT

IATI

VE

, 201

1

The U.S. Offi ce of Naval Research (ONR) has joined forces with the Texas-based National Math and Science Initiative (NMSI) to increase the number of high school students taking rigorous science and math courses. Its $1.1 million grant, announced last month, could prompt a closer look at one controver-sial aspect of the initiative’s approach: paying students to do well on standardized tests.

Launched in 2007 with funding from major corporations and foundations, NMSI gives money to school districts that agree to follow its tightly scripted program. The goal is to boost participation in Advanced Place-ment (AP) classes, a curriculum designed by the nonprofi t College Board to be on par with entry-level college courses. The NMSI program, which targets low-achieving and low-income high schools with large minority popu-lations, assumes that these students can handle more challenging mate-rial if given the chance. NMSI offi -cials say that is already happening: In the midst of a nationwide surge in AP test-taking, the number of minority and female students pass-ing AP tests at 228 NMSI-sponsored schools this year increased four to 10 times faster than for the country as a whole (see graphic).

Most of NMSI’s approach rep-resents education orthodoxy: extra class time for students, additional resources, and special training for teachers. But there is less evi-dence behind another core element, namely, cash incentives. Students receive from $100 to $500 for pass-ing an end-of-the-year test, and their teachers also get bonuses for each stu-dent who succeeds and for teaching AP classes. NMSI pays for success in only English, math, and science courses (tests are offered in 34 subjects). But students may take half a dozen or more AP courses in those subjects during their high school careers, so the money can add up.

ONR’s investment puts the Navy ahead of the curve on national education policy. The Obama Administration has no offi-cial position on the use of cash incentives, and Congress has never addressed the topic in legislation. At the same time, Education Secretary Arne Duncan has voiced support for the concept as a tool for raising student

achievement, and a small departmental pro-gram to increase participation in AP courses allows offi cials to give money directly to stu-dents. However, only two of the 55 current grantees are doing so.

The Navy’s STEM (science, technology, engineering, and mathematics) education budget is expected to double in 5 years, to more than $100 million, refl ecting its con-cern about fi lling science- and technology-based Navy jobs. Michael Kassner, head of the Navy’s STEM offi ce, says NMSI could become “a big part of that investment” if the 3-year pilot, which supports three pub-lic schools in Virginia and Hawaii that enroll large numbers of students from military

families, is able to help prime the pump. “We know that students whose parents are in the military are more likely to go into the military,” he adds.

Kirabo Jackson, a labor economist at Northwestern University in Evanston, Illi-nois, has looked at the Texas program that spawned NMSI. His 2008 study, perhaps the only one to examine the role of incentives in the population that NMSI is targeting, found that the strategy increased AP participation rates and boosted the number of students with high scores on national college entrance tests. At the same time, the program didn’t increase

high school graduation rates or the number of students taking college entrance exams. That fi nding suggests it’s more likely to help high achievers already headed to college than to raise the aspirations of those who hadn’t planned to continue their education.

Still, his overall assessment is posi-tive: “It’s one of the few programs that does something good for these students,” Jackson says. “Most programs haven’t been evalu-ated rigorously. And I don’t know if it can be expanded to other settings, with other popu-lations. But if a school district had $1 mil-lion to spend, I think a program like this is a good investment.”

A 2010 study by economist Roland Fryer Jr. of Harvard University delivers a much more sobering message about the value of cash incentives for younger students. Fryer conducted a randomized trial of experimen-tal programs in four large urban school dis-tricts involving 38,000 children from grades two through nine. Although the program ele-ments varied greatly from one district to the next, he found that paying for outputs, such as test results, didn’t work and in some cases resulted in lower scores. On the other hand, paying students for inputs—showing up for class, staying on task, reading a certain num-ber of books—had a positive effect, he writes in a paper posted by the National Bureau of Economic Research.

Teacher unions have generally been very leery of incentives or bonuses. In addition to clashing with most labor agreements, they are regarded as undermining the learning process. “And almost none of these incentive programs have worked,” says a spokesperson for the United Federation of Teachers, which represents New York City schools.

A program serving 31 New York City schools that is otherwise modeled after the Texas and NMSI efforts omitted the teacher incentives after the union declined to participate. And this month offi cials for the program, called REACH (Rewarding Achievement), dropped the student incen-tives in response to a fi nancial squeeze. “We found that it prompted more students to take AP courses, but we didn’t see the magnitude effect that we had hoped,” says Kathrine Mott, REACH’s executive director. REACH will continue to offer professional devel-opment for teachers, Saturday classes, and classroom grants, she adds.

–JEFFREY MERVIS

Navy Dives Into Program Offering Cash for Good ScoresU. S . S C I E N C E E D U C AT I O N

Students taking AP tests

Pe

rce

nta

ge

inc

rea

se o

f pa

ssin

g s

co

res

FEMALES*

AFRICAN-AMERICAN

AND HISPANIC**

OVERALL**

0

50

100

150

200

250

23%

124%

50%

216%

20%

144%

*Math, Science**Math, Science, and English

U.S.Total

NMSISchools

Advancing AP. NMSI offi cials say that test results demon-strate the value of cash incentives in boosting the number of high school students taking and passing AP courses.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 301

NEWS&ANALYSISC

RE

DIT

: HO

NG

MIN

JIN

In China’s booming economy, there are many ways to get rich. For a husband-and-wife team on Hainan, an island off China’s southern coast, scientific publishing was their cash cow. For 7 years, Guo Hong and Fu Li operated 20-some journals, collecting a reported $1.5 million in publication fees from thousands of contributors. They solic-ited papers through elaborate Web sites, offering a discount on the publication fees common in China. But the journals were fake, provincial authorities allege. Upon receiving submissions, the couple would print up only a few copies—journal titles included Chinese Applied Nursing and Chinese Medicine Forum—to send to the author. Guo and Fu were detained in March; prosecutors have not yet fi led charges.

The highly publicized takedown is one of several recent efforts to clean up China’s academic publishing industry. In a country where low publication standards abound and every university or institute, it seems, has its own journal, the Chinese government is get-ting serious about raising standards. Although the Hainan journals fraud is an outlier, it’s symptomatic of a larger problem: slapdash and irrelevant publications read by next to no one. At most Chinese journals, “the academic level is not high,” Li Dongdong, vice director of the General Administration of Press and Publication (GAPP), which regulates publi-cations in China, noted in a speech in Decem-ber. She estimated that two-thirds of journals are “not market-oriented.”

As Chinese science barrels ahead, a few of its journals are getting international atten-tion, and leading Western publishers have set up shop on the mainland in response. But the country’s 4700 scientifi c periodicals include a hefty number of what the Chinese press refers to as “trash journals.” Despite being second only to the United States in total papers pub-lished from 2006 to 2010, China ranked at the bottom of the top 20 countries for citations per article over the same period, with just 1.47 citations on average, according to Elsevier’s SciVerse Scopus database and SciVal Spot-light country matrix, compared with 5.16 for the United States. It doesn’t help that many institutions in China offer fat rewards for pub-lishing in overseas journals with high impact factors. For example, according to its Web site, Guangzhou Medical University doles out 300,000 yuan ($47,000) to lead authors

on papers in journals with impact factors of at least 15—a level no Chinese journal has attained. (In Thomson Reuters’s Journal Citation Reports, top-ranked journals in cat-egories such as cell biology and biochemistry can show impact factors of over 30.)

As an antidote, GAPP has begun rolling out a series of reforms aimed at boosting the prestige of Chinese publishing. “GAPP has been given heaps of money to spend,” says Torsten Weise, a Berlin-based consultant who advises foreign publishers on operating in China. In the past 2 years, GAPP has secured billions of dollars in loans from state banks, with the “major goal,” Weise says, of inter-nationalization: building journals and pub-lishers capable of becoming multinational.

China’s 12th 5-year plan, in effect since March, sets a heady goal for journals. It calls for making cultural production—including media and publishing—a “pillar” industry. GAPP has moved swiftly. Earlier this year, the agency closed six obscure publications and reprimanded two others for violations that included indiscriminate printing of up to 200 papers per issue, over the limits set by publishing licenses. Then last summer, offi -cials unveiled China Science and Technology Media Group, one of a handful of fl agship publishers due to be rolled out over the next few years to compete with foreign rivals such as Wiley, Elsevier, and Springer. Li has talked about GAPP supporting a group of select academic journals; editors are unclear when funding might materialize.

The malaise has deep roots. The prolifera-tion of journals is tied to the danwei, or work unit, system put in place after 1949. As the government brought institutions under cen-tral control, academic work units—often uni-versity departments or institutes—launched journals to publish their scholars’ work.

Fast-forward to the 1990s. With Chi-nese science opening up, academic centers shifted course and began rewarding scientists for publishing in journals listed in indices that track citation rates. Pressure to publish piled up, and although only a few thousand Chinese papers a year then made it into jour-nals indexed by Thomson Reuters, authors seeking to get into print in both Chinese and English-language outlets proliferated. Medi-ocre danwei-linked journals gladly solicited papers from outside scientists and began charging steep publication fees. (These can

now top $1000.) And so the early journals persisted, constituting what Cong Cao, a scholar of Chinese science at the University of Nottingham, U.K., calls a “phenomenon with Chinese characteristics.”

Today, the few stars that have emerged are published in English and focus on areas in which China is strong, such as cell biol-ogy, nanoscience, and materials science. They have risen quickly: In 1999, the high-est impact factor of any Chinese journal was 0.5 (Science, 26 November 1999, p. 1683). Today, China’s top indexed journal, Cell Research, has an impact factor of 9.4.

China’s leading journals have made their mark by bringing on international editorial boards, wooing editors from top-shelf West-ern publications—Cell Research poached deputy editor in chief Li Dangsheng from Cell in 2006—and taking stabs at branding, such as shedding China-specific names in favor of more international monikers. Oth-ers, including Chinese Medical Journal and Science China Life Sciences, are experiment-ing with open-access platforms. “We have to change the way journals are run,” says Gang Pei, editor-in-chief of Cell Research. He has cultivated relationships with societies and accelerated response time for submissions from leading researchers.

The vast majority of journals have little hope of following that recipe. Their day of reckoning is not long off. “There is no need to keep poor-quality journals around,” says Meng Zhao, development editor at Neural Regeneration Research. “By administrative measures or by market measures,” Pei says, “there will be some kind of cleanup.”

–MARA HVISTENDAHL

China Looks to Purge Academia Of ‘Trash Journals’

S C I E NT I F I C P U B L I S H I N G

Raising the bar. Gang Pei’s Cell Research is China’s top indexed journal.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org 302

NEWS&ANALYSIS

CR

ED

ITS:

CE

NTE

R F

OR

TH

E S

TUD

Y O

F TH

E F

IRST

AM

ER

ICA

NS,

TE

XA

S A

&M

UN

IVE

RSI

TY

The experienced hunter hurled or thrust his spear at the left side of the 3-ton adult male mastodon. The bone point passed through 25 to 30 centimeters of hide and muscle, then pierced a rib bone. That wound alone likely did not kill the massive animal, but under the onslaught of a group of hunters, the mastodon eventually fell on its left side. The victorious hunters then retrieved most of their weapons and butchered the animal’s right side. But the valuable spear point remained inaccessible, buried under the giant carcass.

That ancient hunter’s loss is a gain for researchers on a different kind of hunt: the search for clues to the peopling of the Americas. Back in the 1970s, archaeologists found the mastodon’s remains, complete with rib bone and embed-ded point, at the Manis Mastodon site on Washington’s Olympic Peninsula, near the Juan de Fuca Strait. Now on page 351 of this issue, researchers led by Michael Waters of Texas A&M Uni-versity in College Station use DNA and radiocarbon dating to demonstrate that the point came from a mastodon bone shaped into a weapon by humans and used a startling 13,800 years ago. That’s nearly 1000 years before the Clovis culture, known for its distinctive stone spear points and long considered to be the fi rst culture in the New World.

The fi nd adds to the wave of recent com-pelling evidence demonstrating an earlier, pre-Clovis settling of the Americas (Science, 25 March, p. 1512). Although a few Clovis-fi rst holdouts remain unconvinced, the early bone point also suggests that the extinction of large mammals such as mastodons and mammoths may have begun long before the Clovis people came on the scene. “This is signifi cant because we have so few widely accepted pre-Clovis sites,” says anthropolo-gist Daniel Sandweiss of the University of Maine, Orono. The solid dating combined with the strong evidence for pre-Clovis hunt-ing on a site near the coast make the results particularly important, he adds.

Waters’s team subjected the rib and bone barb to a battery of tests, from DNA sequenc-ing and protein analyses to radiocarbon dating and a CT scan. They determined that the barb comes from another mastodon and appears to resemble the sharpened bone points used to kill mammoths, mastodons, and other large

animals in Beringia, the land now partly submerged in the region around the Bering Strait. The first Americans likely migrated from Beringia as the last ice age gave way to warmer temperatures and glaciers retreated.

For many archaeologists, the debate over whether pre-Clovis peoples roamed the Americas is over. “Manis is another pre- or non-Clovis site on the map,” says anthropol-ogist Tom Dillehay of Vanderbilt University

in Nashville, who announced his own pre-Clovis site in Chile back in 1997.

Megafauna like the mast-odon and its relative the mam-

moth disappeared quickly after the arrival of Clovis points 13,000 years ago, prompting some scientists to speak of a

“blitzkrieg”: a rapid hunting of these giant animals to extinction. But combined with evidence of pre-Clovis mammoth hunting at two other North American sites, Waters argues that human hunters were already at work on killing megafauna before the debut of Clovis-style weapons. Other researchers agree: “The notion of the blitzkrieg move-ment died with the Clovis-fi rst paradigm,” Dillehay asserts.

Although a well-dated point embedded in a mastodon rib seems like a smoking gun—or spear—for the pre-Clovis case, a handful

of Clovis-fi rst advocates remain skeptical. They say that so far all the evidence, includ-ing that from Manis, has problems. “There may have been a period of ‘pre-Clovis’ human presence in North America, but I wish I could see a solid demonstration of that presence somewhere that doesn’t have nagging problems,” says archaeologist Gary Haynes of the University of Nevada, Reno. He questions whether the bone point was really a human-shaped weapon, saying that it might have been a piece of bone acci-dentally driven into the rib. Archaeologist Stuart Fiedel of the Louis Berger Group Inc. in Richmond, Virginia, praises the team’s sophisticated techniques but questions the dating. He notes that the mastodon’s envi-ronment had sources of “old” carbon, including the ocean and geothermal pools,

that could give a falsely ancient date if the mastodon ingested food or water from those sources. Waters calls this a “red herring,” however, because the surrounding sediment age closely matches the bone dating.

Haynes adds that the oldest Clovis sites are “only” 8 centuries younger than Manis. Thus the rib “may actu-ally indicate the earliest beginning of the Clovis era, or an immediately proto-Clovis stage of human disper-sal,” he says. “Proto-Clovis” peoples in small numbers may have fi ltered south from Beringia as early as 14,000 years ago, he says, although their impact was negligible until the arrival of Clovis technology. To Waters, such talk of “proto-Clovis” amounts to “grasping at straws.”

Sandweiss says the implications of the paper in fact go beyond the Clovis–pre-Clovis wrangling. Many scientists argue that pre-Clovis peo-ple moved south along the Pacific

Coast, possibly by boat (Science, 4 March, p. 1122). In his view, the location of Manis near the ocean is an intriguing hint favor-ing this idea. “The Manis site supports early occupation of the coastal zone,” he says. Waters, who has reported pre-Clovis inte-rior sites as well, is cautious. “We can’t say these were coastal folks,” he says of the Manis hunters. But the paper’s findings, Sandweiss says, “point to a more wide-spread and complex early settlement system than some might have suspected.”

–ANDREW LAWLER

Pre-Clovis Mastodon Hunters Make a PointA R C H A E O LO G Y

Big game. Archaeologists say that pre-Clovis hunters in Washington state used a bone spear point (seen above embedded in a rib) to pierce a mastodon.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 303

NEWS&ANALYSISC

RE

DIT

S: (L

EFT

) CH

RIS

TOPH

ER

PA

LA; (

RIG

HT)

NA

SA’S

MO

DIS

RA

PID

RE

SPO

NSE

SY

STE

M (2

)

TASTUBEK, KAZAKHSTAN—At a fi shing camp near this village on the Northern Aral Sea, a dozen small boats recently returned to shore bathed in a soft morning light. Their nets bulged with carp, pike, fl ounder, perch, and a half-dozen other species, all edible and ready for sale. It was a scene unimagi-nable just 6 years ago, when the Aral—once the world’s fourth-largest lake—had shrunk to one-tenth of its original size. Soviet-era planners had diverted most of the water that fl owed into the lake from two rivers to irri-gate cotton, creating three smaller lakes that became too salty for most fi sh. Catches that once totaled more than 50,000 tons a year plummeted to just 52 tons in 2004.

Now, since the 2005 construction of a $65 million dike, the northern part of the Aral has become a remarkably healthy fi shery. The biomass, or weight of all the fi sh in the Northern Aral, has soared from an estimated 3500 tons in 2005 to 18,000 tons today, says Zaualkhan Yermakhanov, the Kazakh government’s regional fi sheries director. “And it’s still growing.”

“It’s been an amazingly fast recovery,” says Philip Micklin, a retired geographer from Western Michigan University who has been studying the Aral Sea since the 1970s. During an expedition last month around the Rhode Island–sized lake, Micklin found that salinity levels have dropped and oxy-gen levels increased since the Kazakh gov-ernment, backed by the World Bank and

other donors, built the 13-kilometer earthen dike along the Northern Aral’s southeastern edge. The dike traps water from one feeder river, the Syr Darya, and has raised the lake’s water level by 2 meters and expanded its surface area by some 900 square kilo-meters (Science, 14 April 2006, p. 183). During his expedition, Micklin found that the fresh water had reduced salinity from 12.3 grams of salt per liter in 2005 to 8 grams and increased water clarity. As a result, not only are fish becoming more abundant, but aquatic plants and reeds are spreading fast.

Prior to construction of the dike, only a hardy fl ounder introduced from the Black Sea was able to survive in the northern part of the lake. Now, about two dozen species of native freshwater fi sh that high salinity had driven into the Syr Darya’s delta and adja-cent lakes have returned to deeper waters and are reproducing at a rapid clip. And the rebound has been even bigger than World Bank planners expected: In the early 2000s, says the bank’s Masood Ahmad, a feasi-bility study had concluded that a revived Northern Aral would ultimately produce about 2000 tons of fi sh a year—but fi shing fl eets are already catching nearly twice that amount. To make sure the stocks continue to fl ourish, Yermakhanov says he is commit-ted to limiting yearly catches to about one-third of the total biomass—much less than what is allowed in many managed fi sheries.

“I know we could fi sh more,” he says, “but I want to make sure we can grow the biomass to at least 40,000 tons.”

Kazakhstan, flush with income from oil and minerals, is now considering tak-ing the rescue effort a step further. One project, backed by Yermakhanov and local f ishers, would raise the dike enough to increase water levels by more than 6 meters, expanding the lake surface by about 50%, to 5000 square kilometers. That plan “would be the best for the ecosystem,” Micklin says, but it isn’t as politically or economi-cally attractive as an alternative. That would involve digging a canal to bring water back to the historic port of Aralsk, which was left high and dry by the desiccation. The canal plan “would benefi t more people,” Micklin says. For the moment, however, “there isn’t enough water fl ow for both.” The govern-ment announced last month that it would appoint an expert panel to decide between the two options.

–CHRISTOPHER PALAChristopher Pala is a writer based in Almaty, Kazakhstan.

In Northern Aral Sea, Rebound Comes With a Big Catch

E C O LO G I C A L R E STO R AT I O N

Aral redux. More freshwater in the Northern Aral (above, at top) has enabled fi shers to once again pick a rich catch out of their gill nets (left).

2004

Tastubek

North Aral Sea

2011

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org 304

CR

ED

IT: L

OS

ALA

MO

S N

ATI

ON

AL

LAB

OR

ATO

RY

BLACKSBURG, VIRGINIA—Unlike old sol-diers, some scientifi c concepts seem never to fade away. Take the hypothetical subatomic particle called the “sterile neutrino,” which would be about the oddest bit of matter imag-inable. For 15 years, researchers have accu-mulated hints from particle physics, nuclear physics, astrophysics, and cosmology that the particle—a more-elusive cousin of the nearly undetectable neutrinos—might be out there. But most physicists have found the evidence unconvincing, as most of the results pointing toward sterile neutrinos are of marginal statis-tical signifi cance.

Recently, however, the case for sterile neu-trinos has grown stronger, bolstered by a new analysis of data from nuclear reactors. So last month 60 physicists from around the world gathered here* to hash out the arguments for

and against the existence of sterile neutrinos and to try to decide whether it’s worth stag-ing a dedicated experiment to settle the matter.

Performing such an experiment won’t be easy. The hypothetical neutrinos are called sterile because they do not interact at all with known particles. “You’re trying to prove the existence of something with no interactions,” says Patrick Huber, a theorist here at Virginia Polytechnic Institute and State Uni-versity (Virginia Tech). “It’s like try-ing to prove the existence of God.” Still, he says, it’s time to fi gure out what it will take to discover or rule out sterile neutrinos once and for all. “I’m afraid we’ll have the same workshop 15 years from now and will just have more [inconclusive] results that don’t make the situation any clearer.”

Some researchers say the case for a sterile neutrino is still half-baked. “I’m quite skepti-

cal,” says Yves Déclais, a neutrino physicist at the University of Lyon in France. “Each piece of evidence itself is not completely self-consistent,” he says. “So I’m really con-cerned that there should be more work to understand each anomaly itself instead of trying to put together a dedicated experiment to look for sterile neutrinos.”

Abundant in theoryOrdinary neutrinos are already weird. Nearly massless and hardly interacting with other matter, they are born in “weak” nuclear decays and interactions. For example, a neutron decays into a proton by emitting an electron and an antineutrino. A neutrino can emerge when a nucleus of the isotope beryllium-7 turns into lithium-7 by capturing an electron and releasing a neutrino. Trillions of neutrinos stream through each of us every second.

Weirder still, neutrinos come in three “fl avors”—electron neutrinos, muon neutri-nos, and tau neutrinos—that can morph into one another. For example, when cosmic rays strike the atmosphere, they create particles called muons that decay much as neutrons do, to produce muon neutrinos. The muon neutri-nos can then “oscillate” or “mix” into other fl avors before reaching Earth, as observed in 1998 by physicists using a giant subterranean detector called Super-Kamiokande in Japan. Electron neutrinos from the sun also change fl avor, as physicists at the Sudbury Neutrino Observatory in Canada showed in 2001.

A sterile neutrino would be even more elu-sive than an ordinary neutrino. It would not participate in weak interactions and would arise only from ordinary neutrinos oscillating into a sterile form. As sterile neutrinos would not interact themselves, physicists could detect them only indirectly, by observing ordinary neutrinos disappearing or appearing where they are not expected.

Theorists have been think-ing about sterile neutrinos since the late 1960s, when they fi rst suspected that neutrinos from the sun oscillated. The morph-ing meant that neutrinos were not massless, and sterile neutri-nos would help explain how the

wispy particles put on weight.Flavor-changing oscillations prove that

neutrinos have mass because a massless par-ticle must travel at light speed, and, accord-ing to Einstein’s theory of relativity, at light speed time stands still, making change impossible. The standard model of particle

Mystery machine. The guts of the LSND detector, which may have seen sterile neutrinos.

Onlinesciencemag.org

Podcast interview with author Adrian Cho.

The Sterile Neutrino: Fertile Concept or Dead End?Dozens of physicists gathered recently to debate whether the phantom particle exists and if it’s worth hunting it

NEWSFOCUS

*Sterile Neutrinos at the Crossroads, 26–28 September.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 305

NEWSFOCUSC

RE

DIT

S (T

OP

TO B

OTT

OM

): N

ASA

/WM

AP

SCIE

NC

E T

EA

M; M

. BLA

NTO

N A

ND

TH

E S

LOA

N D

IGIT

AL

SKY

SU

RV

EY

(SD

SS) C

OLL

AB

OR

ATI

ON

, HTT

P://

WW

W.S

DSS

.OR

G

physics assumes that neutrinos are massless, but most extensions of the theory that fix that problem include sterile neutrinos, says Paul Langacker, a theorist at the Institute for Advanced Study in Princeton, New Jersey. “The sterile neutrino is not something bizarre or exotic,” he says.

The details involve another key fact: As far as physicists know, all neutrinos spiral to the left, like footballs thrown by left-handed quarterbacks, and all antineutrinos spiral to the right. That would be fi ne if neutrinos traveled at unobtainable light speed. But as neutrinos have mass and travel slower, it’s possible in principle for an observer to over-take a left-handed neutrino. The neu-trino would then appear to travel the opposite way, as a right-handed neutrino—a particle not found in nature.

Theorists have found two ways around this problem. First, when overtaken, an ordi-nary left-handed neutrino could appear instead as a heavy right-handed sterile neutrino. Or second, the overtaken neutrino could appear as a right-handed antineutrino. Even then, a theoreti-cal “seesaw mechanism” would require heavy sterile neutrinos to explain why ordinary neu-trinos are so light.

That’s plenty of reason to think sterile neu-trinos are out there. However, most theories assume that sterile neutrinos are far heavier than ordinary neutrinos. For that reason and others, theory doesn’t generally allow an ordi-nary neutrino to just morph into a sterile neu-trino in the way some experiments indicate, Langacker says. So it’s not clear that theorists and experimenters are stalking the same beast.

Evidence of all sortsThe strongest experimental evidence for ster-ile neutrinos comes from the Liquid Scintil-lator Neutrino Detector (LSND), which ran at Los Alamos National Laboratory in New Mexico from 1993 through 1998. Using a par-ticle accelerator, physicists generated muon antineutrinos that streamed through a detector fi lled with 167 tons of mineral oil.

Those low-energy muon antineutrinos should have passed right through. However, an electron antineutrino could interact with the detector by merging with a proton to cre-ate a positron and a neutron—essentially, the weak decay of the neutron run backward. Thus, physicists could spot electron anti-neutrinos appearing in a beam of muon anti-neutrinos. And they spotted 88 of them, give or take 23. “Lo and behold, we saw an excess of events,” says Los Alamos’s William Louis.

Reported in 1996 and 2001, the LSND results might seem to show muon anti-neutrinos mixing into electron antineutrinos. But it couldn’t be that simple, Louis says. Dif-ferent fl avors of neutrinos mix at a rate that depends on the difference in their masses: The bigger the mass difference, the faster the mixing. Studies of atmospheric and solar neutrinos had placed limits on the mass dif-ferences among the three neutrino fl avors, and the values were too low to explain the lickety-split mixing that LSND saw as the particles fl ew just 30 meters, Louis says.

LSND researchers could explain their results, however, if muon antineutrinos oscil-lated fi rst into sterile antineutrinos and then into electron antineutrinos. The sterile neu-trinos would have to be heavier than ordinary neutrinos by 1 electron volt—about 100 times the differences among ordinary neutrinos.

Hints of extra neutrinos also come from the heavens. For example, cosmologists think the universe burst into existence in the big bang as an ultrahot, ultradense soup of particles. Tiny fl uctuations in the density of the soup then stretched to immense propor-tions during a faster-than-light growth spurt known as infl ation and seeded the formation

of galaxies. The fl uctuations also limit the number of neutrino types, theorist Kevork Abazajian of the University of California, Irvine, said at the meeting.

The density fl uctuations can be thought of as waves of various wavelengths randomly piled on one another. The fl uctuations’ inten-sity and gravitational pull grow stronger as their wavelengths decreases; then the intensity peaks at a certain wavelength and starts to fall again. The position of that peak in a graph of intensity versus wavelength depends on the relative amounts of radiation and matter in the early universe. And because lightweight neutrinos acted like an addi-

tional form of radiation, the position of the peak also reveals the number

of types of light neutrinos.To deduce the distribution

and the peak in it, scientists measure tiny variations in the afterglow of the big bang—the

cosmic microwave background radiation—across the sky, as

NASA’s space-borne Wilkinson Microwave Anisotropy Probe did from

2001 to 2010. Scientists also measure the dis-tribution of the galaxies, as the Sloan Digi-tal Sky Survey has done using a telescope at the Apache Point Observatory in New Mex-ico. The results suggest a fourth neutrino, Abazajian says. “There’s a 1-in-20 chance that it’s a statistical fl uctuation, and those sorts of things go away all the time,” he says. “Still, it’s intriguing that the data’s converging to that value” of four types of neutrinos.

The newest bit of evidence comes from nuclear reactors. Earlier this year, a team of theorists argued that reactors are putting out more electron antineutrinos than detectors tens of meters away show. The result suggests that some antineutrinos escape detection by morphing into sterile antineutrinos.

Within a nuclear reactor, nuclei of the isotopes uranium-235, uranium-238, plutonium-239, and plutonium-241 split randomly to make myriad smaller nuclei that release copious antineutrinos. For example, a uranium-235 nucleus can split to make a nucleus of krypton-89. Krypton-89 then changes identity to rubidium-89, strontium-89, and yttrium-89, as one neu-tron after another in the nucleus spits out an electron and an antineutrino and turns into a proton. Thousands of other chains or “branches” of decays also occur.

David Lhuillier of France’s Alternative Energies and Atomic Energy Commission in Saclay and colleagues kept track of all those branches in a new calculation. Previ-ous calculations showed that a score of reactor

Far out! Cosmic microwave background (top) and the map of the galaxies hint at an extra neutrino.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org 306

NEWSFOCUS

CR

ED

IT: A

RM

EL

BR

UC

ELL

E/S

YG

MA

/CO

RB

IS

measurements taken over decades observed 97.6%, give or take 2.4%, of the expected antineutrino fl ux—fi ne agreement with the prediction. With the new calculation, the measurements average 94.3%, plus or minus 2.3%—a signifi cant difference that suggests neutrinos are disappearing. “Before, all the experiments were in agreement with the pre-diction,” Lhuillier says. “Now everybody is below the prediction.” Physicists say this sin-gle result triggered the workshop.

The disparate hints are tantalizing, says Joseph Formaggio of the Massachusetts Insti-tute of Technology in Cambridge. “What’s nice is that these anomalies come from differ-ent directions,” he says.

Signs of discordEach clue comes with caveats, however. For example, starting in 2002, physicists tested the LSND result with the Mini-Booster Neutrino Experiment (MiniBooNE) at Fermi National Accel-erator Laboratory (Fermi-lab) in Batavia, Illinois. First, they fi red muon neu-trinos—instead of LSND’s muon antineutrinos—450 meters into a detector f illed with 800 tons of mineral oil. In 2007, they saw signs of electron neu-trinos appearing in the muon neutrino beam, but with the wrong energy to mirror the process seen in LSND with anti-neutrinos. The result dampened enthusiasm for sterile neutrinos.

But last year, the researchers reported that using muon antineutrinos, they see electron antineutrinos appearing as LSND did, albeit at lower statistical signifi cance. “The excess in the MiniBooNE antineutrino data agrees beautifully with what you would expect from LSND,” says Louis, who also works on Mini-BooNE. But it also makes matters more com-plicated. To explain why the effect appears only for antineutrinos, physicists need to add two sterile neutrinos to their theory.

The cosmological evidence for sterile neutrinos also comes with qualifi cations, says Yvonne Wong of RWTH Aachen University in Germany. The unknown particles scien-tists might be glimpsing in cosmic radiation are signifi cantly lighter than the sterile neu-trinos hinted at by LSND and MiniBooNE. They aren’t even necessarily true neutrinos, Wong says, but could be any feebly inter-acting particle.

Even the newfound “reactor anomaly” has not bowled skeptics over. Petr Vogel, a theo-rist at the California Institute of Technology in Pasadena, who worked on the original reactor calculations 30 years ago, says the new cal-culations are undoubtedly more thorough and realistic than the old ones. However, they still leave out important details that might make the falloff in neutrinos less impressive, Vogel says. “I think what has been done is state of the art, and the shift [in the prediction] looks rea-sonable to me,” Vogel says. “But whether the error is really 2.3% remains to be seen.”

Finally, the signs of sterile neutrinos may not agree with one another, says Thomas Schwetz-Mangold of the Max Planck Institute for Nuclear Physics in Heidelberg, Germany, who presented a “global fi t” to all the data. In particular, if ordinary neutrinos quickly oscil-

late into sterile neutrinos, then experiments that send muon neutrinos to distant detectors should see a decrease in the total number of neutrinos reaching their detectors. But experi-ments such as Fermilab’s Main Injector Neu-trino Oscillation Search, which fi res neutrinos 735 kilometers to a detector in Minnesota, see no such loss. “If I take everything at face value, then the probability is less than a percent that it all fi ts together,” Schwetz-Mangold says.

The killer experimentIn spite of the odds, some experimenters are still eager to hunt sterile neutrinos. Plans vary widely, but physicists generally agree on what a killer experiment must do. If ordinary neu-trinos morph into sterile neutrinos and back, then the number of ordinary neutrinos in a beam should go up and down as the neutri-nos fl y away from their source. So scientists would have to spot that spatial oscillation over tens of meters.

The easiest way would be to add a sec-ond detector to the MiniBooNE experiment

closer to the neutrino source or to move the existing detector. The rate at which electron antineutrinos appear should then change. Building a second detector 200 meters from the source would cost $10 million, Geoffrey Mills, a MiniBooNE team member from Los Alamos, said at the conference. Alter-natively, researchers could move the current detector for about $5 million, he reported.

Adding the second detector to MiniBooNE is a must-do, some researchers say. But Roxanne Guenette of Yale University warned that a defi nitive measurement would likely take two more-expensive new detectors.

Others want to look for the oscillation of electron neutrinos by putting an intense radio-active source inside a jumbo detector. The number of electron neutrino detections should go up and down as the distance within the

detector from the source increases. Virginia Tech’s Jonathan Link proposes placing a chromium-51 source in the center of the Sudbury Neutrino Obser-vatory, which is a sphere filled with 1000 tons of heavy water.

The source should cost less than $3 million, Link says. “I do believe that this is the cheapest option that has some chance of mak-ing some sort of statement about the LSND-type ster-ile neutrino,” he says. Gio-acchino Ranucci of Italy’s

National Institute of Nuclear Physics in Milan presented a proposal to place a source under the 270-ton Borexino detector in Italy’s sub-terranean Gran Sasso National Laboratory.

Hanging over all of this is the question of money, as the United States particle phys-ics budget has been stuck at $800 million for years. Virginia Tech’s Ramaswamy Raghavan is developing the Low Energy Solar Neutrino Spectrometer detector, which would study solar neutrinos and, with a radioactive source, could look for sterile neutrinos. It would cost $50 million to $75 million. “Can you predict in the current fi scal situation in the U.S. that this is going to happen?” Raghavan says.

To help make the case for funding, con-ference attendees plan to write a white paper laying out the options. There’s some urgency, says Huber, the Virginia Tech theorist who helped organize the meeting. “I don’t want to do sterile neutrinos my whole career,” he says. He doesn’t say whether the ephemeral beast will continue to entice him if no defi nitive answer is quick in coming. –ADRIAN CHO

MIA. New calculations suggest that nuclear reactors put out more neutrinos than are observed.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 307

NEWSFOCUSC

RE

DIT

: PA

NO

S IP

EIR

OTI

S &

DA

HN

TA

MIR

, NE

W Y

OR

K U

NIV

ER

SITY

/GO

OG

LE F

USI

ON

TA

BLE

S

It’s a problem that all social scientists face. You have a brilliant idea for a study. You have the experimental design all worked out, and your university’s review board has approved it. But you still have to recruit hundreds of people as subjects for the experiment.

Gabriel Lenz, a political scientist at the University of California, Berkeley, faced this problem last year when he and collaborators wanted to follow up on another group’s study of voting behavior (Science, 10 June 2005, p. 1623). For that study, Americans were shown photographs of past U.S. congressional candidates and asked to rate the politicians on various characteristics, such as competence and attractiveness. Even though the study subjects had no information beyond an image of the candidates’ faces, their snap judgments were a signifi cant predictor of who actually won the races. Lenz wanted to see if that sur-prising result collapsed when those evaluating the photos come from cultures different from those of the candidates. But how to recruit people living in multiple countries?

Lenz and his research assistant Michael Myers had an idea: Why not order research subjects through Amazon.com? The company runs an online marketplace called Mechani-cal Turk for people across the world avail-able to do work on computers. (The name is a reference to an 18th century chess-playing “machine” that actually worked by virtue of a man hidden inside.) For tiny sums, anyone can hire people to perform almost any kind of simple task, such as tagging items in images. Lenz’s experiment required people to look at photographs of Brazilian political candidates and fi ll in a data sheet.

But first, he and his colleagues had to decide on how much they would pay each participant. Those offering a job through MTurk, known as requestors, compete with each other to recruit Turkers, the 500,000

people currently registered with the MTurk site as available for work. The task of rating the political candidate photos required about 4 minutes. “We played around with various payment rates,” Lenz says. For Turkers based in India, the researchers started low, offering 15 cents. In just 4 days, they received data from 100 people. Then for a control group, they recruited more than 300 Americans for between 20 and 50 cents each. The total cost? About $160, and that includes the 10% fee Amazon charges.

In just a few weeks, Lenz had all the data his group needed. In spite of the cultural dif-ferences, the snap-judgment effect persisted: American and Indian subjects predicted the winners of Brazilian political races based on nothing more than a mug shot, the research-ers reported last year in the social science journal World Politics.

As others follow Lenz’s lead, many more social science papers using MTurk will appear in the coming years, predicts Adam Berinsky, a political scientist at the Massachusetts Insti-tute of Technology in Cambridge. “Everyone I know is using it,” he says. For example, social scientists used 10,000 Turkers to create a tool for tracking the emotional content of Twitter messages (Science, 30 September, p. 1814).

For now, most researchers are using MTurk for pilot studies, quickly and cheaply testing online versions of experiments that they then perform with subjects face to face. But the use of MTurk subjects will eventually become mainstream, Berinsky says. The obvious advantage is the speed and cost. “Generally, we pay $8 for a 15- to 20-minute experiment in a lab. We can run the same study on MTurk for 75 cents to a dollar.”

There are other advantages. “Turkers are amazingly focused research subjects,” Berinsky says. Unlike the typical univer-sity undergraduates used for social science

studies, Turkers get paid only if they generate usable data. This is neces-sary to eliminate not only people who don’t understand the task but also “spammers,” people who try to exploit MTurk by skimming through the jobs and giving random responses wherever possible to accelerate the process.

For example, Lenz had to reject about 20% of his American and 50% of his Indian Turkers for those rea-

sons. But that is a manageable problem, Berinsky says. A counterintuitive solution is to keep the price low. “If you offer more than a dollar, you attract the spammers who sort jobs by level of pay,” he says. “You have to fi nd the sweet spot where the payment is not too high but still attractive enough for most Turkers.” So far, that sweet spot seems to be between 15 and 50 cents for a 10-minute job.

Even if MTurk is cheap and fast, doubts will linger about interpreting data from research subjects whom you never meet. To address those concerns, Berinsky and Lenz are teaming up with Gregory Huber, a politi-cal scientist at Yale University, to study the Turker population. And of course, they are using MTurk to do so. They recently repli-cated two classic survey experiments and a political science experiment. In each case, the data obtained with MTurk were consis-tent with published studies that tested people in laboratories.

The scientists have found some differ-ences, too. Turkers “are younger and more ideologically liberal than the U.S. public,” Berinsky says. However, they are more repre-sentative of the U.S. population than a typical cohort of university undergraduates.

There is one long-term concern: the “super-Turkers,” people who are essentially professional workers on MTurk, some of them logging more than 20 hours per week. Many social science experiments rely on the sub-jects not knowing the researchers’ intentions. Berinsky says super-Turkers could potentially skew experiments if they try too hard to please researchers. There is incentive to do that because MTurk uses a reputation system. If a Turker does not have at least a 95% positive approval rating from their requestors, they’ll often go unhired.

“Mechanical Turk seems like the prover-bial goose that lays the golden eggs,” Berinsky says. “But I worry that in the rush for cheap research subjects, we’re going to trample the goose to death.” –JOHN BOHANNON

Global pool. This map shows a 10% sample of workers (red) available on Amazon.com’s Mechanical Turk.

Social scientists are turning to online retail giant Amazon.com to cheaply recruit people around the world for research studies

Social Science for PenniesH U M A N S U B J E C T R E S E A R C H

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org 308

CR

ED

ITS:

(TO

P LE

FT T

O R

IGH

T) J

OSL

IN M

OO

RE

; CH

EN

GJI

N C

HU

; (B

OTT

OM

) ELI

ZAB

ETH

BO

RE

R

In 2005, a handful of young researchers in Santa Barbara, California, were fed up with their inability to answer a major ecological question by reviewing the literature. So they decided to take matters into their own hands and created a network of small experiments. In the past 6 years, the network has spread to six continents and is now poised to make substantial contributions to ecology. “We’re on the edge of something big,” says John Orrock of the University of Wisconsin, Mad-ison, a network co-founder.

The half-dozen Ph.D. students and post-docs were part of a workshop at the National Center for Ecological Analysis and Synthesis (NCEAS) in Santa Barbara. The group was investigating fundamental infl uences on the structure of grasslands, such as herbivory and nutrients. Trying to ana-lyze data from far-fl ung places, the group was stymied by a common obstacle. “It’s really frustrating because everyone does their studies differently,” says Elizabeth Borer, who is now at the University of Minnesota, Twin Cities.

During a coffee break at NCEAS, Borer

and a few others hatched a plan: They would each set up a small research plot, use the same methods, then pool their data. The vision was a network of sites that would be quick and cheap to set up without the need for major grants, enabling simple experiments around the world. “It’s like big science on a shoe-string,” says Scott Collins of the University of New Mexico, Albuquerque, who later joined the network.

The collaboration, called the Nutrient Network—now known as NutNet—has grown far beyond initial expectations, with scientists volunteering at 68 sites in 12 coun-tries. In part, it’s popular because the simple experiments are designed to answer a broad set of questions about how grasslands respond

to global change—without disproportionate effort by any one individual. “It’s not a brand-new idea, but it’s novel that they’ve pulled it off,” says Alan Townsend of the University of Colorado, Boulder, who is not involved. The network also provides an easy way for young faculty members, postdocs, and grad students to get involved in a large collaboration and contribute to high-profi le papers.

So far, the effort has been funded with just a single $322,000 grant from the U.S. National Science Foundation (NSF) for coor-dinating data and analysis, yet already the fi rst few papers have been published over the past year. The most recent, which appeared in Science last month (23 September, p. 1750), challenged a long-standing idea in ecology about plant diversity and productivity. Doz-ens more papers are in the works, and ecol-ogists enthuse about the network’s potential for cost-effective, rapid results. “NutNet has tremendously improved on the way we’ve done things,” says Alan Knapp of Colorado State University, Fort Collins, another ecolo-gist who is not involved. “I’ve been incred-ibly impressed.”

Keep it simpleResearch networks aren’t new to ecology, of course. The Long Term Ecological Research (LTER) network, for example, is composed of 26 research sites and stations, almost all in the United States, that have been collecting data for 30 years. And construction began this fall on some of the 20 U.S. observatories that will make up the $434 million National

Ecological Observatory Network. These hefty networks require a fair amount of money to operate, because staff members collect hun-dreds of types of data, often year-round.

During the NCEAS workshop, NutNet’s f o u n d e r s q u i c k ly sketched an alternative vision: Each researcher would conduct the same few experiments

Open-Source Ecology Takes Root Across the WorldA new collaboration of volunteer research sites is running simple yet powerful experiments to shed light on global change in grasslands

N E T WO R K S C I E N C E

Experimental

Observationalonly

Diversity. NutNet sites include 1747 plant taxa in many ecosystems, such as (see photos, left to right) subalpine grassland, alpine meadow, desert, pasture, sagebrush steppe, and savanna.

Standardized. Researchers worldwide add nutri-ents and measure plots the same way.

China

Australia

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 309

NEWSFOCUSC

RE

DIT

S: (T

OP

LEFT

TO

RIG

HT)

SC

OTT

CO

LLIN

S; Y

AN

N H

AU

TIE

R; P

ETE

R A

DLE

R; T

. MIC

HA

EL

AN

DE

RSO

N; (

MID

DLE

) TIM

RU

MM

ELH

OFF

; (B

OTT

OM

) CO

URT

ESY

OF

STA

N H

AR

POLE

in several plots of 25 square meters. They would add combinations of three crucial plant nutrients—nitrogen, phosphorus, and potassium—and they would fence part of the plots to exclude deer, zebras, kangaroos, and other herbivores.

By measuring changes in biomass and spe-cies composition, they would try to tease apart the relative impact of herbivores and nutrients on the structure of the com-munity. “Ecologists have been fascinated by this question for a long time,” Orrock says. Moreover, the experiments simulate the impacts of anthro-pogenic global change. Nutri-ent levels have been boosted dramatically by fertilizers and pollution from fossil fuels. At the same time, humans have altered the density of herbi-vores in many places through farming or indirectly by hunt-ing of predators.

Several attendees at the NCEAS workshop immedi-ately volunteered to partici-pate. One of the fi rst was Helmut Hillebrand of the Carl von Ossietzky University of Old-enburg in Germany, who set up a NutNet site, even though he’s a plankton ecologist. “I think it’s the next generation of ecologi-cal experiments,” he says. The site he started is located in an old fi eld 5 minutes from his parents’ house, so he drops by to collect data while visiting.

Borer and the others also invited a few colleagues to join, and the idea began to spread by word of mouth. Sensing potential, the group sent an e-mail in November 2006 to just about every grassland ecologist they knew. By the time data started arriving the next year, there were 51 sites.

Members of the network agree to submit data immediately to a central database. All participants—now about 100, including a dozen or so graduate students—have access to the data. Simply by contributing data, they can be an author on high-profi le papers that address the project’s big questions. The net-work is already making a mark: Last month’s paper in Science showed that a textbook idea about the relationship between plant productivity and species richness in fact

occurs rarely. Other key papers, based on the experimental results of adding nutrients and excluding herbivores, are still being written.

NutNet participants must propose papers on additional ideas to the whole group. The goal is to avoid duplication and allow other members to contribute to analysis or writing the manuscript. Jennifer Firn of the Queensland University of Technology

in Brisbane, Australia, for example, wanted to look at invasive species in the plots. “The process of turning this idea into a paper was the best learning experience I have ever had,” says Firn, who became an assistant professor in February. “I had more than 30 authors and co-authors, so it meant so much advice and expertise were available.” Published in Ecology Lettersin March, the paper showed that non-native plants, some invasive, don’t all spread like the worst weeds. Instead, most species in the NutNet

plots were about as common in their new environment as in their native range. That suggests that regulators of plant imports might want to focus on screen-ing out plants that are highly abundant overseas.

Network members decide among themselves what kinds of additional data to gather. “This is like an indie garage band, a cooperative without all the top-down headaches,” says co-founder W. Stanley Harpole, an assistant pro-fessor at Iowa State Univer-sity in Ames. (Others make analogies to the development of open-source software or start-up companies.) Eighteen members are analyzing regular deliveries from other participants, who col-lect everything from soil microbes to arthro-pods and leaf litter. “It is simple, mail-order sampling,” says co-founder Eric Seabloom of the University of Minnesota, Twin Cities. “The person in the fi eld doesn’t have to do that much.”

Facing the futureAn all-volunteer approach may have its limi-tations, however. So far, the majority of sites are in the United States. Peter Adler of Utah State University in Logan, a co-founder, says the group tried to recruit scientists in South America without much success. “Maybe it’s just [bad] luck,” he says. Townsend expects that more researchers in less developed coun-tries will eventually sign up, as word spreads about the network and its publications. Ear-lier this month, several sites in India agreed to provide observational data, and a few more will also conduct experiments.

A larger question is how long a volun-teer effort can be sustained. “In absence of external funding, I fear that the good will of those individuals and their institutions may not persist,” says Michael Willig of the Uni-versity of Connecticut, Storrs, who is not a participant in the network. But co-founder Melinda Smith of Yale University predicts that interest will remain high as long as the network produces high-impact papers. Harpole points out that each plot has space reserved for experiments not yet planned. “We’re banking for the future,” he says.

The looming danger is the expiration of the NSF grant in January 2013. These funds pay for collaboration meetings and

for a postdoc, Eric Lind of the University of Minnesota, Twin Cities, who runs the cen-tral database. “The death of the Nutrient Network will be when the funding for that post-doc position runs out,” Adler says. The steering committee hopes to cover those expenses with future research grants for more ambitious analyses.

Even if the NutNet peters out, the founders hope it will be a model. To Borer, the suc-cess so far shows that individ-ual scientists at any stage of their career can help answer

big questions even if they haven’t landed a major grant. “We’re out to change the cul-ture,” she says. “The success of this model could empower other groups to address equally important ecological problems at a global scale.”

–ERIK STOKSTAD

“We’re out to change the culture.”

—ELIZABETH BORER,UNIVERSITY OF MINNESOTA,

TWIN CITIES

“This is like an indie garage band.”

—STAN HARPOLE,IOWA STATE UNIVERSITY

USA USASwitzerland Tanzania

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org 310

LETTERSedited by Jennifer Sills

LETTERS I BOOKS I POLICY FORUM I EDUCATION FORUM I PERSPECTIVES312

A multiple personality fable Bending light with antenna arrays

317

COMMENTARY

Martial Arts Research: Prudent Skepticism A. DIAMOND AND K. LEE’S REVIEW “Interventions shown to aid executive func-tion development in children 4 to 12 years old” (special section on Investing Early in Education, 19 August, p. 959) leaves the impression that martial arts training as usu-ally delivered enhances executive functions. This is far from established. Martial arts training is a heterogeneous independent vari-able with average effects that may be negli-gible or even negative.

Diamond and Lee cite two studies in sup-port of martial arts. In the Trulson study (which was based on 34 students and 1 instructor),

the only outcome measures are the self-report personality inventories completed by the “delinquent” students (1). Trulson concluded that the meditation, contemplation of goals, and other noncombat components of martial arts are helpful, but pure competitive fi ght training is harmful. The Lake and Hoyt study (207 students and 1 instructor) found the most positive effects on a measure of behavior dur-ing completion of an obstacle course (2). With teacher ratings, however, insignifi cant effects were reported for four out of fi ve variables, including self-control.

Longitudinal studies observing the results of many instructors lead to skepticism about the effects of martial arts training. Endresen and Olweus (3), using a longitudinal design, reported that “participation in power sports

Editorial Expression of ConcernIN THE 4 JUNE 2010 ISSUE, SCIENCE PUBLISHED THE REPORT “SPHK1 REGULATES PROINFLAM-matory responses associated with endotoxin and polymicrobial sepsis” by P. Puneet et al. (1). After the receipt of an anonymous e-mail on 22 March 2011, Science learned that authori-ties at the authors’ principal institutions at the time of publication (University of Glasgow and National University of Singapore) and the University of Liverpool (corresponding author A.J.M.’s more recent affi liation) were investigating allegations of fi gure manipulation in the Science Report and in a paper published in Nature Immunology [Nature Immunology 12, 344 (2011)] also by P. Puneet et al. The Nature Immunology paper was subsequently retracted after an investigation by the University of Liverpool, but we have been informed that the investiga-tion into the Science Report has not yet reached a conclusion, despite indications that it was near completion.

On 14 January 2011, Science published a Correction to two of the fi gures in the Puneet et al. Report, after correspondence with A.J.M. In light of the continuing investigation, we can no longer be confi dent in the reliability of the corrected record. Pending the results of the investiga-tions, Science is publishing this Editorial Expression of Concern to alert our readers to the fact that serious questions have been raised about the validity of fi ndings in the Puneet et al. paper.

BRUCE ALBERTS

Editor-in-Chief

Reference 1. P. Puneet, C. T. Yap, L. Wong, L. Yulin, D. R. Koh, S. Moochhala, J. Pfeilschifter, A. Huwiler, A. J. Melendez, Science 328, 1290

(2010).

Published online 3 October 2011; 10.1126/science.1214735

[including martial arts] actually leads to an increase or enhancement of antisocial involve-ment in the form of elevated levels of violent as well as non-violent antisocial behavior out-side sports.” We analyzed data from a large, nationally representative sample (4). The out-come variable was teacher-rated behavior, including self-control and attention. In each of our two main outcome analyses, we found that martial arts had no effect on behavior.

In a world beset by violence, there is irony and pathos in hoping that our children will be improved by teaching punching, kicking, and tripping. Unless the evidence for benefi t is robust, it is prudent to be skeptical.

JOSEPH M. STRAYHORN1* AND JILLIAN C. STRAYHORN2

1Department of Psychiatry, Drexel University College of Medicine, Philadelphia, PA 19129, USA. 2Undergraduate, Department of Psychology, Cornell University, Ithaca, NY 14853, USA.

*To whom correspondence should be addressed. E-mail: joestrayhorn@gmail.com

References 1. M. E. Trulson, Hum. Relat. 39, 1131 (1986). 2. K. D. Lakes, W. T. Hoyt, Appl. Dev. Psychol. 25, 283 (2004). 3. I. M. Endresen, D. Olweus, J. Child Psychol. Psych. 46,

468 (2005). 4. J. M. Strayhorn, J. C. Strayhorn, J. Child Adolesc. Psychiatr.

Ment. Health 3, 32 (2009).

Martial Arts Research: Weak Evidence

THE REVIEW “INTERVENTIONS SHOWN TO aid executive function development in chil-dren 4 to 12 years old” by A. Diamond and K. Lee (special section on Investing Early in Education, 19 August, p. 959) cited work that close examination shows to be weak. Some of the studies (1, 2) were randomized, but they failed to meet other criteria such as blind-ing of teachers and parents to pupils’ treat-ment groups. Studies involving martial arts and physical exercise were particularly weak on isolation of variables. One study on mar-

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 311

318 321

Signs of ocean eddies

Information capacity in cell signaling

CR

ED

IT: S

TUA

RT M

ON

K/I

STO

CK

PHO

TO.C

OM

tial arts training for children (1) com-pared a treatment group who wore spe-cial uniforms, medi-tated, bowed to their instructor, and were reminded of self-awareness and self-control, to a control group who continued with their ordinary physical education

activities; these authors concluded that when some improvement on some scales occurred for the treatment group, the change was caused by the self-awareness and self-control messages, rather than by other ways the two groups differed. Another study (2) compared children who did “sport stacking,” a bimanual

physical task, with a control group that did not experience any exciting new activity, and con-cluded that improvement on one of two read-ing measures was caused by the stacking task.

A relevant volume dealing with treatments for developmental disabilities (3) stressed the weakness of evidence for special education interventions and described some such con-ditions as “fad magnets.” Unfortunately, early educational interventions seem to be similarly weak in evidence. The stakes are high and the resources scarce in both cases. JEAN MERCER

Richard Stockton College, Pomona, NJ 08240, USA. E-mail: jean.mercer@stockton.edu

References 1. K. D. Lakes, W. T. Hoyt, Appl. Dev. Psychol. 25, 283 (2004). 2. T. A. Uhrich, R. L. Swalm, Percept. Mot. Skills 104, 1935

(2007). 3. J. W. Jacobson, R. M. Foxx, J. A. Mulick, Controversial

Therapies for Developmental Disabilities (Erlbaum, Mahwah, NJ, 2005).

ResponseWE AGREE WITH STRAYHORN AND STRAYHORN that modern and traditional versions of mar-tial arts differ. We tried to emphasize that modern American martial arts (which empha-size “punching and kicking” and competi-tion) have been found to make unproductive behaviors worse, whereas evidence indicates that traditional martial arts [which emphasize self-control, self-defense, patience, waiting for the other person to make an error, con-centration, respect, and humility (1)] improve executive functions.

We agree with Mercer about weaknesses in many studies thus far published on execu-tive function interventions. We reviewed only peer-reviewed studies and provided detailed information about them (see tables S1 to S3 in the supporting online material) to give read-ers an opportunity to judge the evidence for themselves. We disagree with Mercer about the martial arts study being particularly weak. First studies are designed to determine whether there is an overall difference. Follow-up stud-ies can then try to dissect which aspect(s) of a program had the most effect. That said, the martial arts study by Lakes and Hoyt (1) is to be commended. It used random assignment, pre- and post-testing, an intervention imple-mented during regular school hours (making it feasible to reach many children), an active control group that also engaged in physical activity, and incrementally increasing levels of diffi culty in the martial arts condition, and it provided evidence that executive-function improvements generalized to multiple con-texts. Unlike many studies that have targeted disadvantaged children and/or those behind on executive function, children in this study were socioeconomically advantaged, making the fi ndings especially impressive.

ADELE DIAMOND* AND KATHLEEN LEE

Department of Developmental Cognitive Neuroscience, University of British Columbia, Vancouver, BC V6N 3L6, BC, Canada.

*To whom correspondence should be addressed. E-mail: adele.diamond@ubc.ca

Reference 1. K. D. Lakes, W. T. Hoyt, Appl. Dev. Psychol. 25, 283

(2004).

Letters to the EditorLetters (~300 words) discuss material published in Science in the past 3 months or matters of gen-eral interest. Letters are not acknowledged upon receipt. Whether published in full or in part, Let-ters are subject to editing for clarity and space. Letters submitted, published, or posted elsewhere, in print or online, will be disqualifi ed. To submit a Letter, go to www.submit2science.org.

CORRECTIONS AND CLARIFICATIONS

Review: “Interventions shown to aid executive function development in children 4 to 12 years old” by A. Diamond and K. Lee (special section on Investing Early in Education, 19 August, p. 959). The journal cited in reference 28 should have been Appl. Dev. Psychol.

Education Forum: “Mathematics teachers’ subtle, complex disciplinary knowledge” by B. Davis (24 June, p. 1506). The number line on the right in part C of the fi gure was mis-numbered. The correct panel is shown here.

TECHNICAL COMMENT ABSTRACTS

Comment on “How Cats Lap: Water Uptake by Felis catus”Michael NauenbergReis et al. (Reports, 26 November 2010, p. 1231) reported on the mechanism by which cats lap and gave a theo-retical and experimental analysis of their observations. Their explanation for the cat’s lapping frequency, however, is based on an incorrect application of the principles of fl uid dynamics. The revised analysis given here agrees with their observations and predicts a similar lapping frequency for cats and dogs.Full text at www.sciencemag.org/cgi/content/full/334/6054/311-b

Response to Comment on “How Cats Lap: Water Uptake by Felis catus”Roman Stocker, Jeffrey M. Aristoff, Sunghwan Jung, Pedro M. ReisWe return to the physics of cat lapping to show that our proposed scaling analysis predicts the functional depen-dencies revealed by the experimental data more accurately than a recently proposed alternative description by Nauenberg. Experimental verifi cation of functional dependencies, rather than single numerical values, represents the appropriate test for any scaling argument.Full text at www.sciencemag.org/cgi/content/full/334/6054/311-c

C+4

+2

+1

0

–1

–2

–3

–4

+3

+8

+4

+2

0

–2

–8

–6

–4

+6

Factor a (multiplier):adjustment ratio

Factor b (multiplicand):starting position Product c:

corresponding position

on the adjusted number

line

BOOKS ET AL.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org 312

CR

ED

IT: (

1)/I

MA

GE

CO

URT

ESY

BE

N H

AR

RIS

In 1972, incorporation papers for Sybil, Incorporated, were drawn up by a patient, her therapist, and a journalist who had

just fi nished writing a book about her case. Flora Rheta Schreiber’s Sybil ( 1) told the story of a young woman who was cured of having her consciousness split into 16 separate person-alities. The corporation’s pur-pose was to share the profi ts from the book, a planned fi lm, and spin-offs that included T-shirts, dolls, a jigsaw puzzle, and a board game. Although the corporation did not last and the ancillary products did not materialize, the book sold millions of copies. It was dramatized twice in made-for-television movies. In the fi rst, Sybil’s psychiatrist was played by Joanne Woodward, who had previously won the Academy Award for Best Actress for her per-formance in the role of a patient with mul-tiple personality disorder in Nunnally John-son’s 1957 fi lm The Three Faces of Eve.

The most notable product of Sybil, Inc.’s principals was the epidemic of multiple personality disorder that swept the United States in the 1970s and 1980s. Unlike most previous cases, these late-20th-century variants featured personalities stuck in various stages of childhood. The patients’ “alters” revealed memories of childhood sexual abuse—often horrifi c—elicited by hypnosis. Today, researchers and clinicians are skeptical about most cases of multiple personality disorder, which has been recon-ceptualized and renamed dissociative iden-tity disorder.

Over the past two decades, revelations have emerged to raise doubts about the validity of the case of Sybil. First, a clinician who had met the patient disputed the diag-nosis promoted in the book. Then, tapes left by the book’s author suggested that she col-luded with the therapist to create symptoms to fi t their diagnosis.

In Sybil Exposed, journalist Debbie Nathan chronicles the rise and fall of Sybil

as the paradigm-setting case of multiple per-sonality disorder. She does so in three inter-twined biographies, beginning with that of the patient, Shirley Mason. Mason’s upbring-

ing was strict (Seventh Day Adventist), and although her mother was odd and subject to mood swings, she was not the sexual sadist depicted in Sybil. Starting at age 22, Mason was treated by Corne-lia Wilbur, an ambitious psy-chiatrist who progressed from treating traumatized World War II soldiers with hypno-sis and sodium pentothal to inducing traumatic memories

in Mason with the same tools. This she did on and off for more than two decades, fi rst in Nebraska and then in New York City. In New York, Wilbur often saw Mason daily, making house calls in the evening with a satchel of drugs and a portable electroconvulsive ther-apy machine. Beginning in 1963, the patient and her therapist collaborated with Schreiber, a journalist with a tendency to make her mag-azine stories more salable by massaging bio-graphical details.

Nathan offers a compelling account of the creation, packaging, and selling of this case of medical and journalistic malpractice. Her sources include transcripts of therapy sessions, letters by the patient to childhood friends and former roommates, and inter-views with acquaintances and colleagues of all three women. Nathan’s credentials suit this topic well, as she coauthored an ear-lier book ( 2) that helped reverse the fl ood of false memories implanted in children by prosecutors and therapists inspired by Sybil. As a feminist, she was dismayed that a seg-ment of the women’s movement channeled its social concerns into a hunt for psycho-sexual demons that unjustly targeted teach-ers and day care workers.

Analyzing the significance of Sybil, Nathan shows how the dilemmas faced by post–World War II women helped shape that case and gave it cultural resonance. Wilbur, she explains, saw herself as a psychiatric Betty Friedan, encouraging female patients to try out new social roles and, indeed, even new personalities. Similarly, journalist Schreiber saw Mason as a country girl from a stifl ing background who found a new iden-tity in the big city. Both Wilbur and Schreiber had struggled to find acceptance in male-dominated professions and reveled in the sta-tus and remuneration that Sybil brought them.

Looking beyond the three women who created Sybil, Nathan explores the institu-tional and professional context of both that case and the epidemic that followed. With its shifting boundaries and history of diag-nostic uncertainty, psychiatry was a medi-cal specialty with little resistance to the fad-dishness and yearning for breakthroughs that fueled the multiple personality disor-der fervor. Adding to clinicians’ enthusiasm was the embrace of multiple personality disorder by celebrities, journalists, and tele-vision producers.

In concluding her “cautionary tale of this great American multiplicity,” Nathan regrets that desire for personal change went awry at a “fractured moment in history.” The result was that “women and their social struggles were reduced to a bizarre illness. The cure was not critical inquiry or protest marches or efforts at the polls. Instead the cure was drugs [and] hypnosis.” Reading Sybil Exposed, that con-clusion seems warranted.

References 1. F. R. Schreiber, Sybil (Regnery, Chicago, 1973). 2. D. Nathan, M. R. Snedeker, Satan’s Silence: Ritual Abuse

and the Making of a Modern American Witch Hunt (Basic, New York, 1995).

Sybil, Inc.PSYCHIATRY

Ben Harris

The reviewer is at the Department of Psychology, Univer-sity of New Hampshire, Durham, NH 03824, USA. E- mail: bh5@unh.edu.

Sybil ExposedThe Extraordinary Story

Behind the Famous

Multiple Personality Case

by Debbie NathanFree Press (Simon and

Schuster), New York, 2011.

328 pp. $26, C$29.99, £16.95.

ISBN 9781439168271.

10.1126/science.1212843

The account that made the story famous. The cover of the fi rst edition of Sybil.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 313

EDUCATIONFORUM

For academic researchers and educators, summer months are often among the most productive times of year. Unfortu-

nately, at many minority-serving institutions (MSIs) that serve a high proportion of under-represented minorities (URMs), the summer often brings a “brain drain” that threatens to erode capacity for conducting research. Sum-mer research programs (SRPs) recruit stu-dents from MSIs to spend a summer working at a research-intensive university. SRPs can expose students to resources and experiences they may not as easily access at their home MSI. Among benefits to host institutions, SRPs help fulfill some “broader impacts” (i.e., nonresearch, such as education and out-reach) that are required by many funders. MSI faculty may benefi t from skills the SRP students bring back. But this brain drain can undermine the ability of MSI faculty to effec-tively teach research skills and to develop tal-ent among a diverse pool of students.

In 2008, ~25% of U.S. science, technol-ogy, engineering, and math (STEM) bacca-laureate degrees awarded to URM students were from MSIs ( 1). These institutions enroll ~30% of all URM students ( 2). Although the percentage of URMs earning baccalaureate degrees has declined ( 1), those who complete their degrees show Ph.D. graduation rates similar to non-Hispanic white counterparts ( 1). Many MSIs are among the leading STEM bachelor’s degree–granting institutions for Hispanic and African American students who complete Ph.D. degrees ( 3). But whereas interest in diversity underlies undergradu-ate SRPs, questions remain about intended recruitment and retention of URM students in graduate programs. Despite a steady increase in graduate program graduation rates over the past decades ( 4), URM participation in STEM graduate programs is lower than expected on the basis of U.S. demographics ( 5).

Erosion of the Capacity to Develop TalentDespite SRP’s benefits, they may hinder the ability of talented individuals to access the skill set required to succeed in research careers. Institutionalized “one-way bridges,” like SRPs, can siphon well-trained talented

students from MSIs. As a result, research goals of MSI faculty may not advance as well as they could; well-trained personnel are key to producing good-quality data that become the basis for publications and grant applica-tions. In turn, reviews of research grant appli-cations from MSIs may criticize productivity levels, the amount of preliminary data, and the adequacy of academic environments to implement research programs. Such percep-tions of MSIs may erode funding for facili-ties, equipment, and supplies.

A vicious cycle may unfold. Adequately supported MSI faculty can develop talent among URM students with high potential, but lacking research experience. Early exposure to research ( 6) and long-term experiences ( 6, 7) have been correlated with outcomes such as persistence in research careers. Applicant training is a key element in successful grant applications. Early and consistent access to research mentors and adequate resources may provide cumulative advantages ( 8). Although it is not clear what factors contribute to differ-ential success rates for URM faculty in obtain-ing research funds ( 8, 9), lack of funding in turn affects the overall capacity to develop and nurture local students, leading to a shortage of well-qualifi ed, diverse talent.

Ensuring Long-Term SustainabilityAlthough some promising programs exist ( 10), we must invest in sustainable develop-ment of talent across a wider spectrum of aca-demic institutions. This requires a shift away from the prevailing system at research-inten-sive institutions ( 11) and MSIs ( 12) to provide research-intensive experiences for all stu-dents. Collaborations between research-inten-sive institutions and MSIs could be rebalanced for more “two-way bridge” partnerships.

For example, there are well-equipped MSI research facilities supported by initiatives from funders such as the National Institutes of Health and National Science Foundation (e.g., Spelman College’s Center for Health Disparities Research and Education, and the Ponce School of Medicine–Moffitt Cancer Center Partnership). Thoughtful integration of research and teaching training activities, guided by aligned interests, may allow a post-doctoral fellow to receive training in teaching at a MSI partner school. Graduate students may gain valuable experience in communicat-

ing science ( 13) and acting as mentors to URM colleagues at partner MSIs. Students who act as role models demonstrate increased learn-ing and tend to make stronger commitments to their studies ( 14). Students from MSIs will share experiences with groups at SRP host institutions. This may increase their sense of belonging, shown to help students overcome the uncertainty ( 15) that discourages many from pursuing a STEM career ( 16).

The benefi ts of diversity at all stages of the academic pipeline ( 17) are well docu-mented and recognized by funding agen-cies ( 10). Amid calls for supporting MSIs to expand their effective recruiting and retention rates ( 18), while establishing basic indicators of student outcomes to enable institutions to assess their effectiveness ( 12), now is the time to rethink our approach to developing a diverse, talented STEM workforce.

References and Notes 1. National Science Foundation (NSF), Women, Minorities,

and Persons with Disabilities in Science and Engineering: 2011 (NSF, Arlington, VA, 2011).

2. X. Li, Characteristics of Minority-Serving Institutions and Minority Undergraduates Enrolled in These Institutions (NCES 2008-156, U.S. Department of Education, Wash-ington, DC, 2007).

3. National Academies and Institute of Medicine, Expanding Underrepresented Minority Participation: America’s Science and Technology Talent at the Crossroads (National Acad-emies Press, Washington, DC, 2010).

4. NSF, Doctorate Recipients from U.S. Universities: 2009 (NSF, Arlington, VA, 2010); www.nsf.gov/statistics/nsf11306.

5. P. Einaudi, Science Resources Statistics InfoBrief (NSF11-319, NSF, Arlington, VA, 2011).

6. S. H. Russell et al., Science 316, 548 (2007). 7. A. L. Zydney et al., J. Eng. Educ. 91, 151 (2002). 8. D. K. Ginther et al., Science 333, 1015 (2011). 9. L. A. Tabak, F. S. Collins, Science 333, 940 (2011). 10. E. M. August et al., J. Cancer Educ., (2011); 10.1007/

s13187-011-0265-4. 11. W. A. Anderson et al., Science 331, 152 (2011). 12. L. E. Malcom et al., Tapping HSI-STEM Funds to Improve

Latina and Latino Access to the STEM Professions (Univ. of Southern California, Los Angeles, CA, 2010).

13. D. F. Feldon et al., Science 333, 1037 (2011). 14. J. M. Good et al., J. Negro Educ. 69, 375 (2000). 15. G. Cohen, J. García, Curr. Dir. Psychol. Sci. 17, 365 (2008). 16. G. M. Walton, G. L. Cohen, J. Pers. Soc. Psychol. 92, 82

(2007). 17. P. Gurin et al., Harv. Educ. Rev. 72, 330 (2002). 18. E. M. Bensimon, L. E. Malcom, B. Dávila, (Re)constructing

Hispanic-serving institutions: Moving beyond numbers toward student success (EP3: Education Policy and Prac-tice Perspectives 6, Iowa State Univ., Ames, IA, 2010).

19. The author thanks undergraduate researchers (supported by Department of Defense grant W911NF-09-1-0219); particularly, D. Kiehart, T. Littleton, and C. Doe, who participated in the two-way collaboration. The author was supported by NSF.

Rethink Summer Student ResearchEDUCATION

Franklin A. Carrero-Martínez

Research and training at institutions serving minority students may suffer as top students leave for other schools each summer.

Department of Biology, University of Puerto Rico, Mayagüez, Mayagüez, PR 00681, USA. E-mail: franklin.carrero@upr.edu 10.1126/science.1209555

A Journal with Impact from AAAS, the publisher ofScience

Science Translational MedicineIntegrating Medicine and ScienceA recent journal article features the sequencing of fetal DNA from plasma

of a pregnant woman to permit prenatal, noninvasive genome-wide

screening to diagnose fetal genetic disorders.

Sci Transl Med 8 December 2010:

Vol. 2, Issue 61, p. 61ra91

DOI: 10.1126/scitranslmed.3001720

Recommend aninstitutional subscriptionto your library today!ScienceOnline.org/recommend

ScienceTranslationalMedicine.org

Indexed inMEDLINE/PubMed

Academic Opportunities inEuropean Science

In This IssueScientists who wish to pursue academic careers in Europehave much with which to contend, especially now amidst anuncertain fiscal landscape. However, buttressed by the EuropeanCommission, several member states, including Germany, theUnited Kingdom, and France, as well as Scandinavia, have plansin place to bolster scientific research and innovation, and makethose regions attractive destinations for early and mid-careerprofessionals looking for academic positions.

See full story on page 397.

Upcoming Features:Neuroscience: Emerging Fields—November 4Focus on China—December 9BS/MS Scientists (online only)—January 13

FOCUS ON EUROPEBrought to you by the Science/AAAS Custom Publishing Of�ce

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 315

PERSPECTIVES

Electrical synapses have a rich and sometimes controversial history. During the early 20th century, the

question of whether chemical or electrical synapses underlie the main mode of signal-ing in the mammalian central nervous sys-tem (CNS) was hotly debated ( 1). However, after the discovery of neuronal inhibition in the early 1950s, it was accepted that trans-mission of information via neurotransmit-ter molecules (chemical synapses) repre-sents the major form of signaling among CNS neurons. Although a consensus devel-oped that electrical synapses are present in a subset of neuronal connections, it was unknown whether these synapses exhibit selectivity and plasticity as shown for most chemical synapses, or whether neuronal activity can result in long-term change of synaptic strength of electrical synapses ( 2) in the mammalian CNS. On page 389 of this issue, Haas et al. ( 3) demonstrate that the strength of electrical synapses among spe-cifi c neurons in the thalamus of the mam-malian brain affects long-term depression (LTD), a process important for learning and memory. Given the proposed role that elec-trical synapses play in synchronizing neuro-nal activity, these fi ndings suggest a mech-anism for controlling the coordination of neuronal activity.

Chemical synapses selectively connect different types of neurons to provide a spe-cific path of communication among CNS neurons ( 4, 5). It was discovered more than 10 years ago that electrical synapses are formed with exquisite selectivity among specifi c classes of inhibitory neurons ( 6, 7). Moreover, another hallmark of synapses—modulation by neurotransmitters—has been also demonstrated in both the thalamus and hippocampus ( 8, 9).

To investigate the effects of neuronal activity on the strength of electrical syn-apses in the mammalian CNS, Haas et al. recorded electrical activity from pairs of neurons in the rat thalamic reticular nucleus (TRN). All sensory messages are conveyed to the cortex via the thalamus (see the fi gure) ( 10). The TRN is located between the thala-mus and the cortex and, unlike other groups

of thalamic neurons (thalamic nuclei), is composed of inhibitory neurons that are connected by electrical synapses ( 11). These electrical synapses, together with inhibi-tory chemical synapses (those that transmit γ-aminobutyric acid), can synchronize fi r-ing of TRN neurons. Because TRN neurons project to other thalamic nuclei, synchro-nized activity of TRN neurons can entrain the activity of other thalamic regions ( 10).

To measure the strength of electrical cou-pling, Haas et al. injected depolarizing cur-rent into one TRN neuron and measured the response in both the injected cell and the noninjected neuron. The ratio of the two electrical responses is defi ned as the cou-pling coeffi cient (cc). The coupling coeffi -cient among TRN neurons is large (indicat-ing strong electrical coupling), and often a burst of electrical activity (action potential) in one neuron will produce action potential in the coupled neuron. The coupling coef-ficient can be estimated in two ways: by injecting current in cell 1 (i.e., cc12) or by injecting current in cell 2 (i.e., cc21). Inter-estingly, Haas et al. found that a substantial number of neuron pairs exhibited asymmet-

rical coupling. That is, the distribution of the ratio cc1/cc2 was skewed. In vivo, TRN neurons often fi re bursts of electrical activ-ity (spikes). The authors found that bursts of spikes in pairs of coupled TRN neurons resulted in LTD of the strength of electri-cal synapses. Burst activity in both neurons produced symmetrical LTD such that cc12 = cc21. However, when cell 1 was induced to burst while the coupled cell (cell 2) was pre-vented from bursting (by current injection), the resulting LTD was asymmetrical. Under these conditions, the coupling measure from cell 1 to cell 2 (cc12) decreased much less than the coupling from the inactive cell (cc21) that was prevented from bursting. Although Hass et al. observed a modest decrease of synap-tic strength, they show that LTD of electrical coupling is suffi cient to prevent propagation of bursting activity among neurons.

What are the mechanisms underlying activity-dependent LTD of electrical syn-apses? The results of Haas et al. suggest that sodium-dependent action potentials may play a role (in response to a stimulus, sodium channels open and allow sodium into the neuron, which triggers the f ir-

NEUROSCIENCE

After burst activity

Inactive

Brain

Depression of electrical synapses. (Enlarged view, left side) Inhibitory neurons (circles) in the mamma-lian thalamic reticular nucleus (TRN) are connected via electrical (small black arrows) and inhibitory chemi-cal synapses (not illustrated). External excitatory inputs impinge on TRN neurons and can activate them. (Enlarged view, right side) When TRN neurons generate bursts of action potentials, electrical synapse strength decreases. The decrease of strength can be asymmetrical between cells that do not generate bursts (white arrows) and those that do (gray arrows).

CR

ED

IT: B

. STR

AU

CH

/SC

IEN

CE

Department of Comparative Medicine, Stanford University, Stanford, CA 94305, USA. E-mail: shaul.hestrin@stanford.edu

The Strength of Electrical SynapsesShaul Hestrin

The strength of electrical synapses in the mammalian brain can be modulated by neuronal ativity.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org 316

PERSPECTIVES

Watery DisksPLANETARY SCIENCE

Rachel Akeson

The Herschel Space Observatory has detected a large reservoir of water stored as ice in the disk surrounding a nearby star.

ing of an action potential). In addition, an increase in cytoplasmic calcium concen-tration could activate a protein kinase that modifi es (by phosphorylation) gap junction channels,which bridge electrically coupled neurons. Alternatively, partial replacement of nonrectifying gap junctions with rectify-ing gap junctions (thus allowing electrical current to pass preferentially in one direc-tion) is a possible mechanism.

The functional consequences of LTD produced by electrical synapses are of great interest. By producing asymmetrical changes of electrical coupling, the outbound connec-tion of a group of bursting neurons will be stronger than the inbound connection from coupled neurons that did not burst. The func-

tional consequences resulting from the gen-eration of asymmetry of electrical synapses within the TRN remains to be explored. The impact of reducing the strength of electrical synapses at the circuit level and their poten-tial effect on rhythmic activity in the TRN is also of interest. In vivo, TRN neurons can switch between burst and tonic fi ring modes ( 10). If these activities can induce LTD, one may expect to fi nd changes in the strength of electrical coupling in the TRN that correlate with different behavioral states during sleep and wakefulness.

References 1. W. M. Cowan, E. R. Kandel, in Synapses, W. M. Cowan,

T. C. Südhof, C. F. Stevens, Eds. (Johns Hopkins Univ. Press, Baltimore, 2001), pp. 1–87.

2. A. E. Pereda et al., Proc. Natl. Acad. Sci. U.S.A. 95,

13272 (1998). 3. J. S. Haas, B. Zavala, C. E. Landisman, Science 334, 389

(2011). 4. S. P. Brown, S. Hestrin, Curr. Opin. Neurobiol. 14, 415

(2009). 5. H. Ko et al., Nature 473, 87 (2011). 6. M. Galarreta, S. Hestrin, Nature 402, 72 (1999). 7. J. R. Gibson, M. Beierlein, B. W. Connors, Nature 402, 75

(1999). 8. C. E. Landisman, B. W. Connors, Science 310, 1809

(2005). 9. V. Zsiros, G. Maccaferri, J. Neurosci. 28, 1804

(2008). 10. S. M. Sherman, R. W. Guillery, in The Synaptic Orgnaiza-

tion of the Brain, G. M. Shepherd, Ed. (Oxford Univ. Press, Oxford, ed. 5, 2004), pp. 311–359.

11. C. E. Landisman et al., J. Neurosci. 22, 1002 (2002).

10.1126/science.1213894

The paradigm for star formation is understood to center around the for-mation of a rotating disk from a cloud

of gas and dust. The circumstellar disk fun-nels material onto the newly formed central star and also serves as a reservoir of mate-rial from which a planetary system may arise. Determining the physical and chemical com-position of these disks is necessary to under-stand the formation and evolution of planets. Previous observations have detected the pres-ence of molecules within the disk, thereby demonstrating an active chemical network. However, this chemistry is harder to trace in the majority of the disk where low tempera-tures result in the molecules freezing out onto grains. On page 338 of this issue, Hoger heijde et al. ( 1) use the Herschel Space Observatory to detect cold water vapor in one of the closest young stars, TW Hydrae. The source of that water vapor is likely to be a large reservoir of ice grains.

In the past few decades, the basic physical information we have learned about circum-stellar disks—their mass, size, and lifetimes—come from observations of micrometer-sized dust grains within them. However, the vast majority of the material is gaseous, primarily molecular hydrogen. Transitions of molecular hydrogen are very diffi cult to observe, and the only H2 detections in disks are of warm gas

in the center. Attention has therefore turned to other molecules such as CO, HCO+, and CN to trace the gaseous component.

Water is the main constituent of the mantles on grains found in the low-density medium that fills the space between stars and can serve as an oxygen reservoir in the gas phase. Thus, water may be one of the key components in the chemistry and ther-mal balance of both the parent cloud and the circumstellar disk. Spectroscopic obser-

vations of water vapor from ground-based telescopes are hindered by water vapor fea-tures in Earth’s atmosphere. In the inner disk, where water is present as hot vapor and can be observed from the ground, observations have shown water vapor at abundances above the levels measured in molecular clouds ( 2). These observations include the so-called habitable zone where terrestrial planets are believed to have formed in our solar system. However, most of the mass within the disk is

NASA Exoplanet Science Institute, California Institute of Technology, Pasadena, CA 91125, USA. E-mail: rla@ipac.caltech.edu

Rocky planetesimals Icy planetesimals

Snow line

A frozen line of division. Interior to the snow line, planetesimals forming in the disk are rocky, while out-side they are icy. In our solar system, this line corresponds to the divide between the inner terrestrial planets and the outer gas giants. C

RE

DIT

: P. H

UE

Y/S

CIE

NC

E

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 317

PERSPECTIVES

below the 150 K condensation temperature for water, and thus most water exists as ice on grains. Spectroscopic features from water ice have been seen ( 3, 4), but these techniques have only been applied to a few objects with favorable viewing geometries. Observations of the water isotope HDO (D is deuterium) in the gas phase of the outer disk led to mod-els of ultraviolet photons from the central star desorbing water molecules from the icy grains back into the gas phase ( 5). The obser-vations of Hogerheijde et al. are consistent with this cycle. Their calculations show that a large population of icy grains, equivalent to several thousand Earth oceans, is necessary to maintain the observed level of water vapor on the surface of the disk.

Water can also play a critical role in the formation and fi nal surface composition of planets. Ice enhances the solid material in the cold outer part of a protoplanetary disk, which promotes the formation of cores of gaseous planets ( 6). The disk radius where ice can condense is often termed the “snow line,” and the location of this line is a property of the stellar mass and disk evolutionary state (see the fi gure). The location and evolution of this snow line may affect the formation rate of large planets ( 7).

The distribution of water ice in the cir-cumstellar disk can also help address the issue of where Earth’s water originated. While forming, Earth is believed to have been too hot to have liquid water and would have retained little water vapor from the gas-eous component of the disk. Thus, the water we have now arrived later, most likely from ice-covered comets or asteroids from the outer parts of the solar system. In addition to the water abundance, the spectra obtained by Hogerheijde et al. allow determination of the spin isomer ratio, where the spin refers to the alignment of the hydrogen proton spin vectors (that is, the ratio of the amount of para- to ortho-hydrogen in the water mole-cules). They found a ratio much lower than that measured for solar system comets, sug-gesting that material from multiple locations in the TW Hydrae disk is mixed before incor-poration into larger bodies. Evidence for such radial transport in the early solar sys-tem includes results from the Stardust mis-sion that returned comet samples containing material formed at high temperatures ( 8).

As the number of planets discovered around other stars expands to include many systems with multiple planets, it is clear that the universe includes many planetary sys-

tem architectures very different from that of our own solar system. To constrain models of planet formation, including the chemi-cal composition, we need to understand the distribution and evolution of molecules in the disk, including water, a key catalyst for life on Earth. The next several years will pro-vide many opportunities to progress in this study as Herschel and other observatories will make spectroscopic observations of a much larger sample of disks covering a range of stellar age and mass. Also, new facilities, such as the Atacama Large Millimeter Array, will greatly expand on the current sensitivity levels to allow spatially resolved observations of molecules in the disk.

References 1. M. R. Hogerheijde et al., Science 334, 338 (2011). 2. J. S. Carr, J. R. Najita, Science 319, 1504 (2008). 3. K. M. Pontoppidan et al., Astrophys. J. 622, 463 (2005). 4. M. Honda et al., Astrophys. J. 690, L110 (2009). 5. C. Ceccarelli, C. Dominik, E. Caux, B. Lefl och, P. Caselli,

Astrophys. J. 631, L81 (2005). 6. C. Hayashi, K. Nakazawa, Y. Nakagawa, Protostars and

Planets II (University of Arizona Press, Tucson, 2005), pp. 1100–1153.

7. G. M. Kennedy, S. J. Kenyon, Astrophys. J. 673, 502 (2008).

8. T. Nakamura et al., Science 321, 1664 (2008).

10.1126/science.1213752

Antenna-Guided LightAPPLIED PHYSICS

Nader Engheta

Compact arrays of gold nanoantennas can be used to create optical structures that bend the path of light in unusual ways.

The bent appearance of a stick half-sub-merged in water is caused by the dif-ference in refractive indices of air and

water—light travels more slowly in water than in air (see the fi gure, panel A) and refracts and refl ects off the air-water interface. Snell’s law ( 1) lets us calculate the bending angle if we know the geometry and the refractive indices. In complex optical instruments, where sev-eral lenses, mirrors, and other components may be present, designers control the bending by keeping track of the phase shifts imposed along the wavefront of the light; for exam-ple, a light beam can be focused by different phase shifts that occur along a curved lens. These optical components are much larger than the wavelength of light, which limits the minimum size of devices. On page 333 of this issue, Yu et al. ( 2) show how arrays of struc-

tures smaller than the wavelength of light, V-shaped nanoantennas made of gold, bend light by creating abrupt phase shifts through the excitation of resonances. The authors show that these compact “metasurfaces” fol-low a more general version of Snell’s law that accounts for the bending of a light beam in unconventional but potentially useful ways.

Conventionally, the bending of light may occur at an abrupt interface of two media (e.g., air and water), or through a gradual change of refractive index (e.g., air above the hot desert roads causes mirage; see the figure, panel B). However, it is possible to obtain the desired phase shift along the wavefront by tailoring planar interfaces. One of the early examples is the Fresnel lens, in which a set of concentric lenses are cut to different curvatures and impose dif-ferent phase shifts. Although a Fresnel lens is much thinner than an equivalent conven-tional lens, its thickness is still far greater than the wavelength of light.

Light does not always simply pass through a medium; it can also excite reso-nances that can lead to absorption and emis-sion. For the much longer wavelengths of “light” used in radio and microwave com-munications, antennas called refl ectarrays ( 3) and transmitarrays ( 4) contain multiple antenna elements that act as resonators to control the direction in which signals are received or broadcast. However, the reso-nant elements responsible for the required phase shift and their arrangement in peri-odic arrays are still comparable in size to the wavelength of operation ( 3– 5). These devices often operate over only a narrow range of frequencies.

For shorter-wavelength light, such as infrared and visible light, plasmonic phenom-ena—the excitation of collective oscillations of electrons in materials such as gold and silver—can allow subwavelength objects to undergo resonance responses in the scatter-ing process. Yu et al. designed subwavelength

University of Pennsylvania, Department of Electrical and Systems Engineering, Philadelphia, PA 19104, USA. E-mail: engheta@ee.upenn.edu

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org 318

PERSPECTIVES

gold antennas with a V shape; they varied the scattering of light by changing the length of the arm and the angle and the orientation of these “V’s.” The phase difference between the scattered and incident fi elds is tailored over a small distance along the light’s path, that is, the structures are optically thin.

Yu et al. printed planar arrays of such V-shaped nanoantennas in suitably designed patterns on a silicon wafer and demonstrated several intriguing light-bending scenarios at these metasurfaces, including unconventional refl ection and refraction angles, total internal reflections with two critical angles (rather than only one), and refl ected light becoming evanescent (diminishing in amplitude with distance away from the interface, rather than propagating) at certain angles. None of these effects are predicted from the conventional Snell’s law, but they do follow a generalized version derived by the authors that allows for desired variations of the change of phase on the interface.

These arrays of nanoantennas, which could include movable sections, could be used to design photonic components such as lenses and mirrors that are ultrathin, confor-mal (angle-preserving), and even deform-able. Reconfigurable couplers and wave-guides, which could be driven by electric, magnetic, or optical stimuli, may be envi-sioned that could guide and mix light beams through almost arbitrary paths chosen along a surface. Yu et al. have also created optical vortices with orbital angular momentum ( 6) by impinging a beam at normal incidence on the specially designed planar metasurface of these V-shaped nanoantennas. Such vortices could fi nd use in applications such as opti-cal tweezers.

Metasurfaces ( 7) are the planar version of metamaterials that are engineered to con-trol and tailor the light interaction in uncon-ventional ways (for example, creating mate-rials with optical band gaps that completely refl ect light over a given frequency range). In the three-dimensional metamaterials, it can be diffi cult to engineer a structure that maintains its designed performance and avoids performing like a bulk material. Meta-

surfaces may offer advantages in this regard because their constituent resonant elements are all distributed in a planar surface and more readily assembled. This type of two-dimensional structure will add another tool to the fi eld of transformation optics ( 8, 9), in which a prescribed change (such as a phase shift or amplitude variation) is designed into the light path for applications such as cloak-ing, or where metasurfaces are used to creat-ing highly confi ned cavity modes ( 10, 11) of potential interest in quantum optics.

References 1. M. Born, E. Wolf, Principles of Optics (Pergamon,

Oxford, 1980). 2. N. Yu et al., Science 334, 333 (2011). 3. D. M. Pozar, S. D. Targonski, H. D. Syrigos, IEEE Trans.

Antenn. Propag. 45, 287 (1997). 4. C. G. M. Ryan et al., IEEE Trans. Antenn. Propag. 58,

1486 (2010). 5. N. Bliznyuk, N. Engheta, Mic. Opt. Tech. Lett. 40, 361

(2004). 6. M. Padgett, J. Courtial, L. Allen, Phys. Today 57, 35

(2004). 7. E. F. Kuester, M. A. Mohamed, M. Piket-May, C. L. Hol-

loway, IEEE Trans. Antenn. Propag. 51, 2641 (2003). 8. J. B. Pendry, D. Schurig, D. R. Smith, Science 312, 1780

(2006). 9. U. Leonhardt, Science 312, 1777 (2006). 10. M. Caiazzo, S. Maci,, N. Engheta, IEEE Antenn. Wirel.

Propag. Lett. 3, 261 (2004). 11. C. L. Holloway, D. C. Love, E. F. Kuester, A. Salandrino, N.

Engheta, IET Microwave Antenn. Propag. 2, 120 (2008).

A

C DB

L >> λ L >> λ

d ≤ λ 5 μm

Bending light, big and small. Several mechanisms for bending light are depicted. The optical structures shown in (A) and (B) are much larger than the wavelength of light. In (A), an interface between two media with two different indices of refraction bends light. In (B), light is bent by a material that gradually changes refractive index with distance. Yu et al. caused the bending of light in unusual ways (C) with thin metasur-faces. These metasurfaces contain distributed arrays of gold nanoantennas (D) that are smaller than the wavelength of light. In such arrays, the proper patterns of phase changes created by resonant nanostructures lead to bending effects not anticipated by conventional laws of refl ection and refraction in optics.

10.1126/science.1213278

The advent of satellite-based remote sensing of ocean color in the late 1970s ( 1) provided the first large-

scale views of chlorophyll distributions in the upper ocean. These distributions are a proxy for the biomass of phytoplankton, which drive oceanic productivity. More recently, ocean color measurements have been combined with satellite data on sea-surface height (SSH) and other physical properties of the ocean to elucidate the pro-cesses that regulate primary production in

the sea. On page 328 of this issue, Chelton et al. ( 2) further advance this fi eld by showing that ocean eddies exert a strong infl uence on near-surface chlorophyll.

Initial comparisons ( 3, 4) of satellite ocean color measurements and SSH data showed that some of the variability in ocean color was associated with large-scale SSH patterns that propagate westward in extra-tropical latitudes. The authors attributed these patterns to planetary or Rossby waves, which are freely propagating modes of large-scale variability in the ocean. Four basic processes have been proposed to explain the observed relations, including lateral

Eddies Masquerade as Planetary Waves

OCEANS

Dennis J. McGillicuddy Jr.

Variabilities in sea-level and upper-ocean chlorophyll reveal the systematic infl uence of nonlinear eddies.

Woods Hole Oceanographic Institution, Woods Hole, MA 02543–1541, USA. E-mail: dmcgillicuddy@whoi.edu

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 319

PERSPECTIVES

advection of the mean chlorophyll gradi-ent, uplift of the deep chlorophyll maximum into the surface layer, enhancement of phy-toplankton biomass stimulated by upwelling of nutrients, and accumulation of material in convergence zones within the planetary wave fi eld ( 5– 7).

These early studies focused on large-scale signals characteristic of Rossby waves by processing the satellite measurements with scale-selective fi lters. This processing was intended to remove seasonal variability as well as the effects of mesoscale (tens to hundreds of kilometers) eddies—ubiquitous features of ocean circulation (sometimes referred to as the internal weather of the sea) that result from both direct forcing and inter-nal instability processes.

Merging data from multiple satellite mis-sions led to altimetric data sets ( 8– 10) with higher resolution than used previously, but it remains diffi cult to differentiate between Rossby waves and eddies in the merged data sets. Single Rossby waves (see the fi gure, panel A) are rarely if ever observed in the ocean, but superposition of multiple Rossby waves can result in eddy-like features (panel A, inset) that are similar to the patterns seen in altimeter data (panel B). However, plane-tary waves and eddies have different degrees of nonlinearity: Nonlinear eddies trap fl uid inside them, whereas linearly propagating wavelike disturbances do not. The degree of nonlinearity can be estimated as the ratio between an eddy’s swirl velocity and its translation speed.

Previously, Chelton and co-workers used this insight to show that mid-latitude SSH variability is dominated by westward-prop-agating nonlinear eddies, and developed automated tracking algorithms to compile a

global synthesis of eddy trajectories ( 9, 10). Chelton et al. have now overlaid those eddy tracks on the westward-propagating signals previously attributed to Rossby waves in the fi ltered SSH and ocean color data. The results strongly suggest that eddies are driv-ing these signals ( 2).

How might mesoscale eddies masquer-ade as larger-scale Rossby waves? Due to the latitudinal dependence of the effects of Earth’s rotation, both types of features move westward at roughly the same speed. More-over, Chelton et al. show that the statisti-cal properties of a patchwork of westward-propagating eddies are qualitatively similar to those expected for Rossby waves. This observation explains why eddies can pass through the filters intended to eliminate them in earlier studies.

Chelton et al.’s fi ndings require reassess-ment of the underlying mechanisms used to explain satellite observations of variability in SSH and upper-ocean chlorophyll. Although the same four basic processes of biomass modulation mentioned above for Rossby waves remain valid for eddies, lateral advec-tion of the mean chlorophyll gradient is the dominant mechanism revealed in Chelton et al.’s analysis. However, the relative impor-tance of each of the four processes can vary with oceanographic regime and scale, rang-ing from the mesoscale down to the submeso-scale ( 11– 13). At present, submesoscale fea-tures are not resolved by operational remote-sensing technology for SSH.

Although higher-resolution data are expected in the future for both SSH and ocean color, in situ observations will con-tinue to be critical for those variables that cannot be measured from space. Moreover, because near-surface waters are depleted in

nutrients over large areas of the mid-lati-tudes, key aspects of the biological response to physical perturbations take place too deep to be detected by satellite ocean color imag-ery ( 14). Although Chelton et al.’s results must be interpreted with that caveat, their fi ndings constitute a key step forward in our understanding of physical-biological inter-actions in the ocean, with important rami-fi cations for both ecosystem dynamics and biogeochemical cycling.

References and Notes 1. J. F. R. Gower, K. L. Denman, R. J. Holyer, Nature 288,

157 (1980). 2. D. B. Chelton, P. Gaube, M. G. Schlax, J. J. Early, R. M.

Samelson, Science 334, 328 (2011); 10.1126/science.1208897.

3. B. M. Uz, J. A. Yoder, V. Osychny, Nature 409, 597 (2001).

4. P. Cipollini, D. Cromwell, P. G. Challenor, S. Raffaglio, Geophys. Res. Lett. 28, 323 (2001).

5. P. D. Killworth, P. Cipollini, B. M. Uz, J. R. Blundell, J. Geophys. Res. 109, C07002 (2004).

6. Y. Dandonneau, A. Vega, H. Loisel, Y. du Penhoat, C. Menkes, Science 302, 1548 (2003).

7. G. Charria, F. Mélin, I. Dadou, M.-H. Radenac, V. Garçon, Geophys. Res. Lett. 30, 1125 (2003).

8. A. Pascual, Y. Faugère, G. Larnicol, P.-Y. Le Traon, Geophys. Res. Lett. 33, L02611 (2006).

9. D. B. Chelton, M. G. Schlax, R. M. Samelson, Prog. Oceanogr. 91, 167 (2011).

10. D. B. Chelton, M. G. Schlax, R. M. Samelson, R. A. de Szoeke, Geophys. Res. Lett. 34, L15606 (2007).

11. M. Lévy, P. Klein, A.-M. Treguier, J. Mar. Res. 59, 535 (2001).

12. E. R. Abraham, Nature 391, 577 (1998). 13. D. A. Siegel, P. Peterson, D. J. McGillicuddy Jr., S. Mari-

torena, N. B. Nelson, Geophys. Res. Lett. 38, L13608 (2011).

14. D. J. McGillicuddy Jr. et al., Science 316, 1021 (2007). 15. J. C. McWilliams, G. R. Flierl, Deep-Sea Res. 23, 285

(1976). 16. Altimeter data were produced and distributed by AVISO

(www.aviso.oceanobs.com) as part of the SSALTO ground-processing segment.

17. I thank NSF and NASA for support and L. Anderson for preparing the fi gure.

Rossby waves Satellite altimeter

Single wave

Superposition of 4 waves

Longitude (˚W)

Lati

tude

(˚N

)A

70 65 60 55 70 65 60 55

29

32

35

Longitude (˚W)

Sea-

leve

l ano

mal

y (c

m)

B

29

32

35

–20

–10

0

10

20

Rossby waves and eddies. (A) A model of a single Rossby wave propagating through a still ocean leads to a highly regular sea-level anomaly pattern. If sev-eral such waves are superposed, an eddy-like pattern results (inset) ( 15). (B) Mapped observations ( 16) of sea-level anomaly for a region of the western North

Atlantic on 17 June 2005. Satellite ground tracks are shown as dotted lines. Although the patterns seen in the altimeter data resemble those of the super-posed Rossby waves, Chelton et al. show that they are in fact caused by nonlinear mesoscale eddies.

10.1126/science.1208892

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org 320

PERSPECTIVES

Cells, and the organelles within them, are surrounded by lipid-bilayer membranes that compartmentalize

biochemical reactions and pathways. Mem-brane-embedded proteins control the fl ux of molecules, energy, and information such that the segregated compartments func-tion as a unifi ed living cell. Traditionally, membrane proteins were pictured as fl oat-ing around quite independently of the sur-rounding lipids, yet when this fl uid-mosaic model was described, Singer and Nicolson qualifi ed that “a small fraction of the lipid may interact specifi cally with the membrane proteins” ( 1). It has taken 40 years to fully appreciate the importance of this assertion. On page 380 of this issue, Zhou et al. ( 2) report mass spectrometry of intact integral membrane protein complexes solubilized from bilayers. The results show that specifi c structural lipids remain bound in the gas phase and can be counted.

Despite some technical hurdles, integral membrane proteins and their complexes have proven amenable to the soft ionization technologies of biological mass spectrom-etry. Intact complexes are isolated by gen-tle solubilization of membranes with mild nonionic detergent, followed by biochemi-cal isolation within a detergent micelle for solubility. Earlier studies next used organic solvents to break noncovalent associations with lipids and between subunits. Chroma-tography in aqueous/organic mixtures cou-pled to mass spectrometry provided intact molecular mass profi les of individual sub-units and their covalent modifications on low-resolution analyzers ( 3, 4). Top-down dissociation experiments on high-resolution Fourier-transform instruments yielded fully assigned primary structures with high statis-tical confi dence ( 5, 6).

However, if the complexes could be ana-lyzed intact, subunit stoichiometry and the presence of intimate cofactors such as spe-cific lipid molecules could also be deter-mined. In 2004, Robinson’s group realized that electrospray ionization could be used

to transfer membrane-protein micelles from aqueous solutions into the gas phase, with subsequent depletion of detergent in vacuo ( 7). Development of this native mass spec-trometry approach has enabled the measure-ment of bound lipid stoichiometry in mem-brane lipid-protein complexes described by Zhou et al.

The detection of lipids within integral membrane protein structures requires care-ful solubilization to preserve native asso-ciations. This is important because micelle-forming detergents tend to displace lipids, even while preserving protein subunit inter-actions. For example, successful crystalliza-tion of the cytochrome b6f complex required addition of lipids to isolated complexes, even though preparations retained some native lipids ( 8, 9).

As early as 1973, Santiago et al. argued, based upon lipid-peroxidation/protection data, for the specifi c requirement of the lipid cardiolipin for rat mitochondrial adenosine triphosphatase (ATPase) activity (see the fi gure, left), concluding that specifi c lipids were critical to the function of this integral membrane protein complex ( 10). Apprecia-tion of the importance of specifi cally bound lipids to the structure and function of inte-gral membrane proteins has grown steadily as x-ray crystallography has provided ever more examples from numerous membrane proteins. Cardiolipin was fi rst resolved in the structure of the purple bacterial reaction cen-

ter ( 11) and later shown to provide thermal stability to the structure ( 12). Although the structure of bovine mitochondrial rotary ATP synthase has been solved, cardiolipin was not seen, presumably due to its displacement by detergent ( 13).

The rotary ATP synthases couple the electrochemical energy of a transmembrane proton gradient to synthesis of ATP; proton movement through an integral membrane ring of subunits causes rotation of this rotor relative to a stationary integral stator and associated extrinsic subunits, driving confor-mational changes that facilitate ATP synthe-sis ( 14). The number of rotor subunits varies across living organisms, with those having the least subunits synthesizing the most ATP per proton translocated. Current research explores the role of bound lipids in the rotor structure, and measurements of the number of bound lipids and their binding sites by mass spectrometry should thus help resolve some of the outstanding details of the mecha-nism of rotary coupling.

Zhou et al. use native electrospray-ion-ization mass spectrometry to measure rotor subunit and bound lipid stoichiometry in two intact rotary ATPases from the bacteria Thermus thermophilus and Enterococcus hirae. The results show that the solubilized T. thermophilus complex has a ring with 12 subunits, with 6 phosphatidylethanolamine lipids bound to it, whereas the E. hirae rotor has 10 subunits with 10 cardiolipins. The

Up Close with Membrane Lipid-Protein Complexes

STRUCTURAL BIOLOGY

Julian Whitelegge

State-of-the-art mass spectrometry reveals how many specifi c lipids are bound to membrane proteins.

O

O

PO

–O

O

O

OH

OO

O

O

P O

O–

O

O

OO

Cardiolipin

Along for the ride. Cardiolipin (left) is an unusual phospholipid with four fatty acyl chains. In the integral membrane rotor of the eight-membered vertebrate mitochon-drial ATP synthase (right), each of four cardiolipin molecules binds the four trans-membrane domains of two ring subunits.

CR

ED

IT: A

DA

PTE

D B

Y P

. HU

EY

/SC

IEN

CE

NPI-Semel Institute for Neuroscience and Human Behav-ior, University of California at Los Angeles, Los Angeles, CA 90024–1759, USA. E-mail: jpw@chem.ucla.edu

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 321

PERSPECTIVES

Every Bit CountsCELL SIGNALING

Peter J. Thomas

Information theory is applied to cell biology to determine the processing properties of signaling pathways and the consequences of noise.

authors picture the T. thermophilus complex as six dimers each with a single lipid, con-sistent with electron microscopy images that support a hexameric structure ( 15). The E. hirae complex has 10 monomers, each with a bound cardiolipin, correcting a crystal struc-ture with misassigned lipids ( 16).

These measurements cover a range of 0.5 to 2 lipid fatty acyl chains per transmembrane domain within the rotor, consistent with the value of one inferred from the measurement of four cardiolipin per eight-subunit verte-brate mitochondrial ATPase rotor using 31P nuclear magnetic resonance (NMR) ( 17). The 16 transmembrane domains that, along with four intimately associated cardiolipins, make up this eight-membered ring (see the fi gure, right) function as the smallest natural ATPase rotor, accounting for the maximal effi ciency of ATP synthase activity observed for this complex (13).

Zhou et al. go on to show that gas-phase dissociation patterns are sensitive to pH and the presence of nucleotides, suggesting con-formational changes within the complexes. This conclusion is further supported by

gas-phase ion mobility mass spectrometry experiments that elegantly demonstrate the technological sophistication of the instru-mentation used ( 2). The authors further dis-cuss how rearrangement of lipid-protein interactions might serve to modulate ATP synthase/ATPase activities.

The ability to profile intact membrane lipid-protein complexes and detect organiza-tional changes related to dynamic processes by mass spectrometry, as reported by Zhou et al., will empower our understanding of the functioning of biological membranes. A new class of neopentyl glycol detergents that appear to solubilize membrane lipid-protein complexes with minimal structural pertur-bation bodes well for native mass spectrom-etry ( 18). Future research should also aim to elucidate the role of bilayers in general, as well as the specifi c lipid-protein associations discussed above, in the insertion, folding, assembly, function, and regulation of integral membrane proteins.

References 1. S. J. Singer, G. L. Nicolson, Science 175, 720 (1972). 2. M. Zhou et al., Science 334, 380 (2011).

3. I. M. Fearnley, J. E. Walker, Biochem. Soc. Trans. 24, 912 (1996).

4. J. P. Whitelegge, H. Zhang, R. Aguilera, R. M. Taylor, W. A. Cramer, Mol. Cell. Proteomics 1, 816 (2002).

5. B. Thangaraj et al., Proteomics 10, 3644 (2010). 6. C. M. Ryan et al., Mol. Cell. Proteomics 9, 791 (2010). 7. L. L. Ilag, I. Ubarretxena-Belandia, C. G. Tate, C. V.

Robinson, J. Am. Chem. Soc. 126, 14362 (2004). 8. H. Zhang, G. Kurisu, J. L. Smith, W. A. Cramer, Proc.

Natl. Acad. Sci. U.S.A. 100, 5160 (2003). 9. S. S. Hasan et al., J. Mol. Biol. (2011), 10.1016/

j.jmb.2011.09.023. 10. E. Santiago, N. López-Moratalla, J. F. Segovia, Biochem.

Biophys. Res. Commun. 53, 439 (1973). 11. K. E. McAuley et al., Proc. Natl. Acad. Sci. U.S.A. 96,

14706 (1999). 12. P. K. Fyfe, N. W. Isaacs, R. J. Cogdell, M. R. Jones,

Biochim. Biophys. Acta 1608, 11 (2004). 13. I. N. Watt, M. G. Montgomery, M. J. Runswick, A. G.

Leslie, J. E. Walker, Proc. Natl. Acad. Sci. U.S.A. 107, 16823 (2010).

14. D. Stock, A. G. Leslie, J. E. Walker, Science 286, 1700 (1999).

15. R. A. Bernal, D. Stock, Structure 12, 1789 (2004). 16. T. Murata, I. Yamato, Y. Kakinuma, A. G. Leslie, J. E.

Walker, Science 308, 654 (2005). 17. K. S. Eble, W. B. Coleman, R. R. Hantgan, C. C.

Cunningham, J. Biol. Chem. 265, 19434 (1990). 18. P. S. Chae et al., Nat. Meth. 7, 1003 (2010).

10.1126/science.1214084

The term “signal transduction” has been in use for over 40 years, originating in biological studies of cellular photo-

receptors and chemotaxis, and elaborated through discoveries of signaling systems such as the protein kinases ( 1). In engineer-ing, “signaling” has a precise meaning that defi nes the notion of “information” ( 2), and quantifi es the capabilities of a signaling sys-tem in terms of its channel capacity, measured in bits ( 3). Despite the attractive analogy, a number of obstacles have thwarted attempts to apply information theory quantitatively to cellular signal transduction networks. On page 354 of this issue, Cheong et al. ( 4) suc-ceed in doing so, making rigorous and quan-titative measurements of information capaci-ties in a biochemical signaling system.

A rigorous analysis of signaling requires a well-defined communications channel, including an ensemble of channel inputs as well as outputs that are conditional on each

possible input (see the fi gure). In a biological setting, it is not always clear what defi nes the appropriate ensembles. Cheong et al. use the signaling pathway engaged by tumor necro-sis factor (TNF), which stimulates a host of intracellular responses mediated by two tran-scription factors, nuclear factor kappa B (NF-κB) and activating transcription factor–2 (ATF-2). The TNF concentration serves as the input to the channel, and the NF-κB and ATF-2 responses serve as outputs.

Estimating channel capacities requires estimating the entropy of probability distri-butions, which is a notoriously data-inten-sive undertaking. Channel capacities and quantities such as “information per spike” have been estimated for neural networks in systems with large fi ring (activity) rates ( 5). Previous attempts to accomplish a sim-ilar analysis in a biochemical signaling sys-tem have not been convincing. Using quan-titative immunocytochemistry, Cheong et al. obtained four-dimensional arrays of data (time, dosage, genotype, and pathway) for thousands of individually resolved cells. With this much data, it becomes possible to

estimate the entropies required to establish the channel capacity, accounting for known biases in entropy measurements due to fi nite (but large) sample sizes.

Biochemical systems often form net-works with multiple interactions (“cross talk”) that make the analysis of statistical dependencies, and hence channel capaci-ties, problematic. Here, Cheong et al. mea-sured two readouts in the same system. The amount of information carried by the com-bination of the ATF-2 and the NF-κB path-way depends, in principle, on the topology of interactions between the pathways. If both pathways get their information directly from the incoming signal, the mutual informa-tion of the two responses taken together can potentially be substantially increased. This situation is described as a “bush” topology, because the signals propagate directly along multiple branches from the root of the net-work. Alternatively, the network could have a “tree” topology, in which the input signal passes through a common intermediate (the trunk of the tree) before going to independent readout branches. Remarkably, the differ-

Departments of Mathematics, Biology, and Cognitive Science, Case Western Reserve University, Cleveland, OH 44106, USA. E-mail: pjthomas@case.edu

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org 322

PERSPECTIVES

ence in the mutual information predicted in these two scenarios is large enough, and the data set collected by the authors substantial enough, to unambiguously identify the TNF–NF-κB–ATF-2 network as having a treelike, rather than bushlike, topology. Each pathway alone carries just short of one bit of informa-tion. [In information theory, one bit, or one binary digit, is the information needed to dis-tinguish exactly two equally likely outcomes of an experiment (e.g., heads versus tails in a fair coin toss). Two bits is enough informa-tion to distinguish four outcomes, and N bits is enough information to distinguish 2N out-comes.] However, the joint response of both pathways carries just over one bit, enough to reliably distinguish a single yes-no decision, such as presence or absence of TNF. Indeed, the treelike topology likely refl ects the fact that before branching out to stimulate multi-ple pathways, the TNF response is mediated by a single receptor signaling (and pathway activation) complex. As the authors show, this arrangement limits the capacity of the net-work. Even if an unlimited number of inde-pendent pathways separately read the output of the TNF receptor complex, the information obtained by a single cell about the amount of stimulatory TNF cannot exceed about 1.25 bits, which is only slightly more than from the NF-κB and ATF-2 pathways alone.

The capacity of any channel is intimately related to the nature of the noise affecting the input-output relationship, and it can be dif-fi cult to establish quantitative characteristics of the noise sources present in a biochemical channel. In engineering, one usually consid-ers the properties of a communications chan-nel (such as a telegraph line or fi ber optic

cable) as fixed, and the engineering prob-lem amounts to choosing the best encoding scheme given the (noisy) signal transmission equipment. A decade ago, it was proposed that biological information channels should be sculpted by evolutionary pressures so as to match to the information sources that feed into them ( 6). In sensory neural systems, source-channel matching takes the form of redundancy reduction and effi cient coding ( 7– 9). The pioneering work by Cheong et al. opens the door to similar analyses for bio-chemical signaling systems.

The biggest surprise in the analysis of Cheong et al. is how small the informa-tion capacity of the TNF signaling system appears to be. Noise suppression in bio-chemical systems is notoriously challeng-ing for cells to implement (and for investi-gators to analyze) ( 10). Possible strategies for increasing the information about a signal such as TNF concentration include reduc-ing noise by introducing a negative-feedback loop, averaging over noise by integrating received signals over time, or pooling infor-mation across multiple cells. The authors explore each possibility. Negative feedback proves to be a double-edged sword: It can reduce the amount of noise in transducing a signal, which tends to boost capacity, but it can also restrict the dynamic range of the sig-nal input, reducing capacity. In a mutant cell line lacking a certain negative-feedback loop that normally inhibits the TNF receptor com-plex, the authors found increased informa-tion capacity in the short term (output at 30 min after stimulation) but decreased capac-ity in the long term (4 hours after stimula-tion). In both cases, however, the information

capacity was still approximately one bit. [In a signaling system comprising a rate-modulated Pois-son process—for example, com-municating by directly detecting occurrence of chemical events—it can be proven mathematically that feedback cannot increase the channel capacity ( 11).] Integration over time (or averaging over time) does not fare much better, because once activated, NF-κB activity remains approximately constant in any given cell. In other biochemi-cal systems, time integration could still play a role in noise reduction. Of the strategies explored, pooling information across multiple cells that each independently process the same input signal (an example of the bush topology) holds the most promise. The authors show

that readouts from local pools of 14 cells give twice the information of single cells alone.

We can look forward to more contribu-tions from information theory to cell biol-ogy (and vice versa), particularly in gra-dient sensing the process by which motile eukaryotic cells orient themselves during chemotaxis ( 12). In this process, the ensem-ble of inputs is all the possible directions of the chemoattractant gradient ( 13– 15).

In 2002, Berger asserted that biology had become positioned to profi t meaningfully from an invasion by information theorists ( 6). Nearly a decade later, the invasion is well under way and shows no sign of slow-ing down.

References 1. T. Hunter, B. M. Sefton, Proc. Natl. Acad. Sci. U.S.A. 77,

1311 (1980). 2. C. Shannon, Bell Syst. Tech. J. 27, 379 and 623 (1948). 3. T. M. Cover, J. A. Thomas, Elements of Information Theory

(Wiley, New York, ed. 2, 2006). 4. R. Cheong, A. Rhee, C. J. Wang, I. Nemenman,

A. Levchenko, Science 334, 354 (2011); 10.1126/science.1204553.

5. F. Rieke, D. Warland, R. de Ruyter van Steveninck, W. Bialek, Spikes: Exploring the Neural Code (MIT Press, Cambridge, MA, 1999).

6. T. Berger, IEEE Inf. Theory Soc. Newsl. 53 (March 2003). 7. H. Barlow, in Sensory Communication, W. A. Rosenblith,

Ed. (MIT Press, Boston, MA, 1961), pp. 217–234. 8. B. A. Olshausen, D. J. Field, Nature 381, 607 (1996). 9. E. C. Smith, M. S. Lewicki, Nature 439, 978 (2006). 10. I. Lestas et al., Nature 467, 174 (2010). 11. Y. M. Kabanov, Theory Probab. Appl. 23, 143 (1978). 12. C. A. Parent, P. N. Devreotes, Science 284, 765 (1999). 13. B. W. Andrews, P. A. Iglesias, PLOS Comput. Biol. 3, e153

(2007). 14. J. M. Kimmel, R. M. Salter, P. J. Thomas, Adv. Neural Inf.

Process. Syst. 19, 705 (2007). 15. D. Fuller et al., Proc. Natl. Acad. Sci. U.S.A. 107, 9656

(2010).

Receiver Destination

Noisesource

Signal Receivedsignal

MessageMessageInformationsource

Transmitter

Information flow. Schematic diagram of a general communication system ( 2). The information capacity of a signaling system quantifi es its ability to transmit information (measured in bits), whether it is an engineered cable or a living cell.

10.1126/science.1213834 CR

ED

IT: Y

. HA

MM

ON

D/S

CIE

NC

E

Globalization, Land Use, and the

Invasion of West Nile Virus

A. Marm Kilpatrick

Many invasive species that have been spread through the globalization of trade and travel arepathogens. A paradigmatic case is the introduction of West Nile virus (WNV) into North Americain 1999. A decade of research on the ecology and evolution of WNV includes three findingsthat provide insight into the outcome of future pathogen introductions. First, WNV transmissionin North America is highest in urbanized and agricultural habitats, in part because the hostsand vectors of WNV are abundant in human-modified areas. Second, after its introduction,the virus quickly adapted to infect local mosquito vectors more efficiently than the originallyintroduced strain. Third, highly focused feeding patterns of the mosquito vectors of WNV resultin unexpected host species being important for transmission. This research provides a frameworkfor predicting and preventing the emergence of foreign vector-borne pathogens.

The growth of human populations and the

development of rapid transportation sys-tems have made the world’s biota more

connected than at any time in Earth’s history. Theresult has been a breakdown in biogeographic

barriers and the introduction of species into nov-el habitats. Globally, introduced invasive spe-

cies are estimated to cause >$120 billion indamage annually (1) and include several patho-

gens that have direct impacts on the health ofhumans, livestock, andwildlife. Pathogens spread

by trade and travel in the past 500 years includethose causing the human diseasesmalaria, dengue,

and HIV/AIDS; wildlife and livestock pathogens,such as anthrax, rinderpest, rabies, and avian ma-

laria; and numerous diseases of crops and wildplants, including chestnut blight, potato blight,

and sudden oak death (2, 3). Introductions havecontinued with invasions by novel strains of in-

fluenza, severe acute respiratory syndrome, and

West Nile virus (WNV), amongmany others. Thefactors that determine the outcome and impact of

invasions are frequently complex and poorly un-derstood (4, 5); however, extensive research on

WNVover the past decade has enabled a detailedexploration of its invasion, including pathways of

introduction, interactions with the biotic and abi-otic environment in the new region (Fig. 1), and

impacts on ecosystem health.WNV is a single-stranded RNA virus in the

family Flaviviridae that includes several impor-tant humans pathogens: dengue, Japanese enceph-

alitis, and yellow fever viruses (6). WNV wasfirst isolated in 1937 from a febrile patient in

Uganda, and subsequent studies showed thatWNVtransmission was endemic and widespread across

tropical parts of Africa, southern Asia, and north-ern Australia, and episodic in more temperate

parts of Europe (7). As with other vector-borne

diseases, the warmer temperatures in the tropics

facilitate longer transmission seasons and some-times increased transmission intensity through

faster mosquito and virus development and in-creased biting rates. In some populations in Af-

rica, >80% of people over 15 years old haveantibodies to WNV (8); however, WNV was

previously considered nearly asymptomatic and

in the 1950s was even tried as an anticancer

therapy (9).In 1999, WNV was introduced into North

America, where it spread rapidly with major ec-onomic and public health consequences (7). The

virus reached the west coast in only 4 years(Fig. 2), with regional epidemics in 2002 and

2003 and more localized epidemics occurring inother years. Between 1999 and 2010, ~1.8 mil-

lion people were infected, with ~360,000 ill-nesses, 12,852 reported cases of encephalitis/

meningitis, and 1308 deaths. The threat of WNV

infection has led to the costly implementation ofnational blood donor screening, as well as vac-

cine and drug development (10). Public out-reach campaigns have altered human behavior,

including the time spent outdoors, especiallyby older people, who are at high risk for WNV

disease.The impacts of WNV on wildlife have been

yet more severe than those on humans. Millionsof birds have died from WNV infection, and

regional-scale population declines of >50% havebeen observed for several species (11). The range

of taxa that have suffered declines is surprisinglylarge and includes corvids, chickadees and tit-

mice, wrens, and thrushes (Fig. 1) and probablyothers. Some populations have recovered after

initial declines, whereas others have not (11). The

REVIEW

Department of Ecology and Evolutionary Biology, Univer-sity of California, Santa Cruz, CA 95064, USA. E-mail:akilpatr@ucsc.edu

Fig. 1. An American robin (T. migratorius) and its nestlings. Robins flourish in human-altered landscapesand appear to play a key role in WNV amplification across many regions of North America. [Photo credit:Bruce Lyon]

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 323

ecological and economic consequences of theseregional declines in bird populations have not yet

been elucidated and need further study.

Globalization and the Introduction of Pathogens

The probability of the introduction and establish-ment of introduced species has been shown to

increase with the “propagule pressure,” or therate at which individual organisms are introduced

to a new region (Fig. 3) (5). The yearly propagulepressure and the pathway bywhichWNVreached

North America in 1999 remain unknown, butseveral possibilities have been proposed, including

mosquitoes being transported by shipping, air-planes, or wind; migratory birds or birds in trade;

and humans traveling (12). The large and increas-ing volume of air traffic into NewYork City over

the past five decades makes the transport of in-fected mosquitoes on an airplane a likely path-

way. A close phylogenetic relationship betweenviruses isolated in New York in 1999 and those

circulating in Israel in the previous year sug-gests a possible Middle East origin (6). Trade

and travel have also previously introduced key

mosquito vectors of WNV, Culex pipiens andC. quinquefasciatus, as well as vectors for den-

gue, yellow fever, and chikungunya viruses, suchas Aedes albopictus and A. aegypti (13).

What predictions could have been made in

1999 about the outcome of the introduction ofWNV into New York City that summer? An an-

swer comes from comparing the ecology ofWNVtransmission in the Americas with that in Africa

and Europe (14, 15).

WNV Ecology in Its Native Range

Studies of endemic WNV transmission in Egypt,

Sudan, and South Africa and of Kunjin virus, asubtype ofWNV, in Australia show that the virus

was most frequently isolated fromCulexmosqui-toes. In Australia, most isolations come from

C. annulirostris, which is a competent laboratoryvector (16). In Africa,C. univittatusmakes up the

largest fraction of WNV-infected mosquitoes(8, 17). Interestingly, there is little evidence of

WNV infection in C. pipiens in South Africa, de-spite frequent feeding on avian hosts.C. pipiens is

an important WNV vector in Europe and North

America (15, 18). It is possible that the lowerWNVinfection prevalence observed inC. pipiens

than in C. univittatus can be attributed to its beingless susceptible to infection (17).

Accurate quantification of the contribution ofdifferent host species to viral amplification re-

quires data onmosquito feeding patterns and hostabundance from the same place and time, com-

bined with information on the duration and in-tensity of host infectiousness (19). Host abundance

and mosquito feeding data have never been col-lected simultaneously for WNV hosts and vec-

tors in Africa, Asia, Australia, or Europe and haveonly rarely been collected in North America. As a

result, only tentative conclusions can be drawn

about the relative importance of host species forWNV outside North America and these largely

come from studies of seroprevalence and infec-tiousness based on viremia (concentration of virus

in the blood) observed after experimental infec-tions. In Egypt, hooded crows (Corvus cornix)

and house sparrows (Passer domesticus) hadhigh antibody prevalence and infectiousness (8).

In South Africa, waterbirds (ducks and rails) andpasserine birds in the family Ploceidae (weavers

and Old World sparrows, including house spar-rows) were most infectious and most frequently

had antibodies to WNV (17).

The Vectors, Hosts, and Transmission

of WNV in the Americas

The most important vectors in North Americashare some similarities with those in Africa, Eu-

rope, and Australia. AlthoughC. univittatus is notpresent in the Americas, C. pipiens, C. quinque-

fasciatus, and several other species that take themajority of their blood meals from birds are

found in North America, including C. restuans,

C. tarsalis, and C. nigripalpus. Based on their

feeding ecology and their vector competence, allthese species would be expected to be important

in enzootic (bird-to-bird) transmission (18, 20).In addition, their abundance in anthropogenically

modified areas points to a significant role in hu-man WNVepidemics.

An important insight was gained in the courseof determining the vector species responsible for

transmitting WNV from nonhuman animals to

humans (i.e., “bridge vectors”). Bridge vectorswere initially thought to be mosquito species that

fed frequently on mammals (such as Aedes mos-quitoes), but an integrated analysis of the abun-

dance, infection prevalence, feeding patterns, andvector competence of a wide range of mosquitoes

indicated that C. pipiens and C. restuans mos-quitoes, which frequently take <15%of their blood

meals from humans, may nonetheless be respon-sible for the majority of human infections in sev-

eral regions (18, 21). Their importance results fromtheir higher relative abundance and WNV infec-

tion prevalence than the more anthropophilic mos-quito species. Applying this integrated approach

to other pathogens may simplify targets for vec-tor control, especiallywhen the same species serves

as both the bridge and the enzootic vector.

0

47

0.0

0.2

0.4

0.6

0.8

1.0

Not detected yet199920002001200220032004200520072009

1 2 3 4 5 6 7 8 9 10 11

Years since WNV detection in that state

Years of WNV detection

Rel

ativ

e n

um

ber

of

case

s

47 47 47 47 47 47 46 43 27 12 4

Fig. 2. Spread of WNV throughout the Americas. Themap shows the year that WNV was first detected in astate, province, or country. The box plot shows the temporal pattern of WNV incidence at the state levelafter WNV arrival. The y axis shows the relative number of WNV neuroinvasive cases (the fraction of themaximum observed in that state) that occurred in each state in each year, starting with the year WNV wasfirst detected in birds, mosquitoes, humans, or horses. The number of states included in each column isshown above the box.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org324

REVIEW

Predicting which avian hosts would be im-portant for WNV transmission in the Americas

based on data from Africa or Europe would havefailed, because inferences based solely on abun-

dance, infectiousness, or serological (antibody)prevalence can be misleading (19). Analyses of

data from the mid-Atlantic to Colorado that havecombined host abundance and mosquito feeding

data with host infectiousness suggested that al-though introduced house sparrows (P. domesticus)

and crows (Corvus spp.) are abundant and/orhighly infectious, they appear to be of minor im-

portance in WNV transmission (21–23). Crowsmake up a small fraction ofmosquito bloodmeals,

and house sparrows are rarely fed on by mos-

quitoes relative to their abundance, resulting infew bites per individual and inefficient trans-

mission. Instead, a species of thrush, the Amer-ican robin (Turdus migratorius), appears to be

more important in WNV transmission (Fig. 4)

(21–23). This is primarily because 30 to 80% ofmosquito feedings by the dominant WNV vec-

tors (C.pipiens,C. restuans, andC. tarsalis) are onrobins, despite robins making up only 1 to 20%

of the avian communities studied. Questionsthat arise are why are robins so frequently fed

on by mosquitoes, and do robins share a traitwith other thrushes that makes them generally

important for avian arboviruses? For example,serological studies of the avian Sindbis virus in

Sweden indicated higher exposure of thrushesthan of any other group (24).

Research has also shown that focused feedingon robins amplifies WNV transmission intensity

(22). This raises the following question: If Amer-

ican robins, which have increased 50 to 100% inabundance over the past 25 years with the ur-

banization of the North American landscape (11),were less abundant, wouldWNV transmission be

lower? It’s uncertain, because if mosquito abun-

dance and feeding patterns remain constant, de-creasing host abundance increases the vector:host,

ratio which increases transmission intensity. Inaddition, seasonal decreases in robin abundance

have been correlated with a shift in mosquito feed-ing from birds to humans, which increases hu-

manWNVinfections (25). However, in thewesternand southern United States, where robins are less

abundant, they provide only a small fraction ofmosquito blood meals, and yet mosquito feeding

by another species, C. quinquefasciatus, on hu-mans is no greater than in the east (26). As a

result, the impact of reducing robin abundance onWNV transmission is unknown and would proba-

bly depend on the identity, abundance, and infec-

tiousness of alternate sources of mosquito bloodmeals.

In summary, three important insights havebeen gained in determining the amplification hosts

of WNV in North America. First, abundant hosts

Human land useUrbanization, Agriculture

Globalization of trade & travel

Greenhouse gases

Increases in humancommensal vectorsand hosts

TravelTrade in animalsAnimal migration

Altered CO2,temperatures,and precipitation

Fig. 3. Anthropogenic processes that facilitate the introduction and es-tablishment of novel pathogens and increase their transmission. Trade,travel, and animal movement introduce new pathogens. Climate, hosts,and the abundance and feeding ecology of vectors determine establish-ment and transmission intensity. Land use modifies animal communitiesthat serve as hosts and vectors for pathogens, and climate change alters

pathogen and vector demographic rates. [Image credits: Google and TeleAtlas (aerial photos); Edward Canda (rice paddy); Photos8.com (corn-field); L. Hufnagel (air traffic map); Dori (dori@merr.info) (smokestacks);Joe Hoyt (left mosquito); Andrew Flemming (right mosquito); Richard Kuhn,Purdue Department of Biological Sciences (virus); NASA (clouds); MarmKilpatrick (others)]

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 325

REVIEW

may be fed on infrequently by vectors, making

them less important in transmission. Second, theimportance of hosts may be determined more by

how frequently mosquitoes feed on them than byvariation in their infectiousness. Finally, variation

in the abundance of key avian hosts can haveunpredictable impacts on transmission, especially

to humans.

Hosts and Vectors as Ecological Niches

for Pathogens

Studies of WNV amplification hosts show how

the feeding patterns and competence of insectvectors and vertebrate hosts create ecological

niches for introduced vector-borne pathogens. Char-acterization of these niches can inform predictions

of establishment probability for pathogen intro-ductions (19) and augment projections that are

frequently based on climatic conditions alone.Such predictions can be used to guide manage-

ment decisions in allocating resources towardprevention of pathogen introductions (such as

vaccine development and testing and quarantineof imported animals). One example of the insight

gained from host and vector competence studiescomes from an elegant comparison of WNV and

St. Louis encephalitis virus (SLEV, a flavivirus

native to theAmericas) by Reisen et al. (27), which

showed that WNV is more infectious in hosts tobiting vectors than SLEV and explained why

WNVepidemics aremore severe than those causedby SLEV.

Land Use and WNV Transmission

Recent evidence has suggested that at the countyscale in eastern and western North America, hu-

manWNV incidence increases with urbanization

and agriculture, respectively (28) (Figs. 3,4). Thismay result from the habitats used and human-

commensal nature of three important WNVmos-quito vectors,C. pipiens,C. quinquefasciatus, and

C. tarsalis, although the exact mechanisms actingat local scales are not yet known. Nevertheless,

the distribution ofWNVindicates that it is similarto other pathogens whose transmission is linked

with anthropogenic land use and increased abun-dance of domesticated animals and human-tolerant

wildlife species (Figs. 3 and 4). For example,H5N1 avian influenza emerged from poultry in-

tensification; rabies transmission in the Serengetiis maintained by domestic dogs; Lyme disease

increases with the fragmentation of forests ineastern North America; and yellow fever, den-

gue, and chikungunya viruses are all transmitted

by the anthropophilic mosquitoes A. aegypti and

A. albopictus (29–33). Perhaps ecologically basedland-use planning, combined with improved de-

velopment and sanitation, could reduce contactwith and the abundance of human-commensal spe-

cies and hence transmission of their pathogens.

Coevolution of Hosts, Vectors, and Pathogens

Rapid coevolution between WNV and its hosts,

vectors, and other pathogens is expected based

on reciprocal fitness impacts and inmany cases, thelack of shared evolutionary history (11, 27, 34, 35).

Still, it was somewhat surprising that by 2005,the strain ofWNV that was introduced into North

America in 1999 (NY99) had been displacedcontinent-wide by a locally evolved genotype,

WN02 (36). WN02 was first detected in 2001and spread continent-wide between 2002 and

2004. Viruses in the WN02 clade consistentlydiffer from NY99 viruses by only three nucle-

otides that result in one amino acid change.Nonetheless, WN02 viruses are more efficient-

ly transmitted by both C. pipiens and C. tarsalis

mosquitoes, and the difference was found to in-

creasewith temperature in the laboratory, aswouldbe expected if the WN02 viruses replicate at a

higher rate (37, 38).

Fig. 4. WNV ecology across an urbanization gradient in the northeastern andmidwestern United States. WNV is transmitted primarily by C. pipiens mosqui-toes among a wide range of birds, but American robins (outlined) are a key

amplification host. The diversity of avian hosts decreases with urbanization,whereas C. pipiens abundance appears to increase. [Image credit: U.S. Geo-logical Survey (mosquito); Marm Kilpatrick and Ryan Peters (others)]

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org326

REVIEW

The difference between NY99 andWN02 vi-ruses in competence (that is, in magnitude and

duration of infectiousness) in avian hosts has notyet been determined. However, another single-

stranded RNAvirus, Venezuelan equine enceph-alitis virus, has repeatedly evolved to be able to

infect novel hosts and mosquito vectors efficient-ly, and this shows that host adaption is also

possible (33).There may be evolutionary selective pressure

for WNV to kill its avian hosts. Individual birdsthat die from WNV infection have higher vi-

remia, and thus infectiousness to biting mosqui-toes, than individuals that survive (27, 34), and

host illness from infection decreases host de-

fenses against biting mosquitoes, which wouldincrease vector-host contact rates. Both of these

mechanisms increase pathogen fitness by in-creasing host-to-vector transmission. In addition,

in contrast to an assumption made in many mod-els of the evolution of virulence, host death from

WNV does not appear to reduce the length ofthe infectious period of the host: Most avian

hosts that survive WNV infection clear virusfrom their blood between days 4 and 6 after in-

fection, and most individuals that die fromWNVinfection do so at approximately the same time

after infection (27, 34). A key question is whetherviral evolution that increased replication and vi-

rulence in hosts would have deleterious effectsin the vector.

It is also unknown whether North Americanbirds have evolved increased resistance to WNV.

This could be determined by repeatingWNV lab-oratory challenge experiments using individuals

from the same populations in which resistancewas previously measured early in the WNVepi-

demic (27, 34). Ideally, such studies would in-clude a range of host species or populations that

have experienced different selective pressuresexerted on them by WNV; for example, in terms

of WNV transmission intensity or initial suscep-tibility toWNVmortality (for example, doves are

more resistant than corvids) (27, 34).

Outlook: Unanswered Questions

A key question is will WNV follow the boom-

and-bust pattern seen in some plant and animal

species invasions (5)—are the worst WNV epi-demics behind us? WNV epidemics peaked in

many states the year after it arrived, with fewerhuman cases having been observed subsequent-

ly (Fig. 2). This reduction in WNV disease hasled to reduced research focus and less funding

from public health agencies for WNV, and, morerecently, less testing for WNV by health care

providers.Reduced transmission may be a product of

several factors, including elevated immunity inbirds or humans, especially the subset of people

most at risk: the homeless and those spendingmore time outdoors at dusk (7). However, annual

recruitment of young-of-the-year birds apparent-ly fuelsWNV (39), which reduces the importance

of avian host immunity in suppressing transmis-

sion. Instead, it’s possible thatWNV transmissionis modulated primarily by rainfall and temper-

ature, and if so, climatic conditions in 2002 and2003 were especially suitable. Climate is known

to influence many aspects of WNV transmission,including mosquito abundance, biting rate, and

survival as well as viral replication within themosquito (37). If WNV transmission is regulated

by climate, then severe epidemics could recur,especially if they are facilitated by climate change

(Fig. 3). It is notable that the largest number ofneuroinvasiveWNV cases observed in NewYork

State was in 2010, 11 years after the virus wasintroduced. Clearly, determining the relative roles

of climate versus other factors in year-to-year

variation in transmission is important and neces-sary to predict the long-term trajectory of WNV

in North America.A second unanswered question is why haven’t

more morbidity and mortality been reported inhorses, humans, and birds in tropical regions

(7, 40)? Less surveillance is undertaken in theseless-developed countries than in North Amer-

ica, and the presence of other diseases, such asdengue, malaria, and Chagas, whose public health

impact dwarfs that of WNV, could account forlower reporting, despite similar WNV incidence

and illness. Alternatively, cross-protection by anti-bodies or evolved resistance to illness from other

flaviviruses (such as SLEV, dengue, or yellowfever viruses) in humans and horses may decrease

illness, without reducing bird-mosquito trans-mission. In addition, enzootic transmission may

be lower in the tropics than in North America,owing to cross-protecting flavivirus antibodies in

birds, a mismatch between periods of peak mos-quito abundance and susceptible young-of-the-

year birds, or lower infectiousness of tropicalavian hosts. These mechanismsmay be operating

simultaneously.Continual introduction of pathogens to new

regions is inevitable in our globally connectedplanet. It is unclear which vector-borne pathogen

will be the next to cross hemispheres, but manyviruses of public health concern exist in Africa,

Asia, and Europe, including other arthropod-borneviruses such as Japanese encephalitis, Rift Valley

fever, tick-borne encephalitis, and chikungunya

(31). Conversely, there are several pathogens fromthe Americas that could be introduced into the

OldWorld, including Venezuelan equine enceph-alitis and SLEV. Insights gained from studying

the invasion of WNV can be used to help predictwhich are the highest-risk pathogens for estab-

lishment after cross-hemispheric introduction (19).Gaining an understanding of the ecology of zoo-

notic viruses, combined with fast-developing re-combinant vaccine technologies that have already

been applied towildlife (41), could form the basisof a strategy to prevent the emergence of newly

introduced pathogens.

References and Notes1. D. Pimentel, R. Zuniga, D. Morrison, Ecol. Econ. 52,

273 (2005).

2. P. K. Anderson et al., Trends Ecol. Evol. 19, 535 (2004).

3. P. Daszak, A. A. Cunningham, A. D. Hyatt, Science 287,

443 (2000).

4. R. N. Mack et al., Ecol. Appl. 10, 689 (2000).

5. A. K. Sakai et al., Annu. Rev. Ecol. Syst. 32, 305

(2001).

6. R. S. Lanciotti et al., Science 286, 2333 (1999).

7. L. R. Petersen, E. B. Hayes, Med. Clin. North Am. 92,

1307, ix (2008).

8. H. S. Hurlbut, F. Rizk, R. M. Taylor, T. H. Work, Am. J.

Trop. Med. Hyg. 5, 579 (1956).

9. C. M. Southam, A. E. Moore, Am. J. Trop. Med. Hyg.

3, 19 (1954).

10. D. W. C. Beasley, Immunotherapy 3, 269 (2011).

11. S. L. LaDeau, A. M. Kilpatrick, P. P. Marra, Nature 447,

710 (2007).

12. A. M. Kilpatrick et al., Conserv. Biol. 20, 1224 (2006).

13. L. P. Lounibos, Annu. Rev. Entomol. 47, 233 (2002).

14. Z. Hubálek, J. Halouzka, Emerg. Infect. Dis. 5, 643

(1999).

15. M. V. Fyodorova et al., J. Med. Entomol. 43, 552 (2006).

16. R. A. Hall, J. H. Scherret, J. S. Mackenzie, Ann. N.Y. Acad.

Sci. 951, 153 (2001).

17. B. M. McIntosh, P. G. Jupp, D. B. Dickinson, G. M. McGillivray,

J. Sweetnam, S. Afr. J. Med. Sci. 32, 1 (1967).

18. A. M. Kilpatrick et al., Emerg. Infect. Dis. 11, 425 (2005).

19. Supporting material is available on Science Online.

20. M. J. Turell, M. R. Sardelis, M. L. O'Guinn, D. J. Dohm,

in Japanese Encephalitis and West Nile Viruses Vol. 267

Current Topics in Microbiology and Immunology,

J. Mackenzie, A. Barrett, V. Deubel, Eds. (Springer-Verlag,

Berlin, 2002), pp. 241–252.

21. G. L. Hamer et al., Am. J. Trop. Med. Hyg. 80, 268

(2009).

22. A. M. Kilpatrick, P. Daszak, M. J. Jones, P. P. Marra,

L. D. Kramer, Proc. Biol. Sci. 273, 2327 (2006).

23. R. Kent, L. Juliusson, M. Weissmann, S. Evans, N. Komar,

J. Med. Entomol. 46, 380 (2009).

24. J. O. Lundström, K. M. Lindström, B. Olsen, R. Dufva,

D. S. Krakower, J. Med. Entomol. 38, 289 (2001).

25. A. M. Kilpatrick, L. D. Kramer, M. J. Jones, P. P. Marra,

P. Daszak, PLoS Biol. 4, e82 (2006).

26. G. Molaei et al., Am. J. Trop. Med. Hyg. 83, 1269 (2010).

27. W. K. Reisen, Y. Fang, V. M. Martinez, J. Med. Entomol.

42, 367 (2005).

28. S. E. Bowden, K. Magori, J. M. Drake, Am. J. Trop.

Med. Hyg. 84, 234 (2011).

29. J. L. Kyle, E. Harris, Annu. Rev. Microbiol. 62, 71

(2008).

30. B. F. Allan, F. Keesing, R. S. Ostfeld, Conserv. Biol. 17,

267 (2003).

31. S. C. Weaver, W. K. Reisen, Antiviral Res. 85, 328 (2010).

32. T. Lembo et al., J. Appl. Ecol. 45, 1246 (2008).

33. S. C. Weaver, A. D. T. Barrett, Nat. Rev. Microbiol. 2,

789 (2004).

34. N. Komar et al., Emerg. Infect. Dis. 9, 311 (2003).

35. J. N. Thompson, Am. Nat. 153, S1 (1999).

36. C. T. Davis et al., Virology 342, 252 (2005).

37. A. M. Kilpatrick, M. A. Meola, R. M. Moudy, L. D. Kramer,

PLoS Pathog. 4, e1000092 (2008).

38. R. M. Moudy, M. A. Meola, L. L. Morin, G. D. Ebel,

L. D. Kramer, Am. J. Trop. Med. Hyg. 77, 365 (2007).

39. G. L. Hamer et al., Vector Borne Zoonotic Dis. 8,

57 (2008).

40. N. Komar, G. G. Clark, Pan Am. J. Public Health 19,

112 (2006).

41. C. A. Russell, D. L. Smith, J. E. Childs, L. A. Real,

PLoS Biol. 3, e88 (2005).

Acknowledgments: I thank T. Bogich, K. Koelle, I. Parker,

N. Komar, and the Kilpatrick lab for input. NSF grants

EF-0914866 and BCS-0826779 and NIH grant

1R01AI090159-01 provided funding.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/334/6054/323/DC1

SOM Text

Fig. S1

References

10.1126/science.1201010

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 327

REVIEW

The Influence of Nonlinear

Mesoscale Eddies on Near-Surface

Oceanic ChlorophyllDudley B. Chelton,1* Peter Gaube,1 Michael G. Schlax,1

Jeffrey J. Early,2 Roger M. Samelson1

Oceanic Rossby waves have been widely invoked as a mechanism for large-scale variability ofchlorophyll (CHL) observed from satellites. High-resolution satellite altimeter measurements haverecently revealed that sea-surface height (SSH) features previously interpreted as linear Rossbywaves are nonlinear mesoscale coherent structures (referred to here as eddies). We analyze 10 yearsof measurements of these SSH fields and concurrent satellite measurements of upper-ocean CHLto show that these eddies exert a strong influence on the CHL field, thus requiring reassessmentof the mechanism for the observed covariability of SSH and CHL. On time scales longer than 2to 3 weeks, the dominant mechanism is shown to be eddy-induced horizontal advection of CHL bythe rotational velocities of the eddies.

Adecade of concurrent satellite measure-ments of sea surface height (SSH) and

upper-ocean chlorophyll (CHL) is en-abling studies of physical-biological interaction

that are not feasible from ship-based observations.Although satellites provide only near-surface in-

formation about ocean physics and biology, theyare the only practical means of obtaining dense,

global observations. Altimetric measurements of

SSH reveal that westward propagation is ubiqui-tous (1) with characteristics similar to the linear

Rossby waves by which the ocean adjusts to windand thermal forcing (2). Westward propagation

is also evident in CHL estimates derived from sat-ellite measurements of ocean color. The wide-

spread interpretation of the westward-propagatingSSH variations as Rossby waves led naturally to

interpretations that the CHL variations are alsoinduced by Rossby waves (3–5).

The mechanism for Rossby wave influenceon CHL has been debated (4–9), in part because

of inconsistency in the lag between variations ofSSH and CHL. The most widely accepted view

is that the covariability between SSH and CHLarises from cyclical advection of CHL by the

horizontal velocity field associated with passingRossby waves (7–9).

The prevailing view before the recent focuson Rossby wave influence was that CHL concen-

tration is influenced by nonlinear eddies (10–15).Investigations of this eddy influence have con-

tinued in parallel with Rossby wave studies.Here, we show that the copropagation of CHL

and SSH previously interpreted as having been

caused by Rossby waves is in fact attributableto eddies.

Nonlinearity of SSH variability. High-resolution SSH fields produced by merging the

measurements from two simultaneously operat-ing satellite altimeters (16) reveal that westward-

propagating features previously believed to belinear Rossby waves are actually nonlinear rotat-

ing coherent structures (“eddies”) with radii of

~100 km (17, 18). Because such mesoscale fea-tures propagate westward with approximately

the speed of long Rossby waves (17–19), they canmasquerade as Rossby waves in low-resolution

SSH fields constructed from measurements by asingle altimeter.

The degree of nonlinearity of a mesoscale fea-ture is characterized by the ratio of the rotational

fluid speed U to the translation speed c of thefeature. When U/c > 1, the feature is nonlinear,

which allows it to maintain a coherent structureas it propagates (20). This requires that all of

the wavelength components of the feature prop-agate at the same speed, i.e., nondispersively. With

linear Rossby wave dynamics, features that areinitially spatially compact quickly lose their co-

herent structure through dispersion (21).At latitudes higher than 25°, 98% of the fea-

tures tracked for ≥10 weeks have U/c > 1 (fig.S2). The degree of nonlinearity is slightly less

at lower latitudes where the propagation speedsc are faster (17, 18). But even in the latitude range

15° to 25°, 95% of the tracked features haveU/c > 1 (fig. S2).

Westward copropagation of SSH and CHL.

The revised interpretation of westward-propagating

SSH as nonlinear eddies mandates a reassessmentof past conclusions that the westward copropa-

gation of CHL and SSH is indicative of Rossbywave influence on CHL. The alternative hy-

pothesis that CHL variability is eddy-induced is

examined here from 10 years of concurrent mea-

surements of SSH and CHL in the southeasternPacific (SEP) near 20°S that has been a focus of

past studies (6, 7).The trajectories of mesoscale eddies (18) in

the SEP are shown in Fig. 1A. Compared witheddies observed globally in the latitude range

15° to 25°, their mean amplitude is smaller (3.2 cmversus 6.2 cm) but their mean radius is the same

(110 km). Because U is approximately propor-tional to eddy amplitude, eddies in the SEP are

less nonlinear (fig. S2); 87% have U/c > 1.

The mean CHL distribution has a generallynorthward gradient over most of the SEP (Fig.

1B). The influence of eddies is evident from thesinuous character of the CHL field at any par-

ticular time (Fig. 1, C and D). The distortionsof an otherwise smoothly varying CHL field

are most apparent in regions of strong CHLgradient.

Eddy influence on the CHL field becomesclearer after filtering to remove the large-scale

and seasonally varying CHL and SSH (22). West-ward copropagation of CHL and SSH is apparent

from time-longitude plots of the resulting anom-aly fields (Fig. 2, A and B). The trajectories of

the centroids of clockwise (CW) and counter-clockwise (CCW) rotating eddies in the SEP co-

incide, respectively, with negative and positiveextrema of westward-propagating SSH (Fig. 2A).

The positive lag of maximum positive correla-tion in Fig. 2C indicates that the SSH extrema at

the eddy centroids lag the extrema of westward-propagating CHL by ~1 month in the eastern

SEP, decreasing to ~0.5 month in the west. Thereis a weaker negative correlation at negative lags

of 1 to 1.5 months.Eddy influence on CHL. To interpret the lagged

correlations in Fig. 2C, anomaly CHL was com-posite averaged within eddy interiors in a trans-

lating and rotated coordinate system in which thelarge-scale CHL gradient vector is oriented at a

polar angle of 90° (22). The CHL anomaly com-posites consist of dipoles with opposing signs

and with different orientations in CW and CCWrotating eddies (Fig. 3A). As indicated by the

ratio r in Fig. 3A, the dipoles are asymmetric in

both cases with larger magnitudes in the left halfof each composite, corresponding to the leading

half of these westward-propagating eddies. Thedisplacements from the eddy centroid are smaller

for these primary poles than for the secondarypoles of opposite sign in the trailing (right) halves

of the eddies (see also fig. S4).The negative extremum of SSH at the cen-

troids of CW rotating eddies in the SEP is strad-dled by negative and positive poles of CHL to the

west and east, respectively. The opposite occursin CCW rotating eddies, for which the positive

extremum of SSH at the centroids is straddledby positive and negative poles of CHL to the

west and east, respectively. The parallel bandsof positive and negative lagged correlations in

Fig. 2C thus arise from a combination of west-

RESEARCHARTICLES

1College of Oceanic and Atmospheric Sciences, 104 COAS Ad-ministration Building, Oregon State University, Corvallis, OR97331–5503, USA. 2Northwest Research Associates, Post Of-fice Box 3027, Bellevue, WA 98009, USA.

*To whom correspondence should be addressed. E-mail:chelton@coas.oregonstate.edu

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org328

ward propagation and the zonal alignments ofthe monopole extrema of SSH and dipole ex-

trema of CHL.The geographical patterns of anomalous CHL

within the eddy interiors in this region of gen-erally northward CHL gradient are indicative of

horizontal advection of CHL by the rotationalvelocities of the eddies. For CW rotating eddies

(Figs. 3A and 4A, top panels), the northwardvelocity in the western half of each eddy advects

low CHL from south to north, resulting in anom-alously low CHL in the northwest quadrant. The

southward velocity in the eastern half of eachsuch eddy advects high CHL from north to south,

resulting in anomalously high CHL in the south-

east quadrant. The opposite rotational sense ofCCW rotating eddies (Figs. 3A and 4A, bottom

panels) results in anomalously high and lowCHL in the southwest and northeast quadrants,

respectively.The importance of composite averaging

the CHL in a rotated coordinate system is clearfrom Fig. 4, A and B. The dipole patterns of

anomaly CHL result from a combination ofthe rotational sense of the eddies and the di-

rection of the CHL gradient. These dipoles aremanifest as distortions of the total CHL field

(Fig. 1, C and D).The CHL anomaly within the trailing half

of the eddy is generally weaker and noisierthan within the leading half because the trailing

half encounters a CHL field that has beendistorted by the leading half. The noisiness

of the secondary poles accounts for the some-what weaker negative correlations in Fig. 2C,

as well as their smaller composite average mag-nitudes compared with the primary poles (Fig. 3A).

Within an individual eddy, the structure ofthe dipole of anomaly CHL from eddy-driven

advection varies, depending on the strengthand orientation of the geographically and tem-

porally varying gradient of CHL, the degreeof eddy nonlinearity U/c, and the influence of

other eddies that have recently perturbed theCHL field. Past confusion about geographi-

cal and temporal variations of the lag relation-

ships between SSH and CHL is therefore notsurprising.

Rotational advection of CHL by eddies isreproduced in numerical model simulations of

random westward-propagating eddies in a tracerfield with a northward gradient (22). The dipoles

of anomalous tracer concentration for weaklynonlinear eddies (Fig. 3B) are very similar to the

dipoles of observed CHL in Fig. 3A. The modelreproduces the asymmetry of the magnitudes of

the dipoles, as well as the smaller offset betweenthe eddy centroid and the primary pole in the

leading halves of the eddies compared with thesecondary pole in the trailing halves. The higher

rotational velocities within strongly nonlinear ed-dies result in tracer dipoles with larger magni-

tudes and with centers advected farther aroundthe eddy interiors (Fig. 3C).

Composite averaging separately for the SEPeddies east and west of 108°W (figs. S3 and S5)

reveals that the longitudinal variations of thecouplet of positive and negative lagged correla-

tions in Fig. 2C are attributable to longitudinalvariations of the structures of the CHL dipoles.

The shorter lag of maximum positive correlationin the west is due to smaller displacements

between the eddy centroids and the primarypoles of CHL anomaly in the leading halves

of the eddies. The near-symmetry of the lagsof positive and negative correlation bands in the

east is consistent with the near-symmetric dis-

placements of the dipole centers from the eddycentroids.

The geographical patterns of eddy-inducedCHL anomalies in the SEP are similar to the

patterns found in other regions of northwardgradient of CHL. Composite averages of anomaly

CHL computed globally between latitudes of15° and 45° are shown in Fig. 3D for the

tracked eddies within regions of northward CHLgradient (22). The telltale asymmetric dipole

patterns from opposing meridional advectionin opposite halves of the eddies are readily

apparent.Anomaly CHL was composite averaged for

regions of generally southward CHL gradientby rotating the translating coordinate system

Fig. 1. Geographical characteristics of observedSSH and CHL in the SEP. (A) The trajectories froma 16-year data record of eddies that rotate clock-wise (CW, blue lines) and counterclockwise (CCW,red lines), with the starting locations shown bysolid circles. (B) The 10-year average log10(CHL)for CHL in units of mg m–3, with a contour in-terval of log10(CHL) = 0.1, increasing northward,and with the thick line corresponding to log10(CHL) =–1.3. (C) An example map for 7 March 2001showing log10(CHL) in color with contours ofpositive and negative anomaly SSH (solid anddashed lines, respectively) at intervals of 2 cm,excluding the zero contour. (D) The same as (C),except showing anomaly SSH in color with con-tours of log10(CHL) at the same interval as in (B).The horizontal lines in each panel are the sectionalong which the time-longitude plots in Fig. 2 andthe spectra in Fig. 5, A to C, were computed.

A

B

C

D

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 329

RESEARCH ARTICLES

so that the large-scale CHL gradient vector isoriented at a polar angle of –90°. The axes of

the CHL dipoles in these composite averages(Fig. 3E) are rotated ~180° relative to those

in regions of northward CHL gradient. Thesemirrored dipoles still have larger magnitude in

the leading (left) half of the westward-propagatingeddies.

When U/c > 1, fluid is trapped within theeddy interior and transported along the eddy tra-

jectory (20, 21). This nonlinear signal of sustainedtransport of fluid trapped in the eddy core is not

A B C D E

Fig. 3. Composite averages of filtered fields in a rotated and normalizedcoordinate system (22) within the interiors of CW (top panels) and CCWrotating eddies (bottom panels). (A) log10(CHL) in the region 18°S to 22°S,130°W to 80°W. (B) A tracer field in a model simulation seeded with weaklynonlinear Gaussian eddies. (C) A tracer field in the model seeded with stronglynonlinear Gaussian eddies; (D) log10(CHL) globally between 15° and 45°latitude in regions of northward CHL gradient. (E) log10(CHL) globally be-tween 15° and 45° latitude in regions of southward CHL gradient. The outer

perimeter of each circle corresponds to twice the eddy radius scale Ls (22).The vectors in each panel are the gradient of the composite average SSH,which is proportional to the geostrophic velocity. The number N of eddyrealizations in the composite average and the magnitude r of the ratio of theprimary pole in the leading (left) half of each composite to the secondarypole in the trailing (right) half are labeled on each panel. The estimated95% confidence intervals along profiles connecting the dipole extrema ineach of these composite averages are shown in fig. S4.

Fig. 2. Spatial and tem-poral variability of filteredSSH and log10(CHL) obser-vations (22) along 20°S be-tween 130°W and 80°W.Time-longitude sections ofwestward-only propagationover a 3-year portion of the10-year period analyzedhere are shown for (A) SSHwith eddy tracks within T2°of 20°S overlaid (dashed andsolid lines for CW and CCWrotating eddies, respectively);(B) log10(CHL) with the sameeddy tracks overlaid; and (C)the lagged cross-correlationbetween log10(CHL) at timet and SSH at time t + lag,calculated over the full 10-year data record; the white areas correspond tocorrelations smaller than the estimated 95% significance level of 0.083 (22).Positive lags correspond to log10(CHL) leading SSH, and the contour interval

is 0.2 with the zero contour omitted for clarity. Analogous time-longitudesections and lagged cross correlations are shown for the SSH and tracerfields from a quasi-geostrophic model in fig. S6.

A B C

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org330

RESEARCH ARTICLES

the signature revealed in the CHL observationsdescribed here. The extrema of the dipole CHL

features occur near the radius of maximum rota-tional velocity that defines the outer edge of the

eddy core (21). As thewestward-translating eddiesimpinge on preexisting CHL gradients, ambient

fluid is advected part way around the portion ofthe eddy outside of the core region of closed

streamlines in the translating reference frame(20, 21).

Spectral analysis of SSH and CHL. Wavenumber–frequency spectra of SSH and CHL

provide further evidence that the westward propa-gation of CHL is induced by nonlinear eddies.

The nondispersive character of nonlinear eddiesimplies that the spectral variance should fall ap-

proximately along a straight line in wave number–frequency space. In contrast, the spectral variance

for linear Rossby waves would be constrained tofrequencies defined by their dispersion relation.

The wave number–frequency spectrum ofSSH in the SEP (Fig. 5A) is clearly inconsistent

with the theoretical Rossby wave dispersionrelation, regardless of whether the classical the-

ory is modified to account for effects of meancurrents or rough bottom topography. Where-

as the dispersion relations from these theoriesall flatten with increasingly negative wave num-

ber, the spectral variance of observed SSHextends to higher frequencies, straddling the

Fig. 5. Zonal (east-west) wave number–frequencyspectra. (A) Filtered SSH for 130°W to 80°W along20°S (see fig. S1 for SSH spectra along four otherzonal sections). (B) SSH for Gaussian approxima-tions of each tracked eddy for 130°W to 80°W along20°S [different color bars are used to accommodatethe smaller variance of the Gaussian approximations(23)]. (C) Filtered log10(CHL) for 130°W to 80°Walong 20°S. (D) SSH from days 3000 to 15,000 of amodel simulation with purely linear dynamics. (E)SSH from days 3000 to 15,000 of a model sim-ulation with nonlinear quasi-geostrophic dynam-ics (see also fig. S7, A to C). (F) A tracer field inthe same model simulation as in (E) (see also fig.S7, D to F). The negative wave numbers correspondto westward propagation. The units are cm for SSH,log10 of mg m–3 for CHL, and arbitrary for thetracer field. The straight lines are the mean eddyspeeds from observed SSH [–4.9 cm s–1, (A) to (C)]and from SSH in the model [–4.3 cm s–1, (D) to(F)]. The curved lines correspond to the dispersionrelations for linear Rossby waves from the classicaltheory for a flat bottom, no mean currents, andzero meridional (north-south) wave number (2)(solid), a theory that accounts for mean currents(25) (dashed), and a theory for rough bottomtopography and no mean currents (26) (dotted).The latter two are irrelevant to the SSH fields fromthe flat-bottom model with no mean currents in (D)to (F).

A B C

D E F

Fig. 4. Schematic diagram of eddy-driven horizontal advection ofCHL for CW and CCW rotating eddies (top and bottom, respectively)propagating westward in regions where the CHL gradient is (A)northward and (B) northeastward. An otherwise smooth contour ofCHL (dashed lines) is distorted by the rotational velocity field withinthe eddy, as shown by the solid lines. Advection of CHL within thelarge-scale background CHL gradient results in the positive andnegative CHL anomalies shown by the red and purple regions, re-spectively. The dependence of the locations of these CHL anomalieson the direction of the large-scale background CHL gradient that isevident from comparison of (A) and (B) was accounted for by com-posite averaging in a coordinate system rotated for each eddy (22).

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 331

RESEARCH ARTICLES

straight line of constant (nondispersive) prop-agation speed.

We tested our hypothesis that the spectralcharacteristics of SSH in Fig. 5A are attributable

to eddies with compact structures by construct-ing SSH fields consisting only of the interiors

of the tracked eddies and zero outside of the ed-dies. Each eddy was approximated as an axi-

symmetric Gaussian feature with amplitude andscale estimated from the automated tracking

procedure (18). The spectral variance of the re-sulting eddy-only SSH fields (Fig. 5B) strad-

dles the same straight line as the spectrum oftotal SSH (23, 24). The discrepancies between

the wave number–frequency spectra of SSH and

the Rossby wave dispersion relations are thusconsistent with SSH variability being attributa-

ble to a field of propagating eddies.The importance of nonlinearity is clarified

from model simulations. The spectral variance ofmodel SSH with linear dynamics is restricted to

frequencies below the dispersion relation forRossby waves with zero meridional wave num-

ber (Fig. 5D). The spread of variance to frequen-cies below this dispersion relation is from Fourier

components with finite meridional wave num-bers (2).

With nonlinear quasi-geostrophic dynamics,the spectral variance of model SSH primarily rep-

resents long-lived, coherent eddy features, in whichthe small-scale spectral components are phase-

locked to the large-scale components (21). Theresulting spectrum (Fig. 5E) consists of a non-

dispersive band along a straight line that extendsto frequencies higher than are allowed by linear

Rossby wave dynamics, very similar to the spec-trum of observed SSH in Fig. 5A.

A short “spur” of spectral variance centeredat about 0.007 cycles per day (cpd) extends along

the dispersion relation in the model SSH spec-trum (Fig. 5E). This spur arises from small-

amplitude SSH variability outside the nonlinearcores of the eddies. The extent of the spur is

found to decrease with increasing nonlinearityof the eddies (fig. S7). Although not apparent in

the spectrum of observed SSH in Fig. 5A, sug-gestions of similar spurs exist in SSH spectra in

other regions (fig. S1).

Determination of the spectrum of CHL var-iability is more challenging than for SSH (22).

The spectrum of CHL in the SEP (Fig. 5C) isnonetheless similar to the spectrum of SSH. In

particular, the spectral variance is concentratedalong the same straight line of nondispersive

variability to higher frequencies than are allowedby linear Rossby waves theories. Moreover, the

spectral variance is restricted to smaller negativewave numbers than would be the case if the

CHL variability were induced by linear Rossbywaves. A short spur of spectral variance straddles

the dispersion relation at about 0.008 cpd forthe same reason discussed above for the model

SSH spectra. Similar spurs are found in spectraof tracer fields in quasi-geostrophic model sim-

ulations (Fig. 5F and fig. S7, D to F). The re-

stricted wave number extents of these spurs arean important distinction from the spectral char-

acteristics in Fig. 5D that would be found if theCHL or tracer variability were attributable to

linear Rossby waves.Conclusions.Westward copropagation of CHL

and SSH that has previously been attributed tolinear Rossby waves is actually caused by non-

linear eddies that were not resolvable in the SSHfields analyzed in past studies. This eddy influ-

ence is manifest as distortions of the CHL fieldand is most evident in regions of strong gradi-

ents of the CHL field. Eddy influence on CHLbecomes clearer after filtering. The distinctly

different dipole structures of the resulting anom-

alous CHL distribution within the interiors ofCW and CCW rotating eddies (Fig. 3, A, D,

and E) and their similarity to the dipoles of atracer field in model simulations (Fig. 3, B and C)

are evidence that the dominant mechanism foreddy-induced CHL variability on the time scales

>2 to 3 weeks considered here (22) is horizontaladvection of CHL by the rotational velocity with-

in the interior of each eddy.The dominance of horizontal advection as

the mechanism for the observed westward prop-agation of CHL variability has been suggested

previously (7–9) but has been attributed to hori-zontal advection by linear Rossby waves, rather

than to nonlinear eddies. [The role of eddies indefining the dipole structure of CHL anomalies

has been suggested for the central North At-lantic (14, 15).] The distinction between linear

Rossby waves and eddies is important becausethe dynamics of nonlinear eddies differ fun-

damentally from those of linear Rossby waves.While distinct from the rotational advection

identified here, eddies can trap and transportfluid parcels and their associated water proper-

ties (20, 21), including nutrients, CHL, and zoo-plankton. They can also upwell nutrient-rich water

by various mechanisms (10–15), thus stimulat-ing the growth of phytoplankton and increasing

CHL in the eddy core. Such eddy-induced en-hancements often occur deep in the euphotic zone

where they can be undetectable in the satellite-based measurements of near-surface ocean color

(12–15).

Because of the unique trapping of fluid inthe cores of nonlinear eddies, it is perhaps sur-

prising that the CHL distribution associated withthe eddies consists of dipoles with extrema out-

side of the eddy cores, rather than monopoles ofpositive or negative CHL anomalies trapped at

the eddy centers. Monopole structures with veryactive physical-biological interaction are some-

times observed within eddy cores. In contrast tothe ubiquitous presence of rotational advec-

tion around the outer portion of the eddies iden-tified here, however, such monopole structures

are usually episodic, often with time scales short-er than the 2- to 3-week filtering applied here.

Although it is unclear whether the rotationaladvection that dominates the variability on longer

time scales has important biological consequences,

the results presented here clarify the dynamicalimportance of eddies to the observed CHL

distribution.

References and Notes1. D. B. Chelton, M. G. Schlax, Science 272, 234 (1996).

2. A. E. Gill, Atmosphere-Ocean Dynamics (Academic Press,

Cambridge, 1982).

3. P. Cipollini, D. Cromwell, P. G. Challenor, S. Raffaglio,

Geophys. Res. Lett. 28, 323 (2001).

4. B. M. Uz, J. A. Yoder, V. Osychny, Nature 409, 597 (2001).

5. D. A. Siegel, Nature 409, 576 (2001).

6. Y. Dandonneau, A. Vega, H. Loisel, Y. du Penhoat,

C. Menkes, Science 302, 1548 (2003).

7. P. D. Killworth, P. Cipollini, B. M. Uz, J. R. Blundell,

J. Geophys. Res. 109, C07002 (2004).

8. G. Charria et al., J. Mar. Res. 64, 43 (2006).

9. E. Gutknecht, I. Dadou, G. Charria,

P. Cipollini, V. Garçon, J. Geophys. Res. 115,

C05004 (2010).

10. P. G. Falkowski, D. Ziemann, Z. Kolber, P. K. Bienfang,

Nature 352, 55 (1991).

11. D. J. McGillicuddy Jr. et al., J. Geophys. Res. 104, 13381

(1999).

12. D. A. Siegel, D. J. McGillicuddy Jr., E. A. Fields,

J. Geophys. Res. 104, 13359 (1999).

13. D. J. McGillicuddy Jr. et al., Science 316, 1021 (2007).

14. D. A. Siegel et al., Deep Sea Res. Part II Top. Stud.

Oceanogr. 55, 1218 (2008).

15. D. A. Siegel, P. Peterson, D. J. McGillicuddy Jr.,

S. Maritorena, N. B. Nelson, Geophys. Res. Lett. 38,

L13608 (2011).

16. N. Ducet, P.-Y. Le Traon, G. Reverdin, J. Geophys. Res.

105, 19477 (2000).

17. D. B. Chelton, M. G. Schlax, R. M. Samelson,

R. A. de Szoeke, Geophys. Res. Lett. 34, L15606 (2007).

18. D. B. Chelton, M. G. Schlax, R. M. Samelson, Prog.

Oceanogr. 91, 167 (2011).

19. J. C. McWilliams, G. R. Flierl, J. Phys. Oceanogr. 9, 1155

(1979).

20. G. R. Flierl, Geophys. Astrophys. Fluid Dyn. 18, 39 (1981).

21. J. J. Early, R. M. Samelson, D. B. Chelton, J. Phys. Oceanogr.

41, 1535 (2011).

22. Information on methods is available as supporting online

material on Science Online.

23. The somewhat smaller variance is due to unavoidable

biases in the amplitudes of compact mesoscale

features as estimated by the automated procedure (18).

24. The large spectral variance at small wave numbers

(long wavelengths) in the spectrum of eddy-only SSH

fields is easily understood qualitatively from consideration

of a single Gaussian eddy. Because the Fourier

transform of a Gaussian in space is a Gaussian in wave

number, the associated spectrum is dominated by

variance at small wave numbers. It is evident from

Fig. 5B that this dominance of spectral variance at

small wave numbers for a single Gaussian feature is

maintained in the spectrum of a superposition of

many Gaussian eddies.

25. P. D. Killworth, D. B. Chelton, R. A. de Szoeke, J. Phys.

Oceanogr. 27, 1946 (1997).

26. R. Tailleux, J. C. McWilliams, J. Phys. Oceanogr. 31, 1461

(2001).

Acknowledgments: We thank D. McGillicuddy and T. Farrar

for discussions throughout the course of this study.

We also thank L.-L. Fu, V. Combes, D. Siegel, P. Cipollini,

E. Shulenberger, and two anonymous reviewers for

helpful comments on the manuscript. This research was

supported by NASA grants NNX08AI80G and

NNX08AR37G and NSF Award 0621134.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/science.1208897/DC1

Materials and Methods

Figs. S1 to S7

References

25 May 2011; accepted 26 August 2011

Published online 15 September 2011;

10.1126/science.1208897

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org332

RESEARCH ARTICLES

Light Propagation with Phase

Discontinuities: Generalized Laws of

Reflection and RefractionNanfang Yu,1 Patrice Genevet,1,2 Mikhail A. Kats,1 Francesco Aieta,1,3 Jean-Philippe Tetienne,1,4

Federico Capasso,1* Zeno Gaburro1,5*

Conventional optical components rely on gradual phase shifts accumulated during lightpropagation to shape light beams. New degrees of freedom are attained by introducing abruptphase changes over the scale of the wavelength. A two-dimensional array of optical resonatorswith spatially varying phase response and subwavelength separation can imprint such phasediscontinuities on propagating light as it traverses the interface between two media. Anomalousreflection and refraction phenomena are observed in this regime in optically thin arrays of metallicantennas on silicon with a linear phase variation along the interface, which are in excellentagreement with generalized laws derived from Fermat’s principle. Phase discontinuities providegreat flexibility in the design of light beams, as illustrated by the generation of optical vorticesthrough use of planar designer metallic interfaces.

The shaping of the wavefront of light withoptical components such as lenses and

prisms, as well as diffractive elements suchas gratings and holograms, relies on gradual phase

changes accumulated along the optical path. Thisapproach is generalized in transformation optics

(1, 2), which uses metamaterials to bend lightin unusual ways, achieving such phenomena as

negative refraction, subwavelength-focusing, andcloaking (3, 4) and even to explore unusual ge-

ometries of space-time in the early universe (5).A new degree of freedom of controlling wave-

fronts can be attained by introducing abrupt phaseshifts over the scale of the wavelength along the

optical path, with the propagation of light gov-erned by Fermat’s principle. The latter states that

the trajectory taken between two points A and Bby a ray of light is that of the least optical path,

∫B

An(→r )dr, where n(→r ) is the local index of re-

fraction, and readily gives the laws of reflection

and refraction between two media. In its mostgeneral form, Fermat’s principle can be stated as

the principle of stationary phase (6–8); that is,the derivative of the phase ∫

B

Adϕ(→r ) accumu-

lated along the actual light path will be zero withrespect to infinitesimal variations of the path. We

show that an abrupt phase shift F(→rs) over thescale of the wavelength can be introduced in the

optical path by suitably engineering the interface

between two media; F(→rs) depends on the co-ordinate →

rs along the interface. Then, the total

phase shift F(→rs) þ ∫B

A

→

k ⋅ d→

r will be stationaryfor the actual path that light takes;

→

k is the wave

vector of the propagating light. This provides ageneralization of the laws of reflection and re-

fraction, which is applicable to a wide range ofsubwavelength structured interfaces between two

media throughout the optical spectrum.Generalized laws of reflection and refraction.

The introduction of an abrupt phase shift, de-noted as phase discontinuity, at the interface be-

tween two media allows us to revisit the laws ofreflection and refraction by applying Fermat’s

principle. Consider an incident plane wave at anangle qi. Assuming that the two paths are infi-

nitesimally close to the actual light path (Fig. 1),then the phase difference between them is zero

½koni sin(qi)dx þ (F þ dF)� −

½kont sin(qt)dx þ F� ¼ 0 ð1Þ

where qt is the angle of refraction; F and F+dF

are, respectively, the phase discontinuities at the

locations where the two paths cross the interface;dx is the distance between the crossing points; niand nt are the refractive indices of the two media;and ko = 2p/lo, where lo is the vacuum wave-

length. If the phase gradient along the interface isdesigned to be constant, the previous equation

leads to the generalized Snell’s law of refraction

sin(qt)nt − sin(qi)ni ¼lo

2p

dF

dxð2Þ

Equation 2 implies that the refracted beam can

have an arbitrary direction, provided that a suit-able constant gradient of phase discontinuity along

the interface (dF/dx) is introduced. Because ofthe nonzero phase gradient in this modified Snell’s

law, the two angles of incidence Tqi lead to dif-ferent values for the angle of refraction. As a

consequence, there are two possible critical an-

gles for total internal reflection, provided that

nt < ni:

qc ¼ arcsin þ−

nt

ni−

lo

2pni

dF

dx

� �ð3Þ

Similarly, for reflection we have

sin(qr) − sin(qi) ¼lo

2pni

dF

dxð4Þ

where qr is the angle of reflection. There is anonlinear relation between qr and qi, which is

markedly different from conventional specular re-flection. Equation 4 predicts that there is always a

critical angle of incidence

q′

c ¼ arcsin 1 −lo

2pni

���� dFdx

����� �

ð5Þ

above which the reflected beam becomes

evanescent.In the above derivation, we have assumed that

F is a continuous function of the position alongthe interface; thus, all the incident energy is trans-

ferred into the anomalous reflection and refraction.However, because experimentally we use an array

of optically thin resonators with subwavelengthseparation to achieve the phase change along

the interface, this discreteness implies that thereare also regularly reflected and refracted beams,

which follow conventional laws of reflectionand refraction (dF/dx = 0 in Eqs. 2 and 4). The

separation between the resonators controls

the amount of energy in the anomalously re-flected and refracted beams. We have also

assumed that the amplitudes of the scatteredradiation by each resonator are identical, so that

the reflected and refracted beams are plane waves.In the next section, wewill showwith simulations—

which represent numerical solutions of Maxwell’s

1School of Engineering and Applied Sciences, Harvard Uni-versity, Cambridge, MA 02138, USA. 2Institute for QuantumStudies and Department of Physics, Texas A&M University,College Station, TX 77843, USA. 3Dipartimento di Fisica eIngegneria dei Materiali e del Territorio, UniversitàPolitecnicadelle Marche, via Brecce Bianche, 60131 Ancona, Italy. 4Lab-oratoire de Photonique Quantique et Moléculaire, Ecole Nor-male Supérieure de Cachan and CNRS, 94235 Cachan, France.5Dipartimento di Fisica, Università degli Studi di Trento, viaSommarive 14, 38100 Trento, Italy.

*To whom correspondence should be addressed. E-mail:capasso@seas.harvard.edu (F.C.); gaburro@seas.harvard.edu (Z.G.)

Fig. 1. Schematics used to derive the generalizedSnell’s law of refraction. The interface between thetwo media is artificially structured in order to in-troduce an abrupt phase shift in the light path,which is a function of the position along the in-terface. F and F + dF are the phase shifts wherethe two paths (blue and red) cross the boundary.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 333

RESEARCH ARTICLES

equations—how, indeed, one can achieve theequal-amplitude condition and the constant phase

gradient along the interface through suitable de-sign of the resonators.

There is a fundamental difference between theanomalous refraction phenomena caused by phase

discontinuities and those found in bulk designermetamaterials, which are caused by either negative

dielectric permittivity and negative magneticpermeability or anisotropic dielectric permittivity

with different signs of permittivity tensor com-ponents along and transverse to the surface (3, 4).

Phase response of optical antennas.The phaseshift between the emitted and the incident radia-

tion of an optical resonator changes appreciably

across a resonance. By spatially tailoring the geom-etry of the resonators in an array and hence their

frequency response, one can design the phasediscontinuity along the interface and mold the

wavefront of the reflected and refracted beams innearly arbitrary ways. The choice of the reso-

nators is potentially wide-ranging, from electro-magnetic cavities (9, 10) to nanoparticle clusters

(11, 12) and plasmonic antennas (13, 14). Weconcentrated on the latter because of their widely

tailorable optical properties (15–19) and the easeof fabricating planar antennas of nanoscale thick-

ness. The resonant nature of a rod antenna madeof a perfect electric conductor is shown in Fig.

2A (20).

Phase shifts covering the 0-to-2p range areneeded to provide full control of the wavefront.

To achieve the required phase coverage whilemaintaining large scattering amplitudes, we used

the double-resonance properties of V-shaped an-tennas, which consist of two arms of equal length

h connected at one end at an angle D (Fig. 2B).We define two unit vectors to describe the ori-

entation of aV-antenna: ŝ along the symmetry axisof the antenna and â perpendicular to ŝ (Fig. 2B).

V-antennas support “symmetric” and “antisym-metric”modes (Fig. 2B, middle and right), which

are excited by electric-field components along ŝ

and â axes, respectively. In the symmetric mode,

the current distribution in each arm approximates

Fig. 2. (A) Calculated phase and amplitude ofscattered light from a straight rod antenna made ofa perfect electric conductor (20). The vertical dashedline indicates the first-order dipolar resonance ofthe antenna. (B) A V-antenna supports symmetricand antisymmetric modes, which are excited, re-spectively, by components of the incident field alongŝ and â axes. The angle between the incident po-larization and the antenna symmetry axis is 45°.The schematic current distribution is representedby colors on the antenna (blue for symmetric andred for antisymmetric mode), with brighter colorrepresenting larger currents. The direction of cur-rent flow is indicated by arrows with color gradient.(C) V-antennas corresponding to mirror images ofthose in (B). The components of the scattered elec-tric field perpendicular to the incident field in (B)and (C) have a p phase difference. (D and E) An-alytically calculated amplitude and phase shift ofthe cross-polarized scattered light for V-antennasconsisting of gold rods with a circular cross sectionand with various length h and angle between therods D at lo = 8 mm (20). The four circles in (D) and(E) indicate the values of h and D used in exper-iments. The rod geometry enables analytical cal-culations of the phase and amplitude of the scatteredlight, without requiring the extensive numericalsimulations needed to compute the same quan-tities for “flat” antennas with a rectangular cross-section, as used in the experiments. The opticalproperties of a rod and “flat” antenna of the samelength are quantitatively very similar, when therod antenna diameter and the “flat” antennawidth and thickness are much smaller than thelength (20). (F) Schematic unit cell of the plasmonicinterface for demonstrating the generalized laws ofreflection and refraction. The sample shown in Fig. 3Ais created by periodically translating in the x-y planethe unit cell. The antennas are designed to haveequal scattering amplitudes and constant phasedifferenceDF = p/4 between neighbors. (G) Finite-difference time-domain (FDTD) simulations of thescattered electric field for the individual antennascomposing the array in (F). Plots show the scat-tered electric field polarized in the x direction fory-polarized plane wave excitation at normal in-cidence from the silicon substrate. The siliconsubstrate is located at z ≤ 0. The antennas are equally spaced at a sub-wavelength separation G/8, where G is the unit cell length. The tilted redstraight line in (G) is the envelope of the projections of the spherical wavesscattered by the antennas onto the x-z plane. On account of Huygens’s

principle, the anomalously refracted beam resulting from the superposi-tion of these spherical waves is then a plane wave that satisfies thegeneralized Snell’s law (Eq. 2) with a phase gradient |dF/dx| = 2p/G alongthe interface.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org334

RESEARCH ARTICLES

that of an individual straight antenna of lengthh (Fig. 2B, middle), and therefore the first-order

antenna resonance occurs at h ≈ leff/2, whereleff is the effective wavelength (14). In the anti-

symmetric mode, the current distribution in eacharm approximates that of one half of a straight

antenna of length 2h (Fig. 2B, right), and thecondition for the first-order resonance of this

mode is 2h ≈ leff/2.The polarization of the scattered radiation

is the same as that of the incident light whenthe latter is polarized along ŝ or â. For an ar-

bitrary incident polarization, both antenna modesare excited but with substantially different am-

plitude and phase because of their distinctive reso-

nance conditions. As a result, the scattered lightcan have a polarization different from that of the

incident light. These modal properties of theV-antennas allow one to design the amplitude,

phase, and polarization state of the scattered light.We chose the incident polarization to be at 45°

with respect to ŝ and â so that both the symmetricand antisymmetric modes can be excited and

the scattered light has a substantial componentpolarized orthogonal to that of the incident light.

Experimentally, this allows us to use a polarizerto decouple the scattered light from the excitation.

As a result of the modal properties of theV-antennas and the degrees of freedom in choosing

antenna geometry (h and D), the cross-polarizedscattered light can have a large range of phases

and amplitudes for a given wavelength lo; ana-

lytical calculations of the amplitude and phaseresponse of V-antennas assumed to be made of

gold rods are shown in Fig. 2, D and E. In Fig.2D, the blue and red dashed curves correspond to

the resonance peaks of the symmetric and anti-symmetric modes, respectively. We chose four

antennas detuned from the resonance peaks, asindicated by circles in Fig. 2, D and E, which

provide an incremental phase of p/4 from left toright for the cross-polarized scattered light. By

simply taking the mirror structure (Fig. 2C) of anexisting V-antenna (Fig. 2B), one creates a new

antenna whose cross-polarized radiation has an ad-ditional p phase shift. This is evident by observing

that the currents leading to cross-polarized radia-

tion are p out of phase in Fig. 2, B and C. A set ofeight antennas were thus created from the initial

four antennas, as shown in Fig. 2F. Full-wave sim-ulations confirm that the amplitudes of the cross-

polarized radiation scatteredby the eight antennas arenearly equal,with phases inp/4 increments (Fig. 2G).

A large phase coverage (~300°) can also beachieved by using arrays of straight antennas (fig.

S3). However, to obtain the same range of phaseshift their scattering amplitudes will be substan-

tially smaller than those of V-antennas (fig. S3).As a consequence of its double resonances, the

V-antenna instead allows one to design an arraywith phase coverage of 2p and equal, yet high,

scattering amplitudes for all of the array elements,leading to anomalously reflected and refracted

beams of substantially higher intensities.

Experiments on anomalous reflection and

refraction. We demonstrated experimentally the

generalized laws of reflection and refractionusing plasmonic interfaces constructed by peri-

odically arranging the eight constituent antennasas explained in the caption of Fig. 2F. The spacing

between the antennas should be subwavelengthso as to provide efficient scattering and to prevent

the occurrence of grating diffraction. However, itshould not be too small; otherwise, the strong near-

field coupling between neighboring antennaswould perturb the designed scattering amplitudes

and phases. A representative sample with thedensest packing of antennas,G = 11 mm, is shown

in Fig. 3A, where G is the lateral period of the

antenna array. In the schematic of the experimen-tal setup (Fig. 3B), we assume that the cross-

polarized scattered light from the antennas on theleft side is phase-delayed as compared with the

ones on the right. By substituting into Eq. 2 –2p/Gfor dF/dx and the refractive indices of silicon

and air (nSi and 1) for ni and nt, we obtain theangle of refraction for the cross-polarized beam

qt,⊥ = arcsin[nSisin(qi) – lo/G]

Figure 3C summarizes the experimental results

of the ordinary and the anomalous refraction forsix samples with different G at normal incidence.

The incident polarization is along the y axis inFig. 3A. The sample with the smallest G corre-

sponds to the largest phase gradient and the most

Fig. 3. (A) Scanning electron microscope (SEM)image of a representative antenna array fabricatedon a silicon wafer. The unit cell of the plasmonicinterface (yellow) comprises eight gold V-antennasof width ~220 nm and thickness ~50 nm, and itrepeats with a periodicity of G = 11 mm in the xdirection and 1.5 mm in the y direction. (B) Schematicexperimental setup for y-polarized excitation (electricfield normal to the plane of incidence). (C and D)Measured far-field intensity profiles of the refractedbeams for y- and x-polarized excitations, respective-ly. The refraction angle is counted from the normalto the surface. The red and black curves are mea-sured with and without a polarizer, respectively, forsix samples with different G. The polarizer is used toselect the anomalously refracted beams that arecross-polarized with respect to the excitation. Theamplitude of the red curves is magnified by a factorof two for clarity. The gray arrows indicate thecalculated angles of anomalous refraction accordingto Eq. 6.

ð6Þ

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 335

RESEARCH ARTICLES

efficient light scattering into the cross-polarizedbeams. We observed that the angles of anoma-

lous refraction agree well with theoretical pre-dictions of Eq. 6 (Fig. 3C). The same peak

positions were observed for normal incidence,with polarization along the x axis in Fig. 3A (Fig.

3D). To a good approximation, we expect that theV-antennas were operating independently at the

packing density used in experiments (20). Thepurpose of using a large antenna array (~230 mm

by 230 mm) is solely to accommodate the sizeof the plane-wave–like excitation (beam radius

~ 100 mm). The periodic antenna arrangementis used here for convenience but is not necessary

to satisfy the generalized laws of reflection and

refraction. It is only necessary that the phasegradient is constant along the plasmonic interface

and that the scattering amplitudes of the antennasare all equal. The phase increments between

nearest neighbors do not need to be constant, ifone relaxes the unnecessary constraint of equal

spacing between nearest antennas.The angles of refraction and reflection are

shown in Fig. 4, A and B, respectively, as afunction of qi for both the silicon-air interface

(black curves and symbols) and the plasmonicinterface (red curves and symbols) (20). In the

range of qi = 0° to 9°, the plasmonic interface

Fig. 4. (A) Angle of refraction versus angle of incidence for the ordinary (black curve and triangles)and anomalous refraction (red curve and dots) for the sample with G = 15 mm. The curves aretheoretical calculations made by using the generalized Snell’s law for refraction (Eq. 2), and thesymbols are experimental data extracted from refraction measurements as a function of the angle ofincidence (20). The shaded region represents “negative” refraction for the cross-polarized light, asillustrated in the inset. The blue arrows indicate the modified critical angles for total internal reflection.(B) Angle of reflection versus angle of incidence for the ordinary (black curve) and anomalous (redcurve and dots) reflection for the sample with G = 15 mm. The top left inset is the zoom-in view. Thecurves are theoretical calculations made by using Eq. 4, and the symbols are experimental dataextracted from reflection measurements as a function of the angle of incidence (20). The shadedregion represents “negative” reflection for the cross-polarized light, as illustrated in the bottom rightinset. The blue arrow indicates the critical angle of incidence above which the anomalously reflectedbeam becomes evanescent. Experiments with lasers emitting at different wavelengths show that theplasmonic interfaces are broadband, anomalously reflecting and refracting light from l ≈ 5 mm to l ≈10 mm.

Fig. 5. (A) SEM image of a plasmonic interface thatcreates an optical vortex. The plasmonic patternconsists of eight regions, each occupied by oneconstituent antenna of the eight-element set of Fig.2F. The antennas are arranged so as to generate aphase shift that varies azimuthally from 0 to 2p, thusproducing a helicoidal scattered wavefront. (B) Zoom-inview of the center part of (A). (C and D) Respectively,measured and calculated far-field intensity distributionsof an optical vortex with topological charge one. Theconstant background in (C) is due to the thermal ra-diation. (E and F) Respectively, measured and calcu-lated spiral patterns created by the interference of thevortex beam and a co-propagating Gaussian beam. (Gand H) Respectively, measured and calculated interfer-ence patterns with a dislocated fringe created by theinterference of the vortex beam and a Gaussian beamwhen the two are tilted with respect to each other. Thecircular border of the interference pattern in (G) arisesfrom the finite aperture of the beam splitter used tocombine the vortex and the Gaussian beams (20). Thesize of (C) and (D) is 60mm by 60mm, and that of (E)to (H) is 30 mm by 30 mm.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org336

RESEARCH ARTICLES

exhibits “negative” refraction and reflection forthe cross-polarized scattered light (schematics are

shown in the bottom right insets of Fig. 4, A andB). The critical angle for total internal reflection

is modified to ~–8° and +27° (Fig. 4A, blue ar-rows) for the plasmonic interface in accordance

with Eq. 3, comparedwith T17° for the silicon-airinterface; the anomalous reflection does not exist

beyond qi = –57° (Fig. 4B, blue arrow).At normal incidence, the ratio of intensity R

between the anomalously and ordinarily refractedbeams is ~0.32 for the sample with G = 15 mm

(Fig. 3C). R rises for increasing antenna packingdensities (Fig. 3, C and D) and increasing angles

of incidence [up to R ≈ 0.97 at qi = 14° (fig.

S1B)]. Because of the experimental configuration,we are not able to determine the ratio of intensity

between the reflected beams (20), but we expectcomparable values.

Vortex beams created by plasmonic interfaces.

To demonstrate the versatility of the concept of

interfacial phase discontinuities, we fabricated aplasmonic interface that is able to create a vortex

beam (21, 22) upon illumination by normally in-cident linearly polarized light. Avortex beam has

a helicoidal (or “corkscrew-shaped”) equal-phasewavefront. Specifically, the beam has an azi-

muthal phase dependence exp(ilϕ) and carries anorbital angular momentum of L ¼ lℏ per photon

(23). Here, the topological charge l is an integer,indicating the number of twists of the wavefront

within one wavelength; ϕ is the azimuthal anglewith respect to the beam axis; and ℏ is the

reduced Planck constant. These peculiar statesof light are commonly generated by using a spiral

phase plate (24) or a computer-generated holo-gram (25) and can be used to rotate particles (26)

or to encode information in optical communica-tion systems (27).

The plasmonic interface was created by arrang-ing the eight constituent antennas as shown in

Fig. 5, A and B. The interface introduces a spiral-like phase shift with respect to the planar wave-

front of the incident light, creating a vortex beamwith l = 1. The vortex beam has an annular in-

tensity distribution in the cross section, as viewedin a mid-infrared camera (Fig. 5C); the dark re-

gion at the center corresponds to a phase singu-

larity (22). The spiral wavefront of the vortexbeam can be revealed by interfering the beam

with a co-propagating Gaussian beam (25), pro-ducing a spiral interference pattern (Fig. 5E). The

latter rotates when the path length of theGaussianbeam was changed continuously relative to that

of the vortex beam (movie S1). Alternatively, thetopological charge l = 1 can be identified by a

dislocated interference fringe when the vortex andGaussian beams interfere with a small angle (Fig.

5G) (25). The annular intensity distribution andthe interference patterns were well reproduced in

simulations (Fig. 5, D, F, and H) by using thecalculated amplitude and phase responses of the

V-antennas (Fig. 2, D and E).Concluding remarks. Our plasmonic inter-

faces, consisting of an array of V-antennas, im-

part abrupt phase shifts in the optical path, thusproviding great flexibility in molding of the op-

tical wavefront. This breaks the constraint of stan-dard optical components, which rely on gradual

phase accumulation along the optical path tochange the wavefront of propagating light. We

have derived and experimentally confirmed gen-eralized reflection and refraction laws and studied

a series of intriguing anomalous reflection andrefraction phenomena that descend from the

latter: arbitrary reflection and refraction anglesthat depend on the phase gradient along the

interface, two different critical angles for totalinternal reflection that depend on the relative

direction of the incident light with respect to the

phase gradient, and critical angle for the reflectedbeam to be evanescent. We have also used a

plasmonic interface to generate optical vorticesthat have a helicoidal wavefront and carry orbit-

al angular momentum, thus demonstrating thepower of phase discontinuities as a design tool

of complex beams. The design strategies presentedin this article allow one to tailor in an almost

arbitrary way the phase and amplitude of anoptical wavefront, which should have major im-

plications for transformation optics and integratedoptics. We expect that a variety of novel planar

optical components such as phased antenna arraysin the optical domain, planar lenses, polarization

converters, perfect absorbers, and spatial phasemodulators will emerge from this approach.

Antenna arrays in the microwave andmillimeter-wave regions have been used for the

shaping of reflected and transmitted beams inthe so-called “reflectarrays” and “transmitarrays”

(28–31). These typically consist of a double-layerstructure comprising a planar array of antennas

and a ground plane (in the case of reflectarrays)or another array (in the case of transmitarrays),

separated by a dielectric spacer of finite thick-ness. Reflectarrays and transmitarrays cannot be

treated as a single interface for which one canwrite down the generalized laws because they

rely on both antenna resonances and the propa-gation of waves in the spacer to achieve the

desired phase control. The generalization of thelaws of reflection and refraction we present is

made possible by the deeply subwavelength thick-

ness of our optical antenna arrays and their asso-ciated abrupt phase changes, with no contribution

from propagation effects. These generalized lawsapply to the whole optical spectrum for suitable

designer interfaces and can be a guide for thedesign of new photonic devices.

References and Notes1. J. B. Pendry, D. Schurig, D. R. Smith, Science 312, 1780

(2006).

2. U. Leonhardt, Science 312, 1777 (2006).

3. W. Cai, V. Shalaev, Optical Metamaterials: Fundamentals

and Applications (Springer, New York, 2009).

4. N. Engheta, R. W. Ziolkowski, Metamaterials: Physics and

Engineering Explorations (Wiley-IEEE Press, New York,

2006).

5. I. I. Smolyaninov, E. E. Narimanov, Phys. Rev. Lett. 105,

067402 (2010).

6. S. D. Brorson, H. A. Haus, J. Opt. Soc. Am. B 5, 247 (1988).

7. R. P. Feynman, A. R. Hibbs, Quantum Mechanics and

Path Integrals (McGraw-Hill, New York, 1965).

8. E. Hecht, Optics (Addison Wesley, Boston, ed. 3,

1997).

9. H. T. Miyazaki, Y. Kurokawa, Appl. Phys. Lett. 89, 211126

(2006).

10. D. Fattal, J. Li, Z. Peng, M. Fiorentino, R. G. Beausoleil,

Nat. Photonics 4, 466 (2010).

11. J. A. Fan et al., Science 328, 1135 (2010).

12. B. Luk’yanchuk et al., Nat. Mater. 9, 707 (2010).

13. R. D. Grober, R. J. Schoelkopf, D. E. Prober, Appl. Phys.

Lett. 70, 1354 (1997).

14. L. Novotny, N. van Hulst, Nat. Photonics 5, 83

(2011).

15. Q. Xu et al., Nano Lett. 7, 2800 (2007).

16. M. Sukharev, J. Sung, K. G. Spears, T. Seideman,

Phys. Rev. B 76, 184302 (2007).

17. P. Biagioni, J. S. Huang, L. Duò, M. Finazzi, B. Hecht,

Phys. Rev. Lett. 102, 256801 (2009).

18. S. Liu et al., Opt. Lett. 34, 1255 (2009).

19. J. Ginn, D. Shelton, P. Krenz, B. Lail, G. Boreman,

Opt. Express 18, 4557 (2010).

20. Materials and methods are available as supporting

material on Science Online.

21. J. F. Nye, M. V. Berry, Proc. R. Soc. London A Math.

Phys. Sci. 336, 165 (1974).

22. M. Padgett, J. Courtial, L. Allen, Phys. Today 57, 35

(2004).

23. L. Allen, M. W. Beijersbergen, R. J. C. Spreeuw,

J. P. Woerdman, Phys. Rev. A 45, 8185 (1992).

24. M. W. Beijersbergen, R. P. C. Coerwinkel, M. Kristensen,

J. P. Woerdman, Opt. Commun. 112, 321 (1994).

25. N. R. Heckenberg, R. McDuff, C. P. Smith, A. G. White,

Opt. Lett. 17, 221 (1992).

26. H. He, M. E. J. Friese, N. R. Heckenberg, H. Rubinsztein-

Dunlop, Phys. Rev. Lett. 75, 826 (1995).

27. G. Gibson et al., Opt. Express 12, 5448 (2004).

28. D. M. Pozar, S. D. Targonski, H. D. Syrigos, IEEE Trans.

Antenn. Propag. 45, 287 (1997).

29. J. A. Encinar, IEEE Trans. Antenn. Propag. 49, 1403

(2001).

30. C. G. M. Ryan et al., IEEE Trans. Antenn. Propag. 58,

1486 (2010).

31. P. Padilla, A. Muñoz-Acevedo, M. Sierra-Castañer,

M. Sierra-Pérez, IEEE Trans. Antenn. Propag. 58, 2571

(2010).

Acknowledgments: The authors acknowledge helpful

discussion with J. Lin, R. Blanchard, and A. Belyanin.

The authors acknowledge support from the National

Science Foundation (NSF), Harvard Nanoscale Science

and Engineering Center (NSEC) under contract

NSF/PHY 06-46094, and the Center for Nanoscale

Systems (CNS) at Harvard University. This work was

supported in part by the Defense Advanced Research

Projects Agency (DARPA) N/MEMS S&T Fundamentals

program under grant N66001-10-1-4008 issued by the

Space and Naval Warfare Systems Center Pacific

(SPAWAR). Z.G. acknowledges funding from the

European Communities Seventh Framework Programme

(FP7/2007-2013) under grant agreement PIOF-GA-

2009-235860. M.A.K. is supported by NSF through a

Graduate Research Fellowship. Harvard CNS is a

member of the National Nanotechnology Infrastructure

Network. The Lumerical (Vancouver, BC, Canada)

FDTD simulations in this Research Article were run

on the Odyssey cluster supported by the Harvard

Faculty of Arts and Sciences Sciences Division Research

Computing Group.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/science.1210713/DC1

Materials and Methods

SOM Text

Figs. S1 to S6

References (32–39)

Movie S1

5 July 2011; accepted 19 August 2011

Published online 1 September 2011;

10.1126/science.1210713

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 337

RESEARCH ARTICLES

Detection of the Water Reservoir in a

Forming Planetary SystemMichiel R. Hogerheijde,1* Edwin A. Bergin,2 Christian Brinch,1 L. Ilsedore Cleeves,2

Jeffrey K. J. Fogel,2 Geoffrey A. Blake,3 Carsten Dominik,4 Dariusz C. Lis,5

Gary Melnick,6 David Neufeld,7 Olja Panić,8 John C. Pearson,9 Lars Kristensen,1

Umut A. Yıldız,1 Ewine F. van Dishoeck1,10

Icy bodies may have delivered the oceans to the early Earth, yet little is known about water inthe ice-dominated regions of extrasolar planet-forming disks. The Heterodyne Instrument for theFar-Infrared on board the Herschel Space Observatory has detected emission lines from both spinisomers of cold water vapor from the disk around the young star TW Hydrae. This water vaporlikely originates from ice-coated solids near the disk surface, hinting at a water ice reservoirequivalent to several thousand Earth oceans in mass. The water’s ortho-to-para ratio falls wellbelow that of solar system comets, suggesting that comets contain heterogeneous ice mixturescollected across the entire solar nebula during the early stages of planetary birth.

Water in the solar nebula is thought tohave been frozen out onto dust grains

outside ∼3 astronomical units (AU)(1, 2). Stored in icy bodies, this water provided a

reservoir for impact delivery of oceans to theEarth (3). In planet-forming disks, water vapor is

thought to be abundant only in the hot (>250 K)inner regions, where ice sublimates and gas-phase

chemistry locks up all oxygen in H2O. Emissionfrom hot (>250 K) water has been detected from

several disks around young stars (4, 5). In thecold (∼20 K) outer disk, water vapor freezes out,

evidenced by spectral features of water ice in afew disks (6, 7). However, (inter)stellar ultravi-

olet radiation penetrating the upper disk layersdesorbs a small fraction of water ice molecules

back into the gas phase (8), suggesting that cold(<100K)water vapor exists throughout the radial

extent of the disk. The detection of this watervapor would signal the presence of a hidden ice

reservoir.We report detection of ground-state rotation-

al emission lines of both spin isomers of water(JKAKC

110-101 from ortho-H2O and 111-000 from

para-H2O) from the disk around the pre–main-

sequence star TW Hydrae (TW Hya) using theHeterodyne Instrument for the Far-Infrared (HIFI)

spectrometer (9) on board the Herschel SpaceObservatory (10) (Fig. 1) (11, 12). TW Hya is

a 0.6 M⊙ (solar mass), 10-million-year-old TTauri star (13) 53.7 T 6.2 pc away from Earth. Its

196-AU-radius disk is the closest protoplanetarydisk to Earth with strong gas emission lines. The

disk’smass is estimated at 2 × 10−4 to 6 × 10−4M⊙

in dust and, using different tracers and assump-

tions, between 4×10−5 and 0.06M⊙ in gas (14–16).The velocity widths of the H2O lines (0.96 to

1.17 km s−1) (table S1) exceed by ∼40% those ofcold CO (14). These correspond to CO emission

from the full 196-AU-radius rotating disk inclined

at∼7°with only little (<65m s−1) turbulence (17).The wider H2O lines suggest that the water emis-

sion extends to ∼115 AU, where the gas orbitsthe star at higher velocities compared with

196 AU.To quantify the amount of water vapor traced

by the detected lines, we performed detailed sim-ulations of the water chemistry and line for-

mation using a realistic disk model matchingprevious observations (12, 18). We adopted a

conservatively low dust mass of 1.9 × 10−4 M⊙

and, using a standard gas-to-dust mass ratio of100, a gas mass of 1.9 × 10−2 M⊙. We explored

the effects of much lower gas-to-dust ratios. Wefollowed the penetration of the stellar ultraviolet

and x-ray radiation into the disk; calculated theresulting photodesorption of water and ensuing

gas-phase chemistry, including photodissociation;and solved the statistical-equilibrium excitation

and line formation. The balance of photodesorptionof water ice and photodissociation of water vapor

results in an equilibrium column of water H2Ovapor throughout the disk (Fig. 2). Consistent

with other studies (19), we find a layer of max-imum water vapor abundance of 0.5 × 10−7 to

2 × 10−7 relative to H2 at an intermediate height inthe disk. Above this layer, water is photodis-

sociated; below it, little photodesorption occursandwater is frozen out, with an ice abundance, set

by available oxygen, of 10−4 relative to H2.In our model, the 100- to 196-AU region

dominates the line emission, which exceeds ob-servations in strength by factors of 5.3 T 0.2 for

H2O 110-101 and 3.3 T 0.2 for H2O 111-000. Alower gas mass does not decrease the line in-

tensities, if we assume that the water ice, from

1Leiden Observatory, Leiden University, Post Office Box 9513,2300 RA Leiden, Netherlands. 2Department of Astronomy, Uni-versity of Michigan, Ann Arbor, MI 48109, USA. 3Division ofGeological and Planetary Sciences, California Institute of Tech-nology, Pasadena, CA 91125, USA. 4Astronomical InstituteAnton Pannekoek, University of Amsterdam, 1098 XH Am-sterdam, Netherlands. 5Division of Physics, Mathematics, andAstronomy, California Institute of Technology, Pasadena, CA91125, USA. 6Harvard-Smithsonian Center for Astrophysics,Cambridge, MA 02138, USA. 7Department of Physics and As-tronomy, Johns Hopkins University, Baltimore, MD 21218,USA. 8European Southern Observatory, 85748 Garching, Ger-many. 9Jet Propulsion Laboratory, California Institute of Tech-nology, Pasadena, CA 91109, USA. 10Max-Planck-Institut fürExtraterrestrische Physik, 85748 Garching, Germany.

*To whom correspondence should be addressed. E-mail:michiel@strw.leidenuniv.nl

Fig. 1. Spectra of para-H2O111-000 (A) and ortho-H2O 110-101 (B) obtained with HIFI onthe Herschel Space Observatorytoward the protoplanetary diskaround TWHya after subtractionof the continuum emission. Thevertical dotted lines show thesystem’s velocity of +2.8 km s−1

relative to the Sun’s local en-vironment (local standard ofrest).

REPORTS

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org338

which the water vapor derives, formed early inthe disk’s evolution, before substantial gas loss

occurred, and remains frozen on grains. Themostplausible explanation involves a difference in the

relative location of small, bare grains regulatingthe ultraviolet radiative transport and larger, ice-

carrying grains. Differential settling of large grainsrelative to small grains moves much of the ice

reservoir below the reach of the ultraviolet ra-diation, resulting in less water vapor and weaker

lines. Ourmodel matches the observations if only12% of the original ice content remains above

this line (20). A radially increasing degree ofsettling of icy grains explains the observed H2O

line widths.

The detected water vapor, resulting fromphotodesorption, implies an ice reservoir in the

giant planet formation zone and beyond. In oursimulations, the 7.3 × 1024 g of detected water

vapor (equivalent to 0.005 times the mass ofEarth’s oceans) originate from a total ice reser-

voir of 9 × 1027 g (or several thousands of Earth’soceans) throughout the disk. The size of this res-

ervoir is tied to the dust mass contained in thedisk, for which we adopt a conservatively low

value. Although the ice reservoir is only observedindirectly, no known mechanism can remove it

from the regions probed by Herschel. Any small-er ice reservoir implies the corresponding ab-

sence of elemental oxygen that efficiently reformswater ice on the grains.

The detection of both spin isomers of watervapor allows its ortho-to-para ratio (OPR) to be

derived, because our simulations indicate that thelines are optically thin. An OPR of 0.77 T 0.07

matches our observations (12). This value is

much lower than the OPR range of 2 to 3 ob-served for solar system comets (21). It is com-

mon practice to associate the OPR with a spintemperature Tspin at which a Boltzmann distri-

bution reproduces the ratio of spin isomers. Ourderived OPR corresponds to Tspin = 13.5 T 0.5 K,

whereas the range for solar system comets yieldsa Tspin of >20 K.

Radiative conversion between spin isomers isnot allowed in the gas phase, preserving the OPR

for long time scales. Gas-phase formation of wa-ter occurs through exothermic reactions leading

to an OPR of 3. On grains, water forms and sur-vives at low temperatures, and it is tempting to

equate Tspin with the grain temperature. However,

the energetics of water formation and ortho-to-para exchange on grains are poorly understood

(22), and the water OPR may be changed byphotodesorption. This process starts by dissociat-

ing water to H and OH in the ice and continueswith the energetic H kicking out a neighboring

H2O molecule from the ice matrix or with the Hand OH recombining in the ice to form H2Owith

sufficient internal energy to sublimate (23). Thelatter route drives the OPR to at least unity, im-

plying an even lower original ice OPR, to yield aresulting OPR of 0.77. Cometary volatiles are

released through thermal sublimation, and theirmeasured OPRs are interpreted to reflect the OPR

of their ice constituents. Equating Tspin with thephysical temperature of the grain on which the

ice formed is supported by the similarity of mea-sured Tspin of NH3 and H2O in several individual

solar system comets (24).Solar system comets consist of a heteroge-

neous mixture of ices and solids, likely assem-

bled in the giant planet formation zone bymixinglocal material with material that drifted in from

larger radii (25). Our water vapor observationsprobe cold, ice-coated precometary grains resid-

ing beyond >50 AU, representing the bulk of thelatter material. The presence in comets of crys-

talline silicates, requiring formation temperatures>800 K (26), together with CO and H2O ices that

condense at 20 to 100 K, argues for transport ofhot material from near the star to the icy outer

regions of the solar nebula (27). Provided thatspin temperatures reflect formation histories, the

different Tspin inferred for the water ice in TWHya (<13 K) and solar system comets (>20 K)

indicates a similar mixing of volatiles throughout

the entire solar nebula, blending water formed at>50 K and an OPR of 3 with water formed at 10

to 20 K and an OPR < 1 probed by our observa-tions. In this case, the range of Tspin values of the

cometary inventory reflects the stochastic natureof transport and mixing.

Our Herschel detection of cold water vapor inthe outer disk of TW Hya demonstrates the pres-

ence of a considerable reservoir of water ice inthis protoplanetary disk, sufficient to form several

thousand Earth oceans’ worth of icy bodies. Ourobservations only directly trace the tip of the ice-

berg of 0.005 Earth oceans in the form of watervapor.

References and Notes1. One astronomical unit (AU) is the mean distance between

Earth and the Sun of 1.49598 × 1011 m.

2. C. Hayashi, Prog. Theor. Phys. 70 (suppl.), 35

(1981).

3. T. Matsui, Y. Abe, Nature 322, 526 (1986).

4. K. M. Pontoppidan, C. Salyk, G. A. Blake, H. U. Käufl,

Astrophys. J. 722, L173 (2010).

5. J. S. Carr, J. R. Najita, Astrophys. J. 733, 102

(2011).

6. H. Terada et al., Astrophys. J. 667, 303 (2007).

7. M. Honda et al., Astrophys. J. 690, L110

(2009).

8. C. Dominik, C. Ceccarelli, D. Hollenbach, M. Kaufman,

Astrophys. J. 635, L85 (2005).

9. T. de Graauw et al., Astron. Astrophys. 518, L6

(2010).

10. G. L. Pilbratt et al., Astron. Astrophys. 518, L1

(2010).

11. Water in Earth’s atmosphere obstructs ground-based

detection of cold water vapor in planet-forming disks.

Although Herschel cannot spatially resolve even the

closest disk in water ground-state emission lines, HIFI

spectrally resolves the H2O line profiles. Comparison

with previous spectrally and spatially resolved

observations of CO confirms the disk origin of the

H2O lines.

12. Materials and methods are available as supporting

material on Science Online.

13. R. A. Webb et al., Astrophys. J. 512, L63 (1999).

14. J. H. Kastner, B. Zuckerman, D. A. Weintraub, T. Forveille,

Science 277, 67 (1997).

15. N. Calvet et al., Astrophys. J. 568, 1008 (2002).

16. U. Gorti, D. Hollenbach, J. Najita, I. Pascucci, Astrophys. J.

735, 90 (2011).

17. A. M. Hughes, D. J. Wilner, S. M. Andrews, C. Qi,

M. R. Hogerheijde, Astrophys. J. 727, 85 (2011).

18. W. Thi et al., Astron. Astrophys. 518, L125 (2010).

19. P. Woitke, W. Thi, I. Kamp, M. R. Hogerheijde,

Astron. Astrophys. 501, L5 (2009).

20. R. Meijerink, K. M. Pontoppidan, G. A. Blake,

D. R. Poelman, C. P. Dullemond, Astrophys. J. 704, 1471

(2009).

A

C D

B

Fig. 2. Adopted model for the TW Hya protoplanetary disk. (A) H2 number density, (B) dust temperature,(C) the number density of water vapor molecules and contours of volume-averaged water ice abundancedecreasing from white to black as 2 × 10−4, 2 × 10−5, 2 × 10−6, 2 × 10−7, and 2 × 10−8, relative to H2,and (D) one quadrant of the resulting water emission line intensity from the near face-on disk, in arbitraryunits. In (A) and (B), the blue contour delineates the layer of maximum water vapor abundance.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 339

REPORTS

21. B. P. Bonev et al., Astrophys. J. 661, L97

(2007).

22. H.-H. Limbach et al., ChemPhysChem 7, 551

(2006).

23. S. Andersson, E. F. van Dishoeck, Astron. Astrophys. 491,

907 (2008).

24. Y. Shinnaka et al., Astrophys. J. 729, 81 (2011).

25. S. J. Weidenschilling, Mon. Not. R. Astron. Soc. 180, 57

(1977).

26. S. A. Sandford et al., Science 314, 1720

(2006).

27. D. H. Wooden, Space Sci. Rev. 138, 75

(2008).

Acknowledgments: Herschel is a European Space Agency

space observatory with science instruments provided by

European-led principal investigator consortia and with

important participation from NASA. This work was

partially supported by Nederlandse Organisatie voor

Wetenschappelijk Onderzoek grant 639.042.404, NSF

grant 0707777, and, as part of the NASA Herschel HIFI

guaranteed time program, NASA. The data presented

here are archived at the Herschel Science Archive,

http://archives.esac.esa.int/hda/ui, under OBSID

1342198337 and 1342201585.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/334/6054/338/DC1

Materials and Methods

Table S1

References (28–39)

25 May 2011; accepted 20 September 2011

10.1126/science.1208931

Supramolecular Linear Heterojunction

Composed of Graphite-Like

Semiconducting Nanotubular SegmentsWei Zhang,1,2 Wusong Jin,3 Takanori Fukushima,1,4* Akinori Saeki,5 Shu Seki,5 Takuzo Aida1,2*

One-dimensionally connected organic nanostructures with dissimilar semiconducting properties areexpected to provide a reliable platform in understanding the behaviors of photocarriers, which areimportant for the development of efficient photon-to-electrical energy conversion systems. Althoughbottom-up supramolecular approaches are considered promising for the realization of such nanoscaleheterojunctions, the dynamic nature of molecular assembly is problematic. We report asemiconducting nanoscale organic heterojunction, demonstrated by stepwise nanotubularcoassembly of two strategically designed molecular graphenes. The dissimilar nanotubular segments,thus connected noncovalently, were electronically communicable with one another over theheterojunction interface and displayed characteristic excitation energy transfer and charge transportproperties not present in a mixture of the corresponding homotropically assembled nanotubes.

Heterojunctions, occurring between two

dissimilar semiconducting materials, areexpected to provide peculiar electronic

properties that are hard to realize by homojunc-tions. Heterojunctions of varying dimensions

are readily fabricated from inorganic semicon-ductors and lead to many applications, including

solid-state lasers, diodes, solar cells, and tran-sistors (1–6). Organic heterojunctions are of im-

portance in the development of organic thin-film

solar cells (7, 8). However, most that have beenstudied are so-called bulk heterojunctions, which

are formed only coincidentally from donor/acceptormixtures upon phase separation (9–12). Although

bottom-up supramolecular approaches (13, 14)are a potent tool for the formation of organic het-

erojunctions, such studies have just started withmolecularly engineered donor/acceptor couples

(15–19). From a fundamental viewpoint, one chal-lenge would be to tailor a linear organic hetero-

junction at the nanoscale by joining together

dissimilar semiconducting one-dimensional mo-

lecular objects, because one has to overcome theessential problem arising from the dynamic

nature of molecular assembly (13, 14, 20–24).We reported that a Gemini-shaped hexa-peri-

hexabenzocoronene (HBC) derivative, bearingtwo triethylene glycol-appended phenyl groups

on one side of the HBC core and dodecyl sidechains on the other, self-assembles into a semi-

conducting nanotube with inner and outer diam-

eters of 14 and 20 nm, respectively (25, 26). Arecent structural analysis using a synchrotron

x-ray diffraction technique revealed that the nano-tube is composed of a graphite-like bilayer wall

consisting of helically twisted columnar arraysof p-stacked HBC units (26). For the realiza-

tion of a nanotubular heterojunction using thisself-assembling motif, essential requisites are (i)

the formation of a morphologically stabilizedseed nanotube and (ii) the design of a second

graphene monomer capable of tubularly assem-bling from the extremely thin facets of the seed

nanotube termini. Further issues include how tocope with a high dispersibility of the seed nano-

tube and a solubility of the second monomerunder assembling conditions.

A keen examination, taking into account allthe above requisites, led us to HBC derivatives

1 and 2 (27) as the monomers for the seed andsecond nanotubular segment, respectively (Fig.

1). HBC 1 carries two bipyridine (bpy) units, inorder for the resulting seed nanotube (Fig. 1A)

to be morphologically stabilized by wrapping

with a metal-coordination network (Fig. 1 B)(28). The charged surface of the resultant seed

also merits its homogeneous dispersion by anelectrostatic repulsion (29). On the other hand,

HBC 2 bears four electron-withdrawing fluorinesubstituents, so it can adhere electronically to

the seed termini and self-assemble selectivelyfrom their nanotubular facets. When these HBC

molecules coassemble stepwise (Fig. 1C), theresultant connecting segments are electronically

dissimilar to one another (Fig. 1D).As a typical example of the preparation of the

seed nanotube (NT1•Cu, Fig. 1B), a 5-ml glassvial containing a tetrahydrofuran (THF) solution

(2.0 ml) of HBC 1 (0.5 mg, 1.5 × 10−4 M) wasplaced in a 50-ml glass vial containing 10 ml of

methanol (MeOH) and allowed to stand at 25°C,whereupon a yellow suspension gradually formed.

Absorption spectroscopy of the suspension aftera 24-hour incubation (fig. S1, A and B, broken

curve) showed red-shifted absorption bands at426 and 459 nm characteristic of J-aggregated

HBCs (25, 26). Scanning electron microscopy(SEM, fig. S2A) and transmission electron mi-

croscopy (TEM, fig. S2B) of an air-dried sampleof the suspension allowed for visualizing nano-

tubes (NT1) with a uniform diameter of 20 nm,although they were heavily bundled (Fig. 1A)

just like other HBC nanotubes (25, 26). We in-vestigated the metal-coordination capability of

NT1 by using Cu2+, because bpy is known to bind

to Cu2+, affording a bpy2•Cu2+ complex. As soon

as a MeOH solution (1.0 ml) of copper(II) tri-

fluoromethanesulfonate [Cu(OTf )2, 0.5 mg, 1.5 ×10−6 mol; 5.0 equivalents to HBC 1] was added,

the suspension containing bundled NT1 becameclear, suggesting that Cu2+ ions are bound to the

surface bpy groups (Fig. 1, A to B) and makethe nanotubes (NT1•Cu) electrostatically repulsive

from one another (29). When an air-dried sam-ple of NT1•Cu, isolated by filtration and washed

with MeOH to remove free Cu(OTf )2, was sub-jected to SEM, highly dispersed nanotubes were

observed (fig. S3A). Complete coordination ofCu2+ with bpy-appended HBC 1 was confirmed

by matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry of

isolated NT1•Cu (fig. S4); no peaks attributableto metal-free 1 were detected, but those assign-

able to 1•Cu, dissociated from NT1•Cu, were. Themetal coordination of NT1 did not give rise to

any shift of the J-aggregate absorption bands (426and 459 nm; fig. S1B, solid curve). Hence, the

p-stacking geometry of the HBC units is intact

1Functional Soft Matter Research Group, RIKEN AdvancedScience Institute, 2-1 Hirosawa, Wako, Saitama 351-0198,Japan. 2Department of Chemistry and Biotechnology, School ofEngineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku,Tokyo 113-8656, Japan. 3College of Chemistry, Chemical Engi-neering and Biotechnology, Donghua University, 2999 NorthRenmin Road, Songjiang, Shanghai 201620, People’s Republicof China. 4Chemical Resources Laboratory, Tokyo Institute ofTechnology, 4259 Nagatsuta, Midori-ku, Yokohama 226-8503,Japan. 5Graduate School of Engineering, Osaka University, 2-1Yamadaoka, Suita, Osaka 565-0871, Japan.

*To whom correspondence should be addressed. E-mail:fukushima@riken.jp (T.F.); aida@macro.t.u-tokyo.ac.jp (T.A.)

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org340

REPORTS

even with the formation of a metal-coordinationnetwork on the nanotube surface.

We investigated whether HBC 2 alone has theability to form a nanotubular assembly. On the

basis of a computational study using density func-tional theory (DFT), the core of HBC 2 with four

fluorine substituents most likely adopts a slightlytwisted, concave structure (fig. S5). Probably be-

cause of this skeletal distortion, HBC 2 was un-able to assemble into nanotubes under a variety of

conditions used previously (16, 18, 25, 26, 28, 29).However, we eventually found that, in acetone,

2 self-assembles into tubules. This observationwas unexpected because none of other reported

HBC derivatives assemble to form nanotubes inacetone. For the nanotubular assembly of 2, an

acetone suspension of this HBC (1.5 × 10−4 M)was heated to 50°C, and the resultant clear solu-

tion was allowed to cool to 25°C. The solu-tion gradually became turbid and displayed a

red-shifted absorption spectrum characteristic ofJ-aggregated HBCs (fig. S1, C and D). SEM of

an air-dried sample of the suspension, obtained

after a 12-hour incubation, showed the presenceof heavily bundled cylindrical nanostructures (fig.

S6A). TEM revealed that the cylinders are actu-ally nanotubes (NT2) with a uniform diameter

of 20 nm (fig. S6B).NT1•Cu dispersed individually in acetone,

where the nanotubular assembly of HBC 2 canoccur. As a typical example of the successful het-

erojunction (Fig. 1C), the acetone dispersionof NT1•Cu, used as the seed (Fig. 1B), was pre-

sonicated for 5 to 10 min so that the nanotubeswere cut into short pieces for enhancing the prob-

ability of linear heterojunction (fig. S3C). Theresultant dispersion of NT1•Cu (1.5 × 10−4 M,

1.0 ml) was mixed at 50°C with an acetone solu-tion of HBC 2 (1.5 × 10−4 M, 1.0 ml). When

the mixture was allowed to cool and stand at25°C, the assembly of HBC 2 took place. The

self-assembly of 2 without NT1•Cu gave rise to asuspension of bundled NT2. However, the co-

assembling mixture in the presence of NT1•Cu

remained clear even after 12 hours. When an air-

dried sample of this clear dispersion was studiedby SEM, micrometer-long nanotubular objects

with bright and dark segments were observed(Fig. 2A and fig. S7). Most of the nanotubes ap-

peared to consist of two block segments, butsome were composed of three block segments.

Likewise, scanning TEM allowed for visualiz-ing the presence of segments with different con-

trasts in single nanotubes (Fig. 2C). By means ofelement mapping using TEM energy-dispersive

x-ray spectroscopy (TEM-EDX), we confirmed

that copper is localized in the bright segments(Fig. 2, E and F), whereas carbon populates over

the entire nanotube (Fig. 2, D and F). These ob-servations demonstrate the occurrence of a linear

heterojunction to give block-NT1•Cu/NT2 (Fig.1C). Considering the electronic characters of the

two HBC molecules, the heterotropic p-stackinginteraction seems stronger than the homotropic

one. This drives the preferential assembly ofHBC 2 on the nanotubular facets of seed NT1.Cu.

Because multiblock heterojunction nanotubes suchas NT1•Cu–NT2–NT1•Cu–NT2 and NT1•Cu–NT2–

NT1•Cu were not detected, postconnection of NT2

with NT1•Cu is unlikely. In tapping-mode atomic

force microscopy (AFM) on a silicon wafer, thesetwo block segments exhibited different height

profiles, 16 to 18 nm and 8 to 10 nm (Fig. 2B),

Fig. 2. Microscopic imaging of block-NT1•Cu/NT2 ([2]/[1] = 1.0). The sample was prepared by drop-castingits acetone dispersion and air dried. (A) SEM (scale bar, 500 nm), (B) tapping-mode AFM (scale bar,200 nm), and (C) scanning TEM (scale bar, 50 nm) micrographs. TEM-EDX mapping of (D) carbon (scalebar, 50 nm), (E) copper (scale bar, 50 nm), and (F) carbon and copper elements (scale bar, 50 nm).

Fig. 1. Molecular structures of HBCs 1 and 2 and schematic illustrations of the preparation of (A) NT1(bundled) by MeOH vapor diffusion into a THF solution of HBC 1, (B) seed NT1•Cu (dispersed) by post-functionalization of NT1 with Cu(OTf)2 in MeOH, and (C) block-NT1•Cu/NT2 (dispersed) by cooling a hotacetone solution of HBC 2 in the presence of NT1•Cu as the seed. (D) Schematic illustration of anidealized cross section of block-NT1•Cu/NT2 at the heterojunction interface.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 341

REPORTS

that were the same as those observed for sepa-

rately prepared homotropic NT1•Cu and NT2

(fig. S8). The exceptionally small height profile

of NT2 indicates a low structural robustness ofits graphite-like wall because of the nonplanar

HBC core of 2 (fig. S5).As shown in fig. S9, NT1 (black solid curve)

and NT2 (black broken curve), when photoex-cited at 365 nm, fluoresced most intensely at 585

and 594 nm, respectively (16, 18, 29). In sharpcontrast, NT1•Cu luminesced much less intensely

(fig. S9, red) as a consequence of possible flu-

orescence quenching by Cu2+ in the coordination

network around the nanotube. These contrastingluminescence features allowed us to investigate

whether the two graphite-like nanotubular seg-ments in block-NT1•Cu/NT2 (Fig. 1C) communi-

cate with one another by excitation energy transferover the heterojunction interface. When an ace-

tone dispersion of block-NT1•Cu/NT2 preparedby using the mole ratio [2]/[1] of 0.5 for the step-

wise coassembly was photoexcited at 365 nm,the observed fluorescence (Fig. 3B, pink) was

as weak as that of homotropically assembled

NT1•Cu (Fig. 3B, red). Even when the mole ra-tio [2]/[1] used for the stepwise coassembly was

increased from 0.5 to 1.0 (orange), 1.5 (green),and 2.0 (blue) (Fig. 3A), the fluorescence in-

tensity of block-NT1•Cu/NT2 remained almostunchanged at a very low level (Fig. 3B). In con-

trast to block-NT1•Cu/NT2 (Fig. 3B inset), a mix-ture of homotropic NT1•Cu and NT2 ([2]/[1] =

1.0) (Fig. 3C, orange) displayed a rather brightfluorescence (Fig. 3D inset), where the observed

intensity was the sum of those of NT1•Cu andNT2 (Fig. 3D, orange). When the mixing ratio of

NT2 to NT1•Cu was varied ([2]/[1] = 0.5 to 2.0)(Fig. 3C), the overall fluorescence intensity changed

depending on the amount of NT2 (Fig. 3D). To-

gether with the results of these control experi-ments, the fluorescing properties observed for

block-NT1•Cu/NT2 indicate that the NT1•Cu andNT2 segments communicate efficiently over the

heterojunction interface (Fig. 1D) by excitation en-ergy transfer. The results also suggest that the

amount of unconnected NT2, if any formed in thestepwise coassembly (Fig. 1C), is negligibly small.

Because the excitation energy transfers overthe heterojunction interface from one nanotu-

bular segment (NT2) to the other (NT1•Cu) inblock-NT1•Cu/NT2 (Fig. 1C), we were motivated

to explore the behaviors of charge carriers, ifgenerated, in block-NT1•Cu/NT2. For this pur-

pose, we used a flash-photolysis time-resolvedmicrowave conductivity (FP-TRMC) method (30),

which allows for evaluating the intrinsic prop-erties of charge carriers without electrodes. At

first, we investigated the FP-TRMC profilesof homotropically assembled NT1•Cu and NT2.

Upon laser excitation at 355 nm (photon den-sity, 4.7 × 1015 cm–2), both nanotube samples in

the solid state displayed TRMC signals, indicat-ing that these nanotubes are photoconductive.

The TRMC signals thus observed for NT1•Cu andNT2 decayed at comparable rates (fig. S10, A

and B) with lifetimes (t1/e) of 1.4 × 10−6 and 2.5 ×10−6 s, respectively (Fig. 4). Of particular inter-

est to note here, the charge carriers generated inblock-NT1•Cu/NT2 ([2]/[1] = 1.0) were long-

lived (fig. S10D): The observed lifetime (t1/e =8.8 × 10−6 s) was roughly five times longer than

those of NT1•Cu and NT2 (Fig. 4). Taking into

account the energy levels of 1 (16) and 2 (fig.S11) estimated from their electrochemical and

spectral data, it is most likely that the NT1•Cu

and NT2 segments preferentially accommodate

hole and electron, respectively. As a consequenceof such preferential localization of hole and elec-

tron in block-NT1•Cu/NT2, the probability ofcharge recombination could be reduced. In sharp

contrast, a mixture of homotropic NT1•Cu and NT2

([2]/[1] = 1.0) (fig. S10C) hardly showed suppres-

sion of charge recombination (t1/e = 2.0 × 10−6 s)(Fig. 4).

The behaviors of excitons and charge car-riers thus observed for semiconducting block-

NT1•Cu/NT2 are remarkable considering thatthey are brought about only by noncovalent con-

nection of two different homotropic blocks with

Fig. 3. (A) Absorption and (B) fluorescence spectra in acetone at 25°C of block-NT1•Cu/NT2 preparedby using different molar ratios of 2 to 1 for their stepwise coassembly {[1] = 1.5 × 10−4 M; [2] = 0(red), 0.75 × 10−4 M (pink), 1.5 × 10−4 M (orange), 2.3 × 10−4 M (green), and 3.0 × 10−4 M (blue)}. (Binset) A photograph of an acetone dispersion of block-NT1•Cu/NT2 ([2]/[1] = 1.0) upon 365-nmexcitation. (C) Absorption and (D) fluorescence spectra in acetone at 25°C of mixtures of NT1•Cu andNT2 at different molar ratios of 2 to 1 {[1] = 1.5 × 10−4 M; [2] = 0 (red), 0.75 × 10−4 M (pink), 1.5 ×10−4 M (orange), 2.3 × 10−4 M (green), and 3.0 × 10−4 M (blue)}. (D inset) A photograph of an acetonesuspension of a mixture of NT1•Cu and NT2 ([2]/[1] = 1.0) upon 365-nm excitation. The baseline uprisein (C) upon increment of [2] is due to light scattering by bundled NT2.

Fig. 4. Lifetimes (t1/e) of charge carriers generatedby laser excitation of NT1•Cu, NT2, a mixture ofNT1•Cu and NT2 ([2]/[1] = 1.0), and block-NT1•Cu/NT2([2]/[1] = 1.0) in the solid state. t1/e is defined by thetime when the FP-TRMC transient decays down to 1/eof its maximum value (fig. S10).

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org342

REPORTS

an extremely thin (~3 nm) facet. In particular,the nearly perfect fluorescence quenching of its

NT2 block substantiates that electronic effectsof heterojunctions can indeed propagate over a

micrometer-long distance through a great numberof p-stacks in semiconducting organic materials.

References and Notes1. M. T. Björk et al., Nano Lett. 2, 87 (2002).

2. S. Banerjee, S. S. Wong, Nano Lett. 2, 195 (2002).

3. O. Harnack, C. Pacholski, H. Weller, A. Yasuda,

J. M. Wessels, Nano Lett. 3, 1097 (2003).

4. B. Tian et al., Nature 449, 885 (2007).

5. A. I. Hochbaum, P. Yang, Chem. Rev. 110, 527 (2010).

6. F.-S. Tsai et al., Appl. Phys. Express 4, 025002 (2011).

7. S. Günes, H. Neugebauer, N. S. Sariciftci, Chem. Rev.

107, 1324 (2007).

8. B. C. Thompson, J. M. J. Fréchet, Angew. Chem. Int. Ed.

47, 58 (2008).

9. G. Yu, J. Gao, J. C. Hummelen, F. Wudl, A. J. Heeger,

Science 270, 1789 (1995).

10. P. Peumans, S. Uchida, S. R. Forrest, Nature 425, 158 (2003).11. G. Li et al., Nat. Mater. 4, 864 (2005).

12. J. Y. Kim et al., Science 317, 222 (2007).13. T. F. A. De Greef et al., Chem. Rev. 109, 5687 (2009).

14. F. J. M. Hoeben, P. Jonkheijm, E. W. Meijer, A. P. H. J.

Schenning, Chem. Rev. 105, 1491 (2005).

15. F. Würthner et al., J. Am. Chem. Soc. 126, 10611

(2004).

16. Y. Yamamoto et al., Science 314, 1761 (2006).

17. A. L. Sisson et al., Angew. Chem. Int. Ed. 47, 3727 (2008).

18. Y. Yamamoto et al., Proc. Natl. Acad. Sci. U.S.A. 106,

21051 (2009).

19. N. Sakai, R. Bhosale, D. Emery, J. Mareda, S. Matile,

J. Am. Chem. Soc. 132, 6923 (2010).

20. By using crystallizable core units, the formation of rodlike

block co-micelles has been demonstrated (21–24).

21. X. Wang et al., Science 317, 644 (2007).

22. T. Gädt, N. S. Ieong, G. Cambridge, M. A. Winnik,

I. Manners, Nat. Mater. 8, 144 (2009).

23. J. B. Gilroy et al., Nat. Chem. 2, 566 (2010).

24. S. K. Patra et al., J. Am. Chem. Soc. 133, 8842

(2011).

25. J. P. Hill et al., Science 304, 1481 (2004).26. W. Jin et al., J. Am. Chem. Soc. 130, 9434 (2008).

27. Materials and methods are available as supporting

material on Science Online.

28. W. Zhang, W. Jin, T. Fukushima, N. Ishii, T. Aida, Angew.

Chem. Int. Ed. 48, 4747 (2009).

29. G. Zhang et al., J. Am. Chem. Soc. 129, 719

(2007).

30. A. Saeki, T. Fukumatsu, S. Seki, Macromolecules 44,

3416 (2011).

Acknowledgments: We thank E. Ohta (RIKEN) for DFT

calculation of HBC 2. W.Z. thanks the Japan Society for

the Promotion of Science Young Scientist Fellowship

(21•9925). T.F. thanks Ministry of Education, Culture,

Sports, Science, and Technology, Japan, for financial

support (21350108).

Supporting Online Materialwww.sciencemag.org/cgi/content/full/334/6054/340/DC1

Materials and Methods

Figs. S1 to S11

References (31–33)

27 June 2011; accepted 14 September 2011

10.1126/science.1210369

Dynamics of the Reaction of Methane

with Chlorine Atom on an Accurate

Potential Energy SurfaceGábor Czakó* and Joel M. Bowman

The reaction of the chlorine atom with methane has been the focus of numerous studies thataim to test, extend, and/or modify our understanding of mode-selective reactivity in polyatomic systems.To this point, theory has largely been unable to provide accurate results in comparison with experiments.Here, we report an accurate global potential energy surface for this reaction. Quasi-classical trajectorycalculations using this surface achieve excellent agreement with experiment on the rotationaldistributions of the hydrogen chloride (HCl) product. For the Cl + CHD3 → HCl + CD3 reaction at lowcollision energies, we confirm the unexpected experimental finding that CH-stretch excitation is no moreeffective in activating this late-barrier reaction than is the translational energy, which is in contradictionto expectations based on results for many atom-diatom reactions.

Decades of experimental and theoreticalstudies of atom-diatom reactions led to a

well-validated framework for predictingthe effect of vibrational excitation on the ensuing

dynamics (1, 2). Earlier fundamental research dem-onstrated the importance of the reaction barrier

location on the efficacy of partitioning the total

energy between internal excitation of the diatomicmolecule and relative translational energy of the

reactants. The careful and correct analysis of thesereactions led to the “Polanyi rules” (3), which

state that vibrational energy is more efficient thanis translational energy in activating a late-barrier

reaction, whereas the reverse is true for an early-barrier reaction. Recent studies have investigated

the generality and validity of these paradigms forpolyatomic systems. The X + methane (CH4

and deuterated isotopologues) reactions (whichreplace the diatomic with a five-atom molecule)

have played a central role in this research, in

which, for example, the choice of X as H, O, F,and Cl has permitted the height and location of

the reaction barrier to vary widely. Recent ex-periments by Liu and co-workers (4–8) on the F

and Cl + methane reactions have uncovered sur-prising departures from expectations that present a

strong challenge to theory, which ultimately pro-

vides detailed understanding of chemical reactiondynamics. A rigorous theoretical approach to re-

action analysis consists of two components. Thefirst is to determine the global potential energy

surface (PES) (9), which governs the nuclear mo-tion, and the second is to perform dynamics

calculations with the PES. We succeeded in carry-ing out this process recently for the early-low-

barrier F + CHD3 reaction, for which we were ableto illuminate the surprising experimental result of

the enhancement of the DF + CHD2 channel byexciting the CH-stretch (10, 11). In this Report,

we take the same approach to address and inter-pret experiments on the late-high-barrier Cl +

CHD3 reaction by Liu and co-workers (5), whichalso uncovered a surprising result, namely that at

low collision energies (Ecoll) vibrational excita-tion of the CH-stretch was no more effective than

was translational energy in promoting the reaction.

This result, as pointed out by these authors, contra-

dicts the rule of thumb of reaction dynamics(Polanyi rules). Other interesting experimental re-

sults of this reaction are also successfully addressed.As with the previous accurate PES for F +

CH4 (10), the Cl + CH4 PES is a permutationallyinvariant fit (12, 9) to roughly 16,000 high-level

ab initio electronic energies. The selection of con-figurations for the PES is quite similar to proce-

dures used for the F + CH4 PES (10, 12), anddetails are given in (13). A key part of this ap-

proach is the use of an electronic structuremethodthat gives accurate energies, especially for the

barrier height, reaction enthalpy, and the entranceand exit channel van derWaals (vdW)wells. These

wells, which result from long-range attractiveinteractions, are ubiquitous in chemistry, and as

we show here, the prereactive one has a substan-tial effect on the Cl + CHD3 reaction dynamics at

low collision energies. The inclusion of the spin-orbit (SO) correction is also essential in the present

case because it effectively increases the barrierheight and reaction endoergicity by 0.8 kcal/mol

and has a substantial effect on the entrance chan-nel vdW well. (The SO correction is a relativistic

effect, which lowers the energy of the halogen

atoms and has about twice as large an energy shifton the heavier Cl than on F. The widely applied

nonrelativistic electronic structure computationsneglect this effect.) There is also a substantial

basis set–superposition error in this region, whichhas to be corrected. Thus, the goal for the present

PES is to take all of this into account. In order toachieve this goal, we used a composite electronic

structure method, which provides accurate en-ergies with affordable computational time. The

general composite approach (14), which is wide-ly used, combines results from several levels of

ab initio method and basis (15, 16). Second, inorder to account for the SO effect differences be-

tween the SO and non-SO ground-state electronicenergies, obtained by means of multireference

configuration interaction (MRCI) with Davidsoncorrection (MRCI + Q) with basis set aug-cc-

pVTZ (MRCI + Q/aug-cc-pVTZ), were added to

Cherry L. Emerson Center for Scientific Computation and De-partment of Chemistry, Emory University, Atlanta, GA 30322,USA.

*To whom correspondence should be addressed. E-mail:czako@chem.elte.hu

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 343

REPORTS

the composite non-SO energies at 1598 ClCH4

configurations in the entrance channel, where the

SO corrections are substantial and nonconstant.(2000 Cl + CH4 fragment data were also shifted

by the constant SO correction of the Cl atom.)Third, counterpoise correction for entrance chan-

nel basis set–superposition error at the above-mentioned 1598 configurations was applied. The

present PES is based on electronic energies withbasis set effects up to aug-cc-pCVTZ (correlation-

consistent polarized core-valence triple-zeta basisaugmented with diffuse functions) (17); electron

correlation up to the “gold standard” CCSD(T)(coupled-cluster with single and double and per-

turbative triple excitations) method (18); correla-

tion effects of the core-core and core-valenceelectrons; and SO and counterpoise corrections

for the entrance channel. As a result, this presentPES corresponds to the SO ground state, ac-

curately describes the vdW regions both in theentrance and exit channels, and describes both

the abstraction (HCl + CH3) and substitution(H + CH3Cl) channels. Thus, it represents a con-

siderable advance over previous PESs for Cl +CH4, including several semiempirical (19, 20)

and ab initio–based surfaces in 2 (21), 3 (22),and 12(full) (23) dimensions.

A schematic of the PES, given in Fig. 1, showsthe stationary-point structures and energies of

the abstraction and substitution reactions Cl(2P3/2,2P) + CH4→HCl + CH3 and H + CH3Cl, where

Cl(2P3/2) is the SO ground state of the Cl atomlying below the non-SO energy of Cl(2P) by

0.8 kcal/mol. The vdWwells in the entrance chan-nel for the CH–Cl and HC–Cl linear bond ar-

rangements are below Cl(2P3/2) + CH4(eq) by 0.3and 0.6 kcal/mol, respectively. (That the most

favorable vdWorientation is theHC–Cl one is bothmeaningful and easily understood from a simple

sum-of-pairs interaction because C is more polar-izable than H.) The abstraction reaction has a late

(product-like) saddle-point of energy 7.6 kcal/molrelative to Cl(2P3/2) + CH4(eq) and a vdW well

in the product channel with a dissociation energy(De) of 2.4 kcal/mol. The PES reaction endoer-

gicity for the HCl + CH3 channel is 5.7 kcal/mol.These values are in excellent agreement with our

benchmark energies of 7.6, 2.4, and 6.0 kcal/mol,

respectively, which were obtained by using thefocal-point analysis (FPA) approach (14) [see com-

putational details in (24)]. The PES reaction en-doergicity is also in excellent agreement with the

experimental value, deduced to be 6.0 kcal/mol (13).None of the previous PESs are as accurate in

all of these key energies. In particular, focusingon the important entrance and exit channel vdW

wells, the present PES contains an accurate de-scription of the latter, (CH3—HCl), with a sub-

stantial De value of 2.4 kcal/mol. This contrastswith the conclusion based on a semiempirical PES

with a De of 0.3 kcal/mol that “the existence ofthis product complex is questionable” (20). With

respect to the entrance channel vdW well, thereare even greater contrasts with previous PESs.

This vdW region in particular is affected by SO

2P 0.8

2P3/2 0

H + CH3Cl

HCl + CH3

6.0 (5.7)

3.6 (3.3)

Cl + CH4

7.6 (7.6)

−−−−0.3

25.2 (26.0)

42.1 (43.0)

−−−−0.6

(CH3---HCl) vdW

(H--CH3--Cl)SP

(CH3--H--Cl)SP

Reaction coordinate

Rel

ativ

e en

erg

y (k

cal/m

ol)

Fig. 1. Schematic of the global SO and non-SO ground-state PESs of the Cl + CH4 reaction showing theaccurate benchmark energies and the PES values in parentheses; for example, 7.6 (7.6) shows theexcellent agreement between the benchmark and PES barrier heights. All the energies are relative tothe SO ground state Cl(2P3/2) + CH4(eq). The negative energies correspond to the attractive region ofthe vdW well in the entrance channel on the SO surface (Fig. 2).

3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0-0.5

0.0

0.5

1.0

1.5

2.0

2.5

E

SO3

SO2

SO1

A1

MRCI+QH

3CH---Cl (C

3v)

En

erg

y (k

cal/m

ol)

2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

E

A1

SO3

SO2

SO1

MRCI+QHCH

3---Cl (C

3v)

En

erg

y (k

cal/m

ol)

R(C---Cl) / Å

3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0

A1

SO1

H3CH---Cl (C

3v)

PES

2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0

SO1

A1

HCH3---Cl (C

3v)

PES

R(C---Cl) / Å

Fig. 2. Potential energy curves of CH4–Cl as a function of the C–Cl distance along the C3v axis withfixed CH4(eq) geometry and (top) CH–Cl or (bottom) HC–Cl linear bond arrangements that were (left)computed at the MRCI+Q/aug-cc-pVTZ level or (right) obtained from the non-SO– and SO-correctedground-state PESs. A1 and E denote the ground and excited non-SO electronic states, respectively, andSO1, SO2, SO3 are the three SO states. The energies are relative to Cl(2P3/2) + CH4(eq).

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org344

REPORTS

interaction, as shown in Fig. 2, where potentialcurves both with and without SO coupling are

shown for the entrance channel. Our multi-reference configuration interaction results show

that the non-SO electronic ground state potentialhas minima with CH–Cl and HC–Cl bond ar-

rangements with depths of 0.3 kcal/mol and 0.9

kcal/mol, respectively (25). The SO interactionhas a minor effect on the former, whereas it de-

creases the depth of the latter by 0.3 kcal/mol;but, the HC–Cl orientation still remains the deeper

minimum. It is critical that the present PESdescribes these vdWregions accurately, as shown

in Fig. 2, because they play an important role in

the low Ecoll dynamics. [Additional comparisonsbetween the PES and benchmark properties, in-

cluding data for the substitution channel, arepresented in (13).]

Having shown the high accuracy of the PESby comparison with benchmark ab initio data, we

applied it to simulations of the collision dynamicsof the title reaction. We performed more than 2

million quasi-classical trajectories (QCTs) for thereactions of Cl(2P3/2) with CH4(v = 0), CHD3(v =

0), and CHD3(vk = 1) [k = 1, 3, 6, 5], where v = 0,v1 = 1, and vk = 1 [k = 3, 6, 5] denote the vi-

brational ground state, CH-stretch, and three dif-ferent bend excitations, respectively. Details of

the QCT calculations are given in (13). Classical

zero-point leak and vibrational energy relaxationfrom vk = 1was investigated thoroughly for CHD3

and found not to be a serious issue, as alreadyreported in (8, 11) and discussed further in (13).

All the results presented below correspond tothe SO-corrected PES. Computations on the non-

SO PES show that the inclusion of the SO cor-rection in the PES decreases the cross sections by

Fig. 3. Computed and experimental HClrotational distributions for the Cl + CH4 re-action at a collision energy of 3.7 kcal/mol.Theory uses quasi-classical trajectory calcu-lations on the SO-corrected PES consideringtrajectories in which CH3 has at least zero-point vibrational energy. Experimental dataare taken from (28). On the basis of analysisof two batches of trajectories, the estimatedstatistical uncertainty of the computedresults is less than 15%.

Fig. 4. (A) Computed cross sections for the groundstate (v = 0) and reactant CH-stretch (v1 = 1) andbend (vk = 1) [k = 3, 6, 5] excited Cl + CHD3(v) →HCl + CD3 reactions and (B) their ratios as a func-tion of Ecoll. (C) Cross section ratios at equivalentamount of total energy (Etot = Ecoll + Ev), where thevibrational energies (Ev) are 0, 8.6, and 3.0 kcal/mol(relative to zero-point energy) for the ground stateand stretch- and bend-excited reactions, respec-tively. (Right) Snapshots of a nonreactive Cl +CHD3(v1 = 1) trajectory illustrating the stereo-dynamics in the vdW region causing the unexpectedsv/sg < 1 ratio at Etot = 9.6 kcal/mol, as seen in (C).

Cl + CHD3(v1=1)Ecoll = 1.0 kcal/mol

0 2 4 6 8 10 12 14 16 18 200

10

20

30

40

50

60

0 2 4 6 8 10 12 14 16 18 200

1

2

3

4

v6=1

v3=1

v5=1

v6=1v

3=1v

5=1

v1=1

Ecoll

(kcal/mol)

σ v / σ

g

0 2 4 6 8 10 12 14 16 18 200

1

2

3

4

5

v3=1

v5=1

v6=1

v1=1

v=0

Cl + CHD3(v) −−>−−>−−>−−> HCl + CD

3

Ecoll

(kcal/mol)

σ (b

ohr2 )

6 8 10 12 14 16 18 200.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

v3=1

v6=1

v5=1

v1=1

σ v / σ

g

Etot

(kcal/mol)

A

B

C

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 345

REPORTS

a factor of 1.5 to 2.5 at low Ecoll because of theincrease in the barrier height, but it has no sub-

stantial effect on the final state distributions.First, consider the HCl(v = 0, J ) rotational

distributions for the Cl + CH4(v = 0) reaction.These were reported by three experimental groups

showing extremely cold rotational populations(26–28). Theoretical simulations have been strug-

gling to reproduce this rotational distribution formany years; the previous work overestimates the

rotational temperature of the HCl product en-semble (19, 23, 29). In Fig. 3, we present QCT

results obtained by using the present PES and, asseen, the agreement between theory and experi-

ment (27, 28) is excellent. We also computed the

HCl rotational distributions for Cl + CHD3(v1 = 1)showing cold rotational distribution of the

stretch-excited product HCl(v = 1, J ), similarto HCl(v = 0, J ) from Cl + CH4(v = 0), and sub-

stantially hotter rotational temperature of thevibrationally ground state HCl in qualitative

agreement with an older experiment (which hadlarge uncertainties) (30).

Next, we considered the effect of vibrationalexcitation inCHD3,mentioned already. ThePolanyi

rules (3) state that for late-barrier (atom + diatom)reactions, the reactant vibrational energy is more

efficient than is the translational energy in sur-mounting the barrier. However, as noted already

a recent crossed molecular beam experiment (5)found that this picture could not be simply ex-

tended for the Cl + CHD3 reaction. To investigatethis finding, we calculated cross sections for the

reactant ground state and bend and CH-stretch–excited Cl + CHD3(vk) → HCl + CD3 reactions

and show the results in Fig. 4 as a function ofboth Ecoll and total energy (Etot). At the same

value of Ecoll, we see that all the bending andespecially the CH-stretch excitations enhance the

reaction relative to Cl + CHD3(v = 0) (Fig. 4, Aand B). Thus, in this sense there is enhancement

of the reaction by excitation of these modes.However, as noted in the experimental study, at

the same Etot translational energy is more effectivethan the excitation of the reactive CH-stretch or

bend at low Ecoll. As the Ecoll increases, the in-tuitively expected enhancement of reactivity is

seen upon vibrational excitation. This is best seen

by plotting cross section ratios (sv/sg) as a func-tion of Etot (Fig. 4C). As seen, sv/sg is less than

1 if the Etot is below 11 and (15, 15, 9) kcal/molfor the CH-stretch and (v3, v6, v5) bendingmodes,

respectively. OnlyCD3(v= 0) products were probedexperimentally, whereas theory shows the total re-

activity. Correlating the QCT cross sections toCD3(v = 0) results in a decrease of the sCH/sgroundratio, improving the agreement with the measureddata (fig. S1). Furthermore, the experiment applied

thermal bending excitation; thus, the measuredbending cross sections show the average effect

of the three bending modes (5). Theory predictsthat the v5(e) (CD3 deformation) bending mode

is the most efficient to drive the reaction (Fig.4B), similar to F + CHD3. Overall, both theory and

experiment show that the same amount of total

energy distributed among different nuclear motionshas different effects on chemical reactivity.

In order to gain insight into these results, weexamined the trajectories for the ground and stretch-

excited CHD3 at the sameEtot of 9.6 kcal/mol, thuscorresponding to a low Ecoll for CHD3(v1 = 1). We

determined the distributions of the smallest H-Cland C-Cl distances for nonreactive trajectories

and, as shown in fig. S2, most of the trajectoriesfor the CH-stretch–excited reaction do not reach

the transition state. Instead, the Cl atom turns backat the vdW region in the rC-Cl = 2.9 to 3.5 Å range.

This occurs because at low Ecoll, the CHD3 rotatesto the energetically favorable, but nonreactive,

H-C–Cl orientation (Fig. 4, right). Thus, at lowEcoll

the entrance channel vdWwell orients the reactantsin an unreactive configuration. At higher collision

energies, this effect is diminished, and the expectedenhancement of the reaction for the stretch-excited

CHD3 (over translational energy) is seen.Next, we considered the experimental results

on the vibrational distribution ofHCl (5). As notedin (5), vibrationally adiabatic theory predicts that

the ground state and CH-stretch–excited Cl +CHD3 reactions produce exclusively HCl(v = 0)

and HCl(v = 1) products, respectively. Experi-ment found the breakdown of this simple theory

for the excited reaction because the measuredfraction of HCl(v = 1), correlated to CD3(v = 0),

was only 45% (5). Our dynamics calculations showthat at higherEcoll, above the energetic threshold for

HCl(v = 1), the ground-state reaction still producesmainlyHCl(v = 0), and the fraction of HCl(v = 1) is

only 1% with only a slight Ecoll dependence. Thissmall ratio increases to about 2% if the results are

correlated to CD3(v = 0), which is in quantitativeagreement with experiment (5). The reactant CH-

stretch excitation increases the fraction ofHCl(v=1)to 10% and 30 to 50% for all the CD3 states and

CD3(v = 0), respectively, which is again in goodagreement with the above-mentioned correlated

experiment (5). The computations support theexperimental observation: The ground state reaction

is vibrationally adiabatic, whereas the CH-stretchexcited reaction is nonadiabatic.

Last to be considered was the very recent ex-periment (6) on steric control of Cl +CHD3(v1 = 1).

We have performedQCTcalculations with aligned

CHD3(v1 = 1) in which the CH-stretch is parallelor perpendicular to the initial relative velocity

vector of the reactants. We found that the totalreactivity of H-abstraction is higher at parallel

alignment relative to that at perpendicular ori-entations, which is in agreement with experiment

(6). The trajectories show that the initial orien-tation is maintained while the Cl approaches

CHD3(v1 = 1), supporting the recent experiment(6). However, at the turning point the QCTs show

substantial energy transfer causing rotational ex-citation of CHD3, and the prealignment is not

conserved after the collision (Fig. 4, right).

References and Notes1. L. Che et al., Science 317, 1061 (2007).

2. D. Skouteris et al., Science 286, 1713 (1999).

3. J. C. Polanyi, Science 236, 680 (1987).

4. J. J. Lin, J. Zhou, W. Shiu, K. Liu, Science 300, 966 (2003).

5. S. Yan, Y.-T. Wu, B. Zhang, X.-F. Yue, K. Liu, Science

316, 1723 (2007).

6. F. Wang, J.-S. Lin, K. Liu, Science 331, 900 (2011).

7. W. Zhang, H. Kawamata, K. Liu, Science 325, 303 (2009).

8. G. Czakó, Q. Shuai, K. Liu, J. M. Bowman, J. Chem. Phys.

133, 131101 (2010).9. J. M. Bowman, G. Czakó, B. Fu, Phys. Chem. Chem. Phys.

13, 8094 (2011).10. G. Czakó, B. C. Shepler, B. J. Braams, J. M. Bowman,

J. Chem. Phys. 130, 084301 (2009).11. G. Czakó, J. M. Bowman, J. Am. Chem. Soc. 131, 17534 (2009).12. B. J. Braams, J. M. Bowman, Int. Rev. Phys. Chem. 28,

577 (2009).

13. Materials and methods are available as supporting

material on Science Online.

14. A. G. Császár, W. D. Allen, H. F. Schaefer, J. Chem. Phys.

108, 9751 (1998).

15. The composite energies were obtained as E [UCCSD(T)/

aug-cc-pVDZ] + E[AE-UMP2/aug-cc-pCVTZ] – E[UMP2/

aug-cc-pVDZ], where AE denotes correlating all the

electrons. For the entrance channel, counterpoise and

spin-orbit corrections were computed at the AE-UMP2/

aug-cc-pCVTZ and MRCI+Q/aug-cc-pVTZ levels of theory,

respectively. The PES was represented by a polynomial

expansion in yij = exp(−rij/a) (where a = 2 bohr), including

all terms up to total degree six. 3262 coefficients were

determined by a weighted linear least-squares fit of

roughly 16,000 energy points. The root mean square

(RMS)–fitting errors are 0.2, 0.4, and 1.0 kcal/mol for

energy intervals (0, 31), (31, 63), and (63, 143), respectively.

16. For 15 arbitrary geometries with energies in the wide

0 to 36 kcal/mol range, we found that this composite

method gives results comparable with extremely

high-quality all-electron CCSD(T)/aug-cc-pCVQZ

calculations with a RMS of only 0.4 kcal/mol, whereas

the RMS error of high-level CCSD(T)/aug-cc-pVTZ

calculations is 1.1 kcal/mol. Furthermore, the composite

method reduces the computational time by factors of

about 1000 and 5 relative to the above-mentioned

high-level computations, respectively.

17. K. A. Peterson, T. H. Dunning Jr., J. Chem. Phys. 117,

10548 (2002).

18. K. Raghavachari, G. W. Trucks, J. A. Pople, M. Head-Gordon,

Chem. Phys. Lett. 157, 479 (1989).

19. J. C. Corchado, D. G. Truhlar, J. Espinosa-García, J. Chem.

Phys. 112, 9375 (2000).

20. C. Rangel, M. Navarrete, J. C. Corchado, J. Espinosa-García,

J. Chem. Phys. 124, 124306 (2006).

21. S. T. Banks, D. C. Clary, Phys. Chem. Chem. Phys. 9, 933 (2007).22. H.-G. Yu, G. Nyman, J. Chem. Phys. 111, 6693 (1999).

23. J. F. Castillo, F. J. Aoiz, L. Bañares, J. Chem. Phys. 125,

124316 (2006).

24. The benchmark FPA study considers extrapolation to the

complete basis set limit using aug-cc-pVnZ [n = 5 and 6]

bases, electron correlation beyond CCSD(T), core correlation

effects, scalar relativistic effects, and spin-orbit corrections.

25. These relative energies are benchmarked at the

highly accurate CCSD(T)/aug-cc-pCVQZ level of theory

correlating all the electrons, including corrections

for the basis set superposition error.

26. W. R. Simpson, T. P. Rakitzis, S. A. Kandel, T. Lev-On,

R. N. Zare, J. Phys. Chem. 100, 7938 (1996).

27. D. F. Varley, P. J. Dagdigian, J. Phys. Chem. 99, 9843 (1995).

28. C. Murray, B. Retail, A. J. Orr-Ewing, Chem. Phys. 301,

239 (2004).

29. S. J. Greaves et al., Phys. Chem. Chem. Phys. 13, 11438 (2011).30. W. R. Simpson, T. P. Rakitzis, S. A. Kandel, A. J. Orr-Ewing,

R. N. Zare, J. Chem. Phys. 103, 7313 (1995).

Acknowledgments: G.C. thanks the National Science

Foundation (grant CHE-0625237), and J.M.B. thanks

the U.S. Department of Energy (grant DE-FG02-97ER14782)

for financial support.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/334/6054/343/DC1

Materials and Methods

Figs. S1 to S3

Tables S1 to S3

17 May 2011; accepted 9 September 2011

10.1126/science.1208514

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org346

REPORTS

800,000 Years of AbruptClimate VariabilityStephen Barker,1* Gregor Knorr,2 R. Lawrence Edwards,3 Frédéric Parrenin,4,5

Aaron E. Putnam,6 Luke C. Skinner,7 Eric Wolff,8 Martin Ziegler1

We constructed an 800,000-year synthetic record of Greenland climate variability based on thethermal bipolar seesaw model. Our Greenland analog reproduces much of the variability seen inthe Greenland ice cores over the past 100,000 years. The synthetic record shows strong similaritywith the absolutely dated speleothem record from China, allowing us to place ice core recordswithin an absolute timeframe for the past 400,000 years. Hence, it provides both a stratigraphicreference and a conceptual basis for assessing the long-term evolution of millennial-scale variabilityand its potential role in climate change at longer time scales. Indeed, we provide evidence for aubiquitous association between bipolar seesaw oscillations and glacial terminations throughout theMiddle to Late Pleistocene.

Ice core records from Greenland first demon-strated the existence of repeated, large, abrupt

shifts in Northern Hemisphere climate dur-ing the last ice age (1, 2). These shifts are one

expression of a global system that is capable ofdriving major changes in climate components

such as ocean temperatures (3, 4) and monsoonrainfall (5). The Greenland records provide an

archetypal view of abrupt climate variability (6)over the last glacial cycle, which was character-

ized by rapid alternations between cold (stadial)and warmer (interstadial) conditions [known as

Dansgaard-Oeschger (D-O) oscillations]. But iron-ically, the very high temporal resolution of these

records makes it difficult to look farther back intime; the high accumulation rates on the Green-

land ice sheet mean that more than 3000 m of

ice may represent just 100,000 years of climatehistory. Fortunately, climate records preserved in

Antarctic ice (7) enable us to address this funda-mental problem.

The thermal bipolar seesaw model (8, 9) at-tempts to explain the observed relationship between

millennial-scale temperature variability observedin Greenland and Antarctica by calling on var-

iations in the strength of the Atlantic meridionaloverturning circulation (AMOC). The north-

ward heat transport associatedwith this circulation(10) implies that changes in the strength of over-

turning should lead to opposing temperature re-sponses in either hemisphere. According to the

seesaw model, a transition from weak to strongAMOCwould cause an abrupt warming across the

North Atlantic region (a D-O warming event)while temperatures across Antarctica would (in

general) shift fromwarming to cooling. The ocean-atmosphere climate system is an integrated and

synergistic system, and it is important to note thatthe overall concept of the bipolar seesaw we in-

voke here is not restricted to oceanic processes butalso includes atmospheric shifts that may be re-

lated to the variationswe are interested in (9,11–14).According to the thermal bipolar seesaw

model, we should observe an antiphase relation-ship between the Greenland temperature anom-

aly and the rate of change of Antarctic temperature(9). This can be illustrated by a lead/lag analysis

of the methane-tuned temperature records afterremoval of their orbital time scale variability (6)

(Greenland, GLT_hi, and Antarctica, AAT_hi)

and the first time derivative of Antarctic temper-ature, AAT_hi´ (Fig. 1). Comparison of the undif-

ferentiated records illustrates the historical debateas to whether the two signals are positively cor-

related, with a southern lead of 1000 to 1600years, or negatively correlated, with the north lead-

ing by 400 to 800 years (15, 16). However, asimplied in (9) and illustrated in Fig. 1, a near

zero-phase anticorrelation is observed betweenGLT_hi and AAT_hi´. Uncertainties in the ice

age–gas age offset (Dage), which may be up tohundreds of years (17), mean that an exact anti-

phase relationship is unlikely to be observed (6).As described by (18) using a similar approach,

the process of differentiating amplifies noise inthe original temperature record. Smoothing the

record of AAT_hi before differentiating reducesthis noise but will compromise the ability of

AAT_hi´ to replicate the abrupt nature of D-Owarming events and reduce the predicted ampli-

tude of smaller events. The choice of smoothingwindow is therefore a trade-off between these ef-

fects (6) (Fig. 1).The empirical relations illustrated in Fig. 1 of-

fer the possibility of producing a synthetic recordof Greenland climate using the Antarctic record,

with the purpose of reconstructing the natureof northern variability beyond the present limit

of the Greenland records. The record of Green-

land temperature (GLT) is broken down into itsorbital and millennial time scale components,

GLT_lo and GLT_hi, respectively (Fig. 2 and fig.S5), where GLT_lo is a 7000-year smoothing of

GLT (6) and GLT_hi is the difference betweenGLT and GLT_lo. We consider GLT_hi as the north-

ern temperature anomaly with respect to meanbackground conditions. Building on Fig. 1, we

assume that the rate of Antarctic temperaturechange is inversely proportional to the northern

temperature anomaly. We therefore scale the am-plitude of AAT_hi´ to match that of GLT_hi to

produce a synthetic record of northern millennial-scale temperature variability, GLT_syn_hi (Fig. 2C)

(6). It can be seen that a synthetic reconstruction

of GLT could be made by combining GLT_syn_hiwith an estimate for GLT_lo. The orbital time

scale components of the Greenland and Ant-arctic temperature records (GLT_lo and AAT_lo,

respectively) are highly correlated, with the south-ern record leading the north by ~2000 years

(6). We therefore incorporate longer time scalevariations into our reconstruction by substitut-

ing GLT_lo with a scaled version of AAT_lo,shifted by 2000 years (which we call GLT_syn_lo)

(fig. S5) (6). Our full reconstruction, GLT_syn(Fig. 2D), is then the sum of GLT_syn_lo and

GLT_syn_hi.Our formulation of the thermal bipolar see-

saw concept is qualitatively analogous to that of(9) in that it implies the existence of a heat reser-

voir that convolves the northern signal, produc-ing a southern signal with a longer characteristic

time scale. Our approach is slightly different inthat we relate the rate of Antarctic temperature

change directly to the northern temperature anom-aly. Indeed, we note that for some long stadial

events, particularly those associated with gla-cial terminations, Antarctic temperatures appear to

rise unabated until an abrupt warming event oc-curs in the north (19). On the other hand, our

formulation does not imply that Antarctic temper-atures must continue to rise indefinitely when-

ever Greenland is cold, only while it is cold withrespect to background conditions (defined by

the orbital time scale component). We also notethat northern temperature (regardless of back-

ground conditions) is not always constant through-

out stadial events. For example, Greenland warmedsignificantly during cold stadial 21 (Fig. 2D).

By our formulation, the rate of Antarctic tem-perature rise during this event would decrease

correspondingly, in line with observations (Fig.2A) (20).

We used a thresholding approach for predict-ing the occurrence of abrupt Greenland warming

events based on minima in the second time dif-ferential of AAT (AAT´´) (Fig. 2, E and F). This

has an advantage over use of the first differential(decreasing through zero) because it is capable of

distinguishing between events of varying mag-nitude and incorporates information about con-

ditions before and after an abrupt event (6). Using arelatively insensitive threshold (blue dashed line in

Fig. 2F), we are able to identify the largest D-O

1School of Earth and Ocean Sciences, Cardiff University, CardiffCF10 3AT, UK. 2Alfred Wegener Institute, 27570 Bremerhaven,Germany. 3Department of Earth Sciences, University of Min-nesota, Minneapolis, MN 55455, USA. 4Laboratoire de Glaciologie,CNRS and Joseph Fourier University, 38400 Grenoble, France.5Laboratoire Chrono-Environnement, 25000 Besançon, France.6Department of Earth Sciences and Climate Change Institute,University of Maine, Orono, ME 04469, USA. 7Godwin Lab-oratory for Palaeoclimate Research, Department of EarthSciences, University of Cambridge, Cambridge CB2 3EQ, UK.8British Antarctic Survey, Madingley Road, High Cross,Cambridge CB3 0ET, UK.

*To whom correspondence should be addressed. E-mail:barkers3@cf.ac.uk

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 347

REPORTS

temperature shifts recorded in Greenland, whereas“smaller” events, such as D-O 2, require a more

sensitive threshold (red line). Accordingly, we areable to identify almost all of the canonical D-O

events over the past 90,000 years without intro-ducing “spurious events.” We incorporate a cor-

rection, based on the time integrated per 55-cmsample of ice (7), to account for loss of temporal

resolution in the deeper parts of the ice core (6).Our synthetic reconstruction of Greenland tem-

perature closely resembles the observed recordin terms of both its timing and the structure of in-

dividual events (Fig. 2). This is despite probablevariability in the relationships between d18O [for

the Greenland Ice Sheet Project 2 ice core (GISP2)]

or dD [for the European Project for Ice Coringin Antarctica Dome C ice core (EDC)] versus

local temperature through time (21) and othermillennial-scale variability that might not be re-

lated to the bipolar seesaw. We find similar resultsusing alternative Antarctic ice core records (6).

On the basis of the predictive ability of GLT_synover the past ~100,000 years, we can extend our

reconstruction back to ~800,000 years ago (Fig.3 and Fig. 4). In doing so we implicitly assume

that the empirical relationships observed over

the last glacial cycle held during earlier periods.Given the inherent uncertainty in this assump-

tion, we do not claim particular skill at pre-dicting the absolute amplitude of earlier events;

however, we do suggest that to the extent thatthe underlying physical mechanisms did persist

throughout the past 800,000 years, the timingand overall structure of events will be relatively

robust.To test this hypothesis, we compared our syn-

thetic reconstructions with real climate records.The record of Asian monsoon variability derived

from cave deposits (speleothems) in China (5, 22)is one of the best candidates for this task (Fig. 3).

The Chinese speleothem d18O record is thought

to represent changes in the proportion of lowd18O (summer monsoon) rainfall within annual

totals (22) and can be considered a measure ofthe amount of summer monsoon rainfall or mon-

soon intensity. The combined record from severaldeposits taken from a number of caves provides

a continuous, absolutely dated record over thepast ~400,000 years (22). The record is domi-

nated by orbital time scale changes, possibly re-lated to the influence of boreal summer insolation

on the strength of the Asian monsoon (6). Re-

moval of this variability by normalizing to theinsolation curve (6) reveals distinctive millennial

time scale activity that has been shown to cor-respond with D-O variability over Greenland

during the last glacial cycle (5, 22). This corre-spondence is thought to be caused by latitudinal

shifts in the position of the Intertropical Con-vergence Zone (ITCZ) and related atmospher-

ic phenomena in response to variations in theAMOC and related changes in North Atlantic

temperature (11).There is a strong one-to-one correspondence

between inferred weak-monsoon events and ourreconstructed cold events in Greenland (Fig. 3).

Moreover, there are pronounced similarities in

the structure of abrupt events, particularly duringdeglacial episodes (terminations); the multiple

weak-monsoon events associated with glacialterminations of the Late Pleistocene (22) are re-

flected by multiple cold events in our records.Our reconstruction suggests the occurrence of

large-amplitude “D-O–type” oscillations between160,000 and 180,000 years ago [during marine

isotope stage 6 (MIS 6)] (Fig. 3). These may becompared with similar events in the records of

planktonic d18O and tree pollen from a marine

3 2 1 0 1 2 3−0.2

0

0.2

0.4

0.6

0.8

1

Lag (kyr)

North leads South leadsB

Methane

Cor

rela

tion

Cor

rela

tion

North leads South leads

3 2 1 0 1 2 3−0.8

−0.4

0.4

0.8

Lag (kyr)

0

1000yr700yr200yr

C

Vs AAT_hi

-10

0

10

30 40 50 60 70 80AA

T_h

i (‰

)

Age (kyr) (EDC3)

-0.05

0

0.05

-0.01

0

0.01

-0.01

0

0.01

-2

0

2

GL

T_h

i (‰

)

-100

0

100

ED

C C

H4

(ppb

v)

-100

0

100

GIS

P C

H4

(ppb

v)

A

Fig. 1. Ice core records from Greenland (GISP2) and Antarctica (EDC) (all records have orbital timescale variability removed; kyr, thousands of years). (A) Methane (17, 25) with tuning points (crosses)used to place all records on the EDC3 age model (26), Greenland temperature (GLT_hi with 200-yearsmoothing) derived from d

18O of ice (2), and Antarctic temperature (AAT_hi, 200-year smoothing)from dD of ice (7). First derivatives of AAT_hi (AAT_hi´) are shown for various smoothing lengths (in brackets) of the undifferentiated record. (B) Lead/lagcorrelation of the methane records suggests successful tuning of the gas records. (C) Lead/lag correlations between GLT_hi, AAT_hi, and AAT_hi´ reveal thewell-known relation between northern and southern temperature records and the antiphase relationship between GLT_hi and AAT_hi´.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org348

REPORTS

Fig. 2. Reconstructing millennial-scale climatevariability over Greenland using the Antarctic tem-perature record. (A) The record of dD from the EDCice core (AAT) (7) with a 7000-year smoothing ofthe same record (AAT_lo). (B) Removal of orbitaltime scale variability and application of a 700-year smoothing (6) produces AAT_hi. (C) AAT_hi isdifferentiated and then scaled to GLT_hi (greencurve) to produce GLT_syn_hi (orange curve). (D)A synthetic reconstruction of Greenland temper-ature variability (GLT_syn; red curve) constructedby adding GLT_syn_hi to GLT_syn_lo (6). Greencurve is GISP2 d

18O placed on EDC3 via methanetuning (Fig. 1). (E and F) Minima in AAT´´ below athreshold (dashed lines) are used to predict theoccurrence of major warming events in Greenland(F), identified by the corresponding colored dots in(E). A threshold value of –1.2 × 10−5 (red dashedline and dots) has good success at picking canon-ical D-O events [green numbers in (D) and (E)]without introducing spurious events.

1

2

34 5 6 7

8

910

1112

13

141516

17

18

19 20

21

22

0

1 2 3 4 5 6 7 8 91011 1213 14151617 18 19 20 21 220

10 20 30 40 50 60 70 80 90Age kyr (EDC3)

Age kyr (EDC3)10 20 30 40 50 60 70 80 90

GLT

E-44

-40

-36

GL

T (

δ 18

O, ‰

)

D

GLT_syn

-2x10-5

0

2x10-5

F

-10

0

10B

AA

T_h

i

-450

-430

-410

-390

ED

C δ

D (

‰)

AAAT

AAT_lo

-4

-2

0

2

4

C

GLT_hiGLT_syn_hi

GL

T _

hi

Fig. 3. (A to H) Compari-son of reconstructed Green-land climate variability withother records. The normal-ized record of monsoonvariability from China (A)(5, 22), marine records fromthe Iberian Margin [(C)to (E)] (23, 32), and therecord of atmospheric CH4(G) (25) share many fea-tures in common with ourrecords derived from theAntarctic temperaturerecord[(B) and (F)]. Colored dotsin (H) represent the oc-currence of D-O eventspredicted from AAT´´ usinga fixed (red) or variable(blue) threshold (6). Allrecords are on the EDC3time scale (26) except themonsoon record which ison its own absolute timescale (22). Thepollen recordfrom MD95-2042 (32) (C)was placed on EDC3 by tun-ing the corresponding plank-tonic d18O record (24) toGLT_recon. Gray bars indi-cate cold conditions andperiods of weak monsoon.Glacial terminations are indicated by Roman numerals.

400

500

600

700

CH

4 (p

pbv)G-40

-35

-30

GL

T_s

yn (

‰) F

120 140 160 180 200 220 240 260Age (kyr)

0

20

40

Tre

e po

llen

(%)

E

20

40

60

Med

iterr

anea

npo

llen

(%)

D

1

2

3

plan

kton

ic δ

18O

(‰

)

C-4

0

4

GL T

_syn

_hi (

‰)

B

Spe

leot

hem

δ18

O(n

orm

aliz

ed) A

H

T-II T-III

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 349

REPORTS

sediment core taken from the Iberian Margin (23).On the basis of the findings of Shackleton et al.

(24), the Iberian Margin records were tuned tothe EDC dD record via the record of benthic

d18O from the same core (23). The tuning exer-cise did not involve the surface records, which

therefore provide a quasi-objective “target” forcomparing with our reconstruction [which should

be aligned with the surface-ocean records accord-ing to (24)], and there is good agreement in the

timing and structure of the abrupt events during

MIS 6. We also note good agreement betweenour reconstructions and the record of atmospheric

CH4 (25). Our predicted D-O warming events aregenerally aligned with sharp increases in CH4

(similar to the observed relationship during MIS3). This relationship holds for the entire 800,000-

year record (Fig. 4) and provides critical ground-truthing for our reconstruction.

Building on previous studies (22), we usedthe precise and absolutely dated Chinese speleo-

them record to place our reconstruction on an

absolute time scale for the past 400,000 years.We did this by aligning the cold events in our

reconstruction with the inferred weak-monsoonevents in the speleothem record (Fig. 4) (6). The

EDC3 age scale (which remains the fundamen-tal basis for our model) was derived through a

combination of ice flow modeling and variousage markers, including orbital tuning constraints

(26). By tuning the millennial-scale features ofGLT_syn to the speleothem record, we provide

a refinement of the age scale that provides an

Fig. 4. (A and B) 800,000years of abrupt climate varia-bility. Records of North AtlanticIRD (4), monsoon rainfall (5, 22)(normalized) and SST fromthe Iberian Margin (27) allshow strong similarities withour reconstruction of Greenlandclimate variability (GLT_syn_hiand GLT_syn). Glacial termi-nations (identified by Romannumerals) are characterized bycold conditions across Green-land and the North Atlanticand weakened monsoon rain-fall, with a corresponding risein atmospheric CO2 (33), fol-lowed by an abrupt warmingover Greenland, strengtheningof the monsoon, and sharp risein atmospheric CH4 (25). Pinkboxes indicate terminal North-ern Hemisphere cold periods.Red and blue dots are predictedD-O warming events using afixed or variable threshold, re-spectively. Lowermost curves ineach panel are moving win-dows of the standard devia-tion of AAT_hi´, our ”bipolarseesaw activity index“ (notethat orbital time scale varia-tions have been removed; blueis 5000-year window; green is10,000-year window). Increasedmillennial-scale activity is gen-erally observed during transi-tions between climate states,with minimal activity duringinterglacials and glacial max-ima. All records are on the newSpeleo-Age (A) or the EDC3(B) time scale except the ben-thic d18O stack (LR04) of (34),which is on its own time scale.

TI TII TIIIA TIII TIV

TV TVI TVII

400

600

800

CH

4 (p

pbv)

-45

-40

-35

-30

GL T

_syn

(‰

)

-4

0

4

GL T

_syn

_hi

(‰)

AA

T_h

i’ S

D (

x10-3

)

0

5

10

3

4

5 δ18O

(LR

04)

400 450 500 550 600 650 700 750 800Age (kyr) (EDC3)

B

200

300

CO

2 (p

pmv)

400

600

800

CH

4 (p

pbv)

5

10

15

20

Iber

ian

Mar

gin

SS

T (

ºC)-45

-40

-35

-30

GL T

_syn

(‰

)

-4

0

4

GL T

_syn

_hi

(‰)

Spe

leot

hem

δ18

O(n

orm

aliz

ed)

0

50

100

Nor

th A

tlant

icIR

D (

%)

Age (kyr) ‘Speleo-Age’0 50 100 150 200 250 300 350 400

A

3

4

5δ18O

(LR

04)

AA

T_h

i’ S

D (

x10-3

)

0

5

10

200

300

CO

2 (p

pmv)

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org350

REPORTS

alternative to the ultimate dependence on orbitaltuning. In addition to providing an absolute time

scale for the ice and gas records from Antarctica,we can also use our absolutely dated Green-

land reconstruction as a tuning target for otherhigh-resolution paleo-records, such as records

of ice-rafted debris (IRD) from a North Atlanticsediment core (4) and a record of sea surface

temperature (SST) from a core off the IberianMargin (27) (Fig. 4). Each of these records has

been tuned to our reconstruction on its absolutetime scale (6).

Our synthetic records confirm that millennialtime scale variability and abrupt climate oscilla-

tions occurred in Greenland throughout the past

800,000 years, and more specifically they sug-gest that the underlying physical mechanisms rep-

resented by the conceptual thermal bipolar seesawwere relatively invariant throughout this period.

In line with observations for the last glacial pe-riod (28), our reconstructions suggest that higher-

amplitude variability and more frequent D-O–likewarming events occurred when climate was in

an intermediate state or during the transitionsbetween states (Fig. 4). Extending the observa-

tions of (22), we find that glacial terminationsof the Middle to Late Pleistocene in general

were characterized by oscillations of the bipolarseesaw.

This apparently ubiquitous association ofmillennial-scale climate variability with glacial

terminations raises an important question: Is thismode of variability a necessary component of de-

glacial climate change, or merely a complicatingfactor? Previous studies (28, 29) have suggested

that D-O–type variability might represent an in-herent resonance of the climate system, attaining

a high amplitude only within certain windowsof opportunity (i.e., intermediate climate states).

Given that global climate must pass through such

a window during deglaciation, one could arguethat terminal oscillations of the bipolar seesaw are

merely a symptom of deglacial climate change(29). However, the precise correspondence observed

between bipolar seesaw oscillations and changesin atmospheric CO2 during glacial terminations

(Fig. 4) suggests that the bipolar seesaw may playmore than just a passive role in the mechanism

of deglaciation (i.e., through the positive feedbacksassociated with increasing CO2) (14, 19, 22). With

the supercritical size of continental ice sheets asa possible precondition (30), and in combination

with the right insolation forcing (31) and ice al-bedo feedbacks, the CO2 rise associated with an

oscillation of the bipolar seesaw could provide

the necessary additional forcing to promote de-glaciation. In this sense, the overall mechanism

of glacial termination during the Middle to LatePleistocene might be viewed as the timely and

necessary interaction between millennial and or-bital time scale variations.

References and Notes1. W. Dansgaard et al., Science 218, 1273 (1982).

2. M. Stuiver, P. M. Grootes, Quat. Res. 53, 277

(2000).

3. G. Bond et al., Nature 365, 143 (1993).

4. J. F. McManus, D. W. Oppo, J. L. Cullen, Science 283, 971

(1999).

5. Y. J. Wang et al., Nature 451, 1090 (2008).

6. See supporting material on Science Online.

7. J. Jouzel et al., Science 317, 793 (2007).

8. W. S. Broecker, Paleoceanography 13, 119 (1998).

9. T. F. Stocker, S. J. Johnsen, Paleoceanography 18, 1087

(2003).

10. K. E. Trenberth, J. M. Caron, J. Clim. 14, 3433

(2001).

11. M. Vellinga, R. A. Wood, Clim. Change 54, 251

(2002).

12. J. C. H. Chiang, M. Biasutti, D. S. Battisti,

Paleoceanography 18, 1094 (2003).

13. R. F. Anderson et al., Science 323, 1443 (2009).

14. S. Barker et al., Nature 457, 1097 (2009).

15. A. Schmittner, O. A. Saenko, A. J. Weaver, Quat. Sci. Rev.

22, 659 (2003).

16. E. J. Steig, R. B. Alley, Ann. Glaciol. 35, 451

(2002).

17. T. Blunier, E. J. Brook, Science 291, 109 (2001).

18. M. Siddall et al., Quat. Sci. Rev. 25, 3185 (2006).

19. E. W. Wolff, H. Fischer, R. Rothlisberger, Nat. Geosci. 2,

206 (2009).

20. E. Capron et al., Quat. Sci. Rev. 29, 222 (2010).

21. E. Capron et al., Clim. Past 6, 345 (2010).

22. H. Cheng et al., Science 326, 248 (2009).

23. V. Margari et al., Nat. Geosci. 3, 127 (2010).

24. N. J. Shackleton, M. A. Hall, E. Vincent,

Paleoceanography 15, 565 (2000).

25. L. Loulergue et al., Nature 453, 383 (2008).

26. F. Parrenin et al., Clim. Past 3, 485 (2007).

27. B. Martrat et al., Science 317, 502 (2007).

28. M. Schulz, W. H. Berger, M. Sarnthein, P. M. Grootes,

Geophys. Res. Lett. 26, 3385 (1999).

29. A. Sima, A. Paul, M. Schulz, Earth Planet. Sci. Lett. 222,

741 (2004).

30. M. E. Raymo, Paleoceanography 12, 577 (1997).

31. J. D. Hays, J. Imbrie, N. J. Shackleton, Science 194, 1121

(1976).

32. M. F. Sánchez Goñi, F. Eynaud, J. L. Turon, N. J. Shackleton,

Earth Planet. Sci. Lett. 171, 123 (1999).

33. D. Lüthi et al., Nature 453, 379 (2008).

34. L. E. Lisiecki, M. E. Raymo, Paleoceanography 20,

PA1003 (2005).

Acknowledgments: We thank the authors of all of the

studies cited here for making their results available for

this work. Supported by a Philip Leverhulme Prize (S.B.),

Natural Environment Research Council (UK) awards

NE/F002734/1 and NE/G004021/1 (S.B.), and NSF grants

0502535 and 1103403 (R.L.E.). This study is also part

of the British Antarctic Survey Polar Science for Planet

Earth Programme, funded by the Natural Environment

Research Council (UK).

Supporting Online Materialwww.sciencemag.org/cgi/content/full/science.1203580/DC1

Materials and Methods

Figs. S1 to S14

Tables S1 to S3

References

31 January 2011; accepted 26 August 2011

Published online 8 September 2011;

10.1126/science.1203580

Pre-Clovis Mastodon Hunting 13,800Years Ago at the Manis Site, WashingtonMichael R. Waters,1* Thomas W. Stafford Jr.,2,5 H. Gregory McDonald,3 Carl Gustafson,4

Morten Rasmussen,5 Enrico Cappellini,5 Jesper V. Olsen,6 Damian Szklarczyk,6 Lars Juhl Jensen,6

M. Thomas P. Gilbert,5 Eske Willerslev5

The tip of a projectile point made of mastodon bone is embedded in a rib of a single disarticulatedmastodon at the Manis site in the state of Washington. Radiocarbon dating and DNA analysisshow that the rib is associated with the other remains and dates to 13,800 years ago. Thus, osseousprojectile points, common to the Beringian Upper Paleolithic and Clovis, were made and usedduring pre-Clovis times in North America. The Manis site, combined with evidence of mammothhunting at sites in Wisconsin, provides evidence that people were hunting proboscideans atleast two millennia before Clovis.

Recent studies have strengthened the casethat the makers of Clovis projectile points

were not the first people to occupy theAmericas (1–5). If hunting by humans was re-

sponsible for the megafauna extinction at the

end of the Pleistocene (6), hunting pressuresmust have begun millennia before Clovis (7).

Here we reexamine the evidence from the Manissite in the state of Washington (8), an early mas-

todon kill that dates to 800 years before Clovis.

Between 1977 and 1979, a single male mas-

todon (Mammut americanum) was excavated fromsediments at the base of a kettle pond at the

Manis site (figs. S1 to S3) (8–10). Some boneswere spirally fractured, multiple flakes were re-

moved from one long bone fragment, and other

bones showed cut marks (8, 11, 12). The onlydocumented artifact associated with the masto-

don was a foreign osseous fragment, interpretedas the tip of a bone or antler projectile point,

1Center for the Study of the First Americans, Departments ofAnthropology and Geography, Texas A&M University, 4352TAMU, College Station, TX 77843–4352, USA. 2Stafford Re-search, 200 Acadia Avenue, Lafayette, CO 80026–1845, USA.3Park Museum Management Program, National Park Service,1201 Oakridge Drive, Suite 150, Fort Collins, CO 80525, USA.4245 Southeast Derby Street, Pullman, WA 99163–2217, USA.5Centre for GeoGenetics, University of Copenhagen, ØsterVoldgade 5-7, 1350 Copenhagen, Denmark, 6Novo NordiskFoundation Center for Protein Research, Faculty of Health Sci-ences, University of Copenhagen, Blegdamsvej 3b, 2200 Co-penhagen, Denmark.

*To whom correspondence should be addressed. E-mail:mwaters@tamu.edu

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 351

REPORTS

embedded in a rib fragment that was recoveredex situ from sediments excavated when a back-

hoe uncovered the bone bed (Fig. 1 and fig. S4)(8). Organic matter associated with the masto-

don yielded calibrated radiocarbon ages of ~14thousand years ago (ka) (8, 10) (table S1). Over

the past 35 years, the age and evidence for hu-man involvement with the Manis mastodon have

been challenged (13).We obtained 13 accelerator mass spectrom-

etry (AMS) 14C dates from purified bone collagen(4) extracted from the mastodon rib containing

the embedded osseous object and from bothtusks (table S2). All dates were statistically iden-

tical at 1 SD and establish an age of 11,960 T

17 14C years before the present (yr B.P.) for theManis mastodon (Table 1; average of four XAD

fractions; 13,860 to 13,765 calendar yr B.P.) (14).These dates show that the ex situ mastodon rib

and in situ skeleton are contemporaneous.High-resolution x-ray computed tomography

(CT) scanning (15) revealed that the osseousobject embedded in the rib is dense bone shaped

to a point (Fig. 1 and movies S1 and S2). Thepoint penetrated 2.15 cm into the rib; the tip

broke after entering the rib and separated fromthe main shaft. The combined length of the point

fragment (tip length plus the length of the em-bedded and external shaft piece) is 3.5 cm.

The rib with the embedded projectile point isa right 12th, 13th, or 14th rib in a series of 19,

but most likely the 14th rib (Fig. 2). The projec-tile point entered the dorsal surface of the prox-

imal end of the rib immediately distal to the lateralmargins of the two articular facets at approxi-

mately a 45° angle relative to the axis of the headof the rib. The point would have penetrated the

hair and skin and about 25 to 30 cm of super-ficial epaxial muscles (Fig. 2 and fig. S5). Thus it

was at least 27 to 32 cm long, comparable withthe known length of later, Clovis-age thrown and

thrust bone points (16). There is no evidence ofbone growth around the point, indicating that the

mastodon died soon after it was attacked.DNA and protein sequencing were under-

taken on the rib and bone point (supporting on-line material text 4 and 5). Attempts to amplify

a 140–base pair (bp) fragment of the 16S mito-

chondrial DNA (mtDNA) from the rib usinguniversal vertebrate primers (17) produced only

modern (human) contamination. However, re-designing primers for a 69-bp fragment (including

primers, table S8) of D-loop mtDNA producedsequences from both the rib and bone point that

were identical to mastodon and distinct fromother proboscideans (mammoth or elephant) by

nine substitutions.We also obtained high-resolution tandem mass

spectrometry (MS/MS)–based protein sequencesfrom the projectile point and rib, and used another

mastodon sample as a second reference (tablesS3 to S6). The MS/MS spectra from the bone

point matched the reconstructed mastodon col-lagen sequences, with the highest scores being

within a reference set of collagen sequences (table

S7 and supporting table of bone point marker

peptides). These results and controls show thatthe point was fashioned from mastodon bone.

The Manis site provides further evidence ofa human presence in the New World 800 years

before Clovis [13 ka (4)] and shows that people

were hunting with mastodon bone weaponsmade from earlier kills. Evidence for pre-Clovis

hunting also comes from the 14.2-ka Schaefersite and 14.8-ka Hebior site, Wisconsin (18, 19),

where stone artifacts, but no projectile points,were found with the remains of mammoth (Mam-

muthus primigenius). Additional evidence of mega-fauna hunting comes from sites where artifacts

are absent, but taphonomic evidence suggests hu-man butchering, such as at the 13.8-ka Ayer Pond

site (45SJ454), Orcas Island, Washington (20).Studies of the dung fungal spore Sporormiella

from lakes in Indiana and New York imply that

megafauna populations collapsed there between

14.8 and 13.7 ka (7). Thus, the impact of humanhunters on the North American megafauna was

more prolonged than previously hypothesized andwas not a “Clovis blitzkrieg” (21). The absence

of stone projectile points at Manis, Hebior,

Schaefer, and Orcas Island and the presence ofan osseous projectile point at Manis suggest that

osseous projectile points may have been the pre-dominant hunting weapon during the pre-Clovis

period. Bone and ivory points and other toolsare common in the Upper Paleolithic of Siberia

and in late Pleistocene sites in Beringia (22–24).They are durable and lethal hunting weapons

that continued to be used during and after Clovis(16, 23, 25). The invention and spread of a new

hunting weapon at 13 ka—the Clovis lithic point—may have accelerated the demise of or doomed

the last megafaunal species.

Table 1. AMS 14C ages used to date the Manis Mastodon.

Specimen dated Date (14C yr B.P. T 1 SD) Lab number Material dated

Mastodon tusk ivory

sample no. 1

11,975 T 35 UCIAMS-11350 XAD-gelatin

(KOH collagen)

Mastodon tusk ivory

sample no. 1

11,975 T 35 UCIAMS-12046 XAD-gelatin

(KOH collagen)

Mastodon tusk ivory

sample no. 2

11,890 T 35 UCIAMS-11677 XAD-gelatin

(KOH collagen)

Mastodon rib with

embedded bone

projectile point

11,990 T 30 UCIAMS-29113XAD-gelatin

(KOH collagen)

Average of four

radiocarbon

measurements

11,960 T 17 14C yr B.P.

(13,860 to 13,763 calendar yr B.P.)

— n = 4

XAD-gelatin

(KOH collagen)

0 1cm0.5

A

C

B

0 1 2 cm

0 1 2 cm

D

Fig. 1. Mastodon rib with the embedded bone projectile point. (A) Closeup view. (B) Reconstructionshowing the bone point with the broken tip. The thin layer represents the exterior of the rib. (C) CTx-ray showing the long shaft of the point from the exterior to the interior of the rib. (D) The entire ribfragment with the embedded bone projectile point.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org352

REPORTS

References and Notes1. T. Goebel, M. R. Waters, D. H. O’Rourke, Science 319,

1497 (2008).

2. M. T. P. Gilbert et al., Science 320, 786 (2008).

3. M. R. Waters et al., Science 331, 1599 (2011).

4. M. R. Waters, T. W. Stafford Jr., Science 315, 1122 (2007).5. T. D. Dillehay et al., Science 320, 784 (2008).

6. P. S. Martin, in Quaternary Extinctions, a Prehistoric

Revolution, P. S. Martin, R. G. Klein, Eds. (Univ. of

Arizona Press, Tucson, AZ, 1984), pp. 354–403.

7. J. L. Gill, J. W. Williams, S. T. Jackson, K. B. Lininger,

G. S. Robinson, Science 326, 1100 (2009).

8. C. E. Gustafson, D. Gilbow, R. Daugherty, Can. J. Archaeol.

3, 157 (1979).

9. K. L. Petersen, P. J. Mehringer Jr., C. E. Gustafson,

Quat. Res. 20, 215 (1983).10. V. E. Morgan, thesis, Washington State University,

Pullman, WA (1985).11. D. W. Gilbow, thesis, Washington State University,

Pullman, WA (1981).

12. A. L. Runnings, thesis, Washington State University,

Pullman, WA (1984).

13. G. Haynes, The Early Settlement of North America:

The Clovis Era (Cambridge Univ. Press, Cambridge, 2002).

14. P. J. Reimer et al., Radiocarbon 51, 1111 (2009).

15. T. M. Ryan, G. R. Milner, J. Archaeol. Sci. 33, 871 (2006).

16. B. A. Bradley, M. B. Collins, C. A. Hemmings,

Clovis Technology (International Monographs in

Prehistory, no. 17, Ann Arbor, MI, 2010).17. P. G. Taylor, Mol. Biol. Evol. 13, 283 (1996).

18. D. F. Overstreet, in Paleoamerican Origins: Beyond

Clovis, R. Bonnichsen, B. T. Lepper, D. Stanford,

M. R. Waters, Eds. (Center for the Study of the First

Americans, Texas A&M University, College Station, TX,

2005), pp 183–195.

19. D. J. Joyce, Quat. Int. 142-143, 44 (2006).20. S. M. Kenady, M. C. Wilson, R. F. Schalk, R. R. Mierendorf,

Quat. Int. 233, 130 (2011).

21. D. K. Grayson, D. J. Meltzer, J. Archaeol. Sci. 30, 585

(2003).

22. T. Goebel, Evol. Anthropol. 8, 208 (1999).

23. R. D. Guthrie, in Animals and Archaeology: Hunters and

Their Prey, J. Clutton-Brock, C. Grigson, Eds. (British

Archaeological Reports International Series 163, Oxford,

1983), pp. 273–294.

24. C. E. Holmes, Arctic Anthropol. 38, 154 (2001).

25. H. Knecht, in Projectile Point Technology, H. Knecht, Ed.

(Plenum, New York, 1997), pp. 191–212.

Acknowledgments: We thank the North Star Archaeological

Research Program established by J. Cramer and R. Cramer

and the Chair in First Americans Studies for funding.

We thank J. Southon for providing the ultrafiltration14C measurements. Work conducted at the Center for

GeoGenetics was supported by the Danish National

Research Foundation. E.C. is supported by the

European Union with a Marie Curie Intra European

Fellowship (grant number 237227). J.O., D.S., and

L.J. are supported by the Novo Nordisk Foundation

Center for Protein Research. CT scanning was

performed at the High Resolution X-ray CT Facility at

the University of Texas, Austin. J. Halligan prepared

the illustrations. T. Jennings, J. Halligan, T. Goebel,

S. Fiedel, and two anonymous individuals reviewed

the manuscript.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/334/6054/351/DC1

SOM Text

Figs. S1 to S5

Tables S1 to S8

References

Table of bone point marker peptides

29 April 2011; accepted 8 September 2011

10.1126/science.1207663

Fig. 2. Anatomical position of theManis rib. (A) Two vertebrae withthe Manis rib inserted into its cor-rect anatomical position. The bluearrow points to the embedded pointfragment. (B) Side view of mastodonvertebrae with the Manis rib insertedinto its correct anatomical position,with the trajectory of the point indi-cated. (C) Mastodon skeleton show-ing the location of ribs 12 to 14.

0 4 8 cm

transverseprocesses

anterior

posterior

point trajectory

point tip

A B

C

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 353

REPORTS

Information Transduction Capacity ofNoisy Biochemical Signaling NetworksRaymond Cheong,1 Alex Rhee,1 Chiaochun Joanne Wang,1 Ilya Nemenman,2 Andre Levchenko1*

Molecular noise restricts the ability of an individual cell to resolve input signals of differentstrengths and gather information about the external environment. Transmitting informationthrough complex signaling networks with redundancies can overcome this limitation. We developedan integrative theoretical and experimental framework, based on the formalism of informationtheory, to quantitatively predict and measure the amount of information transduced bymolecular and cellular networks. Analyzing tumor necrosis factor (TNF) signaling revealed thatindividual TNF signaling pathways transduce information sufficient for accurate binary decisions,and an upstream bottleneck limits the information gained via multiple integrated pathways.Negative feedback to this bottleneck could both alleviate and enhance its limiting effect,despite decreasing noise. Bottlenecks likewise constrain information attained by networkssignaling through multiple genes or cells.

Signaling networks are biochemical systems

dedicated to processing information aboutthe environment provided by extracellular

stimuli. Large populations of cells can accuratelysense signaling inputs, such as the concentration

of growth factors or other receptor ligands, butthis task can be challenging for an individual cell

affected by biochemical noise (1–3). Noise maps

an input signal to a distribution of possible output

responses, which can cause loss of informationabout the input. For example, a cell cannot re-

liably distinguish different inputs that, because ofnoise, can generate the same output (Fig. 1A).

Conventional metrics related to the standarddeviation or variance of the response distribution

measure noise magnitude (4–8), but fail to elu-cidate how noise quantitatively affects the ac-

curacy of information processing in single cells.By contrast, an information theoretic approach

(Fig. 1B), and themetric ofmutual information inparticular, can quantify signaling fidelity in terms

of the maximum number of input values that acell can resolve in the presence of noise. Such

methods have been commonly used to evaluateman-made telecommunication systems (9) and

more recently in computational neuroscience andin analyses of transcriptional regulatory systems

(10–14), but have not been applied to biochem-ical signaling networks. We developed a general

integrative theoretical and experimental frame-work to predict and measure the mutual infor-

mation transduced by one or more signalingpathways. Applying this framework to analyze a

four-dimensional compendium of single-cell re-sponses to tumor necrosis factor (TNF) (Fig. 1C,

see also SOM section 1), an inflammatory cyto-kine that initiates stochastic signaling at physio-

logic concentrations spanning about four orders

of magnitude (15–21), shows that signaling via anetwork rather than a single pathway can abate

the information lost to noise. Furthermore, an in-formation bottleneck can restrict the maximum

information a network can capture, and negativefeedback potentially but not always relieves this

limitation.The mutual information, I(R;S), measured in

bits, is the binary logarithm of the maximum num-ber of input signal values (S), such as ligand con-

centrations, that a signaling system can perfectlyresolve on the basis of its noisy output responses

(R) (9). One bit of information can resolve twodifferent signal values, 2 bits resolves four val-

ues, etc. More generally,

I(R;S) = ∫S ∫RP(R,S)log2P(R,S)

P(R)P(S)

� �dRdS

ð1Þ

1Department of Biomedical Engineering, Johns Hopkins Uni-versity, 3400 North Charles Street, Baltimore, MD 21218, USA.2Departments of Physics and Biology, Emory University, 400Dowman Drive, Atlanta, GA 30322, USA.

*To whom correspondence should be addressed. E-mail:alev@jhu.edu

A

B

C D

Fig. 1. Information theoretic analysis of cell signaling fidelity. (A) Schematicshowing information loss due to overlapping noisy response distributions. (B)Diagram of the TNF–NF-kB signaling pathway represented in biochemicalform (left) and as a noisy communication channel (right). (C) Experimentalflowchart for using immunocytochemistry to sample the conditional responsedistribution at single-cell resolution and resulting four-dimensional compen-

dium of multiple responses in cells of multiple genetic backgrounds tomultiple TNF concentrations, at multiple time points. The data were collectedin a single experiment, allowing controlled, quantitative comparisons alongeach dimension. (D) Distributions of noisy NF-kB nuclear translocation re-sponses to 30-min TNF exposure (examples shown at top) used to compute thechannel capacity of the TNF–NF-kB pathway. Scale bars, 20 mm.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org354

REPORTS

The joint distribution P(R,S) determines themarginal distributions P(R) and P(S), and hence

also the mutual information, and can be decom-posed as P(R,S) = P(S) P(R|S). The response dis-

tribution, P(R|S), is experimentally accessible bysampling responses of individual isogenic cells to

various signal levels (Fig. 1C), and its spread re-flects the noise magnitude given any specific

input. The signal distribution, P(S), reflects po-tentially context-specific frequencies at which a

cell experiences different signal values. Althoughthe amount of information might thus vary from

case to case, one can also determine the maximalamount of transducible information, given the ob-

served noise (see SOM section 2). This quantity,

known as the channel capacity (9), is a generalcharacteristic of the signaling system and the

signal-response pair of interest and can thereby beexperimentally measured without making as-

sumptions about the (possibly nonlinear) relationbetween R and S, signal power, or noise properties.

Using immunocytochemistry, we assayed nu-clear concentrations of the transcription factor

nuclear factor kB (NF-kB) in thousands of in-dividual mouse fibroblasts 30 min after exposure

to various TNF concentrations (Fig. 1D).We chosethis time point because NF-kB translocation peaks

at 30 min regardless of the concentration used,initiating expression of early-response inflamma-

tory genes (19–22). The NF-kB response valuein a single cell could yield at most 0.92 T 0.01 bits

of information, which is equivalent to resolving20.92 = 1.9, or about 2, concentrations of the TNF

signal, thus essentially only reliably indicatingwhether TNF is present or not. (See SOM sec-

tions 2.2 and 3 regarding the low experimentaluncertainty.) A bimodal input signal distribution,

P(S), with peaks at low and high TNF concentra-tions, maximizes the information (fig. S1), support-

ing the notion of essentially binary (digital) sensing

capabilities of this pathway (18), although we didnot observe bimodal output responses, P(R|S).

Noise also limits other canonical pathways,including signaling by platelet-derived growth

factor (PDGF), epidermal growth factor (23), andG protein–coupled receptors (24), to ~1 bit (fig.

S2, A to C, and table S1). Even the most reliablesystem we examined, morphogen gradient sig-

naling through the receptor Torso in Drosophila

embryos (25), was limited to 1.61 bits (fig. S2D

and table S1), corresponding to about three dis-tinguishable signal levels.

The pathways examined above are examplesof individual biochemical communication chan-

nels (Fig. 1B) that capture relatively low amounts

of information about signal intensity, which wouldallow only limited reliable decision making by a

cell. However, information in biological systemsis typically processed by networks comprising

multiple communication channels, each transduc-ing information about the signal. For instance, a

transcription factor often regulates many genes,a receptor many transcription factors, and a dif-

fusible ligand many cells. The integrated outputsof such multiple channels can provide more in-

formation about the signal than the output of anyone channel (see SOM section 4). Subsequently,

downstream signaling processes that convergeto co-regulate common effectors, biological pro-

cesses, or physiologic functions can provide thepoint needed to integrate the multiple outputs to

realize the benefit of increased aggregate infor-mation (fig. S3). To provide a unified framework

for analyzing such various networks, we first the-oretically investigated the information gained by

network signaling in general, then experimentallytested the predictions made by the theory when

applied to a specific system.We considered two information theoretic mod-

els, similar to models of population coding in

neural systems (26–28), for transmitting a signalS through multiple channels to the responses R1,

R2, …, Rn, under the assumption of Gaussianvariables (see SOM section 5). The bush model

uses independent channels (topologically resem-bling an upside-down shrub) (Fig. 2A), whereas

the tree model signals through a common chan-nel (“trunk”) to the intermediate, C, before di-

verging into independent branches (Fig. 2B). Theinformation resulting from the bush model is

Ibush(R1, :::, Rn; S) =1

2log2 1 þ n

s2S

s2S→R

� �

ð2Þ

wheres2Sis the variance of the signal distribution,

and s2S→R

is the noise (variance) introduced in

each branch. Thus, the information can growlogarithmically with the number of branches

without an upper bound. In contrast, the informa-tion resulting from the tree model is

Itree(R1, :::, Rn; S) =

1

2log2 1 þ

ns2S=s2

C→R

1 þ ns2S→C

=s2C→R

� �ð3Þ

where s2S→C

and s2C→R

are the trunk and branchnoises, respectively (see SOM section 3.3). As the

number of branches increases, the information

asymptotically approaches an upper limit equalto the mutual information between the input

signal and the common intermediate. Thus, theinformation lost to noise in the trunk determines

the maximum throughput of a tree network.The key difference between bush and tree

networks is the absence or presence of this trunk-based information bottleneck. The biochemical

structure of a network can resemble a tree, but ifthere is little loss of information upstream, the

A

B

C D

Fig. 2. Information gained by signaling through a network comprising mul-tiple communication channels. (A) Schematic of a bush network with inde-pendent channels lacking an information bottleneck. (B) Schematic of a treenetwork with channels sharing a common trunk that forms an informationbottleneck. Circles represent noise introduced in the indicated portions of thesignaling network; see text for definition of symbols. (C) Comparison of bushand treemodel predictions for the capacity of the TNF network to experimental

values. At 30 min, the NF-kB and ATF-2 pathways together capture moreinformation about TNF concentration than either pathway alone (bars 1 to 3),and the tree rather than bushmodel accurately predicts this increase (bars 3 to5). The tree model further predicts a receptor-level bottleneck of 1.26 T 0.13bits (bar 6). (D) Joint distribution of NF-kB and ATF-2 responses to 30-minstimulation of TNF. Each data point represents a single cell, and eachconcentration of TNF examined is shown with a distinct color.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 355

REPORTS

bush model lacking a bottleneck might best es-timate the capacity of the network. Additionally,

the bush and tree models make various semi-quantitative predictions (see SOM section 6),

such as the information captured by a networkbased on the capacities of its component path-

ways. For example, for a bush network compris-ing two pathways each with 1-bit responses,

Eq. 2 implies s2S=s2

S→R¼ 3 and that together

they should yield 1

2log2ð1 þ 2ð3ÞÞ ¼ 1:4 bits.

TNF activates the NF-kB and c-Jun N-terminal kinase (JNK) pathways, stimulating nu-

clear localization of NF-kB and phosphorylatedactivating transcription factor–2 (ATF-2) (fig. S4),

respectively (29). To determine if the TNF sig-

naling network contains an appreciable upstream

information bottleneck limiting the informationcaptured by these pathways, we examined wheth-

er the bush (bottleneck absent) or tree (bottleneckpresent) network model better approximates the

network (fig. S5). The models are applicable be-cause the NF-kB (Fig. 1D) and ATF-2 (fig. S6)

response distributions are approximately Gaussianat all TNF concentrations. We found that NF-kB

alone yielded at most 0.92 bits of informationabout TNF concentration, and ATF-2 alone

yielded at most 0.85 T 0.02 bits (fig. S1B andtable S1). Together, the bush model predicts that

these pathways jointly yield 1.27 T 0.01 bits (Fig.2C), and a similar model assuming indepen-

dent pathway responses that are not necessarily

Gaussian likewise predicts an increase to 1.13 T

0.01 bits. The actual information determined bydual-staining immunocytochemistry (Fig. 2D)was

1.05 T 0.02 bits, much lower than both predic-tions (Fig. 2C), demonstrating that the bush mod-

el does not approximate the TNF network well.In contrast, the tree model predicts 1.03 T 0.01

bits, matching the experimental value within er-ror (Fig. 2C), and also correctly predicts the sta-

tistical dependency between the responses giventhe signal (fig. S7).

The correspondence between the tree mod-el predictions and experimental measurements

strongly indicates that the network contains aninformation bottleneck. The tree model predicts

that the maximum information that can pass

through the bottleneck is 1.26 T 0.13 bits (Fig. 2C),

A B C D

EFig. 3. Effect of negative feedback to the bottleneck on information transfer. (A) TNF signaling networkdiagram showing A20-mediated negative feedback to the information bottleneck. (B) Comparison of infor-mation about TNF concentration captured with and without A20 negative feedback. The information is larger at30 min but smaller at 4 hours in wild-type cells as compared to A20−/− cells. (C and D) Schematic of NF-kBdynamics in wild-type and A20−/− mouse fibroblasts exposed to saturating concentrations of TNF. Averagedynamics (black) and the expected magnitudes of the dynamic range (double arrow) and noise (single arrow)are shown. See fig. S9 for experimental support. (E) Comparison of NF-kB responses to zero (basal) or sat-urating concentrations of TNF. Differences in the means with and without TNF indicate the dynamic range, anderror bars (SD) indicate the noise.

A B C

Fig. 4. Information gained by signaling through networks of multiple genes. (A)Plot shows the unique curve (solid black) determined by the tree model (inset),passing through the experimentally determined values (circles), for information asa function of the number of copies of a NF-kB reporter gene. The upper limit,corresponding to the maximum information captured by integrating NF-kB ac-tivity over time, is 1.64 T 0.36 bits (blue dashed line). (B) Expression-level dis-

tributions of clonal cell lines containing different numbers of copies of an NF-kBreporter gene in response to ~10 hours of TNF exposure. (C) Time courses cor-responding to individual cells showing cell-to-cell differences in the onset and rateof NF-kB reporter gene expression (left). In each cell, expression is nearly linearand deterministic in time, as quantified by the correlation coefficient (right) of thetime course after onset of expression (shown schematically in inset on left).

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org356

REPORTS

corresponding to just 21.26 = 2.3 distinguishableTNF concentrations. The known biochemistry of

TNF signaling implies that the bottleneck (trunk)comprises the steps of TNF receptor complex ac-

tivation common to both pathways, including lig-and binding, receptor trimerization, and complex

formation and activation. Because all TNF sig-naling passes through the receptor complex, mul-

tiple pathways in the TNF signaling network,activated at the 30-min time point, only modestly

increase the information about TNF concentra-tion regardless of the number of pathways or their

fidelity.We next explored whether negative feedback,

which can reduce noise (12, 30, 31), might alle-

viate the receptor-level signaling bottleneck. Theinformation captured by a single channel (Eq. 2,

n = 1) can be written as 1

2log2ðs

2R=s2

S→RÞ. Thus,

negative feedback can have equivocal effects on

information, depending on the balance of the tend-encies for negative feedback to reduce both the

dynamic range of the signaling response (32), rep-resented by the response variance s

2R, and noise,

represented bys2S→R

. Indeed, comparison of wild-type cells and cells lacking A20 (fig. S8), an

inhibitor of TNF receptor complexes whose ex-pression is up-regulated byNF-kB (33) (Fig. 3A),

showed that A20-mediated negative feedback in-creases information at the 30-min time point, but

decreases it at 4 hours (Fig. 3B).To understand these different outcomes, we

examined how A20 affects the dynamic rangeand noise at either time point. At the early time

point, constitutively expressed A20 inhibits basalNF-kB activity, but TNF does not induce A20

expression rapidly enough to affect saturatinglevels of NF-kB at 30 min (Fig. 3, C and D, and

fig. S9) (17, 34). Hence, A20 negative feedbackdecreases noise, primarily at low TNF concen-

trations, and also increases the dynamic range bylowering basal NF-kB levels (Fig. 3E and fig.

S10A), explaining why information at 30 min ishigher forwild-type than forA20−/− cells (Fig. 3B).

In contrast, at the late time point, A20 is in-creased in wild-type cells (17, 34). The negative

feedback decreases noise at all TNF concentra-tions but also decreases the dynamic range by

strongly suppressing the maximum inducibleNF-kB activity (Fig. 3E and fig. S10A). The net

effect is lower information for wild-type versusA20−/− cells at 4 hours (Fig. 3B).

We observed that A20 negative feedback sim-ilarly both improves and limits information at the

early and late time points, respectively, for ATF-2alone, or together with NF-kB (Fig. 3B and fig.

S10B), consistent with A20 affecting the portionof the network common to both pathways. Nev-

ertheless, the maximal information about TNF con-centration acquired with or without A20-mediated

negative feedback was still ~1 bit, suggesting

limited advantages for mitigating the informationbottleneck in this pathway by using negative

feedback.We next considered whether networks com-

prising multiple target genes can capture sub-stantial amounts of information through time

integration. If the target gene product lifetime islong compared to its transcription and translation

time scales, the accumulated protein concentra-tion is approximately proportional to the time in-

tegral of signaling activity, thereby averaging outtemporal fluctuations (35, 36). However, the bio-

chemical readout of protein synthesis can intro-duce extra noise, confounding determination of

the information contained in the time integral.Fortunately, the maximum information captured

by a tree network, in which the time integral oftranscription factor activity is the intermediate sig-

nal activating multiple independent target genes(Fig. 4A, inset), is determined by the trunk (time

integration) rather than branch noise (readoutmech-anism). We measured the information captured

by such tree networks in cells stably transfectedwith different copy numbers (1.8-fold difference,

as determined by polymerase chain reaction) ofthe gene coding for a stable green fluorescent

protein (GFP) (37) reporting on NF-kB activity(Fig. 4B). Using the tree model to extrapolate the

extent of the bottleneck, under the assumptionthat ~10 hours of TNF exposure induces similar

expression level and noise for each gene, in-dicates that 1.64 T 0.36 bits is the maximum in-

formation that integrating NF-kB activity overthe experimental time period can yield about

TNF concentration (Fig. 4A), regardless of thereadout mechanism.

To understandwhy information was onlymod-erately higher compared to a single time point

(1.64 versus 0.92 bits), we monitored GFP re-porter gene expression in individual cells, finding

that, for any given cell, GFP accumulated linearlyin time in a nearly deterministic fashion, although

its onset and accumulation rate varied from cell tocell (Fig. 4C). This is consistent with observa-

tions made with live cell probes (18–20) showing

NF-kB dynamics to be essentially deterministicover the experimental time scale within each cell,

but distinct across cells. We thus conclude thatthe ability of time integration to increase the in-

formation about TNF concentration is limited bythe lack of rapid temporal fluctuations that would

otherwise be suppressed by integration over the10-hour response.

Finally, we considered signaling via multiplecells, each considered as separate information

channels within a network (Fig. 5A, inset). Anensemble of cells resembles a bush network if

each cell directly and independently accesses thesame signal, and because bush networks do not

contain trunk-based bottlenecks, substantial in-creases in information might be obtained. To test

this hypothesis, we analyzed the collective TNFresponse of different numbers of cells, as mea-

sured by immunocytochemistry. We varied cellnumber by considering cells within nonoverlap-

ping circular regions of variable diameter (Fig.5B) and used the average NF-kB response with-

in each region to simulate cells contributing to acollective response in proportion to their NF-kB

activity. The bush model predicts (Eq. 2), andthe data confirm (Fig. 5A), that the information

should increase logarithmically with the numberof independently signaling cells functioning

collectively.Moreover, we found that networks of just 14

cells can yield up to 1.8 bits of information, fargreater than the other network types analyzed

above. Because ensembles of this size can plau-

sibly experience a similar concentration of a dif-fusing signal such as TNF and function collectively

(21, 38) [e.g., TNF-activated blood vessel en-dothelial cells (39)], collective cell behavior can

effectively increase the information gained andproduce responses that can discriminate be-

tween many TNF concentrations. Nonetheless,networks relying on cell-cell communication can

still contain bottlenecks. For instance, TNF canbe secreted by macrophages stimulated by lipo-

polysaccharide (LPS) from invading bacteria,with the information about the initial LPS dose

lost within the macrophage signaling networksbefore secretion of TNF.

By treating biochemical signaling systems asinformation theoretic communication channels,

we have rigorously and quantitatively shown that,

Fig. 5. Information gained by signaling through networks of multiple cells. (A) Comparison of ex-perimentally measured information obtained by collective cell responses (circles) versus logarithmic trend(solid black line) predicted the bush model (inset). (B) Schematic of methodology used to measurecollective cell responses.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 357

REPORTS

in a single cell, noise can substantially restrict theamount of information transduced about input

intensity, particularly within individual signalingpathways. The bush and tree network models,

which provide a unified theoretical framework foranalyzing branched motifs widespread in natural

and synthetic signaling networks, further dem-onstrated that signaling networks can be more

effective in information transfer, although bot-tlenecks can also severely limit the information

gained. Receptor-level bottlenecks restrict the TNFand also PDGF signaling networks (fig. S11) and

may be prevalent in other signaling systems.We explored several strategies that a cell

might use to overcome restrictions due to noise.

We found that negative feedback can suppressbottleneck noise, which can be offset by concom-

itantly reduced dynamic range of the response.Time integration can increase the information

transferred, to the extent that the response under-goes substantial dynamic fluctuations in a single

cell over the physiologically relevant time course.The advantage of collective cell responses can

also be substantial, but limited by the number ofcells exposed to the same signal or by the in-

formation present in the initiating signal itself.Responses incorporating the signaling history

of the cell might also increase the information(40, 41). For instance, responses relative to the

basal state (fold-change response) might be lesssusceptible to noise arising from diverse initial

states (23), although this does not necessarilytranslate into large amounts of transferred infor-

mation (table S1). Similarly, for the reporter genesystem described here (fig. S12), ~0.5 bits of ad-

ditional information can be obtained if a cell candetermine expression levels at both early and late

time points. However, noise in the biochemicalnetworks that a cell uses to record earlier output

levels and to later compute the final responsemay nullify the information gain potentially pro-

vided by this strategy. Overall, we anticipate that

the information theory paradigm can extend tothe analysis of noise-mitigation strategies and

information-transfer mechanisms beyond thoseexplored here, in order to determine what specific

signaling systems can do reliably despite noise.

References and Notes1. J. G. Albeck, J. M. Burke, S. L. Spencer, D. A. Lauffenburger,

P. K. Sorger, PLoS Biol. 6, e299 (2008).2. N. Rosenfeld, J. W. Young, U. Alon, P. S. Swain,

M. B. Elowitz, Science 307, 1962 (2005).

3. T. J. Perkins, P. S. Swain, Mol. Syst. Biol. 5, 326

(2009).

4. W. J. Blake, M. Kaern, C. R. Cantor, J. J. Collins, Nature

422, 633 (2003).

5. M. B. Elowitz, A. J. Levine, E. D. Siggia, P. S. Swain,

Science 297, 1183 (2002).6. J. Paulsson, Nature 427, 415 (2004).

7. J. M. Pedraza, A. van Oudenaarden, Science 307,

1965 (2005).

8. J. M. Raser, E. K. O’Shea, Science 304, 1811

(2004).

9. T. M. Cover, J. A. Thomas, Elements of Information Theory

(Wiley, New York, 1991).

10. R. R. de Ruyter van Steveninck, G. D. Lewen, S. P. Strong,

R. Koberle, W. Bialek, Science 275, 1805 (1997).

11. D. Fuller et al., Proc. Natl. Acad. Sci. U.S.A. 107, 9656

(2010).

12. E. Ziv, I. Nemenman, C. H. Wiggins, PLoS ONE 2,

e1077 (2007).

13. G. Tkacik, C. G. Callan Jr., W. Bialek, Proc. Natl. Acad.

Sci. U.S.A. 105, 12265 (2008).

14. P. Mehta, S. Goyal, T. Long, B. L. Bassler, N. S. Wingreen,

Mol. Syst. Biol. 5, 325 (2009).

15. R. Cheong, A. Hoffmann, A. Levchenko, Mol. Syst. Biol. 4,

192 (2008).

16. R. Cheong, C. J. Wang, A. Levchenko, Mol. Cell. Proteomics

8, 433 (2009).

17. S. L. Werner et al., Genes Dev. 22, 2093 (2008).

18. S. Tay et al., Nature 466, 267 (2010).

19. L. Ashall et al., Science 324, 242 (2009).

20. D. E. Nelson et al., Science 306, 704 (2004).

21. R. Cheong et al., J. Biol. Chem. 281, 2945 (2006).

22. A. Hoffmann, A. Levchenko, M. L. Scott, D. Baltimore,

Science 298, 1241 (2002).

23. C. Cohen-Saidon, A. A. Cohen, A. Sigal, Y. Liron, U. Alon,

Mol. Cell 36, 885 (2009).

24. X. R. Bao, I. D. Fraser, E. A. Wall, S. R. Quake,

M. I. Simon, Biophys. J. 99, 2414 (2010).

25. M. Coppey, A. N. Boettiger, A. M. Berezhkovskii,

S. Y. Shvartsman, Curr. Biol. 18, 915 (2008).

26. B. B. Averbeck, P. E. Latham, A. Pouget,

Nat. Rev. Neurosci. 7, 358 (2006).

27. J. W. Pillow et al., Nature 454, 995 (2008).

28. E. Schneidman, W. Bialek, M. J. Berry II, J. Neurosci. 23,

11539 (2003).

29. H. Wajant, K. Pfizenmaier, P. Scheurich, Cell Death Differ.

10, 45 (2003).

30. A. Becskei, L. Serrano, Nature 405, 590 (2000).

31. I. Lestas, G. Vinnicombe, J. Paulsson, Nature 467,

174 (2010).

32. R. C. Yu et al., Nature 456, 755 (2008).

33. I. E. Wertz et al., Nature 430, 694 (2004).

34. E. G. Lee et al., Science 289, 2350 (2000).

35. V. Shahrezaei, P. S. Swain, Proc. Natl. Acad. Sci. U.S.A.

105, 17256 (2008).

36. S. Krishna, M. H. Jensen, K. Sneppen, Proc. Natl. Acad.

Sci. U.S.A. 103, 10840 (2006).

37. S. Thierfelder, K. Ostermann, A. Göbel, G. Rödel,

Appl. Biochem. Biotechnol. 163, 954 (2011).

38. K. Francis, B. O. Palsson, Proc. Natl. Acad. Sci. U.S.A. 94,

12258 (1997).

39. J. Parkin, B. Cohen, Lancet 357, 1777 (2001).

40. I. Nemenman, G. D. Lewen, W. Bialek, R. R. de Ruyter

van Steveninck, PLOS Comput. Biol. 4, e1000025

(2008).

41. S. P. Strong, R. Koberle, R. R. de Ruyter van Steveninck,

W. Bialek, Phys. Rev. Lett. 80, 197 (1998).

Acknowledgments: We thank A. Hoffmann, M. Simon,

S. Shvartsman, C. Cohen-Saidon, and U. Alon for sharing

data and materials; A. Ganesan and H. Chang for

experimental assistance; and P. Iglesias, Y. Qi, and

A. Feinberg for insightful discussions and reviewing

drafts of the manuscript. This work was supported by

NIH grants GM072024 and RR020839 (R.C., A.R., C.J.W.,

and A.L.) and CA132629 (I.N.), the Medical Scientist

Training Program at the Johns Hopkins University (R.C.),

and, in early stages of the work, the Los Alamos

National Laboratory Directed Research and Development

program (I.N.).

Supporting Online Materialwww.sciencemag.org/cgi/content/full/science.1204553/DC1

Materials and Methods

SOM Text

Figs. S1 to S12

Table S1

References (42–55)

18 February 2011; accepted 7 September 2011

Published online 15 September 2011;

10.1126/science.1204553

ER Tubules Mark Sites ofMitochondrial DivisionJonathan R. Friedman,1 Laura L. Lackner,2 Matthew West,1 Jared R. DiBenedetto,1

Jodi Nunnari,2 Gia K. Voeltz1*

Mitochondrial structure and distribution are regulated by division and fusion events.Mitochondrial division is regulated by Dnm1/Drp1, a dynamin-related protein that formshelices around mitochondria to mediate fission. Little is known about what determines sitesof mitochondrial fission within the mitochondrial network. The endoplasmic reticulum (ER)and mitochondria exhibit tightly coupled dynamics and have extensive contacts. We testedwhether ER plays a role in mitochondrial division. We found that mitochondrial division occurredat positions where ER tubules contacted mitochondria and mediated constriction before Drp1recruitment. Thus, ER tubules may play an active role in defining the position of mitochondrialdivision sites.

Regulation ofmitochondrial division is crit-ical to normal cellular function; excess

division is linked to numerous diseases,

including neurodegeneration and diabetes (1, 2).The central player in mitochondrial division is

the highly conserved dynamin-related protein

(Drp1 inmammals,Dnm1 in yeast),which belongsto a family of large guanosine triphosphatases

(GTPases) that self-assemble to regulate mem-brane structure (3). Division dynamins are likely

to work by oligomerizing in a GTP-dependentmanner into helices that wrap around mitochon-

dria; locally controlled assembly-stimulated GTPhydrolysis is thought to provide the mechano-

chemical force that completes fission of the out-er and inner membranes (4). There are additional

proteins required formitochondrial division, suchas the outer membrane proteinMff (mitochondrial

fission factor), which is present only in mam-mals (5). Although general mechanisms exist for

1Department of Molecular, Cellular, and Developmental Biol-ogy, University of Colorado, Boulder, CO 80309, USA. 2De-partment of Molecular and Cellular Biology, University ofCalifornia, Davis, CA 95616, USA.

*To whom correspondence should be addressed. E-mail:gia.voeltz@colorado.edu

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org358

REPORTS

Fig. 1. Mitochondrial constriction and division oc-curs at ER-mitochondrial contacts in yeast. (A) The3D models (left images) of ER (green) and mito-chondria (purple) at contact domains were imagedby EM and tomography of high-pressure frozen yeastcells. Middle images are 2D tomographs of contactsites (second column, ER drawn in green) and thecorresponding 3D models of each (third column).Contact, marked in red, is defined as regions wherethe ER membrane comes within 30 nm of the mito-chondrial membrane, and ribosomes are excluded(third column). Right schematics demonstrate thepercentage of the mitochondrial circumference thatmakes contact with the ERmembrane [red is contact,white is not (19)]. The diameter of each mitochon-drion at positions of ER contact is shown. Regionswhere the mitochondria are constricted (models aand c) have a high percent of ER wrapping. Addi-tional EM tomographs and analysis of constrictionsare shown in fig. S1, A and B. (B) Time-lapse imagesof yeast cells expressing mito-dsRed and GFP-HDEL(ER). A single focal plane is shown. Arrows and arrow-heads indicate sites of mitochondrial division. A cor-responding z-series is shown in fig. S1C. Scale barsindicate, in (A), 200 nm; (B), 2 mm.

A

0s 40s 80s 120s 160s 200s 240s 280sB

GF

P-H

DE

Lm

ito-d

sRE

Dm

ito-d

sRE

D

74%

138 nm

43%

215 nm

91%

146 nm

11%

193 nm

a

b

a

b

a

b

c

d

c

d

c

d

a

b

c

d

GFP-Sec61β

mito-dsRed

mito-dsRed

A B

GFP-Sec61β

mito-dsRed

0s 10s 20s 30s 0s 10s 20s 30s

0s 10s 20s 40sC 0s 10s 20s 30sD

mito-dsRed

Fig. 2. Mitochondrial division occurs at ER-mitochondrial contact sites inmammalian cells. (A to D) Four examples of mitochondrial division over timecourses shown in Cos-7 cells expressing GFP-Sec61b (ER) and mito-dsRed.

The site of mitochondrial division (white arrows) and the position of the newlyformed mitochondrial ends (yellow arrows) are shown. Additional examples areincluded in fig. S2A. Scale bars, 1 mm.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 359

REPORTS

recruitingDnm1 orDrp1 tomitochondria, it is notknown whether there are specific sites on mito-

chondria that are marked for division (6). Fur-thermore, both Dnm1 and Drp1 oligomerize into

helices that are much smaller than the diameter ofmitochondria (Dnm1 helices have reported mean

diameters of 109 nm in yeast and 129 nm in vitro),suggesting that Dnm1 (Drp1)–independent mito-

chondrial constriction may be needed to facilitatemitochondrial division (4, 6–9).

Contact sites exist between mitochondria andthe endoplasmic reticulum (ER) and are impor-

tant for phospholipid synthesis and calcium signal-ing [for review, see (10)]. Based on recent data,

there are likely several types of molecular bridges

that mediate these contacts, such as the ERMES

complex identified in yeast and themitochondrialfusion protein mitofusin 2 (Mfn2) in mammalian

cells (11, 12). These physical contacts are per-sistent and maintained under dynamic conditions

(13), suggesting that the ER-mitochondrial inter-face is vital for function. We have used electron

microscopy (EM) and tomography to analyze thethree-dimensional (3D) structure of contacts be-

tween the ER and mitochondria in the yeastSaccharomyces cerevisiae. We observed the high-

resolution (~4 nm) structure and 3D models offour ER-mitochondrial contacts taken from two

cells (Fig. 1A). In these examples, the ER waswrapped around mitochondria to varying degrees.

In two of the four examples, the ER almost com-

pletely circumscribed the mitochondrial outer

membrane, and mitochondria were constricted atthe point of contact (mitochondrial diameter 138

nm and 146 nm circumscribed versus 215 nmand 193 nm uncircumscribed at ER contact) (Fig.

1A; fig. S1, A and B; and movies S1 and S2).These data suggest that ER tubules associate

with and may mediate mitochondrial constric-tion sites.

We thus examined the role of ER in mito-chondrial division by using fluorescence micros-

copy in live yeast cells transformed with an ERmarker (GFP-HDEL) and mito-dsRed to image

the behavior of ER and mitochondria simulta-neously over time. The vast majority of mito-

chondrial division events were spatially linked to

sites of ER-mitochondrial contact (87%, n = 112

LinescanBFP-KDELmito-EGFPmito-EGFP

mCherry-Drp1BFP-KDEL

mito-EGFPmCherry-Drp1

mito-EGFP

E

Distance (pixels)Rel

ativ

e flu

ores

cenc

ein

tens

ity

B

D

mCherry-Drp1GFP-Sec61β

mito-BFP

mito-BFP / mCherry-Drp1 / GFP-Sec61β

0s 40s 80s 120s

0s 40s 80s 120s

Always

Part of time

Never

Per

cent

of p

unct

ae

n=101

0s 40s 80s 120s

Distance (pixels)Rel

ativ

e flu

ores

cenc

ein

tens

ity

Distance (pixels)Rel

ativ

e flu

ores

cenc

ein

tens

ity

Drp1 localization toER-mitochondrial contacts

GFP-HDELmito-CFPDnm1-

mCherry MergeA

0s

10s

20s

30s

C

Fig. 3. Dnm1- and Drp1-mediated mitochondrial division occurs at ER con-tact sites. (A) Time-lapse images of wild-type yeast cells expressing mito-CFP,GFP-HDEL (ER), and Dnm1-mCherry. A single focal plane is shown. Arrows indi-cate the site of mitochondrial division, which is marked by both ER-mitochondriacontact and Dnm1. (B) Merged image of a live Cos-7 cell expressing GFP-Sec61b(ER), mito-BFP, and mCherry-Drp1. (C) Examples of cells as in (B) that show thatDrp1 punctaemaintain colocalization with positions of ER-mitochondrial contactover time. White arrows indicate Drp1 punctae that maintain contact with boththe ER and mitochondria. Yellow arrows indicate a rare example of Drp1 that

does not contact the ER. (D) The percentage of mitochondrial Drp1 punctae thatcolocalize with the ER membrane over a 2-min time course. (E) Examples ofmitochondrial constrictions at ER contact sites marked by Drp1. Left-hand im-ages show Cos-7 cells expressingmito-EGFP, BFP-KDEL (ER), andmCherry-Drp1,merged as indicated. Right graphs are line scans drawn through the mitochon-dria and show the relative fluorescence intensity of mitochondria (green), ER(blue), and Drp1 (red) along its length. White arrows at constrictions on imagescorrespond to black arrows shown on the line scan. Additional examples areshown in fig. S4. Scale bars for (A), (C), and (E), 1 mm; (B), 5 mm.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org360

REPORTS

from 281 cells) (Fig. 1B). ER tubules crossedover (Fig. 1B, yellow arrows) and wrapped around

mitochondria (Fig. 1B, white arrows, and fig.S1C). At ER-mitochondrial contact sites, mito-

chondrial constriction followed by mitochondrialdivision was observed (Fig. 1B).

We next tested whether ER plays a similarrole inmammalianmitochondrial division by using

fluorescence microscopy of live Cos-7 cells tran-siently transfected with fluorescent markers for

ER (GFP-Sec61b) andmitochondria (mito-dsRed).We imaged regions of the cell periphery where

contacts between the mitochondria and ER werewell resolved and observed that mitochondrial

division events predominantly occurred at sites

of contact between ER and mitochondria (94%,n = 32 from 23 cells) (Fig. 2, fig. S2A, and movies

S3 and S4). Furthermore, the majority of events(88%) were sites of ER tubules crossing over the

mitochondria, suggesting that the structural con-text of the interaction is important. The frequency

of ER-associated mitochondrial division is muchhigher than would be predicted on the basis of

the area of mitochondria covered by crossing ERtubules as determined by colocalization of mito-

chondrial and ER markers (fig. S2B).Thus, in both yeast and mammalian cells, ER

tubules are at mitochondrial division sites andmay be involved inmitochondrial constriction dur-

ing this process. Next, we asked whether mito-

chondrial division occurs in yeast cells that havesubstantially reduced levels of tubules because of

the absence of the membrane shaping proteinsRtns and Yop1 (14, 15). By using both EM and

fluorescence microscopic analyses, we observedthat, in regions of mutant cells in which ER tu-

bules were dramatically reduced, short ER tu-bules extended out of the massive ER cisternae

and associated with mitochondrial constrictionsand division events (fig. S3). Thus, ER tubules

are a consistent feature of ER contact at mito-chondrial constrictions, even under conditions

where most tubules are depleted. Furthermore,Rtns and Yop1 are dispensable for the biogenesis

of the ER tubules that associate with mitochon-

drial division events.To ask whether ER-associated division events

are spatially linked to the mitochondrial divisionmachinery, we determined the relationship of ER-

mitochondrial contacts to the division dynaminsDnm1 and Drp1. Dnm1 and Drp1 assemble into

punctate structures at steady state, and a subsetof these structures are found on mitochondria

and at mitochondrial division sites (6, 16, 17).We imaged live yeast transformed with Dnm1-

mCherry, mito–cyan fluorescent protein (CFP),and GFP-HDEL (ER) and observed that a large

percentage of Dnm1 punctae were at sites ofmitochondrial-ER contact (46%, n = 225). These

Dnm1 punctae could be observed at sites where

ER tubule crossover and mitochondrial divisionoccurred (Fig. 3A). In Cos-7 cells transiently trans-

fected with GFP-Sec61b (ER), mito–blue flu-orescent protein (BFP), and mCherry-Drp1, we

observed that the majority of Drp1 punctae sta-bly associated with mitochondria and localized

to ER-mitochondrial contacts over time (Fig. 3,B to D, and movie S5). Furthermore, a subset

of Drp1 at these contacts was associated with amitochondrial constriction site (78%, excluding

punctae localized to mitochondrial tips, n = 50).The mitochondrial constrictions marked by Drp1

punctae were always either at ER tubule cross-overs (81%) or adjacent to them (19%) (Fig. 3E

and fig. S4). Together, the localization of the

mitochondrial division dynamins in yeast andmammalian cells to regions of ER-mitochondrial

contacts and the observations that these regionsare associated with constricted mitochondria and

subsequent division indicate a direct role of theER in the process of mitochondrial division.

Mff is a mammalian-specific mitochondrialouter membrane protein required for mitochon-

drial localization of Drp1 and division (5, 18).Drp1 andMff colocalize in punctate structures on

mitochondria, and Mff punctae persist in cellswhere Drp1 expression is reduced by RNAi (18).

Thus, Mff punctae may mark the future sites ofmitochondrial division before Drp1 recruitment

(18). In Cos-7 cells transiently transfected with

C

BFP-KDEL

mito-dsRedGFP-MffBFP-KDEL

mito-dsRedmito-dsRed mito-dsRedGFP-Mff

Linescan

Distance (pixels)Rel

ativ

e flu

ores

cenc

ein

tens

ity

Distance (pixels)Rel

ativ

e flu

ores

cenc

ein

tens

ity

BFP-KDEL

βmito-dsRed

mito-dsRedD GFP-Sec61βmito-dsRed

Mff

RN

Ai

Linescan

Rel

ativ

e flu

ores

cenc

ein

tens

ity

Distance (pixels)

Distance (pixels)

Rel

ativ

e flu

ores

cenc

ein

tens

ity

A BFP-KDEL

mito-dsRedGFP-Mff

Drp

1 R

NA

i

GAPDH

Drp1

siRNA

cont

rol

Drp

1

IB:

B

GAPDH

Mff

cont

rol

Mff IB:

At ER tubule crossing 16 (64%)

Adjacent to ER tubule crossing 4 (16%)

Alongside ER tubule 3 (12%)

Not at ER 2 (8%)

Constrictions markedby Mff (Drp1 RNAi)

n = 25GFP-Sec61

Fig. 4. The ER localizes to mitochondrial constrictions before Drp1 andMff recruitment. (A) Examples of mitochondrial constrictions at ER contactsmarked by Mff in Cos-7 cells depleted of Drp1. Left and center images showthese cells expressing mito-dsRed, BFP-KDEL (ER), and GFP-Mff, merged asindicated. Right graphs are line scans drawn through the mitochondria andshow the relative fluorescence intensity of mitochondria (red), ER (blue), andMff (green) along their length. White arrow positions at constrictions corre-spond to black arrows on the line scan. Additional examples are shown in fig.

S6. (B) Western blots with antibody against Drp1 (top) or Mff (bottom) andGAPDH demonstrate depletion of Drp1 in lysates from cells transfected withsiRNA against Drp1 [as in (A)] or Mff [as in (D)] compared with control RNAicells. (C) The number of Mff-localized mitochondrial constrictions in Drp1-depleted cells that colocalize with ER tubules, from 23 cells. (D) As in (A), forcells depleted of Mff and expressing GFP-Sec61b (ER; green on line scan) andmito-dsRed (red on line scan). Scale bars for (A) and (D) large left images, 5 mm;(A) and (D) smaller center images, 1 mm.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 361

REPORTS

GFP-Mff, mCherry-Drp1, and mito-BFP, we ob-served that Mff circumscribed and localized to

punctae on mitochondria, the majority of whichcolocalized with Drp1 (fig. S5, A to C). To test

whether Mff punctae localize to ER contactsindependently of Drp1, we depleted Drp1 from

Cos-7 cells with small interfering RNA (siRNA)and cotransfected these cells with GFP-Mff, mito-

dsRed, and BFP-KDEL (ER). Drp1 was sub-stantially depleted in Drp1 RNA interference

(RNAi) cells in comparison with the controlcells (Fig. 4B). Selective depletion of Drp1 was

further supported by the aberrant and elongatedmitochondrial morphology in Drp1 RNAi cells

(Fig. 4A and fig. S5D). As expected (18), in Drp1-

depleted cells, Mff punctae localized to mitochon-dria (Fig. 4A). We asked whether mitochondria

were constricted at Mff punctae in the absence ofDrp1, and if so, whether these sites localized to

ER contacts. Of the 25 constrictions we resolved,16 were at an ER crossover (64%), and another

4 were adjacent to an ER tubule crossing (16%)(Fig. 4, A and C, and fig. S6). Thus,Mff localizes

in a Drp1-independent manner to mitochondrialconstrictions at sites of ER contact. We next asked

whether the ER localizes to regions of mitochon-drial constriction in the absence of Mff. Cos-7

cells were depleted ofMff by siRNA and cotrans-fected with GFP-Sec61b (ER) and mito-dsRed.

As expected, mitochondrial morphology waselongated in these cells (Fig. 4, B and D, and

fig. S5E). In cells depleted of Mff, we observedmitochondrial constriction at sites of ER contact,

indicating that ER-mitochondrial contacts formand mark positions of mitochondrial constriction

independently of both Mff and Drp1 recruitment(Fig. 4D).

Here, we have shown that ER-mitochondrialcontacts are a conserved feature of mitochondrial

division. We envision two ways that ER contactmight directly regulate mitochondrial division: (i)

ERproteins intimately participate in division, and/or(ii) ER tubules physically wrap around and con-

strict mitochondria to a diameter comparable toDnm1 and Drp1 helices to facilitate their recruit-

ment and assembly to complete fission (fig. S9).The latter is attractive given that the diameter of

Dnm1 helices (~110 to 130 nm) is considerablynarrower than that of mitochondria and is quite

similar to the diameter of constricted mitochon-dria at ER tubule contacts (138 nm and 146 nm)

(4, 6–9). Regardless of the exact mechanism, the

ER appears to mark the division site and is likelyto be an active participant in this process, because

it remains in contact with themitochondria throughthe entire fission event. Many human diseases are

associated with excessivemitochondrial division,raising the intriguing possibility that these diseases

could involve an alteration of ER-mitochondrialcontacts.

References and Notes1. D. H. Cho, T. Nakamura, S. A. Lipton, Cell. Mol. Life Sci.

67, 3435 (2010).

2. Y. Yoon, C. A. Galloway, B. S. Jhun, T. Yu, Antioxid. Redox

Signal. 14, 439 (2011).

3. L. L. Lackner, J. M. Nunnari, Biochim. Biophys. Acta

1792, 1138 (2009).

4. E. Ingerman et al., J. Cell Biol. 170, 1021 (2005).

5. S. Gandre-Babbe, A. M. van der Bliek, Mol. Biol. Cell 19,

2402 (2008).

6. A. Legesse-Miller, R. H. Massol, T. Kirchhausen, Mol. Biol.

Cell 14, 1953 (2003).

7. A. M. Labrousse, M. D. Zappaterra, D. A. Rube,

A. M. van der Bliek, Mol. Cell 4, 815 (1999).

8. Y. Yoon, K. R. Pitts, M. A. McNiven, Mol. Biol. Cell 12,

2894 (2001).

9. J. A. Mears et al., Nat. Struct. Mol. Biol. 18, 20

(2011).

10. O. M. de Brito, L. Scorrano, EMBO J. 29, 2715

(2010).

11. O. M. de Brito, L. Scorrano, Nature 456, 605

(2008).

12. B. Kornmann et al., Science 325, 477 (2009); 10.1126/

science.1175088.

13. J. R. Friedman, B. M. Webster, D. N. Mastronarde,

K. J. Verhey, G. K. Voeltz, J. Cell Biol. 190, 363

(2010).

14. M. West, N. Zurek, A. Hoenger, G. K. Voeltz, J. Cell Biol.

193, 333 (2011).

15. G. K. Voeltz, W. A. Prinz, Y. Shibata, J. M. Rist,

T. A. Rapoport, Cell 124, 573 (2006).

16. H. Sesaki, R. E. Jensen, J. Cell Biol. 147, 699

(1999).

17. E. Smirnova, L. Griparic, D. L. Shurland,

A. M. van der Bliek, Mol. Biol. Cell 12, 2245

(2001).

18. H. Otera et al., J. Cell Biol. 191, 1141 (2010).

19. Materials and methods are available as supporting

material on Science Online.

Acknowledgments: This work is supported by NIH grant

R01 GM083977 and a Searle Scholar award (to G.K.V.),

NIH training grant GM08759 (to J.R.F.), NIH grant

R01 GM062942 and an American Heart Innovative

Research Grant (to J.N.), and grants from the Biological

Sciences Initiative (BURST grant) and the Undergraduate

Research Opportunity Program at the University of

Colorado (to J.R.D.). We thank the Boulder 3D Electron

Microscopy facility for shared equipment and helpful

suggestions.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/science.1207385/DC1

Materials and Methods

SOM Text

Figs. S1 to S9

References (20–29)

Movies S1 to S5

22 April 2011; accepted 17 August 2011

Published online 1 September 2011;

10.1126/science.1207385

Antimicrobial Peptides Keep InsectEndosymbionts Under ControlFrédéric H. Login,1,2 Séverine Balmand,1,2 Agnès Vallier,1,2 Carole Vincent-Monégat,1,2

Aurélien Vigneron,1,2 Michèle Weiss-Gayet,2,3 Didier Rochat,4 Abdelaziz Heddi1,2*

Vertically transmitted endosymbionts persist for millions of years in invertebrates and play animportant role in animal evolution. However, the functional basis underlying the maintenanceof these long-term resident bacteria is unknown. We report that the weevil coleoptericin-A (ColA)antimicrobial peptide selectively targets endosymbionts within the bacteriocytes and regulates theirgrowth through the inhibition of cell division. Silencing the colA gene with RNA interferenceresulted in a decrease in size of the giant filamentous endosymbionts, which escaped from thebacteriocytes and spread into insect tissues. Although this family of peptides is commonlylinked with microbe clearance, this work shows that endosymbiosis benefits from ColA,suggesting that long-term host-symbiont coevolution might have shaped immune effectorsfor symbiont maintenance.

Cooperative associations between animals

and symbiotic bacteria are widespread innature and common in insects that exploit

unusually restricted nutritional resources (1). Inmany insects, intracellular bacteria (endosymbionts)

are transmitted vertically and provide nutrient

supplementation to their hosts, thereby im-

proving their adaptive traits and their invasivepower (2–4).

However, maintaining the beneficial nature ofthis long-term relationship requires both the host

and the symbiont to constrain adaptive interac-

tions. Genomic and evolutionary data have shownthat major deletions andmutations of genes occur

in endosymbionts, some of which are involved inbacterial virulence and host tolerance (5–7). Data

on how host immune systems have evolved totolerate cooperative bacteria remain scarce and

are mainly limited to extracellular associationswith environmental and/or horizontal symbiont

transmission (8, 9).

To protect permanent endosymbionts fromthe host’s systemic immune response, and prevent

competition with opportunistic invaders, sym-bionts are sequestered in bacteria-bearing host

cells, called the bacteriocytes, which, in some spe-cies, group together to form a bacteriome (10). To

investigate the immune specificities of bacterio-cytes, we have studied associations with Sitophilus

1INSA-Lyon, INRA, UMR203BF2I, Biologie Fonctionnelle Insecteset Interactions, F-69621Villeurbanne, France. 2Université de Lyon,F-69003 Lyon, France. 3Université Lyon 1, CNRS UMR5534,Centre de Génétique et de Physiologie Moléculaire et Cellulaire,F-69622 Villeurbanne, France. 4INRA, Université Pierre et MarieCurie, UMR1272 Physiologie de l’Insecte Signalisation et Com-munication, F-78026 Versailles, France.

*To whom correspondence should be addressed. E-mail:abdelaziz.heddi@insa-lyon.fr

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org362

REPORTS

genus. This genus of beetle includes three cerealpest species (Sitophilus oryzae, S. zeamais, and S.

granarius), all of which share intimate intracellularsymbiosiswith aGram-negative g-Proteobacterium,

called Sitophilus primary endosymbiont (SPE)(11, 12). In contrast to Buchnera in aphids, the

Sitophilus association with SPE is relatively re-cent [dating to 20million years ago (Ma)] (13, 14).

The SPE genome has not suffered appreciablegenome shrinkage (15) and retains secretion sys-

tems and bacterial wall elements that are impli-cated in bacterial recognition by the host immune

receptors (6). Transcriptomic data revealed thatSPE induces a strong systemic response when in-

jected into weevil hemolymph (16), whereas per-

manent infection of bacteriocytes with SPE leadsto the up-regulation of only one antimicrobial

peptide–encoding gene coleoptericin-A (colA,fig. S1) (16).

In immunohistochemistry experiments withantibody against ColA (14), we found that, in apo-

symbiotic insects, colA is expressed constitutivelyin epithelial cells surrounding the intestine and in

the fat body, with a high concentration under thecuticle (Fig. 1B and fig. S2). In symbiotic insects,

colA is further expressed in all the tissues housingendosymbionts (Fig. 1B). Similarly, follicular cells

also expressed ColA as a thin layer around theoocytes, and ColA signals were stronger in bacte-

riocytes surrounding bacteriomes (Fig. 1B). Thus,a relatively high expression of colA in tissues facing

the external environment, and at the boundary of

tissues housing endosymbionts, supports the ideathat ColA may either prevent pathogen intrusion

or retain endosymbionts within the bacteriomesand oocytes. Moreover, ColA appeared to colocal-

ize with endosymbionts in bacteriocytes (Fig. 1B).Using immunogold electronmicroscopy,we found

ColA located inside SPE cytoplasm (Fig. 1C), andsomeColA spotswere also attached to the bacterial

membrane surface. Overall, microscopic observa-tions indicate that ColA expression targets endo-

symbionts in both somatic and germ cell lines.We tested the antimicrobial activity of ColA

against microbes. The weevil paralog ColB wasused for comparison because the colB gene,

unlike colA, is down-regulated in bacteriocytes(16) and ColB shows important sequence identity

with ColA (fig. S3). ColA and ColB showed

similar bactericidal activity against the Gram-positive Micrococcus luteus and the Gram-

Fig. 1. ColA peptide distribution in weevil tis-sues. (A) Schemes of larva and adult weevilsshowing bacteriome localizations (in red). (B) (Up-per panel) Tissues from aposymbiotic S. zeamais:Left andmiddle images are cuticle sections stainedwith a preimmune serum [negative controls canbe found in SOM (14)] and with antibody againstColA (anti-ColA), respectively; the right imageis a gut section stained with anti-ColA. ColA

signals are detected in the fat body, with relatively high intensity under the cuticle and within gut epithelial cells. (Middle and lower panels) Tissues fromsymbiotic insects stained with anti-ColA. Middle panel: ColA signals can be seen in oocytes and follicular cells (left), in apical bacteriomes of ovaries(middle), and in adult mesenteric caeca (right); lower panel: ColA signals in larval bacteriome (left), within bacteriocytes (middle), and in bacteriomesquashes (right) (14). Arrows indicate high ColA signals at the periphery of tissues and show ColA colocalizing with SPE in bacteriocytes and bacteriomesquashes. (C). Immunogold staining of SPE with anti-ColA. Bacteriocyte sections are shown with ColA spots inside symbiont cytoplasm and attached tobacterial membranes.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 363

REPORTS

negative Escherichia coli (Fig. 2), whereas theyeast Saccharomyces cerevisiae tolerated both

peptides (table S1). Both peptides exhibited awide range of bacteriostatic activity against Gram-

negative bacteria. SPE persists in the bacteriocytebecause colA expression may lead either to bac-

terial growth inhibition or dose-dependent bacte-rial clearance. At bacteriostatic concentrations,

both coleoptericins halt E. coli growth, but onlyColA inhibited cell division or caused bacterial

gigantism (Fig. 2B). The shape of M. luteus wasnot affected by ColA or by ColB (fig. S4). Thus,

ColA and ColB have distinct functions in weevilimmunity and symbiosis in regard to their general

effect on Gram-negative bacteria.

Since the discovery of coleoptericins (17, 18),their function in the immune system and their

role in symbiosis have not yet been explored.One notable observation is that all the coleop-

teran endosymbionts observed exhibit a similarelongated morphology (Fig. 2C), resembling

symbionts in other associations. For example, inRhizobium-legume symbiosis, Rhizobium elon-

gation has been interpreted to be the result ofrepeated chromosome DNA replication without

cell cytokinesis (19). We measured the relationbetween bacterial size and genome amplifica-

tion in E. coli, SPE, and Nardonella, the ances-tral endosymbiont of weevils (125 Ma) (13). All

bacteria were polyploid, and bacterial size washighly correlated with chromosome number. The

highest scores were seen inNardonella, with 120chromosomes observed in a giant cell of 200 mm

(Fig. 2D). Bacterial division in plants is inhibitedby nodule-specific cysteine-rich peptides (20) that

induce irreversible elongation of bacteria and ren-der them incapable of multiplying in vitro. By

using phylogenetically unrelatedmolecules, plantsand animals target bacterial cytokinesis while

preserving DNA replication, hence “domesticat-ing” the bacteria as symbionts.

To elucidate the mechanism by which ColAreaches the bacterial cytoplasm and elicits cell

elongation, we used far-Western blotting to iden-tify bacterial molecules targeted by ColA and

ColB peptides. ColA specifically interacted withOmpA, OmpC, rp-L2, EF-Ts, and GroEL (fig.

S5 and table S2). No interaction was detected

with Hsp60, the eukaryotic cytosolic homologof GroEL. ColB interacted with OmpC and pro-

teins involved in translation, but not with OmpAor GroEL.

As with colicins and phages (21), it is likelythat Omps are receptors that allow ColA to enter

the cell. Tight attachments of ColA to the SPEmembrane (Fig. 1C) support this assumption, and

SPE genome sequence analysis showed that SPEencodes a functional ompC gene (table S3). Al-

though the genome sequence is not available, wefound that ColA targets Nardonella of the palm

weevil Rhynchophorus ferrugineus (fig. S6), sug-gesting that ColA may have a broad impact on

weevil symbioses. Whether Nardonella has re-tained omps orwhether ColA enters the bacterium

by other mechanisms remains to be determined.

We propose that after entering the cytoplasm,ColA elicits cell elongation through interaction

with GroEL, because ColA, but not ColB, inter-acts with GroEL protein, and because groELmu-

tations in E. coli also trigger cell gigantism (22).The absence of any interaction between ColB and

GroEL also indicates that ColAmay have evolveda specific interaction with GroEL. In this con-

text, it is notable that GroEL is the most abundantprotein in insect endosymbionts (23); however,

selective up-regulation of groEL has often beeninterpreted as an adaptive mechanism for protein

folding in endosymbionts with a high A+T biasin the genome (24).

We used RNA interference (RNAi) to inhibitcolA transcription in the larval weevils. Injection

of dsRNA–colA resulted in a significant reduc-tion in the number of bacteriocyte colA transcripts

and abundance of ColA peptide for more than 2weeks (fig. S7). In contrast to the plant-Rhizobium

interaction, colA inhibition resulted in the SPEpopulation declining by half (Fig. 3A). However,

whether this was due to resumption of cytokinesisor multiplication of small bacteria is unclear. We

Fig. 2. ColA and ColB activities against bacteria. (A) ColA (solid line) and ColB (dashed line) activitiesagainst E. coli (triangles) and M. luteus (squares). ColA and ColB have a similar inhibitory effect(analysis of variance, P =0.76). They show bactericidal activities against M. luteus at low concentrations.For E. coli, low concentrations of ColA and ColB have bacteriostatic activity, and higher concentrationskill this bacterium. (B) Effect of low concentrations of ColA and ColB on E. coli morphology. Bacteriawere incubated in LB broth (left, control), in LB with 8 mM ColA (middle), or in LB with 8 mM ColB (right).Cell gigantism is observed with ColA peptide only. (C) Gram staining of endosymbionts from S. oryzae(rod-shaped, left), S. zeamais (spiral, middle), and R. ferrugineus (filamentous, right). (D) Chromosomevisualization of E. coli treated with 8 mM ColA (left), SPE (middle), and Nardonella (right). SPE andNardonella were isolated from larval bacteriomes of S. oryzae and R. ferrugineus, respectively. Chro-mosome number was highest in Nardonella.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org364

REPORTS

quantified SPE DNA by quantitative polymerasechain reaction (qPCR) and showed that colA in-

hibition did not affect bacterial chromosomereplication (fig. S8). This, and the decreased num-

ber of large-sized cells after RNAi treatment (Fig.3A), supports the former hypothesis (i.e., resump-

tion of cytokinesis and loss of the elongated form)but does not exclude the latter (multiplication of

the small-sized cells) if we consider bacterial turn-over and de novo synthesis processes.

To understand ColA function in symbiosis,we monitored SPE in insects by fluorescence

in situ hybridization (FISH) after dsRNA-colA

injections. Unexpectedly, several SPE cells exitedthe bacteriome on the ninth day after treatment

(Fig. 3B). This phenomenon increased at day 14,when bacteria were found spread throughout the

larval tissues. Nevertheless, symbiont “escape”from the bacteriome did not affect insect mortal-

ity under laboratory conditions (table S4), althoughthese data indicate that ColA acts to prevent

bacterial tissue invasion.The weevil ColA peptide demonstrates sev-

eral properties important to immunity and sym-biosis. It appears to act as a first line of defense in

insects against microbial intrusion, and the range

of bacteriostatic and bactericidal activities ofColA suggests its precise regulation of endosym-

biont number and location. The interaction ofGroEL with ColA (but not with ColB) supports

the idea that long-term coevolution may have se-lected ColA for this symbiotic function.

References and Notes1. N. A. Moran, Curr. Biol. 16, R866 (2006).

2. A. E. Douglas, Annu. Rev. Entomol. 43, 17 (1998).

3. A. Heddi, A. M. Grenier, C. Khatchadourian, H. Charles,

P. Nardon, Proc. Natl. Acad. Sci. U.S.A. 96, 6814 (1999).

4. A. G. Himler et al., Science 332, 254 (2011).

5. S. Shigenobu, H. Watanabe, M. Hattori, Y. Sakaki,

H. Ishikawa, Nature 407, 81 (2000).

6. C. Dale, G. R. Plague, B. Wang, H. Ochman, N. A. Moran,

Proc. Natl. Acad. Sci. U.S.A. 99, 12397 (2002).

7. B. L. Weiss, Y. Wu, J. J. Schwank, N. S. Tolwinski,

S. Aksoy, Proc. Natl. Acad. Sci. U.S.A. 105, 15088 (2008).

8. J. L. Round, S. K. Mazmanian, Nat. Rev. Immunol. 9,

313 (2009).

9. J. V. Troll et al., Environ. Microbiol. 12, 2190 (2009).

10. C. Anselme, A. Vallier, S. Balmand, M. O. Fauvarque,

A. Heddi, Appl. Environ. Microbiol. 72, 6766 (2006).

11. A. Heddi, H. Charles, C. Khatchadourian, G. Bonnot,

P. Nardon, J. Mol. Evol. 47, 52 (1998).

12. H. Charles, A. Heddi, Y. Rahbé, C. R. Acad. Sci. 324,

489 (2001).

13. C. Conord et al., Mol. Biol. Evol. 25, 859 (2008).

14. Supporting material is available on Science Online.

15. H. Charles, G. Condemine, C. Nardon, P. Nardon,

Insect Biochem. Mol. Biol. 27, 345 (1997).

16. C. Anselme et al., BMC Biol. 6, 43 (2008).

17. P. Bulet et al., J. Biol. Chem. 266, 24520 (1991).

18. A. Sagisaka, A. Miyanoshita, J. Ishibashi, M. Yamakawa,

Insect Mol. Biol. 10, 293 (2001).

19. P. Mergaert et al., Proc. Natl. Acad. Sci. U.S.A. 103,

5230 (2006).

20. W. Van de Velde et al., Science 327, 1122 (2010).

21. D. Fourel, C. Hikita, J. M. Bolla, S. Mizushima,

J. M. Pagès, J. Bacteriol. 172, 3675 (1990).

22. E. Chapman et al., Proc. Natl. Acad. Sci. U.S.A. 103,

15800 (2006).

23. H. Charles, A. Heddi, J. Guillaud, C. Nardon, P. Nardon,

Biochem. Biophys. Res. Commun. 239, 769 (1997).

24. N. A. Moran, Proc. Natl. Acad. Sci. U.S.A. 93, 2873 (1996).

Acknowledgments: This work was supported by INRA, INSA

de Lyon, the French ANR-06-BLAN-0316 (EndoSymArt)

and ANR-2010-BLAN-170101 (ImmunSymbArt), a

grant from Région Rhône-Alpes (cluster infectiologie),

and the COST action FA0701 (Arthropod Symbioses).

The data reported in this paper are posted in the

supporting online material (14). colA (EY122872) and

colB (EY122826) sequences are published in GenBank.

SPE omp sequences (GenBank JN575265, JN575266,

JN575267) were provided by C. Dale and R. B. Weiss

(University of Utah) from an ongoing SPE genome

sequencing and annotation project supported by NSF

grant EF-0523818 and the Ministerio de Educación y

Ciencia project BFU2006-06003/BMC to A. Moya and

A. Latorre (University of Valencia). We thank R. Gil and

K. Oakeson for omp sequence analysis; V. E. Shevchik,

G. Condemine, and M. Lemaire for supplying E. coli

and S. cerevisiae strains; W. J. Miller and B. Loppin for

critical reading of the manuscript; and V. James for

correction of English. Electron microscopy was carried

out in Centre Technologique des Microstructures (UCBL).

This paper is an homage to the work of Paul Nardon.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/334/6054/362/DC1

Materials and Methods

Figs. S1 to S9

Tables S1 to S6

References (25, 26)

13 June 2011; accepted 7 September 2011

10.1126/science.1209728

Fig. 3. Effects of colA inhibition with RNAi on SPE size and location. (A) Typical forward scatter–area/side scatter–area plots showing size (x axis) and granularity (y axis) of SPE isolated from larvae injectedwith dsRNA-gfp (left) and with dsRNA-colA (right). Three SPE populations were defined arbitrarily.Small-sized cells (P1) and intermediate-sized cells (P2) significantly increased with dsRNA-colA treatment,whereas the population of large-sized cells (P3) decreased (c2-test, P < 0.0001; see percentage values).The mean size of P2 and P3 significantly decreased (Mann-Whitney test, P < 0.0001), whereas the meansize of P1 was equal in dsRNA-gfp– and dsRNA-colA–injected larvae (Mann-Whitney test, P = 0.37). (B)FISH visualization of SPE 9 days (upper panels) and 14 days (lower panels) after larvae were injected withdsRNA-gfp (left) and dsRNA-colA (right). colA inhibition resulted in SPE escaping the bacteriome (seearrows). See table S4 for experimental details.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 365

REPORTS

Stochastic Pulse Regulation inBacterial Stress ResponseJames C. W. Locke,* Jonathan W. Young,* Michelle Fontes,

María Jesús Hernández Jiménez, Michael B. Elowitz†

Gene regulatory circuits can use dynamic, and even stochastic, strategies to respond to environmentalconditions. We examined activation of the general stress response mediated by the alternative sigmafactor, sB, in individual Bacillus subtilis cells. We observed that energy stress activates sB in discretestochastic pulses, with increasing levels of stress leading to higher pulse frequencies. By perturbingand rewiring the endogenous system, we found that this behavior results from three key features of thesB circuit: an ultrasensitive phosphorylation switch; stochasticity (“noise”), which activates that switch;

and a mixed (positive and negative) transcriptional feedback, which can both amplify a pulse and switchit off. Together, these results show how prokaryotes encode signals using stochastic pulse frequencymodulation through a compact regulatory architecture.

Gene circuits exhibit fluctuations (“noise”)in the concentrations of key components

such as transcriptional regulatory pro-teins (1, 2). Increasingly, noise appears to play

functional roles in some systems (3–5). For ex-ample, noise could enable a subpopulation of

cells to enter a transient antibiotic-resistant state,enhancing their survival (6). However, it remains

unclear how genetic circuits use noise to control

cellular behaviors. To address this issue, we an-alyzed sB, the transcriptional regulator of general

stress response in Bacillus subtilis, at the single-cell level (7–9). Here, we show how s

B controls

its target genes through sustained pulsing, ratherthan continuous activation; how noise enables

this behavior; and how stress levels modulate thefrequency of these pulses (Fig. 1A).

In prokaryotes, alternative sigma factors forma part of the RNA polymerase holoenzyme, di-

recting it to regulons that control distinct regu-latory programs (10). sB is found in Gram-positive

bacteria and impacts pathogenicity in Listeria

monocytogenes and Staphyloccous aureus (11, 12).

In B. subtilis, sB activates more than 150 target

genes in response to diverse stresses (7, 13). sB

is kept inactive by its anti-sigma factor RsbW

and is activated by the anti-anti-sigma factorRsbV, which can be reversibly phosphorylated

(see Fig. 1B for regulatory interactions). Toanalyze s

B activation dynamics, we constructed

reporter strains incorporating a yellow fluores-cent reporter (yfp) for sB activity (Fig. 1B) and

used quantitative time-lapsemicroscopy to followsB activation in individual cells (14). To quantify

sB activity in movies, we computed the PsigB

promoter activity, defined as the rate of produc-

tion of yellow fluorescent protein (YFP) [fig. S1Aand supporting online material (SOM)].

We first measured the response of sB to my-

cophenolic acid (MPA), an energy stress trans-duced by RsbQP (fig. S2) (15). Constant MPA

led to pulses of sB activation in individual cells(Fig. 1, C and D, and movie S1). These pulses

were unsynchronized across the population, spo-radic in time, and sustained, continuing through-

out the movie (about six generations) (fig. S3).Similar behavior was observed with other energy

stresses and during growth in liquid culture (figs.S1 and S4). Pulses reflected changes in s

B ac-

tivity and not intrinsic variability of the PsigB -YFPpromoter (fig. S5). Increasing MPA concentra-

tion caused a strong increase in pulse frequency,with weaker increases in mean pulse amplitude

and duration (Fig. 1E and fig. S6), showing thatsB is regulated predominantly by frequency mod-

ulation (FM) in response to energy stress (Fig. 1A).Pulse amplitudes exhibited broad and monoton-

ically increasing variability with increasingMPA,

Howard Hughes Medical Institute, Division of Biology and Bio-engineering, Broad Center, California Institute of Technology,1200 East California Boulevard, Pasadena, CA 91125, USA.

*These authors contributed equally to this work.†To whom correspondence should be addressed. E-mail:melowitz@caltech.edu

Fig. 1. Energy stress modulates the frequency of stochastic pulses of sB ac-tivation. (A) Schematic of FM pulse regulation. The input signal (black line)controls the frequency of stochastic pulses (blue line, schematic). (B) Sche-matic diagram of sB regulatory interactions and states (7). When RsbV (V) isphosphorylated (OFF state), sB is sequestered by RsbW (W) and inactive (28).Under energy stresses such as MPA, RsbV is dephosphorylated by the RsbQPphosphatase complex (QP) (29). Other stress inputs are mediated by the RsbTUphosphatase complex (not shown; see SOM text for discussion). Dephosphoryl-ated RsbV can bind to RsbW, releasing sB to activate target genes, including

its own operon (30), and the yfp reporter (yellow). (C) Promoter activity of thePsigB -YFP reporter pulses in individual lineages (colored solid lines), and itsmean and standard deviation across all lineages in four data sets (dashed lineand shaded area, respectively). (D) Filmstrip of sB activation at 60 mg/mLMPA.Heterogeneous expression levels of PsigB -YFP reflect pulsing activity. (E) MPAconcentration strongly modulates the mean frequency, while more weakly mod-ulating the mean amplitude and duration, of pulses. Error bars, mean T SEM. (F)Pulse amplitude histograms for varying levels of MPA. In (E) and (F), each datapoint represents data from four microcolonies, acquired on two different days.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org366

REPORTS

with coefficients of variation ranging from 0.16to 0.70 (Fig. 1F). Together, these results provoke

the question of how FM pulse regulation is im-plemented by the sB regulation circuit.

In principle, pulses could be generated intwo qualitatively distinct ways: They could arise

through amplification of an inherently stochasticunderlying process (16, 17). Alternatively, they

could result from a limit cycle oscillator whosedynamics become erratic due to noise (18). Sys-

tematically reducing cellular noise would distin-guish between these possibilities, eliminating pulses

in the first case but making them more regular inthe second case.

To modulate the amplitude of noise in cells,

we created strains that could be induced to growinto long multinucleoid filaments by controlling

expression of FtsW, a cell division protein neces-sary for septation. These cells exhibit similar mean

expression levels of cellular components (fig. S7)but reduced fluctuations (19), allowing us to in-

vestigate how reduced noise affects pulse fre-

quency. In time-lapse movies, we observed asystematic decline in pulse frequency in noise-

reduced (long) cells across a range of energy stresslevels (Fig. 2A and fig. S8), which was also con-

sistent with reduced cell-cell variability in sB ac-tivity in liquid conditions (Fig. 2B). This reduction

in pulse frequency did not reflect reduced sensi-tivity toMPA,which had a similar effect on growth

rate in long and short cells (fig. S8). Together, theseresults rule out limit cycle models and suggest a

noise-dependent mechanism for pulse generation.How, then, does the sB circuit amplify noise

to initiate discrete pulses of sB activity? To ad-dress this question, we analyzed the response of

sB to increased expression of each circuit com-

ponent. Up-regulation of kinase (RsbW) and phos-phatase (RsbQP) expression had much stronger

(opposite) effects on sB activity compared with

up-regulation of RsbV (fig. S9). This result is in-

teresting because opposing kinase and phos-phatase activities can generate sharp, switchlike

responses in the phosphorylation of their sub-

strate. An extreme example is zero-order ultrasen-sitivity, where the phosphatase and kinase operate

at saturation (20–22). We found that sB activityexhibited an ultrasensitive response to inducible

phosphatase concentration (Fig. 2C and fig. S10),with an effective Hill coefficient of 2.12 [95%

confidence interval (CI), nH = 2.09 to 2.15]. Sim-ilar results were also observed with the RsbTU

phosphatase (fig. S11). Moreover, this ultrasen-sitivity was not due to the transcriptional feedback

loop. It could be observed in an “open-loop”strain, in which operon expression was inducible

and independent of sB (fig. S12). In this strain,increasing operon expression led to increasing

ultrasensitivity to phosphatase level. These ef-

fects could be explained by a minimal mathemat-ical model of the phosphoswitch that does not

include the detailed dynamics of the network (seeSOM). Finally, consistent with thismodel, ectopic

expression of the kinase RsbW shifted the switch-ing point to higher phosphatase expression levels

(fig. S13). This ultrasensitive phosphoswitch could

Fig. 2. Pulsing is noise-dependent and involves an ultrasensitive phospho-switch. (A) Pulse frequency in long cells (green) is strongly reduced com-pared with short cells (gray; data replotted from Fig. 1E). Error bars, mean T SEM.(B) Variability in PsigB -YFP expression decreases with increasing cell length(see fig. S8). Equal numbers of cells (represented by dots) are plotted in eachlog-spaced bin (delimited by gray vertical lines). (Inset) Overlay of phase

contrast and PsigB -YFP expression (green) at different cell lengths. Notegreater sB variability in short cells. (C) sB expression is ultrasensitive toRsbQP phosphatase levels. Each dot represents the mean RsbQP-YFP level andPsigB -CFP level of one cell, using the strain shown schematically (table S1). Thered line is a Hill function with Hill coefficient nH = 2.12 (95% CI, nH = 2.09 to2.15).

Fig. 3. A mixed transcriptional feedback loop amplifies and terminatespulses. (A) Schematic diagram of supra- and subthreshold protocols. Beforetime-lapse acquisition (gray region), phosphatase is induced to a constantlevel by addition of xylose. After the start of acquisition, isopropyl-b-D-thiogalactopyranoside (IPTG) is added to induce rsbVWB to levels greater than(solid red line) or less than (dashed red line) the level of phosphatase. Thisresults in pulsed (solid green line) or sustained (dashed green line) sB activitydynamics. (B) sB promoter activity exhibits a transition between sub- and

suprathreshold behaviors. Each trace shows the mean PsigB promoter activityaveraged over four colonies. The promoter activity of the IPTG-inducible sB

operon (x axis) was estimated using a separate strain containing a similarIPTG-inducible yfp reporter. Two repeat movies showed similar behaviors.(Inset) Schematic diagram of strain used for this experiment (table S1). (C) Aminimal mathematical model of the open-loop sB network reproduces themainfeatures of the experimental data. (Inset) In this model, the unphosphorylatedactivator, A, directly activates target genes (see SOM).

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 367

REPORTS

activate sB in response to fluctuations in the phos-phatase/kinase ratio and thereby initiate pulses.

How are sB pulses further amplified and sub-sequently terminated? sB activates its own ope-

ron (rsbV-rsbW-sigB). This feedback loop couldincrease sB activity, due to the activating effects

of RsbV and sB, or it could repress sB activity,

due to increased production of RsbW. We hy-

pothesized that the phosphoswitch sets a thresh-old between activating (phosphatase dominant)

and repressing (kinase dominant) feedback re-gimes. As long as phosphatase activity exceeds

kinase activity, activation of the operon increasesfree sB (positive feedback). However, this also in-

creases production of RsbW kinase. When kinase

activity approaches that of the phosphatase, in-creased operon expression will cause RsbW levels

to cross the threshold, shutting off activation(negative feedback). Thus, autoregulation could

result in a “mixed” (positive and negative) feed-back loop, providing a compact mechanism to

first amplify and then terminate a pulse (23).To test this hypothesis, we constructed an

open-loop strain and quantified the change in sB

activity in response to a step increase in operon

expression. In these experiments, we first estab-lished a basal level of phosphatase activity in

cells and subsequently induced a step, of varyingsize, in s

B operon expression (Fig. 3A). We ob-

served a striking transition between two qualita-tively different responses: At lower operon induction

levels, the system produced a sustained response,whereas at higher induction levels, it exhibited a

pulse (Fig. 3B). These results are consistent with

the mixed feedback model: Initially, increasedoperon induction produces more s

B, which is

active due to the high levels of phosphatase, en-gaging the positive feedback loop. For lower

(subthreshold) operon induction levels, RsbW lev-els never exceed phosphatase levels, so the sys-

tem remains on indefinitely (Fig. 3A, dashed lines).In contrast, at higher induction levels (superthresh-

old), RsbW activity eventually crosses the thresh-old set by the phosphatase and thereby shuts the

system off, resulting in a pulse (Fig. 3A, solidlines). Indeed, RsbW dominated other operon

components at steady state, suppressing sB ac-

tivity in a dose-dependent manner (fig. S9). In

this mechanism, pulse amplitude should be rough-

ly proportional to the difference between the phos-phatase level and the kinase level immediately

after the initiating fluctuation, as confirmed ex-perimentally (fig. S14). A minimal mathematical

model of the circuit exhibited qualitatively simi-lar behavior (Fig. 3C). This mechanism for pulse

initiation, amplification, and termination is sum-marized in Fig. 4, A and B.

These results provoke a final question: Howcan the cell modulate the frequency of pulses in

this system? Systematic changes in the activity ofeither kinase or phosphatase could modulate the

likelihood of threshold-crossing events and there-by control pulse frequency.We first examined the

distribution of RsbQP expression levels, using aPrsbQP-rsbQP-YFP protein fusion that complements

an RsbQP null mutant (fig. S15). In response to40 mg/ml MPA, we observed a ~3-fold increase

in mean RsbQP-YFP levels, and a ~6-fold in-

crease in mean sB activity (fig. S16, A and B).

At the single-cell level, RsbQP-YFP expression

mapped to sB activity (fig. S16C), closely fol-

lowing the (independently determined) ultrasen-

sitive response function (Fig. 2C). These resultssuggest that stress increases sB pulse frequency

by increasing the distribution of RsbQP expres-sion levels and thereby increasing the frequency

with which RsbQP fluctuations cross the thresh-old set by RsbW (Fig. 4C). These results do not

rule out the complementary possibility, suggestedpreviously (24), that some energy stresses may

activate sB by reducing kinase activity.To show that this mechanism is indeed suf-

ficient to enable frequency modulation, we re-

wired the endogenous circuit, replacing RsbQPwith an inducible, constitutively active RsbTU

phosphatase complex that was unaffected by en-ergy stress (Fig. 4D, inset). The rewired system

exhibited stochastic pulsing in response to RsbTUexpression (figs. S17 and S18 and movie S2).

Furthermore, we observed an increased frequen-cy of pulsing in response to increased RsbTU

phosphatase expression (Fig. 4D), with weakereffects on pulse amplitude and duration (fig. S20).

These results, qualitatively similar to those ob-served in wild-type cells under energy stress, also

match an extended mathematical model that in-cludes the wild-type transcriptional feedback

(gray dashed lines in Fig. 4D, figs. S19 and S20)(see SOM). Thus, modulation of phosphatase ex-

pression is sufficient to recapitulate FM pulsing,and no special property of the RsbQP phospha-

tase is required.

Fig. 4. Mechanism of FM pulse con-trol. (A) Schematic time course of phos-phatase RsbQP (denoted P, purple),free s

B (sB, green), and kinase (W,red) during a pulse cycle. Circled num-bers indicate specific steps in (B). (B)Schematic diagram of pulse control.The relative concentration of each com-ponent is indicated by size. (1) Initialstate: System components are at lowlevels, kinase activities exceed phos-phatase activities, and therefore RsbVis mostly phosphorylated. A threshold-crossing upward fluctuation in RsbQPlevel dephosphorylates VP, leading to(2) Pulse Initiation. Activation of sB

(indicated by glowing halo) leads toup-regulation of operon components(operon feedback). (3) Pulse peak: sB

activity peaks just before RsbW kinaseactivity exceeds phosphatase activity.(4) Termination: Rephosphorylationof RsbV shuts the system off. (5) Di-lution: Component levels reset to theoriginal state. (C)Mechanismof frequen-cy modulation. Fluctuations in phos-phatase level (purple arrow from state 1 to 2) can cross the kinase threshold(red line) to initiate a pulse, with amplitude determined by the size of fluc-tuation (dashed line). Increased stress shifts the distribution of phosphataselevels from lower to higher values (dark and light gray, respectively), in-creasing the frequency of threshold-crossing events and thereby increasing

pulse frequency (inset). (D) Tuning of phosphatase expression by IPTG (strainindicated schematically in inset) can regulate pulse frequency. Gray dashedlines show a similar behavior for the mathematical model (fit to data). Eachdata point represents statistics from two colonies. Two repeat data sets showedsimilar trends.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org368

REPORTS

FM pulsing can be implemented by a simplecircuit of three genes (rsbW, rsbV, and sigB), with

input from a phosphatase complex. This systemprovides a fundamental signal-processing capabil-

ity to bacterial cells, enabling them to convertsteady “DC” inputs into pulsatile, predominantly

“AC” outputs. Noise plays a key functional rolein this signal processing system (3). The sB cir-

cuit conserves its core architecture in diversebacteria (7), and other alternative sigma factors

similarly feature both posttranslational regulationby anti-sigma factors and autoregulatory feed-

back. Thus, related stochastic pulse modulationschemes are likely employed more generally in

bacteria (10). The relatively slow time scale of sB

pulses (Fig. 1E) could confer advantages in re-sponding to unpredictable environments andmain-

taining a broad, but dynamic, distribution of statesin the population through bet-hedging (25, 26).

Given the negative effect of sB activation on

growth rate in some conditions, even under energy

stress (27), these results suggest that cells balancethe benefits and costs of sB activation dynami-

cally. It will be interesting to see whether otherdynamic encoding schemes are similarly imple-

mented by relatively simple circuit modules.

References and Notes1. A. Raj, A. van Oudenaarden, Cell 135, 216 (2008).

2. N. Rosenfeld, J. W. Young, U. Alon, P. S. Swain,

M. B. Elowitz, Science 307, 1962 (2005).

3. A. Eldar, M. B. Elowitz, Nature 467, 167 (2010).

4. R. Losick, C. Desplan, Science 320, 65 (2008).

5. L. Cai, C. K. Dalal, M. B. Elowitz, Nature 455, 485

(2008).

6. E. Rotem et al., Proc. Natl. Acad. Sci. U.S.A. 107,

12541 (2010).

7. M. Hecker, J. Pané-Farré, U. Völker, Annu. Rev. Microbiol.

61, 215 (2007).

8. W. G. Haldenwang, R. Losick, Nature 282, 256

(1979).

9. O. A. Igoshin, M. S. Brody, C. W. Price, M. A. Savageau,

J. Mol. Biol. 369, 1333 (2007).

10. T. M. Gruber, C. A. Gross, Annu. Rev. Microbiol. 57,

441 (2003).

11. M. J. Kazmierczak, S. C. Mithoe, K. J. Boor, M. Wiedmann,

J. Bacteriol. 185, 5722 (2003).

12. U. Lorenz et al., Microbes Infect. 10, 217 (2008).

13. M. Hecker, U. Völker, Mol. Microbiol. 29, 1129

(1998).

14. J. C. Locke, M. B. Elowitz, Nat. Rev. Microbiol. 7, 383

(2009).

15. S. Zhang, W. G. Haldenwang, J. Bacteriol. 187, 7554

(2005).

16. A. L. Hodgkin, A. F. Huxley, J. Physiol. 117, 500 (1952).

17. G. M. Süel, J. Garcia-Ojalvo, L. M. Liberman,

M. B. Elowitz, Nature 440, 545 (2006).

18. M. B. Elowitz, S. Leibler, Nature 403, 335 (2000).

19. G. M. Süel, R. P. Kulkarni, J. Dworkin, J. Garcia-Ojalvo,

M. B. Elowitz, Science 315, 1716 (2007).

20. A. Goldbeter, D. E. Koshland Jr., Proc. Natl. Acad.

Sci. U.S.A. 78, 6840 (1981).

21. G. J. Melen, S. Levy, N. Barkai, B. Z. Shilo, Mol. Syst. Biol.

1, 2005.0028 (2005).

22. Z. Cheng, F. Liu, X. P. Zhang, W. Wang, Biophys. J.

97, 2867 (2009).

23. J. C. Ray, O. A. Igoshin, PLOS Comput. Biol. 6, e1000676

(2010).

24. S. Alper, L. Duncan, R. Losick, Cell 77, 195

(1994).

25. M. Acar, A. Becskei, A. van Oudenaarden, Nature 435,

228 (2005).

26. E. Kussell, S. Leibler, Science 309, 2075 (2005).

27. T. Schweder, A. Kolyschkow, U. Völker, M. Hecker,

Arch. Microbiol. 171, 439 (1999).

28. A. Dufour, W. G. Haldenwang, J. Bacteriol. 176, 1813

(1994).

29. M. S. Brody, K. Vijay, C. W. Price, J. Bacteriol. 183,

6422 (2001).

30. A. A. Wise, C. W. Price, J. Bacteriol. 177, 123

(1995).

Acknowledgments: We thank C. Price and D. Rudner for

providing strains. We thank A. Eldar, R. Kishony, C. Price,

N. Wingreen, J. Levine, and other members of M.B.E’s

laboratory for helpful discussions. Work in M.B.E’s

laboratory was supported by NIH grants R01GM079771

and P50 GM068763, U.S. National Science Foundation

CAREER Award 0644463, and the Packard Foundation.

J.C.W.L was supported by the International Human

Frontier Science Program Organization and the European

Molecular Biology Organization.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/science.1208144/DC1

Materials and Methods

SOM Text

Figs. S1 to S20

Table S1

References

Movies S1 and S2

10 May 2011; accepted 1 September 2011

Published online 6 October 2011;

10.1126/science.1208144

Transgenerational EpigeneticInstability Is a Source ofNovel Methylation VariantsRobert J. Schmitz,1,2 Matthew D. Schultz,1,2,3 Mathew G. Lewsey,1,2 Ronan C. O’Malley,2

Mark A. Urich,1,2 Ondrej Libiger,4 Nicholas J. Schork,4 Joseph R. Ecker1,2,5*

Epigenetic information, which may affect an organism’s phenotype, can be stored and stablyinherited in the form of cytosine DNA methylation. Changes in DNA methylation can producemeiotically stable epialleles that affect transcription and morphology, but the rates of spontaneousgain or loss of DNA methylation are unknown. We examined spontaneously occurring variationin DNA methylation in Arabidopsis thaliana plants propagated by single-seed descent for 30generations. We identified 114,287 CG single methylation polymorphisms and 2485 CG differentiallymethylated regions (DMRs), both of which show patterns of divergence compared with the ancestralstate. Thus, transgenerational epigenetic variation in DNA methylation may generate new allelicstates that alter transcription, providing a mechanism for phenotypic diversity in the absence ofgenetic mutation.

Cytosine methylation is a DNA base mod-ification with roles in development and

disease in animals as well as in silencingtransposons and repetitive sequences in plants

and fungi (1). In plants, CG methylation is com-monly found within gene bodies (2–5), whereas

non-CG methylation, CHG and CHH (where His A, C, or T), is enriched in transposons and re-

petitive sequences (1). The RNA-directed DNAmethylation (RdDM) pathway targets both CG

and non-CG sites for methylation and is com-

monly associated with transcriptional silencing(6). This pathway can also target and silence

protein-coding genes, giving rise to epigenetic al-leles or so-called epialleles that can be heritable

through mitosis and/or meiosis (7, 8) and can bedependent on the methylation of a single CG di-

nucleotide (9).Two meiotically heritable epialleles result-

ing in morphological variation are the peloric

(Linaria vulgaris) and colorless non-ripening

(Solanum lycopersicum) loci (10, 11). Both show

spontaneous epigenetic silencing events within

their respective populations (10, 12). However, thefrequency at which such spontaneous meiotically

heritable epialleles naturally arise in populationsis unknown. Although epiallelic variation has been

identified between genetically diverse populationswithin Arabidopsis thaliana (13), it is unclear

whether these identified epialleles are due tounderlying genetic variation. Epialleles have al-

so been artificially generated after mutagenesisor because of mutations in the cellular com-

ponents required for the maintenance of DNAmethylation (14–16).

An A. thaliana (Columbia-0) population, theMA lines, derived by single-seed descent for 30

generations (17) was used to examine the extentof naturally occurring variation in DNA methyla-

tion and the frequency at which spontaneous epi-alleles emerge over time. We used the MethylC-Seq

method (3) to determine the whole-genome baseresolution DNA methylomes for three ancestral

1Plant Biology Laboratory, The Salk Institute for BiologicalStudies, La Jolla, CA 92037, USA. 2Genomic Analysis Labo-ratory, The Salk Institute for Biological Studies, La Jolla, CA92037, USA. 3Bioinformatics Program, University of Californiaat San Diego, La Jolla, CA 92093, USA. 4The Scripps Transla-tional Science Institute and the Department of Molecular andExperimental Medicine, The Scripps Research Institute, LaJolla, CA 92037, USA. 5Howard Hughes Medical Institute, TheSalk Institute for Biological Studies, 10010 North Torrey PinesRoad, La Jolla, CA 92037, USA.

*To whom the correspondence should be addressed. E-mail:ecker@salk.edu

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 369

REPORTS

MA lines (numbers 1, 12, and 19) and fivedescendant MA lines (numbers 29, 49, 59, 69,

and 119) (fig. S1). We refer to lines 1, 12, and 19as ancestors throughout this study, although

they are not direct ancestors because they arethree generations removed from the original

founder line (fig. S1). These specific descendantlines were selected because their genomes have

been sequenced and they have a known level ofspontaneous mutation (18). Biological replicates

(sibling plants) for each leaf methylome weresequenced to an average of ~34-fold coverage,

which allowed for an average per line exami-nation of 39,897,093 (96.35%) uniquely mapped

cytosines and 5,307,077 (98.39%) uniquely mapped

CGs (table S1).A total of 1,730,761 CGs were methylated

(mCGs) in at least one MA line (Fig. 1A), andabout 91% of the covered mCGs were invar-

iably methylated across all eight lines (19). Thevariable mCGs revealed a set of 114,287 high-

confidence CG single methylation polymorphisms(SMPs) that showed a consensus of the meth-

ylation status of CG dinucleotides between bi-ological replicates (Fig. 1A). Next, a reference

MA founder DNA methylome was created bypooling the completely conserved mCG site

calls for all ancestral MA lines and used to de-termine the frequency of discordant CG-SMP

sites within the descendant population (Fig. 1B).Within the descendant lines, ~1.62% of the CG

methylome shows susceptibility to dynamic ac-quisitions and losses of mCGs over time (table

S2). On average, ~66,000 methylated CG-SMPs(mCG-SMPs) were identified for each ances-

tral and descendant line (fig. S2). Although thetotal number of mCG-SMPs was similar be-

tween all lines, the conservation of these poly-morphisms among and between ancestral and

descendant populations was different (Fig. 1Cand table S3). A pairwise comparison of both

populations for methylation conservation, esti-mated by global similarity of mCG-SMP sites

(19), revealed that all of the ancestral lines arehighly similar (table S4). Descendant lines showed

greater similarity in CG-SMPs methylation sta-tus to ancestral lines than to other descendant

lines (table S4).

We calculated an estimate of the epimutationrate per generation in this population by using

linear regression and TREE PUZZLE, which re-vealed 704 and 2876 methylation changes each

generation, respectively (19). We estimated a lowerbound of the epimutation rate with the linear

regression results, which revealed 4.46 × 10−4

methylation polymorphisms per CG site per gen-

eration (P < 0.0000216) (table S5). This findingcontrasts with the previously reported spontane-

ous genetic mutation rate of 7 × 10−9 base sub-stitutions per site per generation for these same

MA lines (18). The TREE PUZZLE analysis re-vealed higher estimated epimutation rates in earlier

generations (19). One possible source of this var-iation could be due to seed age, storage, and/or

selection for seed survival. Therefore, although

DNA methylation is predominantly static overrelatively long periods of time, changes in cyto-

sine methylation do occur and at a frequencygreater than that of mutation observed at the

DNA sequence level.By using CG-SMPs derived from both an-

cestral and descendant populations, we carriedout a genome-wide analysis of differentially meth-

ylated regions (DMRs) and identified 2485 CG-DMRs that ranged in size from 11 to 1110 base

pairs (bp) (Fig. 2A and table S6). Hierarchicalclustering of CG-DMRs in this population, cal-

culated solely on the basis of the methylationdensity, revealed that the ancestral lines segregate

as an independent cluster from the descendant

lines (Fig. 2B and fig. S3). Multivariate distance-based regression (MDMR) (20, 21) confirmed

this finding, indicating a statistically significant(P < 0.00005) association between ancestor or

descendant status and methylation density of theCG-DMR profiles. The ancestor or descendant

status explained 47% of the variance in the dis-similarity in methylation density of CG-DMRs

between pairs of samples, indicating that, overtime, there is a divergence of DNA methylation

patterns in both formation and elimination of CG-DMRs. Furthermore, the genome-wide locations

of these CG-DMRs were not uniformly distributed(P < 2.20 × 10−16), because 60.5% (1504/2485)

were found in genic regions compared with 3.3%(82/2485) and 36.2% (899/2485) located in in-

tergenic regions and transposons, respectively(Fig. 2B).

Next, we performed a genome-wide surveyfor nonCG-DMRs and uncovered a total of 284

among all eight lines (table S7). In general, thenonCG-DMRs were largely localized to inter-

genic regions (141/284) of the genome, becauseonly 57/284 overlapped with genes and 86/284

overlapped with transposons. The size ranges ofthe nonCG-DMRs were similar to those of the

CG-DMRs because the vast majority occurred insmaller segments of the genome (10 to 682 bp).

Therefore, variation in DNA methylation ap-

pears to occur in all three methylation sequencecontexts.

CG methylation is present within gene bodiesand is enriched toward the 3′ end (2–5), whereas

CG and nonCG methylation is associated withheterochromatin, transposons, and repetitive se-

quences (1). In agreement with these findings, weobserved that the 3′ portion of genes contained

the greatest source of CG-DMRs and that themajority of nonCG-DMRs were enriched out-

side of the gene bodies (Fig. 2C). Furthermore,we observed a ~twofold depletion of CG-DMRs

in exons compared with introns (Fig. 2D). Thegenome-wide distributions of CG-SMPs, CG-DMRs,

Fig. 1. Epigenetic variation of CG-SMPs. (A) An example of a CG-SMP. Gold lines indicate CG methyl-ation, maroon rectangle indicates the untranslated regions, and green rectangles indicated exons. (B) Abreakdown of the methylation distribution of CG dinucleotides among all samples. (C) A heatmap indi-cating the number of CG-SMPs that differ between two samples (table S3).

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org370

REPORTS

and nonCG-DMRs were depleted in hetero-chromatic regions in the genome (Fig. 2, E

and F). These depletions were mostly observed atthe pericentromeres and centromeres (Fig. 2, E

and F, and figs. S4 and S5). CG-DMRs areenriched in transposons located in euchromatin

but depleted in transposons present near thecentromere. Because the centromeric regions of

the genome contain the highest density of DNAmethylation (Fig. 2, E and F), these observations

combined with the observations that CG-DMRsare enriched in intron sequences may indicate

that DNA methylation that is associated with nu-cleosomes (22) (i.e., exons or tightly packaged

chromatin in the pericentromeres and centro-

meres) may be maintained at a higher fidelityand that DNA methylation not associated with

nucleosomes may undergo greater epigeneticdrift.

A genome-wide screen for DMRs simulta-neously occurring in all three methylation sequence

contexts (C-DMRs are CG, CHG, and CHH)was performed to assess the extent of epiallelic

variation that is characteristic of RdDM acrossthe MA population. In total, 72 C-DMRs were

identified, of which functional categorization

revealed that two-thirds overlapped with trans-poson and intergenic sequences whereas about

one-third overlapped with gene bodies and pro-moters (Fig. 3A and table S8). To determine

whether transposition-induced methylation couldpotentially give rise to the methylated C-DMRs

(mC-DMRs) (23), genomic DNA encompassingall C-DMRs was amplified and compared in all

ancestral and descendant lines. In every case,the observed amplicon size was identical for

all MA lines and was equal to the expected sizeof the locus (table S8), indicating that these

C-DMRs are unlinked to cis-genetic variationlocated within 500 bp, a distance that would be

expected to reveal methylation induced by trans-

poson insertions at these loci (23). Additionally,none of the genetic variants identified by genome

resequencing of this population (18) overlappedwith any of these C-DMRs. Lastly, restriction

enzyme digestion and Southern blot analyseswere performed to rule out the possibility that

copy number variants were the cause of spon-taneous epiallele formation, as is the case for the

PAI epialleles (24). In all cases examined, the ob-served hybridization pattern and gene copy num-

ber were identical for each of the MA lines

(fig. S6). Therefore, we conclude that the 72C-DMRs represent a set of spontaneously occur-

ring epialleles within the MA lines, because theywere not associated with any genetic variation.

By using a set of C-DMRs that exhibitedan identical methylation status (fig. S7), we de-

termined the frequency of discordance of theancestral state with the descendant lines and found

that 29 of the C-DMRs were highly variable (>1descendant line was discordant with the ances-

tral state) (Fig. 3B). C-DMRs discordant in onlyone of the five descendant lines were the most

frequent class, but there was an unexpectedlyhigh number of C-DMRs (63%) that were dis-

cordant in more than one descendant (Fig. 3B).

Within the set of 576 C-DMRs identified (eightlines by 72 C-DMRs), 7 were discordant between

the biological replicates (table S8). These datasuggest that, although many C-DMRs represent

the formation of spontaneous epialleles, a smallsubset may reflect the presence of “hotspots”

(metastable epialleles).We sequenced small RNA (smRNA) pop-

ulations for all eight lines and found thatsmRNAs [represented as RPKCMs (reads per

kilobase of each C-DMR per million reads) in

Fig. 2. CG-DMRs diverge over time and are enriched ingene bodies. (A) Example CG-DMR present in an unmeth-ylated state in both replicates of line 69. (B) A heatmaprepresentation of a two-dimensional hierarchical cluster-ing based on DMRs. Columns represent samples. Rowsindicate DMRs. The column to the left of the heatmapindicates the genomic location of the DMR (blue, genebody; gold, transposon; gray, intergenic; red, transposonin gene body). (C) The average distribution of CG-DMRs

(red) and nonCG-DMRs (blue) across gene bodies (from the start of the 5′ UTR to the end of the 3′ UTR, including 500 bp up- and downstream). (D) CG gene-body DMRs are specifically depleted in exons. (E) Genome-wide distributions of mCG (red), CG-SMPs (green), and CG-DMRs (blue) across chromosome I. (F)Genome-wide distributions of methylated nonCGs (mnonCG, red) and nonCG-DMRs (green) across chromosome I. The centromere is indicated by the pinkvertical bar for (E) and (F).

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 371

REPORTS

tables S9 to 12] were associated with an in-crease in the average methylation density of

C-DMRs (Fig. 3C). Furthermore, this associationresembled a binary switch, because the most

densely methylated C-DMRs contained abun-dant 24-nucleotide (nt) smRNAs (Fig. 3C).

Of the eight previously documented plant

epialleles resulting in phenotypic variation, all af-fected transcriptional output of the differentially

methylated locus (9–11, 23–28). mRNA abun-dance was measured in all eight lines with quan-

titative reverse transcription polymerase chainreaction (qRT-PCR) at eight C-DMRs that over-

lapped with protein-coding regions. In four ofthese genes, the gain or loss of DNA methylation

was correlated with a large decrease or increasein mRNA abundance, respectively, and with the

presence of 24-nt smRNAs at each silenced epi-allele (Fig. 3, D to F, and fig. S8). These find-

ings reveal that changes in epiallelic state canlead to major effects on transcriptional output

(fig. S9).We also observed that the methylation sta-

tus of one C-DMR resulted in alternative pro-moter usage of ACTIN RELATED PROTEIN 9

(At5g43500) (fig. S10C). The loss of DNA meth-ylation within the 5′ untranslated region (UTR)

of the At5g43500.1 isoform led to an increase in

mRNA expression, whereas expression of iso-

form At5g43500.2, with a transcriptional startsite located further downstream, was unaffected

(fig. S10, D and E).Although epialleles can have major impacts

on phenotypic diversity, until now their identi-

fication was not trivial. Even more puzzling is

the origin of “pure” alleles, which are definedby their formation in the absence of any genetic

variation in cis or trans (8). One route to epi-allele formation may be the failure to correctly

maintain the proper methylation status through-

-1 0 1 2 3 4 5 6 7 8

A B

Transposonsn = 27

Intergenicn = 21

Genesn = 14

Promotersn = 7

ncRNAsn = 2

Pseudogenen = 1

# of descendant lines discordant with ancestral state

02468

1012141618

1 2 3 4 5

MethylatedUnmethylated

D E

1 rep11 rep2

19 rep119 rep212 rep112 rep229 rep129 rep249 rep149 rep259 rep159 rep269 rep169 rep2

119 rep1119 rep2

Anc

esto

rsD

esce

ndan

ts

Log2 fold change in mRNA levels of At5g24240 (relative to line 1)At5g24240

At5g242501

19

12

29

49

59

69

119

Num

ber

of C

-DM

Rs

C

F

24nt23nt22nt21nt

1

19

12

29

49

59

69

119

smRNA levels at At5g24240 C-DMR (RPKCMs)

0 2 4 6 8 10 12 14 16

mC-DMR density quantiles (%)

Ave

rage

sm

RN

A R

PK

CM

s

0

2

4

6

8

10

12

14

10 20 30 40 50 60 70 80 90 100

21nt22nt23nt24nt

Fig. 3. Epiallelic variation at protein-coding loci is associated with transcrip-tional variation. (A) Classification of C-DMRs and their genomic locations. (B)The number of descendant lines discordant with the ancestral C-DMR stateand the C-DMR methylation status. The black portions of the bar indicate thedescendant C-DMRs that became methylated, whereas the white portionsindicate regions that became unmethylated, compared with the ancestral pop-ulation. (C) The 24-nt smRNA levels are associated with increasing methyla-tion density. The 24-nt smRNA RPKCMs for all 576 C-DMRs (8 MA lines by

72 C-DMRs) were ranked and binned into 10% quantiles, and then the aver-age mC densities were plotted. (D) A representative C-DMR at At5g24240 inwhich both biological replicates of descendant line 59 were unmethylated. (E)qRT-PCR analysis of At5g24240 reveals >50-fold increase in mRNA abundancein unmethylated line 59. Error bars indicate SEM. (F) The 24-nt smRNAs areenriched specifically in the MA lines that are transcriptionally silenced in (E)for the At5g24240 locus with the exception of line 59, which is abundantlyexpressed in (E).

0

10

20

30

40

50

60

met1 ddc

PartiallyMethylated

# of mC-DMRs that become

unmethylated in

# of C-DMRs that become

re-methylated in

rdd

Not methylated

in

Col-0

Num

ber

of C

-DM

Rs

Fig. 4. Methylation status of all 72 epialleles in methylation and demethylation mutant backgrounds.Most of the epialleles become unmethylated in met1-3, whereas a smaller number become remeth-ylated in the DNA demethylase triple mutant rdd.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org372

REPORTS

out epigenetic reprogramming that occurs post-fertilization (29, 30). It is noteworthy that 63 of

the 72 C-DMRs overlap with regions previouslyshown to have altered methylation patterns in

methylation enzyme mutants (Fig. 4) (3). Of the14 C-DMRs that overlap with genes, 5 become

reexpressed in met1-3 and 1 transcript becomessilenced in rdd (3). These results suggest that a

failure to faithfully maintain genome-wide meth-ylation patterns by MET1 and/or RDD is likely

one source of spontaneous epiallele formation.Regardless of their origin, the majority of epi-

alleles identified in this study are meiotically sta-ble and heritable across many generations in this

population. Understanding the basis for such trans-

generational instability and the mechanism(s)that trigger and/or release these epiallelic states

will be of great importance for future studies.

References and Notes1. J. A. Law, S. E. Jacobsen, Nat. Rev. Genet. 11, 204 (2010).

2. S. J. Cokus et al., Nature 452, 215 (2008).

3. R. Lister et al., Cell 133, 523 (2008).

4. X. Zhang et al., Cell 126, 1189 (2006).

5. D. Zilberman, M. Gehring, R. K. Tran, T. Ballinger,

S. Henikoff, Nat. Genet. 39, 61 (2007).

6. S. W.-L. Chan et al., Science 303, 1336 (2004).

7. J. Paszkowski, U. Grossniklaus, Curr. Opin. Plant Biol. 14,

195 (2011).

8. E. J. Richards, Nat. Rev. Genet. 7, 395 (2006).

9. K. Shibuya, S. Fukushima, H. Takatsuji, Proc. Natl. Acad.

Sci. U.S.A. 106, 1660 (2009).

10. P. Cubas, C. Vincent, E. Coen, Nature 401, 157 (1999).

11. K. Manning et al., Nat. Genet. 38, 948 (2006).

12. A. J. Thompson et al., Plant Physiol. 120, 383 (1999).

13. M. W. Vaughn et al., PLoS Biol. 5, e174 (2007).

14. F. Johannes et al., PLoS Genet. 5, e1000530 (2009).

15. F. K. Teixeira et al., Science 323, 1600 (2009);

10.1126/science.1165313.

16. A. Vongs, T. Kakutani, R. A. Martienssen, E. J. Richards,

Science 260, 1926 (1993).

17. R. G. Shaw, D. L. Byers, E. Darmo, Genetics 155, 369

(2000).

18. S. Ossowski et al., Science 327, 92 (2010).

19. Additional experiments and descriptions of methods used

to support our conclusions are presented as supporting

material on Science Online.

20. C. M. Nievergelt et al., Am. J. Med. Genet. B. Neuropsychiatr.

Genet. 141B, 234 (2006).

21. M. A. Zapala, N. J. Schork, Proc. Natl. Acad. Sci. U.S.A.

103, 19430 (2006).

22. R. K. Chodavarapu et al., Nature 466, 388 (2010).

23. J. Liu, Y. He, R. Amasino, X. Chen, Genes Dev. 18,

2873 (2004).

24. J. Bender, G. R. Fink, Cell 83, 725 (1995).

25. S. Melquist, B. Luff, J. Bender, Genetics 153, 4017

(1999).

26. S. E. Jacobsen, E. M. Meyerowitz, Science 277, 1100

(1997).

27. H. Saze, T. Kakutani, EMBO J. 26, 3641 (2007).

28. W. J. Soppe et al., Mol. Cell 6, 791 (2000).

29. R. A. Mosher et al., Nature 460, 283 (2009).

30. R. K. Slotkin et al., Cell 136, 461 (2009).

Acknowledgments: We thank M. White, R. Lister, M. Galli,

and R. Amasino for discussions; R. Shaw and E. Darmo

for seeds; J. Nery for sequencing operations; and

M. Axtell for Southern blot protocol. R.J.S. was supported

by an NIH National Research Service Award postdoctoral

fellowship (F32-HG004830). M.D.S. was supported by

a NSF Integrative Graduate Education and Research

Traineeship grant (DGE-0504645). M.G.L. was supported

by an European Union Framework Programme 7

Marie Curie International Outgoing Fellowship

(project 252475). O.L. and N.J.S. are supported by

NIH/National Center for Research Resources grant

number UL1 RR025774. This work was supported by

the Mary K. Chapman Foundation, the NSF (grants

MCB-0929402 and MCB1122246), the Howard Hughes

Medical Institute, and the Gordon and Betty Moore

Foundation (GBMF) to J.R.E. J.R.E. is a HHMI–GBMF

Investigator. Analyzed data sets can be viewed

at http://neomorph.salk.edu/30_generations/browser.

html. Sequence data can be downloaded from National

Center for Biotechnology Information Sequence Read

Archive (SRA035939). Correspondence and requests for

materials should be addressed to J.R.E. (ecker@salk.edu).

Supporting Online Materialwww.sciencemag.org/cgi/content/full/science.1212959/DC1

Materials and Methods

SOM Text

Figs. S1 to S11

Tables S1 to S16

References

22 August 2011; accepted 7 September 2011

Published online 15 September 2011;

10.1126/science.1212959

Computation-Guided BackboneGrafting of a Discontinuous Motifonto a Protein ScaffoldMihai L. Azoitei,1* Bruno E. Correia,1,2* Yih-En Andrew Ban,1† Chris Carrico,1,3

Oleksandr Kalyuzhniy,1 Lei Chen,4 Alexandria Schroeter,1 Po-Ssu Huang,1 Jason S. McLellan,4

Peter D. Kwong,4 David Baker,1,5 Roland K. Strong,3 William R. Schief1,6,7‡

The manipulation of protein backbone structure to control interaction and function is achallenge for protein engineering. We integrated computational design with experimental selectionfor grafting the backbone and side chains of a two-segment HIV gp120 epitope, targeted by thecross-neutralizing antibody b12, onto an unrelated scaffold protein. The final scaffolds bound b12 withhigh specificity and with affinity similar to that of gp120, and crystallographic analysis of a scaffoldbound to b12 revealed high structural mimicry of the gp120-b12 complex structure. The methodcan be generalized to design other functional proteins through backbone grafting.

Computational protein design tests ourunderstanding of protein structure and

folding and provides valuable reagentsfor biomedical and biochemical research; long-

term goals include the design of field- or clinic-ready biosensors (1), enzymes (2), therapeutics (3),

and vaccines (4, 5). A major limitation has beenan inability to manipulate backbone structure;

most computational protein design has involvedsequence design on predetermined backbone struc-

tures or with minor backbone movement (1–5).Accurate backbone remodeling presents a sub-

stantial challenge for computational methodsowing to limited conformational sampling and

imperfect energy functions (6).

Novel recognition modules (7), inhibitors (8, 9),enzymes (2), and immunogens (4, 5, 10, 11) have

been designed by grafting functional constel-lations of side chains onto protein scaffolds of

predefined backbone structure. In all cases, therestriction to using predetermined scaffold back-

bone structures limited the complexity of thefunctional motifs that could be transplanted. For

example, the de novo enzymes could accommo-date grafting of only three or four catalytic groups,

whereas many natural enzymes have six or more(12), and the immunogens were limited to con-

tinuous (single-segment) epitopes even thoughmost antibody epitopes are discontinuous (involv-

ing two or more antigen segments) (13, 14).

To address the challenge of incorporating back-bone flexibility modeling into grafting design, we

developed a hybrid computational-experimentalmethod for grafting the backbone and side chains

of functional motifs onto scaffolds (Fig. 1). Wetested this method by grafting a discontinuous

HIV gp120 epitope, targeted by the broadly neu-tralizing monoclonal antibody b12 (15), onto

an unrelated scaffold. b12 binds to a conservedepitope within the CD4-binding site (CD4bs) of

gp120 (16), an area of great interest for vaccinedesign. We focused on transplantation of two

segments from gp120: residues 365 to 372, knownas the CD4b (CD4 binding) loop (17), and resi-

dues 472 to 476, known as the ODe (outer domain

exit) loop (16). The b12-gp120 interaction in-volves six or seven backbone segments on gp120

(16), but 60% of the buried surface area on gp120lies on the CD4b and ODe loops, and a Rosetta

energy calculation (18) suggested that these two

1Department of Biochemistry, University of Washington, Seattle,WA 98195, USA. 2Ph.D. Program in Computational Biology,Instituto Gulbenkian de Ciência, Oeiras, Portugal. 3Divison ofBasic Sciences, Fred Hutchinson Cancer Research Center, Seattle,WA 98109, USA. 4Vaccine Research Center, National Institute ofAllergy and Infectious Diseases, Bethesda, MD 20892, USA.5Howard Hughes Medical Institute, University of Washington,Seattle, WA 98195, USA. 6IAVI Neutralizing Antibody Center,The Scripps Research Institute, La Jolla, CA 92037, USA. 7De-partment of Immunology and Microbial Science, The ScrippsResearch Institute, La Jolla, CA 92037, USA.

*These authors contributed equally to this work.†Present address: Arzeda Corporation, Seattle, WA 98102,USA.‡To whom correspondence should be addressed. E-mail:schief@scripps.edu

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 373

REPORTS

segments could account for up to 80% of the

binding energy.The work flow (Fig. 1) has four stages: (i)

scaffold search, in which the Protein Data Bank(PDB) (19) is searched for scaffolds suitable to

accept the backbone segments comprising themotif; (ii) scaffold design, in which the motif back-

bone segments replace native scaffold backboneand new connecting segments and surrounding

side chains are built to support the motif confor-mation; (iii) computation-guided library design,

in which a small set of mutagenesis libraries forsequential screening are derived from an ensem-

ble of designs with expanded structural and com-positional diversity in the connecting segments;

and (iv) in vitro screening, in which computation-guided libraries are screened to identify clones

with optimal functional activity.For scaffold search, we developed an algorithm

(Multigraft Match) that exhaustively searched aculled PDB for suitable scaffolds. For all possi-

ble combinations of four insert positions inevery scaffold, Multigraft Match produced a

low-resolution prediction of whether the epitopebackbone segments could be grafted onto the

scaffold while maintaining backbone continuity

and avoiding steric clash (fig. S1). Eleven

scaffolds satisfied the geometrical and stericclash requirements and were selected for design

(table S1).For scaffold design, we developed an algo-

rithm (Multigraft Design) that, given a prelimi-nary rigid-body orientation for a discontinuous

epitope relative to a scaffold, deleted appropriateregions of the scaffold, built new segments to

connect the epitope to the scaffold, and designed

side chains neighboring the epitope and connect-

ing segments to support the graft (fig. S2). Thisinvolved aggressive structural manipulations, in-

cluding replacement of ordered secondary structuremotifs by the epitope segments, flexible back-

bone modeling of two or more connecting seg-ments, and sequence design of 10 or more core

residues. Several design variants of each candi-date scaffold (fig. S3) were tested for expres-

sion and purification in Escherichia coli. Of 62

Scaffoldsearch

Scaffolddesign

Computation-guidedlibrary design

In vitroscreening

Loop closure+

Sequence design

Motif transplantation

Motifselection

Structural ensemble Sequence Profile

Freq

uenc

y (%

)

Scaffold display

Ant

ibod

y bi

ndin

g Round 1 Round 6

ODe LoopCD4b Loop

b12 antibody

gp120

b12 antibody

scaffold

Fig. 1. Combined in silico–in vitro strategy for the transplantation of complex structural motifs to heterologous scaffold proteins. The diagrams illustrate thestages in the design of a non-HIV scaffold presenting two loops from the b12 epitope on HIV gp120.

Table 1. Affinity and kinetics of the interaction between recombinant 2bodx variants and b12. Forall the reported values, the standard error is ≤ T7 of the last significant digit. RL, random library;L1, library 1; L2, library 2; L3, library 3. kon and koff represent the kinetic association and disso-ciation rates, respectively, of the measured interactions.

2bodx variant Origin

b12 interaction parameters (SPR)

kon

(M−1 s−1)

koff

(s−1)

KD (kinetic)

(nM)

KD (equilibrium)

(nM)

03 Initial design >300 × 103

Y3 RL ~30 × 103

42 L1 + L2 1.3 × 106 2.3 × 10−1 177 166.6

43 L1 + L2 + RL 3.0 × 106 1.0 × 10−1 33.3 33.5

44 L1 + L2 + RL + L3 1.9 × 106 3.6 × 10−2 18.9 19.5

45 L1 + L2 + RL + L3 3.8 × 106 3.9 × 10−2 10.3 10.3

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org374

REPORTS

candidates tested, 25 could be solubly expressedand purified (table S1).

Purified designs were tested for b12 bindingby surface plasmon resonance (SPR). One de-

sign, 2bodx_03, which had 39 mutations and 11deletions relative to the parent protein (fig. S4),

bound to b12 weakly (dissociation constant KD ≈

300 mM). The binding was specific, because no

binding was detected for the epitope mutant

Asp114 → Arg (D114R) (20) (fig. S5). A high-resolution (1.3 Å) crystal structure of 2bodx_03

showed no discernible electron density for theepitope or connecting segments (fig. S5 and table

S2), indicating that these regions were flexiblein solution. In an initial attempt to optimize the

b12 affinity of 2bodx_03, a whole-protein ran-dom mutagenesis library was screened by yeast

display (21). Clone 2bodx_R3 was thereby iso-lated with two mutations [Ser177 → Gly (S177G)

and Ala118 → Val (A118V)] from 2bodx_03 and

a factor of 10 higher affinity for b12 (KD ≈ 30 mM)(Table 1, Fig. 2, and fig. S6). This interaction re-

mained three orders of magnitude weaker thangp120-b12 interaction [KD = 20 nM (16)]. The

low affinity was likely due to nonoptimal sequencesand conformations in the connecting segments.

Optimization by targeted random mutagenesisand in vitro screening was not feasible because

allowing 20 amino acids at all 21 positions judgedto be important in the connecting segments would

yield impractical library sizes of 2 × 1027.In computation-guided library design, we

used a structure-sequence diversification protocol(fig. S7) to devise relatively small libraries based

on more complete sampling of low-energy struc-tures and sequences in the connecting segments.

For each connecting segment, 20,000 backbone

conformations were separately generated andsubjected to sequence design while keeping the

rest of the 2bodx_03 structure fixed. Several low-energy models for each segment were exhaustive-

ly recombined in silico and subjected to furthersequence design to identify 2bodx models with

optimal structures and sequences in all connect-ing segments. After a final round of conforma-

tional resampling and design (fig. S7), the best45 models by several Rosetta metrics (18) were

used to generate sequence profiles to identify theamino acids that occurred at each of the 21 po-

sitions in the connecting segments (fig. S8). Thediversity was reduced by eliminating residues

that occurred at low frequency, that were similarin size and chemical nature to more frequent res-

idues, or that were judged likely to bury a polarside chain. The final library allowed mutations at

21 positions and had a theoretical size of 1012.For in vitro screening, we used yeast display.

To overcome the limitations of the library sizesupported by yeast display (107), we constructed

two partially overlapping sublibraries and screenedthem sequentially (figs. S9 and S10). The first

sublibrary (library 1) contained all (4 × 106) ofthe computationally designed ODe loop con-

necting segments combined with eight designvariants of the CD4b loop connecting segments

present in 23 of the 45models. After three roundsof screening, the selected ODe loop variants

(from at least 18 different clones) were combinedwith all (2 × 105) of the computationally designed

CD4b loop variants to create library 2. This sub-

library was screened for three rounds to isolateclone 2bodx_42, which differed from 2bodx_03

by17mutations (fig. S11).Recombinant 2bodx_42bound b12 with aKD of 166 nM, an improvement

by a factor of >1800 over 2bodx_03 (Table 1).Introducing the A118Vmutation from 2bodx_R3

further increased b12 affinity, as the resultingvariant (2bodx_43) bound b12with aKD of 33 nM

(Table 1, Fig. 2, and fig. S6), within a factor of 2of the b12-gp120 affinity (16). Introducing the

D114R mutation on 2bodx_43 resulted in loss ofdetectable b12 binding (fig. S12), demonstrating

that the bindingwas specific to the epitope. Further,2bodx_43 was thermally stable (melting point =

75°C) and monomeric in solution (fig. S13).To assess whether the b12 affinity could be im-

proved further and to evaluate if the computation-

Fig. 2. Isolation of scaffold 2bodx variantswith high b12 affinity and specificity. (A)Screening of the computation-guided librariesled to rapid enrichment of clones withhigh b12 affinity; R1-R3 refer to rounds1 to 3 of selection. (B) SPR equilibrium anal-ysis of the initial computational design(2bodx_03) and the 2bodx variants identi-fied from the directed libraries (Table 1 andfig. S9). (C) 2bodx_43 binds to b12, butnot to CD4 or other antibodies that targetthe CD4bs on gp120.

Fig. 3. Atomic-level recapitulation of the b12-gp120 interface by the b12-2bodx_43 complex. (A)Structure of b12 in complex with 2bodx_43. (B) The conformations of the transplanted loops (yellow) in2bodx_43 (red) mimic their conformations on gp120 (green). (C) Conformations of side chains (sticks)making important contacts in the b12-gp120 complex are preserved at the 2bodx_43-b12 interface;H1, H2 and H3 refer to the CDR loops of b12.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 375

REPORTS

guided libraries restricted the sequence spaceeffectively, we screened a third library based on

2bodx_43 with expanded sampling at seven po-sitions (fig. S10). The highest-affinity clone iso-

lated (2bodx_45) differed from 2bodx_43 bytwo mutations (fig. S4) and bound b12 with a

KD of 10 nM, a factor of 3 better than 2bodx_43and as tightly as gp120 (16) (Table 1, Fig. 2,

and fig S6). Another high-affinity variant se-lected from this library (2bodx_44, KD = 19 nM,

fig. S6) was used to investigate the b12 bindingcontributions of library-selected mutations. We

measured the b12 binding of 2bodx_44 constructsin which the “evolved” residues were individually

reverted to their 2bodx_03 identity. Only 6 of 16

reversions reduced the b12 affinity of 2bodx_44by a factor of 3 or more, and a 2bodx_03 variant

that contained nine of the 2bodx_44 mutationshad only micromolar affinity for b12 (KD =

1.5 mM) (fig. S14). Thus, the selected mutationsmade synergistic contributions to the high b12

affinity of 2bodx constructs.To evaluate the degree to which the 2bodx-

b12 interaction recapitulated the gp120-b12 inter-action, we solved a crystal structure for 2bodx_43

complexed with b12 at 2.07 Å resolution (Fig.3A and table S2). Comparison with the gp120-

b12 complex (16) revealed a high degree of mim-icry; superposition of the epitope and paratope

of both complexes gave an overall backbone rootmean square deviation (RMSD) of 0.71 Å (Fig.

3, B and C) (22). Consistent with good backbonemimicry, important interactions involving b12

heavy-chain residues Tyr53, Tyr98, Trp100, Asn100g,and Tyr100h were recapitulated in the 2bodx_43-

b12 complex (Fig. 3C and tables S3 and S4). Thetotal buried areas in the complexes were also

similar, except for a small additional area on thescaffold outside the epitope (fig. S15).

The CD4bs is a major antibody target in HIVinfection (23). Reagents are desired that bind b12

but not CD4bs-directed non-neutralizing anti-bodies (24) such as b13 that engages gp120

similarly to b12 (25). Of eight CD4bs-directedantibodies tested, 2bodx_43 bound tightly to b12

only (Fig. 2C). Additional SPR analyses showedthat 2bodx_43 binds more tightly to b12 than to

b13 by a factor of >10,000 (fig. S16) (26). Theseresults indicate that b12 epitope-scaffolds are

promising tools for HIV vaccine research and en-courage the application of backbone grafting to

engineer antigens, enzymes, and inhibitors.

References and Notes1. L. L. Looger, M. A. Dwyer, J. J. Smith, H. W. Hellinga,

Nature 423, 185 (2003).

2. J. B. Siegel et al., Science 329, 309 (2010).

3. S. J. Fleishman et al., Science 332, 816 (2011).

4. B. E. Correia et al., Structure 18, 1116 (2010).

5. G. Ofek et al., Proc. Natl. Acad. Sci. U.S.A. 107, 17880 (2010).

6. D. J. Mandell, T. Kortemme, Nat. Chem. Biol. 5, 797 (2009).

7. S. E. Rutledge, H. M. Volkman, A. Schepartz, J. Am.

Chem. Soc. 125, 14336 (2003).

8. S. K. Sia, P. S. Kim, Proc. Natl. Acad. Sci. U.S.A.100, 9756 (2003).

9. L. Martin et al., Nat. Biotechnol. 21, 71 (2002).

10. E. Drakopoulou et al., J. Biol. Chem. 271, 11979 (1996).

11. S. M. Lu, R. S. Hodges, J. Biol. Chem. 277, 23515 (2002).

12. D. Baker, Protein Sci. 19, 1817 (2010).

13. L. Jin, B. M. Fendly, J. A. Wells, J. Mol. Biol. 226, 851 (1992).

14. N. D. Rubinstein et al., Mol. Immunol. 45, 3477 (2008).

15. D. R. Burton et al., Science 266, 1024 (1994).

16. T. Zhou et al., Nature 445, 732 (2007).

17. B. Chen et al., Nature 433, 834 (2005).

18. See supporting material on Science Online.

19. H. M. Berman et al., Nucleic Acids Res. 28, 235 (2000).

20. Y. Li et al., Nat. Med. 13, 1032 (2007).

21. G. Chao et al., Nat. Protoc. 1, 755 (2006).

22. With the overall interface superposition of the 2bodx_43-b12

and gp120-b12 complexes, RMSD values for individual

elements were low, as follows: CD4b loop, 0.8 Å; ODe loop,

1.5 Å; b12 CDRH1 residues 25 to 34, 0.28 Å; H2 residues

52 to 56, 0.6 Å; H3 residues 94 to 101, 0.7 Å.

23. L. Stamatatos, L. Morris, D. R. Burton, J. R. Mascola,

Nat. Med. 15, 866 (2009).

24. R. Pantophlet et al., J. Virol. 77, 642 (2003).

25. L. Chen et al., Science 326, 1123 (2009).

26. The very low affinity of b13 for 2bodx_43 is likely due to

(i) the different conformations of the CD4b and ODe loops

in the b13-gp120 structure compared to the b12-gp120

and b12-2bodx_43 structures, and (ii) important contacts

for b13 at gp120 residues 419, 421, and 425 in the

b20-b21 region not present on the scaffold.

Acknowledgments: Supported by grants from the International

AIDS Vaccine Initiative Neutralizing Antibody Consortium and

the Bill and Melinda Gates Foundation Consortium for

AIDS Vaccine Discovery. B.E.C. was supported by a fellowship

from the Portuguese Fundação para a Ciência e a Tecnologia

(SFRH/774 BD/32958/2006). We thank I. Wilson for

comments on the manuscript; D. Burton, A. Hessell, and

the IAVI Neutralizing Antibody Consortium Reagent

Repository for providing b6, b12, and b13; J. Mascola for

VRC01 and VRC03; Progenics Inc. for CD4-IgG2; D. Dimitrov

for m6, m14, and m18; J. Robinson for 15E; and R. Wyatt

for F105 and HxB2 gp120. We thank G. Nabel, M. Kanekiyo,

and Z.-Y. Yang for experiments on early generations of

b12 scaffolds. Coordinates and structure factors were

deposited in the PDB as entries 3RPT and 3RU8. The

University of Washington has filed a patent application

on the b12 scaffolds and the scaffolding method

developed in this study. Materials and information will

be provided to noncommercial users under the Uniform

Biological Materials Transfer Agreement. The Multigraft

software developed in this work is freely available to

noncommercial users through the Rosetta Commons

agreement (www.rosettacommons.org).

Supporting Online Materialwww.sciencemag.org/cgi/content/full/334/6054/373/DC1

Materials and Methods

Figs. S1 to S16

Tables S1 to S4

References

6 June 2011; accepted 8 September 2011

10.1126/science.1209368

Antagonists Induce a ConformationalChange in cIAP1 ThatPromotes AutoubiquitinationErin C. Dueber,1 Allyn J. Schoeffler,1 Andreas Lingel,1 J. Michael Elliott,2 Anna V. Fedorova,1

Anthony M. Giannetti,3 Kerry Zobel,1 Brigitte Maurer,4 Eugene Varfolomeev,1 Ping Wu,4

Heidi J. A. Wallweber,4 Sarah G. Hymowitz,4 Kurt Deshayes,1 Domagoj Vucic,1*Wayne J. Fairbrother1*

Inhibitor of apoptosis (IAP) proteins are negative regulators of cell death. IAP family memberscontain RING domains that impart E3 ubiquitin ligase activity. Binding of endogenous orsmall-molecule antagonists to select baculovirus IAP repeat (BIR) domains within cellular IAP(cIAP) proteins promotes autoubiquitination and proteasomal degradation and so releasesinhibition of apoptosis mediated by cIAP. Although the molecular details of antagonist–BIRdomain interactions are well understood, it is not clear how this binding event influences theactivity of the RING domain. Here biochemical and structural studies reveal that the unliganded,multidomain cIAP1 sequesters the RING domain within a compact, monomeric structure thatprevents RING dimerization. Antagonist binding induces conformational rearrangements thatenable RING dimerization and formation of the active E3 ligase.

Inhibitor of apoptosis (IAP) proteins are anti-apoptotic factors important in blocking pro-

grammed cell death, or apoptosis, in response

to a variety of stimuli (1, 2). Whether initiated byexternal death signals transduced by specific cell

surface receptors (extrinsic pathway) or by inter-

nal cues of compromised cellular integrity (intrinsicpathway), apoptotic signaling pathways converge

in the activation of caspases (cysteine-dependentaspartyl-specific proteases), which effect wide-

spread proteolytic damage and cell death (3). IAPshold these cellular executioners in check, either

through direct inhibitory interactions or by imped-ing upstream caspase activation pathways (4).

Many cancer cells overexpress IAPs, which al-

lows them to resist cytotoxic therapies (2, 5). Thus,IAPs are potentially important targets for cancer

treatment (2, 5, 6).IAP-targeting therapeutics designed to mimic

the endogenous IAP antagonist, SMAC (secondmitochondrial activator of caspases)–DIABLO

(direct IAP-binding protein with low isoelectricpoint) (7, 8), have recently entered phase I clinical

1Department of Early Discovery Biochemistry, Genentech, 1 DNAWay, South San Francisco, CA 94080, USA. 2Department ofProtein Chemistry, Genentech, 1 DNAWay, South San Francisco,CA 94080, USA. 3Department of Biochemical Pharmacology,Genentech, 1 DNA Way, South San Francisco, CA 94080, USA.4Department of Structural Biology, Genentech, 1 DNA Way,South San Francisco, CA 94080, USA.

*To whom correspondence should be addressed. E-mail:fairbro@gene.com (W.J.F.); domagoj@gene.com (D.V.)

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org376

REPORTS

trials (9). The conserved amino-terminal tetra–amino acid peptide (Ala-Val-Pro-Ile or AVPI) of

mature SMAC binds to a surface groove on bacu-lovirus IAP repeat (BIR) domains (10, 11). SMAC

mimetics show proapoptotic activity by prevent-ing interactions between IAPs and active caspases

(2, 4). Moreover, binding of these small-moleculeantagonists to cellular IAP (cIAP) BIR domains

leads to a second mechanism of antagonism—

rapid proteasomal degradation of the cIAP pro-

teins induced by autoubiquitination (12–14).Although analyses of antagonist–BIR domain

complexes have yielded detailed information re-garding the molecular determinants of these inter-

actions (15–17), how antagonist binding induces

cIAP RING E3 ligase activity is not understood.We sought to define aminimal cIAP1 construct

that would maintain key full-length cIAP1 proper-ties and be suitable for biophysical and structural

studies. As IAP antagonists bind preferentially tothe BIR3 domain of cIAP1 (13), we focused on a

truncated version of the protein that lacks the firsttwo BIR domains (Fig. 1A and fig. S1A). Remov-

al of the BIR1 and BIR2 domains, which includethe tumor necrosis factor receptor–associated

factor 2 (TRAF2)–interacting region of cIAP1(18–20), did not affect antagonist-induced deg-

radation (fig. S1). The resulting cIAP1 construct,comprising the BIR3 domain followed by the

ubiquitin-associated (UBA) domain, the caspaseactivation and recruitment domain (CARD), and

the carboxy-terminal RING domain (BIR3-RING),recapitulates full-length cIAP1 behavior and dem-

onstrates proteasomal degradation that is mediated

by IAP antagonist-induced autoubiquitination ac-tivity (fig. S1).

RING E3 ligases often function as dimers(21, 22), and disruption of RING-mediated di-

merization stabilizes IAPs against IAP antagonists(23). We thus assessed whether IAP antagonists

affect the oligomerization state of the BIR3-RINGconstruct. Monomeric unliganded BIR3-RING

appears as a single, discrete band by native gelanalysis (Fig. 1B). Addition of a bivalent antag-

onist (BV6) that can cross-link the protein throughsimultaneous binding of separate BIR3 domains

(13) is a control for migration of the BIR3-RINGdimer. Analytical size-exclusion chromatography

and multiangle light scattering (SEC-MALS) re-sults confirm that the native gel bands represent

monomer and dimer BIR3-RING species (fig.S2D). Monovalent small-molecule (MV1) and

peptide (AVPW) antagonists shift the mobility of

the BIR3-RING band toward the BV6 dimer con-trol. In contrast, a BIR3-RING construct lacking

RING dimerization residues at the carboxy termi-nus (M266 to G611; BIR3-RINGDC7) (18, 23)

does not band-shift unless presented with the bi-valent antagonist. Together, these findings sug-

gest that binding of monovalent antagonists to theBIR3 domain induces RING dimerization through

an intramolecular, allosteric mechanism.It intrigued us that the mobility of the ligand-

free “apo” BIR3-RINGDC7 band is considerablyretarded compared with apo BIR3-RING, despite

modest differences in molecular weight and netcharge for the two proteins (Fig. 1B). MV1- and

AVPW-bound BIR3-RINGDC7 demonstrate iden-tical band migrations as the apo BIR3-RINGDC7

sample. Further SEC-MALS analysis corrobo-rates the distinct mobilities seen by native gel (fig.

S2D). These results suggest that the dimerization-deficient mutant exists in a more extended con-

formation than themonomericBIR3-RINGproteinand that the carboxy-terminal residues of theRING

Fig. 1.BIR3-binding IAP antagonists induceRING dimerization. (A) cIAP1 constructs. (B)Native gels show the relative mobility ofBIR3-RING and BIR3-RINGDC7 proteins (M,monomer; D, dimer) in the absence andpresence of IAP antagonists. The differingeffects of monovalent antagonists MV1 andAVPWon cIAP1 BIR3-RING proteinmobilityare consistent with their binding half-livesof 0.9 and 4.2 min, respectively, as deter-mined by surface plasmon resonance (SPR)spectroscopy (fig. S2, A and B). For compari-son, the half-life of BV6 binding to BIR3-RING is 10.6 min (fig. S2C).

Fig. 2. Crystal structure of the apo cIAP1 monomer. (A) A ribbon diagram ofBIR3-RINGxtal illustrates the global arrangement of the BIR3 domain, UBAdomain, and CARD around a core RING domain. Inset shows surface views ofthe structure. Protein domains are shown in cyan, green, purple, orange, andyellow as indicated; zinc ions are shown as gray spheres. (B) Structural align-ment with a dimeric cIAP2 RING protomer (red) illustrates how the RINGdimerization elements are sequestered in the BIR3-RINGxtal structure. Spe-

cifically, the RING carboxy-terminal strand (C strand), which self-associates inthe RING:RING interface, interacts with the edge strand of the BIR3 b sheet inthe monomer structure; and the RING amino-terminal helix (N helix) is bent~110° and partially unwound as compared with dimerized, active cIAP2 RINGprotomers and so affects the E2-binding surface (fig. S5). cIAP1 domains otherthan the RING domain are shown faded for visual clarity. Inset depicts the cIAP2RING dimer structure (red and gray protomers, PDB code 3EB5).

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 377

REPORTS

domain are critical for maintaining the more com-pact apo structure.

Crystallization required design of a variant ofthe BIR3-RING protein that eliminated two pro-

tease cleavage sites by deletion of a 10-residuesection within the BIR3 to UBA linker (DS363 to

D372) and a point mutation in which Tyr replacedPhe387 (F387Y) (BIR3-RINGxtal) (figs. S1 and S3).

The structure of BIR3-RINGxtal was solved to1.9 Å resolution by single-wavelength anoma-

lous dispersion (SAD) by using three bound zincions, and the final model was refined to an Rfree

of 23% (fig. S4 and table S1).

Consistent with an allosteric activation hy-pothesis, the apo monomer adopts a compact con-

formation, with the BIR3 domain, UBA domain,CARD, and intervening linkers combining to se-

quester a centralized RING domain (Fig. 2A).Multiple interfaces formed between the RING

domain and the surrounding components togetherbury >4400 Å2 of surface area (table S2). Al-

though portions of the RING domain remainaccessible to solvent, key RING dimerization ele-

ments are buried or displaced in this autoinhibitedstate (23) (Fig. 2B). In contrast, the conformations

of the BIR3 domain, UBA domain, and CARD are

similar to homologous domain structures (fig. S6).The interdomain linkers also form well-ordered

secondary structures that interact with the RINGdomain. Inspection of the interdomain interfaces

reveals specific interactions that appear to con-tribute to the confinement of the RING domain

and thus to suppression of RING:RING dimer-ization. The BIR3 domain forms an extensive

interface with the RING domain (table S2), whichis buttressed by the UBA domain and a section

of the BIR3-UBA linker (BU-linker). The BIR3residue W329, located in a hydrophobic pocket

between these four modules, is positioned to

A B E F

W329

V613

I584

L355

L359L356

L359L360

L558L565

L589

V613

R569

E508 D511

R465

H2OR300

R314N301

R600

C D

K610

R600

R606

K601

E440

E444

E447

E451

R614

AV

P

V613

K610

T612

I A

G611

Gstd

sApoAVPW

BV6ApoAVPW

BV6ApoAVPW

BV6ApoAVPW

BV6ApoAVPW

BV6ApoAVPW

BV6ApoAVPW

BV6ApoAVPW

BV6ApoAVPW

BV6ApoAVPW

BV6

“WT” W329A W329F R300AN301AR314A

E317AS318A

R465AE508AD511A

E440KE444KE447KE451K

K610A K610E T612R

ApoAVPWBV6

ApoAVPWBV6

ΔE440-E452

ΔM545-L556

L356AL359AL360A

ApoAVPWBV6

D

M

H

WT W329A

R300AN301AR314AT612R

L356AL359AL360A

MG132:BV6:

cIAP1: WTW329F

− actin

− cIAP1 (FL)

vect

or

- + -- +-

- + -- + - - + - - + - - + - - + -- +- - +- - +- - +- - +- - +-

--

K305

R314

Fig. 3. Domain-domain interactions. Detailed views of the (A) BIR3:BU-linker:UBA:RING, (B) BU-linker:RING, (C) BIR3:RING (D) CARD:RING, and (E)UC-linker:RING interfaces are shown with key residues highlighted as sticks.Color scheme is the same as that of Fig. 2A, with the exception of electro-static surfaces shown for the UC-linker (top) and RING (bottom) in panel (E).(F) Overlay of the RING C strand (orange sticks) and SMAC-peptide antag-onist (white sticks, PDB code 3D9U) within the BIR3 (cyan surface) peptide-binding groove. Unlike the peptide antagonist, which demonstrates goodsurface complementarity and makes numerous favorable interactions with

the BIR3 domain, the RING C strand merely crosses over the peptide-bindinggroove, making a single hydrogen bond contact with G312 of the BIR3 domainvia residue T612. (G) Native gel analysis reveals mutants of BIR3-RINGxtalthat dimerize in the absence of antagonist (red labels). All gels aligned bystandards shown at far left (stds). Positions of WT monomer (M) and dimer(D) are indicated. (H) The stability of full-length FLAG-tagged cIAP1 variantsin 293T cells (T10 mM MG132 or 2 mM BV6 for 1 hour) was determined byevaluating cell lysates for cIAP1 constructs by Western blotting with anti-Flag antibody. Actin was probed as a loading control.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org378

REPORTS

interact favorably with the helix dipole of anadjacent UBA a helix (Fig. 3A). In addition to

contributing to this pocket, residues P353 to T362of the BU-linker form an amphipathic helix

whose hydrophobic surface pairs with the hydro-phobic face of the RING N helix and a nearby

surface of the RING domain (Fig. 3B). These in-teractions pin the N helix in a bent conformation

(relative to dimeric RING protomers) (Fig. 2B) byeffectively substituting the N helix dimerization

contacts with similar intramolecular interactions.On the opposite side of the RING domain, back-

bone carbonyl groups and residue R600 interactwith R300 and R314, and N301 from the BIR3

domain, respectively (Fig. 3C). The remaining

RING interactions are also polar, with the largeCARD:RING interface (table S2) characterized

by a hydrogen-bonding network that includesresidues R465, E508, D511, and R569, as well as

ordered water molecules (Fig. 3D). Additionally,the UBA-CARD linker (UC-linker) acts as a

negatively charged helical “strut” that extendsfrom the final UBA a helix. The strut flanks a

positively charged surface of the RING domain(Figs. 3E and 2A) and thus positions the CARD

for its interactions with the RING domain. Final-ly, the C strand of the RING associates with the

edge of the BIR3 b sheet. Comparison with astructure of theBIR3 domain bound to the SMAC

amino-terminal peptide reveals that the RING Cstrand partially occludes the antagonist-binding

site (Fig. 3F) (24).To evaluate the contributions of these and

other specific interactions to the compact mono-mer state, we introduced mutations designed to

disrupt these contacts into the BIR3-RINGxtal

construct (fig. S7). Evaluating these variants

with native gels revealed that mutations within

the BIR3:RING and BU-linker:RING inter-faces promote dimerization, whereas muta-

tions targeting other interactions have little tono effect, compared with the near–wild-type

(“WT”) BIR3-RINGxtal control (Fig. 3G). Notethat the cIAP1 monomer was insensitive to per-

turbation of the highly polar CARD:RING andUC-linker:RING interfaces, consistent with pre-

vious gross deletions of regions encompassingthe UBA domain or CARD that showed no

observable effect on cIAP1 autoubiquitinationactivity (18). SEC-MALS analyses of the dimer-

promoting mutants confirm the dimeric state ofthese variants in the absence of antagonist (fig.

S8). Accordingly, addition of AVPW did not

significantly alter the apparent molecular weightof these mutants, whereasWTandW329Fmono-

mer samples shifted to dimeric species under sim-ilar conditions.

We next assessed the effect of these interfacemutations on the stability and, therefore, on the

autoubiquitination activity of full-length cIAP1expressed in human embryonic kidney 293Tcells.

As predicted, WT cIAP1 was considerably morestable than variants expected to dimerize on the

basis of biophysical analysis of the BIR3-RINGconstructs (Fig. 3H). This was particularly evi-

dent for the W329A and T612R mutants; thesecould only be detected after treatment with the

proteasome inhibitor, MG132, which suggeststhat their instability results from elevated auto-

ubiquitination. Similar results were observed onexpression of these variants of the shorter BIR3-

RINGxtal construct (fig. S9).To visualize activated, dimeric cIAP1, we col-

lected small-angle x-ray scattering (SAXS) datafor the AVPW-bound BIR3-RING construct. An

averaged ab initio model of the dimer reveals an

extended, two-fold symmetric complex that iscompatible with placement of the RING:RING

homodimer structure at its center (Fig. 4, A andB). Ab initio SAXS models generated for two

other cIAP1 constructs, an N-terminally MBP-tagged BIR3-RING (MBP-BIR3-RING) dimer

and a shorter UBA-RING (S385 to S618) dimer,support a model in which BIR3 domains are

located at the periphery of an elongated, nearlyplanar BIR3-RING homodimer that forms around

a central RING:RING interface (Fig. 4,A toC, andfigs. S10 and S11). In agreement with the mu-

tational analysis, UBA-RING purifies as a con-stitutive dimer, whereas MBP-BIR3-RING, like

BIR3-RING, requires the addition of antagonist

(AVPW) to dimerize (fig. S12).Taken together, these data suggest an E3 ligase

activation model in which apo BIR3-RING existspredominantly in a closed, inactive monomer state.

Antagonist binding to the BIR3 domain is in-compatible with this conformation and stabilizes

an open state in which the RING domain is ex-posed and free to dimerize. RING dimerization

enables rapid cIAP1 autoubiquitination and sub-sequent degradation by the proteasome (23, 25).

Given the high sequence conservation betweencIAP1 and 2, the proposed model likely reflects a

general mechanism of antagonist-induced cIAPactivation, although recent work suggests that apo

cIAP2 may have a greater propensity to dimerizethan cIAP1 (25). In contrast, modeling suggests

that X-linkedAIP (XIAP), which lacks theCARD,could not adopt the same autoinhibited confor-

mation as cIAP1. Nevertheless, the current workprovides a molecular basis for understanding the

mechanism of action of IAP antagonists and of-fers important insights into the regulation of cIAP

E3 ligase activity.

Fig. 4. SAXS analysis of the cIAP1 dimer. (A) Proposed domain ar-rangements for MBP-BIR3-RING, BIR3-RING, and UBA-RING dimersand (B) two views of averaged ab initio SAXS envelopes (P2 symmetry)

for the same constructs. All data were collected in the presence of1 mM AVPW. (C) Pairwise distance-distribution functions of SAXS datafor each dimer.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 379

REPORTS

References and Notes1. G. S. Salvesen, C. S. Duckett, Nat. Rev. Mol. Cell Biol. 3,

401 (2002).

2. D. Vucic, W. J. Fairbrother, Clin. Cancer Res. 13, 5995

(2007).

3. G. S. Salvesen, J. M. Abrams, Oncogene 23, 2774 (2004).

4. J. N. Dynek, D. Vucic, Cancer Lett. (2010).

5. A. M. Hunter, E. C. LaCasse, R. G. Korneluk, Apoptosis 12,

1543 (2007).

6. S. W. Fesik, Nat. Rev. Cancer 5, 876 (2005).

7. C. Du, M. Fang, Y. Li, L. Li, X. Wang, Cell 102, 33

(2000).

8. A. M. Verhagen et al., Cell 102, 43 (2000).

9. J. A. Flygare, W. J. Fairbrother, Expert Opin. Ther. Pat.

20, 251 (2010).

10. Z. Liu et al., Nature 408, 1004 (2000).

11. G. Wu et al., Nature 408, 1008 (2000).

12. A. Gaither et al., Cancer Res. 67, 11493 (2007).

13. E. Varfolomeev et al., Cell 131, 669 (2007).

14. J. E. Vince et al., Cell 131, 682 (2007).

15. C. Ndubaku et al., ACS Chem. Biol. 4, 557 (2009).

16. T. K. Oost et al., J. Med. Chem. 47, 4417 (2004).

17. K. Zobel et al., ACS Chem. Biol. 1, 525 (2006).

18. J. W. Blankenship et al., Biochem. J. 417, 149 (2009).

19. T. Samuel et al., J. Biol. Chem. 281, 1080 (2006).

20. E. Varfolomeev, S. M. Wayson, V. M. Dixit,

W. J. Fairbrother, D. Vucic, J. Biol. Chem. 281, 29022

(2006).

21. R. J. Deshaies, C. A. Joazeiro, Annu. Rev. Biochem. 78,

399 (2009).

22. P. D. Mace, S. Shirley, C. L. Day, Cell Death Differ. 17, 46

(2010).

23. P. D. Mace et al., J. Biol. Chem. 283, 31633 (2008).

24. R. Kulathila et al., Acta Crystallogr. D Biol. Crystallogr.

65, 58 (2009).

25. R. Feltham et al., J. Biol. Chem. 286, 17015 (2011).

Acknowledgments: We thank C. Yu, P. Lupardus, Shamrock

Structures LLC, and staff at the Advanced Light Source,

the Stanford Synchrotron Radiation Lightsource (SSRL)

and the Advanced Photon Source at Argonne National

Laboratory for their assistance with sample handling and

data collection. The Advanced Light Source is supported

by the Director, Office of Science, Office of Basic

Energy Sciences, of the U.S. Department of Energy

(DOE) under contract no. DE-AC02-05CH11231. SSRL is

a Directorate of SLAC National Accelerator Laboratory

and an Office of Science User Facility operated for the

DOE Office of Science by Stanford University. The SSRL

Structural Molecular Biology Program is supported by

the DOE Office of Biological and Environmental Research

and by the Biomedical Technology Program, National

Center for Research Resources, NIH (P41RR001209). Use

of the Advanced Photon Source, a facility operated for

DOE Office of Science by Argonne National Laboratory,

was supported under contract no. DE-AC02-06CH11357.

We also thank the Microchemistry and Proteomics

Laboratory and the Oligo Synthesis and DNA Sequencing

facilities at Genentech for their technical support.

Atomic coordinates and structure factors have been

deposited in the Protein DataBank with accession code

3T6P. Reagents are available from Genentech subject

to a material transfer agreement.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/334/6054/376/DC1

Materials and Methods

Figs. S1 to S12

Tables S1 and S2

References (26–37)

3 May 2011; accepted 15 September 2011

10.1126/science.1207862

Mass Spectrometry of Intact V-TypeATPases Reveals Bound Lipids and theEffects of Nucleotide BindingMin Zhou,1* Nina Morgner,1* Nelson P. Barrera,2,3* Argyris Politis,1 Shoshanna C Isaacson,1

Dijana Matak-Vinković,2 Takeshi Murata,4 Ricardo A. Bernal,5 Daniela Stock,6,7 Carol V. Robinson1†

The ability of electrospray to propel large viruses into a mass spectrometer is established and isrationalized by analogy to the atmospheric transmission of the common cold. Much less clear is thefate of membrane-embedded molecular machines in the gas phase. Here we show that rotaryadenosine triphosphatases (ATPases)/synthases from Thermus thermophilus and Enterococcus hirae

can be maintained intact with membrane and soluble subunit interactions preserved in vacuum.Mass spectra reveal subunit stoichiometries and the identity of tightly bound lipids within themembrane rotors. Moreover, subcomplexes formed in solution and gas phases reveal the regulatoryeffects of nucleotide binding on both ATP hydrolysis and proton translocation. Consequently, wecan link specific lipid and nucleotide binding with distinct regulatory roles.

Rotary ATPases/synthases are membrane-

associated molecular machines that per-form biological energy conversion. Both

V-type and F-type complexes consist of two re-versible motors: the ion pump/turbine in VO/FOand the chemical motor/generator in V1/F1. Themode of operation is influenced by the ratio of

the two fuels (protons:ATP) that drive the two

motors. (1–3). The membrane-embedded VO/FOdomain mediates the movement of Na+ or pro-tons across the membrane, whereas V1/F1 do-

mains interact with nucleotides and inorganicphosphate either to produce or consume ATP in

the case of the eukaryotic F- and V-type families,respectively. Eubacteria and archaea typically

have only one type of rotary ATPase/synthasefor both functions.Most bacteria have complexes

of the F-type, but some bacteria and all knownarchaea have complexes closely related to eu-

karyotic V-type ATPases [also known as A-typeATPases/synthases (4)]. Whether of F- or V-type,

the physiological function of most prokaryoticcomplexes is ATP synthesis; however, many have

evolved regulatory functions that allow reversalinto ATP-driven proton pumps if required.

F1-FO, and V1-VO, are mechanically coupledby a central rotating shaft and held together by

peripheral stalks (Fig. 1). Structural details derivefrom isolated subcomplexes of F1 and V1 (5–7)

and from membrane embedded proteolipid rings

of various species (8–11). Despite this wealth of

structural information, no high-resolution struc-tures of any intact rotary ATPases/synthases have

been reported. Thus, regulatory allosteric changesthat involve both the soluble head and the mem-

brane sector are lost. In addition heterogeneousinteractions with lipids and nucleotides are dif-

ficult to observe with existing structural biologyapproaches.

We show using electrospray mass spectrom-etry (MS) that rotary ATPases/synthases from

Thermus thermophilus (TtATPase) and Entero-

coccus hirae (EhATPase) can remain intact in the

gas phase. Previously, composite models were as-sembled of the intact TtATPase by low-resolution

electron microscopy (EM) data in combinationwith high-resolution x-ray structures of subunits

(12), whereas the first cryo-EM data revealedviews of the entire membrane-embedded region

(13). We compared the MS of the two complexes,

the TtATPase with that of the less well charac-terized EhATPase. Current models suggest that

the EhATPase has only one peripheral stalk (14),and the stoichiometry of the K subunits in the

membrane ring was determined as seven fromEM (15) and 10 from x-ray analysis (8).

TtATPase was purified as described (16) withdodecyl maltoside (DDM) for solubilization, be-

cause under these conditions the complex is moststable and does not form aggregates. The complex

was introduced into amass spectrometermodifiedfor high-mass complexes (17).Well-resolved charge

states were assigned to the intact particle consist-ing of 26 subunits and nine different proteins

(Fig. 1A). An experimentally determined mass of659,202 (T131) dalton corresponds to that calcu-

lated for the intact complex, on the basis of subunitmasses determined byMS (table S1), plus addition-

al mass due to incomplete desolvation, and lipidand nucleotide binding (fig. S1). Gas-phase ac-

tivation is necessary to release the complex fromits detergent micelle (18, 19), giving rise to peaks

at higher mass/charge ratio (m/z) than the intact

1Department of Chemistry, Physical and Theoretical Chem-istry Laboratory, University of Oxford, Oxford OX1 3QZ, UK.2Department of Chemistry, Lensfield Road, University of Cam-bridge, Cambridge CB2 1EW, UK. 3Department of Physiol-ogy, Pontificia Universidad Católica de Chile, Alameda 340,Santiago, Chile. 4Department of Chemistry, Graduate Schoolof Science, Chiba University, 1-33 Yayoi-cho, Inage, Chiba 263-8522, Japan. 5Department of Chemistry, University of Texasat El Paso, El Paso, TX 79968, USA. 6The Victor Chang CardiacResearch Institute, Lowy Packer Building, 405 Liverpool Street,Darlinghurst NSW 2010, Australia. 7Faculty of Medicine, Uni-versity of New South Wales, Sydney 2052, Australia.

*These authors contributed equally to this work.†To whom correspondence should be addressed. E-mail:carol.robinson@chem.ox.ac.uk

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org380

REPORTS

complex, formed by unfolding and dissociationof highly charged subunits, predominantly sub-

units E, G, and I (20) (fig. S2). At lower m/z,charge states are assigned to the membranous

subcomplex VO (ICL12), their bimodal distribu-tion implying that they are formed both in solu-

tion and gas phases. The corresponding solubleV1 complex is also observed, confirming that un-

der these solution conditions a proportion of thecomplex dissociates spontaneously (21) (fig. S3).

Using similar experimental parameters, werecorded a spectrum for the EhATPase isolated in

DDM, where its functional activity has been es-tablished previously (22, 23). The membrane-

embedded rotor for this complex is larger becauseeach K subunit contains four transmembrane heli-

ces as compared with two for the correspondingL subunits from TtATPase. Under conditions in

which the complex emerges from the micellesurviving intact, the spectrum is not well resolved

(fig. S4). Increases in activation energy lead tobetter desolvation and also appearance of sub-

complexes in which either the membrane ringremains but the peripheral stalks and subunit I

have dissociated, or the stalk subunits are attachedbut the membrane region is disrupted (Fig. 1B

and fig. S5). The number of peripheral stalks wasdetermined as two and the stoichiometry in the

K ring as 10 (175 kD) (fig. S6).The lipid components in the K10 ring were

identified as a series of negatively charged cardio-

lipins (figs. S2 and S7 and table S2). The stoi-chiometry of lipid binding was determined as 10

from the mass of the membrane ring. We iden-tified six cardiolipin isomers and, from quanti-

tative analysis and measurement of the proteinconcentration, deduce specific binding of one

lipid per subunit [EhK subunit 1: 1.2 T 0.1 cardio-lipins (fig. S8)]. Previously, the lipids were lo-

cated based on the atomic structure of the isolatedK10 ring, in which peaks of positive density were

attributed to 20 bifurcated phosphatidylglycerollipids (8). We docked 10 negatively charged sym-

metric cardiolipins inside the K10 ring, proximalto the conserved Lys32. The four hydrophobic

chains are positioned with two chains emanatingfrom both sides of the polar head, thus providing a

hydrophobic lining to the inside of the ring (Fig. 2A).For the TtATPase L12 ring, we identified the

bound lipid as phosphatidylethanolamine (PE)(Fig. 2 and fig. S9). Unexpectedly, dissociated

L subunits were observed either with bound PElipid (holo) or without lipid (apo) after tandem

MS of a subcomplex containing the membranesubunits and peripheral stalks (Fig. 2). Before

disruption, the mass of the subcomplex is con-sistent with binding of six lipids. This means that

apo and holo subunits coexist within the samecomplex.We confirmed this observation by quan-

tifying the protein:lipid ratio as one L subunit:0.55 T 0.1 PE (i.e., one PE lipid per L subunit

dimer) (fig. S10). Six equivalent binding sites forlipids in protein dimers in a 12-membered ring

imply a sixfold symmetrical state. Previously, EM

studies of the same preparation of the TtATPaserevealed a sixfold symmetricmembrane-embedded

ring (Fig. 2 and fig. S11) (16). Both sets of dataare consistent with close packing of two neigh-

boring subunits, forming six dimers each withfour transmembrane helices (TMHs), thus emu-

lating the arrangement of C subunits in eukary-otic V-ATPases in which gene duplication has led

to four TMHs per subunit (24, 25). From model-ing, we find that a rotation of the dimers by

about 60° relative to their original orientationin the ring effectively locks six Glu63 residues

in an occluded position, preserving lipid bindingat the dimer interface. Proton transfer can occur at

the remaining six active glutamates [Fig. 2, (i) to(iii)]; consequently, the proton:ATP ratio for the

rotary enzyme is effectively halved.

A

B

L

K

52+

25+

32+

GGEE

19+

45+

36+

GGEE

-C

-F

-F

- I(EF)2

rela

tive

inte

nsi

ty (

%)

m/z10000 15000 20000 25000

100

0

CD

B A

G

C

E

F

D

B A

G

D

B A

C

E D

B A

G

E

F

D

B A

G

E D

B A

G

m/z

100

26+

34+

-G

-E

6000 9000 12000 15000

57+

45+

rela

tive

inte

nsi

ty (

%)

0

26+

L

C

G D

B A

F

I

L

C

G

E

D

B A

F

I

L

CD

B A

F

I

L

C

I

B

FCD

I

G

E

A

L

C

I

78007000

24+

K

E

FB A

GD

K

C

I

Fig. 1. Mass spectra of the intact rotary ATPases from T. thermophilus and E. hirae. (A) Peaks are assignedto the intact TtATPase complex (stars), loss of the membrane subcomplex (ICL12) in solution and gas phases(dark green and green hexagons, respectively), and dissociation of subunits E and G from the peripheralstalk (blue circles and squares respectively). (B) For EhATPase, the membrane subcomplex is observed incontact with the soluble head (green squares). (Inset) Mass spectra of the K ring in aqueous solution.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 381

REPORTS

In contrast to E. hirae, which has both F- andV-type ATPases, T. thermophilus has only one

type of rotary ATPase that operates in both pro-ton pumping and ATP synthesis modes in vitro

(26). Given that each of the 12 L subunits in theTtATPase membrane ring contains only two

TMHs, alternating between synthesis and pumpingwould involve switching the 12-membered ring

from a high proton:ATP ratio, primed for syn-thesis (27), into a lower one biased for pumping

(16). This latter scenario thus mimics the eu-karyotic V-type enzyme where gene duplication

has provided four TMHs each with one activeglutamate. Consequently, this mechanism for

switching betweenATP synthesis and ion-pumpingmodes is likely assisted by specific lipid binding

for this dual-function rotary ATPase.These different lipid-binding patterns in the

two rotary ATPases studied here were found tobe invariant between repeat preparations and

different detergent concentrations. Together, theyprovide an explanation as to how the disparate

membrane rings L12 and K10 (93.6 and 160 kD)can interact with their respective C subunits (35.8

and 38.2 kD), which are likely conserved. Liningwith cardiolipin, bound specifically to the inside

of the K10 ring, reduces the orifice from 54 to38Å (Fig. 2A and fig. S12). Similarly, converting

a 12- to a 6-membered ring for TtATPase reducesthe orifice from 47 to 39 Å, creating two very

similar inner diameters (38 and 39 Å) (Fig. 2B).To assess the role of nucleotides in changing

subunit interactions, we compared spectra for theTtATPase, with and without addition of 50 mM

ATP. In the presence of ATP, the intact complexand membrane-embedded subcomplex ICL12

are the predominant species formed in solution(Fig. 3A and fig. S13). When ATP is depleted,

loss of subunit B leads to extensive dissociationof the soluble head. In addition, tandem MS of

V1 in the presence of ATP leads predominantly toloss of subunit F, with subsequent loss of subunit

A

B

C

D

CH3O

O

O

OP

O

O

NH3+

O-

O H

CH3

CH3

O

O C OP

OH

O

O

CH3

O

O H

CH

O

O C OP

O

O

OH

CH3

3

O

O H

OH H

(iii)(ii)(i) (iv)

2000 4000

%

0

100

1800016000

34+35+

36+

85008000

3+4+

2+

3+4+5+

6+

2+

12+

11+12+

10+

C

m/z

Fig. 2. Lipid binding and its effect on the membrane rings. (Upper panel)Predominant lipid molecules, one of the six cardioplipins (left) and phos-phatidylethanolamine (PE, right), found in intact EhATPase and TtATPase,respectively. (Central panel) Tandem MS of a subcomplex from TtATPase[ICL12E2G2F (282 740 daltons)] leads to disruption of the L12 ring, releasingproteolipids L T PE (red/green circle, 8539 daltons; red circles, 7849 daltons)and a stripped-complex ICE2G2F (blue squares, 184 242 daltons). Atomicstructure of the K10 ring (8) of EhATPase with docking of 10 cardioplipins to

show reduction in the inner diameter (A) and after docking subunit C (B).Models for sixfold symmetry of the L12 ring with six PE molecules (green)(C) and with subunit C (42) (blue) docked into the ring (D). (Lower panel)schematics of the rotor ring with 12 L subunits each having two TMHs (redcylinders) and one conserved Glu63 (yellow) as seen in EM of two-dimensionalcrystals of isolated L ring (27) (i). Transformation into a sixfold symmetric ring[(ii) and (iii)]. (iv) Comparison of the sixfold symmetrical model with EM datareported previously for the intact TtATPase (iv) (16).

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org382

REPORTS

D (Fig. 3B). When ATP is depleted, however,loss of subunit D before subunit F becomes pos-

sible, leaving a subcomplex with F interactingdirectly with the A3B3 hexamer. Extension of

F toward the soluble head has been proposedpreviously as a “braking” mechanism to prevent

unregulated consumption of ATP after in vivodissociation of the head from the base in yeast

V-ATPase (28). In linewith this proposal, an x-raystructure and cryo-EM reconstruction of the iso-

lated A3B3DF complex from TtATPase and yeastV-ATPase respectively, also reveal interactions

of subunit F with the A3B3 hexamer (7, 29). Anx-ray structure of the auto-inhibited F1 head of

the Escherichia coli F-ATPase showed interac-

tions between the subunits analogous to F1 e

and b in the soluble head (30). Subjecting the

E. coli F1 complex to the same tandem MS pro-cedure outlined above showed that subunit e

makes direct interactions with the soluble head(fig. S14). The similar MS dissociation patterns

observed in E. coli F-ATPase, in which the brakingmechanism is well established, and in TtATPase

suggest an analogous mechanism to preventATP hydrolysis in the uncoupled V1 complex of

TtATPase.We observed further sensitivity of TtATPase

to low ATP concentrations, notably loss of sub-unit I from ICL12 to form CL12 (Fig. 3A). Expan-

sion of the peaks assigned to ICL12 is consistentwith binding of up to six lipids and up to two

nucleotides (ATP or ADP) (fig. S15). This agreeswith proposals that a eukaryotic functional equiv-

alent of subunit I senses cellular nucleotide levels

by binding selectively to ADP and undergoingconformational change (31). To investigate this

conformational change, we applied ion mobilityMS (IM-MS) (32) to the intact TtATPase, ICL12,

and CL12 subcomplexes. Because ions with mul-tiple conformations result in broad arrival-time

distributions (ATDs), we conclude that the rela-tively compact ATDs for both the intact ATPase

and CL12 are consistent with one predominantconformation (fig. S16 and tables S3 and S4). By

contrast, the ATDs for the ICL12 complex aremuch broader than those of the intact TtATPase

and CL12, consistent with multiple conformationsof subunit I in the isolated VO complex (Fig. 4

and table S4). The lack of conformational het-erogeneity in the intact complex is rationalized

by the “tethering” of subunit I by forces exertedby the peripheral stalks (subunits E and G), as

suggested recently (13). Once released from the

intact complex, subunit I in VO is not constrained,and flexibility of the hinge domain, located

between the soluble and transmembrane domains,likely leads to its conformational heterogeneity.

I

D

6000 8000 10000 12000

Inte

nsity

(%

)

L

CG

EA

F

C

I

L

CB

F

D

2000 10000 14000 18000 22000

37+

38+

26+31+

30+

17+

17+19+

7+

15+

L3+

L3+

7+

15+

m/z

ADP

ATP

ATP

ADP

D

B A

F

D

A

F

B

D

B A

F

D

AB

A

F

B

AB

F

D

D

B A

F

D

B A

F

AB

D

D

Δ 14+

Δ 13+

Δ 19+

Δ 7+Δ 12+

Δ 7+

D FFL

C

L

C

IATP

ADP

+

F

56+

56+

26+

25+

19+

L

I

I L

CG

E

D

A

F

A B

B

Fig. 3. Nucleotide binding and its effects on intact TtATPase and themembrane-embedded and soluble head complexes. (A) Depletion of ATPleads to dissociation of the B subunit from the head and subunit I from ICL12.(B) Tandem MS of the soluble head (A3B3DF) reveals sequential loss of sub-units F and D in the presence of 50 mM ATP. In TtATPase solutions containing50 mM ADP, two conformations of subunit F are evident from the bimodal

distribution of charge states formed for the V1 stripped of F (D7+ and D12+)together with a direct loss of subunit D. The D7+ series is similar to thatformed from the ATP-bound complex. The D12+ series is consistent with anextended conformation of subunit F. (Inset) Schematic representation of ef-fects of ATP/ADP on the membrane ICL12 complex and the movement of anextended subunit F in the A3B3DF complex.

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 383

REPORTS

The soluble domain of I in EM density mapsof the intact TtATPase is at 90° to the proton

channel (12, 13). Our IMdata for theVO complex,based on modeling of the ICL12 and CL12 com-

plexes (fig. S17), are consistent with a range ofconformational states with angles from 90° to

135° (fig. S18). We propose, therefore, that the

conformational dynamics demonstrated here, to-gether with preferential binding of ADP to the

proposed site near the hinge region (31), destabi-lize interactions between I and CL12 as evidenced

by the facile loss of subunit I under low-ATPconditions.

Given that cellular nucleotide levels likelyaffect proton translocation in the isolated VO

complex, it might also be anticipated that changesin the proton gradient would induce similar reg-

ulatory effects. To test this hypothesis, we in-creased the pH of the ATPase-containing solution

to mimic reduction of the proton concentration.Mass spectra of the TtATPase incubated at pH

9.0 led to subcomplexes formed by loss of IGE(fig. S19). This observation, together with the

dissociation of subunit I from VO, suggests a reg-ulatory role for subunit I, sensing both proton

and ATP concentrations. Previous proposals in-voked a locking together of the membrane por-

tion of subunit I with the CL12 membrane ring,

preventing relative movement and hence explain-ing the absence of passive H+ translocation in

isolated VO (33, 34). Because we observe facileloss of subunit I under both low–H+ concentra-

tion ([H+]) and low-[ATP] conditions, and giventhe lack of extensive interactions between I and

the membrane ring observed in cryo-EM data

(13), our results point to a mechanism in whichsubunit I moves away from the ring, and the re-

sulting gap could then be sealed with membranelipids (Fig. 4D).

Our results show that lipids with two and fourhydrophobic chains associate with subunits with

two and four TMHs—subunits L and K, respec-tively. In both ATPases, the lipids identified in

situ are not the most prevalent ones in the cell(35, 36), implying that lipids are selected from

the available pool for specific structural rolesand metabolic regulation. This further supports

the proposal that membrane proteins possessspecific lipid binding sites (37) and demon-

strates the ability of lipids to fine tune subunitinteractions by defining the conformations and

inner dimensions of the membrane rings. Ournucleotide-binding experiments show that a de-

crease in cellular ATP concentrations is sensedby V1, not only with the movement of subunit

F but also with changes in interactions at the A:B

interface. Unexpectedly, VO is also sensitive tolow [ATP] and [H+], both of which promote dis-

placement of subunit I.We suggest that membranelipids subsequently seal the proton-conducting

channel. Consequently, both ATP and proton/iongradients are conserved when reversible disso-

ciation takes place in vivo.

The existence of intact rotary ATPases/synthases in vacuum has been demonstrated

previously with LILBID MS (38, 39). At its cur-rent resolution, however, it is not possible to

identify bound lipids or nucleotides or to probetheir effects on subunit interactions. Moreover,

previous ES-MS experiments produced well-resolvedmass spectra for the V1 domain (40, 41),

but lack of an intactVO domain, or any interactionsbetween V1 and VO, is attributed to dissociation

of the complex in the absence of the protectivemicelle (18, 19). By contrast, the ES approach

used here enables interrogation of subunit inter-actions within intact rotary ATPase/synthases

and allows us to probe the synergistic effects oflipid and nucleotide binding.

References and Notes1. K. C. Jefferies, D. J. Cipriano, M. Forgac, Arch. Biochem.

Biophys. 476, 33 (2008).2. C. von Ballmoos, A. Wiedenmann, P. Dimroth, Annu. Rev.

Biochem. 78, 649 (2009).

5000 6000 7000 8000 9000

21+

5

20

15

Dri

ft t

ime

(ms)

10

7

22

17

Dri

ft t

ime

(ms)

12

m/z

A

25+

ADP

ATP

E

r

H+

D no ATP / with ADPwith ATP

membrane lipids

Drift time (ms)5 20 35

%

0

100 10.6210.97

11.49

B

Drift time (ms)

21+22+

23+

%

0

100 16.0516.91

17.9419.48

5 20 35

24+25+26+

C

L

C

L

C

I

Fig. 4. Conformational heterogeneity and dissociation of subunit I from ICL12implies a mechanism for closing the H+ channel. (A) IM-MS of the trans-membrane ICL12 and CL12 complexes formed in solution from intact TtATPase.Charge states used for IM measurement are labeled gray and red (B and C).Broader arrival-time distributions for ICL12 than for CL12 are consistent withconformational heterogenity in subunit I. (D) Possible mechanism to “close”the H+ channel after lateral movement of subunit I within the membrane.

Surrounding lipids effectively block the channel. The direction of protonpumping (green) and rotation of the ring (yellow) for ATP hydrolysis, as seenfrom the top (site of interaction with V1). (E) Coarse-grained and atomic modelof the ICL12 complex generated according to IM restraints and homologymodeling (see supplementary methods). The proposed nucleotide bindingarea in the hinge region of I (orange) and Glu63:Arg563 from subunits L and Iare shown in yellow and green, respectively.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org384

REPORTS

3. S. P. Muench, J. Trinick, M. A. Harrison, Quat. Rev. Biophys.

44, 311 (2011).

4. V. Müller, G. Grüber, Cell. Mol. Life Sci. 60, 474 (2003).

5. J. P. Abrahams, A. G. Leslie, R. Lutter, J. E. Walker, Nature

370, 621 (1994).

6. K. Okazaki, S. Takada, Structure 19, 588 (2011).7. N. Numoto, Y. Hasegawa, K. Takeda, K. Miki, EMBO Rep.

25, 25 (2009).

8. T. Murata, I. Yamato, Y. Kakinuma, A. G. Leslie, J. E. Walker,

Science 308, 654 (2005).

9. T. Meier, P. Polzer, K. Diederichs, W. Welte, P. Dimroth,

Science 308, 659 (2005).

10. A. Dautant, J. Velours, M. F. Giraud, J. Biol. Chem. 285,

29502 (2010).

11. I. N. Watt, M. G. Montgomery, M. J. Runswick, A. G. Leslie,

J. E. Walker, Proc. Natl. Acad. Sci. U.S.A. 107, 16823

(2010).

12. L. K. Lee, A. G. Stewart, M. Donohoe, R. A. Bernal,

D. Stock, Nat. Struct. Mol. Biol. 17, 373 (2010).

13. W. C. Lau, J. L. Rubinstein, Proc. Natl. Acad. Sci. U.S.A.

107, 1367 (2010).

14. M. Yamamoto et al., J. Biol. Chem. 283, 19422 (2008).

15. T. Murata et al., J. Biol. Chem. 278, 21162 (2003).

16. R. A. Bernal, D. Stock, Structure 12, 1789 (2004).

17. H. Hernández, C. V. Robinson, Nat. Protoc. 2, 715 (2007).

18. N. P. Barrera, N. Di Bartolo, P. J. Booth, C. V. Robinson,

Science 321, 243 (2008).

19. N. P. Barrera et al., Nat. Methods 6, 585 (2009).

20. J. L. Benesch, C. V. Robinson, Curr. Opin. Struct. Biol. 16,

245 (2006).

21. K. Yokoyama et al., J. Biol. Chem. 278, 42686 (2003).

22. K. Takase et al., J. Biol. Chem. 269, 11037 (1994).

23. T. Murata, K. Takase, I. Yamato, K. Igarashi, Y. Kakinuma,

J. Biochem. 125, 414 (1999).

24. M. Mandel et al., Proc. Natl. Acad. Sci. U.S.A. 85, 5521

(1988).

25. D. K. Clare et al., Structure 14, 1149 (2006).

26. K. Yokoyama et al., J. Biol. Chem. 275, 13955

(2000).

27. M. Toei et al., Proc. Natl. Acad. Sci. U.S.A. 104, 20256

(2007).

28. P. M. Kane, A. M. Smardon, J. Bioenerg. Biomembr. 35,

313 (2003).

29. Z. L. Hildenbrand, S. K. Molugu, D. Stock, R. A. Bernal,

PLoS ONE 5, e12588 (2010).

30. G. Cingolani, T. M. Duncan, Nat. Struct. Mol. Biol. 18,

701 (2011).

31. A. Armbrüster et al., FEBS Lett. 579, 1961 (2005).

32. C. Uetrecht, R. J. Rose, E. van Duijn, K. Lorenzen,

A. J. Heck, Chem. Soc. Rev. 39, 1633 (2010).

33. J. Zhang, M. Myers, M. Forgac, J. Biol. Chem. 267, 9773

(1992).

34. C. Beltrán, N. Nelson, Acta Physiol. Scand. Suppl. 607,

41 (1992).

35. D. Carson, R. A. Pieringer, L. Daneo-Moore, Biochim.

Biophys. Acta 575, 225 (1979).

36. P. H. Ray, D. C. White, T. D. Brock, J. Bacteriol. 108, 227 (1971).

37. C. Hunte, S. Richers, Curr. Opin. Struct. Biol. 18, 406

(2008).

38. T. Meier et al., Mol. Microbiol. 65, 1181 (2007).

39. J. Hoffmann et al., Phys. Chem. Chem. Phys. 12, 13375

(2010).

40. O. Esteban et al., J. Biol. Chem. 283, 2595 (2008).

41. N. Kitagawa, H. Mazon, A. J. Heck, S. Wilkens, J. Biol.

Chem. 283, 3329 (2008).

42. M. Iwata et al., Proc. Natl. Acad. Sci. U.S.A. 101, 59

(2004).

Acknowledgments: Funding from the Wellcome Trust and

by the PROSPECTS (HEALTHF4-2008-201648) grant

within the Research Framework of the European Union

together with funding from the Royal Society (C.V.R.),

the Australian National Health and Medical Research

Council grant 573712 (D.S.), and the FONDECYT

1100515 (N.P.B.) is acknowledged.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/334/6054/380/DC1

Materials and Methods

Figs. S1 to S19

Tables S1 to S4

References (43–58)

22 June 2011; accepted 15 September 2011

10.1126/science.1210148

Cerebellum Shapes HippocampalSpatial CodeChristelle Rochefort,1* Arnaud Arabo,1*† Marion André,2‡ Bruno Poucet,2

Etienne Save,2* Laure Rondi-Reig1*§

Spatial representation is an active process that requires complex multimodal integration from alarge interacting network of cortical and subcortical structures. We sought to determine the roleof cerebellar protein kinase C (PKC)–dependent plasticity in spatial navigation by recording theactivity of hippocampal place cells in transgenic L7PKCI mice with selective disruption ofPKC-dependent plasticity at parallel fiber–Purkinje cell synapses. Place cell properties wereexclusively impaired when L7PKCI mice had to rely on self-motion cues. The behavioralconsequence of such a deficit is evidenced here by selectively impaired navigation capabilitiesduring a path integration task. Together, these results suggest that cerebellar PKC-dependentmechanisms are involved in processing self-motion signals essential to the shaping of hippocampalspatial representation.

It is well established that rodents build an

internal cognitive map to navigate in their en-vironment. A key neural substrate enabling

such representation is the hippocampus, whichcontains CA1 and CA3 pyramidal cells described

as place cells. Each place cell fires for a restricted

region (the place field) of the environment (1, 2).

Both external cues and self-motion cues (i.e.,vestibular, proprioceptive, and optic flow cues)

control place cell firing (3, 4), which suggests theinvolvement of a large network of cortical and

subcortical structures interacting with the hippo-campus for navigation. Determining the function-

al architecture of such a network is thus essentialto our understanding of how the hippocampal

place cell code is generated. The medial ento-rhinal cortex, a key relay structure between neo-

cortical areas and the hippocampus, contains gridcells with regularly spaced multiple firing fields

(5), which integrate self-motion information andparticipate in path integration (4, 6, 7).

The cerebellum has also been shown to beessential to the processing of self-motion infor-

mation: Cerebellar Purkinje cells respond to vestib-ular signals by transforming head-centered vestibular

afferent information into Earth-reference self-motion and spatial orientation signals (8, 9), and

electrophysiological investigations suggest that

the cerebellum and the hippocampus can be func-tionally connected during eyeblink conditioning

(10, 11). However, it is still unknown whethersuch an interaction is functionally relevant in nav-

igation, and a mechanism that might underliesuch a process has not been identified.

In the transgenic mouse strain L7PKCI, thepseudosubstrate protein kinase C inhibitor (PKCI)

is selectively expressed in Purkinje cells underthe control of the pcp-2 (L7) gene promoter (12).

This results in an impaired long-term depression

(LTD) at cerebellar parallel fiber–Purkinje cellsynapses. Such a plasticity mechanism has been

proposed to work as an error-based (anti-Hebbian)learning process (13, 14) during conditioning

tasks (15) and in optimization of motor responseduring navigation (16).

A total of 506 dorsal CA1 hippocampal cellswere recorded. A subset of 150 place cells was

further analyzed in six L7PKCI mice and fivewild-type littermate control mice. Relative towild-

type mice, L7PKCImice had a significantly lowerproportion of place cells [L7PKCI, n = 53/218

(24.3%); wild type, n = 97/288 (33.7%); c2 = 5.2,df = 1, P < 0.025]. Neural activity was sampled

as the mice freely explored a circular arena con-taining a salient cue (a card with a bottle attached

to it), in standard sessions (S1 and S2) and in-volving cue manipulation in subsequent sessions

(S3 and S4). A last session (S5) similar to ses-sions S1 and S2 was run to determine whether

we could restore the initial firing pattern irre-spective of the changes in cell firing observed

during the cuemanipulation sessions (Fig. 1A) (17).After recording in the standard sessions, we

used two distinct environmental manipulations,cue removal and cue conflict, in which mice are

forced to use self-motion cues. In the cue removalcondition, the arena was in the dark and the cue

1Neurobiologie des Processus Adaptatifs (UMR 7102), Naviga-tion, Memory, and Aging (ENMVI) Team, Université Pierre etMarie Curie–Centre National de la Recherche Scientifique(CNRS), F-75005 Paris, France. 2Laboratory of Neurobiologyand Cognition (UMR 6155), Aix Marseille Université–CNRS, 3place Victor Hugo, 13331 Marseille, France.

*These authors contributed equally to this work.†Present address: Laboratoire de Psychologie et Neurosciencesde la Cognition et de l’Affectivité (EA 4306), Université deRouen, Faculté des Sciences, Place Emile Blondel, 76821Mont-Saint-Aignan Cedex, France.‡Present address: Department of Experimental Neurophys-iology, Ruhr University Bochum, Universitätsstrasse 150, MABF01/551, 44801 Bochum, Germany.§To whom correspondence should be addressed. E-mail:laure.rondi@snv.jussieu.fr

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 385

REPORTS

was removed. A control condition was also per-formed with the cue still present in darkness (Fig.

1B). In the cue conflict condition, we used aprotocol previously developed in rats, in which

the external cue was rotated 180° in the absence(hidden rotation) or in the presence (visible

rotation) of the animal, therefore producing aconflict between visual and self-motion infor-

mation (18). During the conflict, rats maintainplace field stability relative to the standard ses-

sion, thus suggesting the dominant use of self-motion cues (18).

The basic firing properties of place cells in thelight condition were unaffected in L7PKCI mice

(table S1). In addition, place field stability (mea-sured as a correlation between two similar light

sessions)was higher in L7PKCImice (0.78 T 0.05)than in wild-type mice (0.60 T 0.03) (t146 = 3.0,

Fig. 1. The compulsory use of self-motion cues affects hippocampal place cellproperties in L7PKCI mice. (A and B) Schematic diagram of the protocol used toassess the effect of self-motion stimulation on place cell properties. After twoconsecutive standard sessions (S1 and S2), light was turned off (S3 and S4) andobjects were either removed (A) or maintained (B) in the arena. S5 was similar toS1 and S2. (C and D) Examples of color-coded rate maps showing firing activityof wild-type (WT) and L7PKCI single CA1 pyramidal cells over the five con-secutive sessions; color coding ranges from blue (silent) to red (peak activity).

Peak firing rates are indicated for each rate map. (E to H) Analysis of place cellcharacteristics shows that the suppression of external cue inputs significantlyalters both the mean field rate (E) and spatial coherence (F) in L7PKCI micespecifically, whereas the suppression of the visual cue alone has no effect [(G)and (H)]. (I to L) Place field stability, as measured within (I) or across (J) sessions,is affected in L7PKCI mice after suppression of all external cues, but not aftersuppression of the visual cue alone [(K) and (L)]. *P < 0.05, **P < 0.01, ***P <0.001 with a Newman-Keuls post hoc analysis. Error bars represent SEM.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org386

REPORTS

P = 0.003). In sharp contrast, several firing pa-rameters were strongly affected in the dark ses-

sions after cue removal in L7PKCI mice (Fig. 1,C, E, and F, and figs. S1A and S2). The mean

field rate, peak firing rate, and overall mean firingrate declined during the dark sessions in L7PKCI

mice but not in wild-type mice (mean field rate,F1,39 = 11.5, P = 0.002; peak firing rate, F1,39 =

12.1, P = 0.001; overall mean firing rate, F1,39 =8.4, P = 0.006) (Fig. 1E and fig. S2). The field

spatial coherence was also found to be decreasedin the dark (F1,39 = 11.2, P = 0.002) (Fig. 1F).

Finally, place field stability in L7PKCI mice wasmarkedly affected by the dark condition: Where-

as wild-type mice maintained place field stabil-

ity throughout the dark sessions, place cells of

L7PKCI mice showed a progressive decrease ofwithin-session stability (F1,39 = 15.6, P= 0.0003)

as well as between-session stability (F2,78 = 6.2,P= 0.003) (Fig. 1, I and J). These results indicate

that place field stability in L7PKCI mice grad-ually decreased over sessions in the dark (post

hoc analysis; P < 0.01 between S1-S2 and S2-S3correlations,P < 0.001 between S1-S2 and S2-S4

correlations) (Fig. 1J).By contrast, place cell firing properties and

place field stability were restored in session S5(fig. S3). In most instances, all simultaneously

recorded cells behaved homogeneously (i.e., allfields were either stable or remapped together).

The modification of place cell properties of

L7PKCI mice was not due to an impaired explor-

atory activity in the dark, because wild-type andL7PKCI mice displayed similar speed (2.09 T

0.08 cm/s versus 1.97 T 0.05 cm/s, t19 = 1.7, P >0.05, t test) and similar traveled distance (14.24 T

0.59 cm versus 13.10 T 0.35 cm, t19 = 1.4, P >0.05, t test) (Table 1).When the cue was available

in the dark, the firing parameters and place fieldstability were not affected in L7PKCI mice

(Fig. 1B and fig. S1B) (P > 0.05 for all param-eters analyzed).

These results suggest that in the dark and inthe absence of the cue, the place cell system of

L7PKCI mice failed to use self-motion informa-tion to maintain stable place fields. Consistent

with this finding, the mice were able to maintain

stable place fields when they could update theirposition by using the cue. As a consequence,

the number of place fields away from the object(>20 cm)was drastically reduced in L7PKCImice

relative to wild-type mice [L7PKCI, n = 1/16cells (6%); wild type, n = 20/37 cells (46%); c2 =

10.67, df = 1, P = 0.0011]. The relative powerof the hippocampal theta band (5 to 10 Hz) was

similar in L7PKCI and wild-type mice (F1,21 =1.72, P > 0.05) in both light and dark cue re-

moval conditions (F1,21 = 3.86, P > 0.05), whichsuggests that alteration of path integration was

not caused by a modification of theta rhythm(fig. S4).

To further investigate the respective influenceof self-motion and external information on spatial

firing pattern in L7PKCI mice, we conducted aconflict condition protocol (Fig. 2A). After two

standard sessions, a 180° hidden rotation of thecue resulted in similar rotation of the place fields

in both wild-type and L7PKCI mice (Fig. 2, B

Fig. 2. Field locations are not efficiently controlled by self-motion cues inL7PKCI mice. (A) Schematic diagram illustrating the protocol used to assessthe effect on place cell firing of a 180° rotation of the cue in the absence(hidden rotation) or presence (visible rotation) of the mouse in the arena. (B)Color-coded rate maps showing firing activity of WT and L7PKCI single CA1pyramidal cells over the five consecutive sessions. (C and D) Histograms

showing the intersession similarity coefficient score associated to a 0° or 180°field rotation after a hidden (C) or a visible (D) rotation of the cue. Fieldstability significantly decreased in L7PKCI after a visible rotation of the cue(D). (E and F) Polar distribution of the place field rotation angles after thevisible rotation in WT mice (E) and L7PKCI mice (F). *P < 0.05, Student t test.Error bars represent SEM.

Table 1. General sensory-motor abilities of WT and L7PKCI mice in the dark (means T SEM). Nosignificant differences between WT and L7PKCI mice were revealed by the different sensory-motor tasks(t test, P > 0.05 for all parameters) assessed in the dark (i.e., using primarily the vestibular system).

Task MeasureL7PKCI

(n = 7)

WT

(n = 6)Mann-Whitney P

Spontaneous

locomotor activity in

the dark

Speed (cm/s) 1.97 T 0.05 2.09 T 0.08 0.10

Distance traveled (cm) 13.10 T 0.35 14.24 T 0.59 0.09

Rearing frequency

(number/min)

4.29 T 0.73 6.20 T 0.73 0.18

Dynamic balance in

the dark

Falling latency (s) 180 180 —

Distance traveled (cm) 629 T 77 542 T 51 0.45

Static balance in the dark Falling latency (s) 144 T 18 105 T 17 0.10

Motor coordination in

the dark (rotarod)

5 rpm walking time (s) 141 T 21 138 T 12 0.45

10 rpm walking time (s) 141 T 28 157 T 17 0.63

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 387

REPORTS

and C, and fig. S1C), indicating that the cue ef-ficiently controlled place cell activity. Visible ro-

tation of the cue was then performed, producing aconflict between external and self-motion sen-

sory information. During the conflict, 63% ofplace cells in wild-type mice maintained their

place field stability relative to the previous ses-sion (0° T 30° rotation), which suggests that

the mice resolved the conflict by relying on self-motion cues (18) (Fig. 2, B, D, and E, and fig.

S1C). The distribution of place field rotationangles after the visible rotation was therefore

concentrated around the same position (Fig. 2E;Z = 13.53,P< 0.001, Rayleigh test). In contrast, a

majority of place cells in L7PKCI mice exhibited

remapping at a different location, leading to ahomogeneous distribution of place field rota-

tion angles (Fig. 2, B, D, and F; Z = 1.55, P= 0.2,Rayleigh test): 30% of place fields remained

stable, suggesting a control by self-motion cues;20% exhibited a 180° T 30° rotation, suggesting

a control by the external cue; and the remaining50% rotated at various angles. As a result, field

stability (as measured by intersession similaritycoefficient at a rotation angle of 0°) between

sessions S4 and S5 was significantly lower inL7PKCI mice than in wild-type mice (Fig. 2D;

t54 = –2.0, P < 0.05). The inability to maintainstable place fields in L7PKCI mice strengthens

the idea of a deficit in the use of self-motion cues.We next examined the ability of L7PKCI

mice to navigate in the dark (i.e., using self-motion cues). L7PKCI mice were trained to find

an escape platform at a constant location with aconstant departure point in the water maze (Fig.

3A). Path optimization was analyzed in lightand dark conditions (17) (table S3). In the light,

L7PKCI mice learned to reach the platform asrapidly and accurately as their control littermates

(Fig. 3, B and C). Both groups increased the useof direct trajectories across training sessions and

decreased other nonoptimal trajectories (Fig. 3, Dand E). In sharp contrast, navigation performance

in the dark was impaired in L7PKCImice (Fig. 3,B and C). Escape latencies and heading were

significantly greater in L7PKCI mice than inwild-type mice (genotype effect, F1,28 = 4.98,

P = 0.034, and F1,28 = 9.63, P = 0.004, respec-

tively), even though there was no difference inswimming speed (F1,28 = 2.01, P = 0.2), circling

(F4,112 = 0.56, P = 0.7) (fig. S5), or other be-havioral parameters that could interfere with nav-

igation (Table 1 and table S2; P > 0.05) (19).Thus, assessment of navigation abilities in the

dark demonstrates impaired path integration per-formances in the L7PKCI mice.

The trajectories of the mutant mice were lessefficient than those of their control littermates in

darkness, as highlighted by the differences in thetype of trajectory used (Fig. 3, D and E). The

importance of the dark context on the deficit ex-hibited here by the L7PKCI mice was reinforced

by the absence of significant genotype effectobserved during a control trial that took place in

the light condition during dark session 3 (D3T1

in Fig. 3, B and C) (table S3) (17). Accordingly,comparing this trial with the mean of the last

trial (L5) in the light condition revealed no sig-nificant differences. This indicates that the dis-

turbed trajectories displayed by transgenic micein the dark cannot be attributed either to a deficit

in the use of task rules, or to altered motivation.The fundamental finding of our study is that

mice lacking PKC-dependent cerebellar LTDshowed disrupted hippocampal place cell proper-

ties and impaired goal-directed navigation in con-ditions in which self-motion information must be

predominantly used. We previously suggested arole of PKC-dependent mechanisms in the link-

age between the spatial context and the motor

response characterized by the animal’s trajectory(16, 20). Here, we demonstrate an additional and

complementary role of PKC-dependent cerebel-lar LTD in self-motion–based hippocampal rep-

resentation and path integration. Although the

cerebellum is classically viewed as a motor struc-ture, a growing body of evidence indicates that

cerebellar circuitry is well suited to act as anadaptive filter of sensory information (21–23). In

particular, vestibular information is combinedwith proprioceptive inputs in the cerebellar fas-

tigial nucleus to generate appropriate internalestimates of the animal’s self-motion (24). In ad-

dition, cerebellar Purkinje cells from lobules IXand X transform vestibular head-centered signals

into self-motion and spatial orientation signalsrelative to the external world (8, 9). It thus ap-

pears that, beyond its role in motor adaptationduring navigation (16, 20), cerebellar LTD con-

tributes to the representation of the relation of

the body to the external world, thereby shapinghippocampal spatial representation.

Recent data show a clear contribution of thevestibular system to hippocampal-dependent spa-

tial memory (25, 26) as well as to spatial firing of

Fig. 3. Inactivation of PKC-dependent cerebellar LTD deteriorates path integration. (A) Design of theexperimental space developed to evaluate navigation abilities using self-motion cues. (B and C) Quan-tification of escape latencies (B) and heading (C) in WT and L7PKCI mice during both light and darkconditions. In the light condition, WT and L7PKCI mice improved their performances significantly oversessions without genotype effect. In the dark condition, both groups improved their performance overtime, but the performance of LKPCI mice was significantly poorer than that of their control littermates. (Dand E) Swim path analyses during both light and dark conditions. The direct trajectory was significantlyimpaired in L7PKCI mice during the dark condition (D). L7PKCI mice cannot perform optimal trajectoriesduring path integration, as highlighted by the differences between WT and L7PKCI mice in the type oftrajectory used in the dark but not in the light condition (E). The P values indicated in (B) to (D) correspondto the genotype effect. *P < 0.05 with Newman-Keuls post hoc analysis. Error bars represent SEM.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org388

REPORTS

hippocampal neurons (27). However, examina-tion of the vestibular-associated motor activity of

L7PKCI mice in both light and dark conditionsrevealed no deficit. Our data do not suggest a

vestibular implication underlying the observedalterations of place cell firing and navigation in

L7PCKI mice. Rather, they demonstrate that cer-ebellar LTD is also involved in processing self-

motion cues. The cerebellum may thereforecontribute to two major circuits crucial for the

representation of space in the hippocampal sys-tem. The first is the retrosplenial cortex, which

is closely associated with vestibular function (27).The second is the parietal cortex, which integrates

self-motion and external information and receives

input from the deep cerebellar nuclei (28, 29).Our study demonstrates the crucial role of PKC-

dependent cerebellar LTD in the preprocessing ofself-motion information required for optimal hip-

pocampal representation. This process appears tobe essential for path integration.

References and Notes1. H. Eichenbaum, Curr. Biol. 10, R785 (2000).

2. J. O’Keefe, L. Nadel, The Hippocampus as a Cognitive

Map (Clarendon, Oxford, 1978).

3. P. J. Best, A. M. White, A. Minai, Annu. Rev. Neurosci. 24,

459 (2001).

4. K. J. Jeffery, Hippocampus 17, 775 (2007).

5. T. Hafting, M. Fyhn, S. Molden, M. B. Moser, E. I. Moser,

Nature 436, 801 (2005).

6. B. L. McNaughton, F. P. Battaglia, O. Jensen, E. I. Moser,

M. B. Moser, Nat. Rev. Neurosci. 7, 663 (2006).

7. F. Sargolini et al., Science 312, 758 (2006).

8. T. A. Yakusheva et al., Neuron 54, 973 (2007).

9. D. E. Angelaki, T. A. Yakusheva, A. M. Green,

J. D. Dickman, P. M. Blazquez, Cerebellum 9, 174 (2010).

10. L. C. Hoffmann, S. D. Berry, Proc. Natl. Acad. Sci. U.S.A.

106, 21371 (2009).

11. J. Wikgren, M. S. Nokia, M. Penttonen, Neuroscience

165, 1538 (2010).

12. C. I. De Zeeuw et al., Neuron 20, 495 (1998).

13. J. S. Albus, Math. Biosci. 10, 25 (1971).

14. D. Marr, J. Physiol. 202, 437 (1969).

15. M. Ito, M. Kano, Neurosci. Lett. 33, 253 (1982).

16. E. Burguière et al., Nat. Neurosci. 8, 1292 (2005).

17. See supporting material on Science Online.

18. A. Rotenberg, R. U. Muller, Philos. Trans. R. Soc.

London Ser. B 352, 1505 (1997).

19. L. Rondi-Reig et al., Neuroscience 104, 207 (2001).

20. E. Burguière, A. Arabo, F. Jarlier, C. I. De Zeeuw,

L. Rondi-Reig, J. Neurosci. 30, 13265 (2010).

21. S. Pasalar, A. V. Roitman, W. K. Durfee, T. J. Ebner,

Nat. Neurosci. 9, 1404 (2006).

22. C. C. Bell, V. Han, N. B. Sawtell, Annu. Rev. Neurosci. 31,

1 (2008).

23. P. Dean, J. Porrill, C. F. Ekerot, H. Jörntell,

Nat. Rev. Neurosci. 11, 30 (2010).

24. J. X. Brooks, K. E. Cullen, J. Neurosci. 29, 10499

(2009).

25. Y. Zheng, M. Goddard, C. L. Darlington, P. F. Smith,

Hippocampus 19, 480 (2009).

26. M. Goddard, Y. Zheng, C. L. Darlington, P. F. Smith,

Behav. Neurosci. 122, 448 (2008).

27. R. W. Stackman, A. S. Clark, J. S. Taube, Hippocampus

12, 291 (2002).

28. S. Giannetti, M. Molinari, Brain Res. Bull. 58, 481

(2002).

29. D. M. Clower, R. A. West, J. C. Lynch, P. L. Strick,

J. Neurosci. 21, 6283 (2001).

Acknowledgments: We thank C. De Zeeuw for providing

L7PKCI mice; S. Quet and F. Jarlier for technical

supports; C. Lamouroux for animal care; G. Dallérac

and K. Benchenane for helpful comments on the

manuscript and programming; and C. Léna for

scientific insights. Supported by grants from Agence

Nationale de la Recherche (ANR Young Researcher

07-JCJC-0108-01 and ANR-09-EMER-005-02)

(L.R.R.) and by Action Concertée Incitative (ACI) grant

NIC0027 (B.P. and E.S.).

Supporting Online Materialwww.sciencemag.org/cgi/content/full/334/6054/385/DC1

Materials and Methods

Figs. S1 to S5

Tables S1 to S3

References

22 April 2011; accepted 13 September 2011

10.1126/science.1207403

Activity-Dependent Long-TermDepression of Electrical SynapsesJulie S. Haas,1,2* Baltazar Zavala,2 Carole E. Landisman1,2*

Use-dependent forms of synaptic plasticity have been extensively characterized at chemicalsynapses, but a relationship between natural activity and strength at electrical synapses remainselusive. The thalamic reticular nucleus (TRN), a brain area rich in gap-junctional (electrical)synapses, regulates cortical attention to the sensory surround and participates in shifts betweenarousal states; plasticity of electrical synapses may be a key mechanism underlying these processes.We observed long-term depression resulting from coordinated burst firing in pairs of coupledTRN neurons. Changes in gap-junctional communication were asymmetrical, indicating thatregulation of connectivity depends on the direction of use. Modification of electrical synapsesresulting from activity in coupled neurons is likely to be a widespread and powerful mechanismfor dynamic reorganization of electrically coupled neuronal networks.

The thalamic reticular nucleus (TRN) is

a shell comprising a homogenous pop-ulation of parvalbumin (PV)–positive g-

aminobutyric acid (GABA)–releasing (GABAergic)neurons surrounding the dorsal thalamus (1, 2).

These cells provide powerful inhibition to thala-mocortical relay neurons (3) upon integration of

their corticothalamic and thalamocortical inputs.In addition to its proposed role in focusing the

neural spotlight of attention (4, 5), the TRN is

strongly involved in regulating states of arousal

(6, 7) by means of alternation between burstand tonic modes of firing. Burst firing in the

TRN is a prominent component of sleep spin-dles (8, 9) and absence seizures (9, 10), both of

which aremarked by dramatic changes in corticalattention and behavioral responsiveness to sen-

sory input.In central mammalian neurons, electrical

(gap-junctional) synapses appear all over thebrain (11, 12) and mainly couple GABAergic

neurons of similar subtype (13–15). Electricalsynapses contribute to synchrony in coupled net-

works (11, 16–21), although computational studiessuggest that the precise role of gap junctions in

synchrony can be complex (22–24).Cells in the TRN are densely and powerfully

connected by electrical synapses (17, 18) that

persist into adulthood (25) and, as in other areas,

participate in its synchronous activity (18). Theexperimentally isolated TRN generates spindle

rhythms in the absence of other inputs (26), sug-

gesting that electrical synapses are likely to bekey players in TRN synchrony and in behavioral

switching between firing states.Activity-dependent forms of plasticity have

been extensively described at excitatory (gluta-matergic) chemical synapses (27, 28) and, to a

lesser extent, at inhibitory (GABAergic) chemi-cal synapses (29–31). Although the issue has

received far less attention than plasticity of chem-ical synapses, modifications of electrical synapses

have been documented in a handful of reports(32, 33). Because electrical synapses are likely to

play a major role in coordinating TRN activity,we sought to investigate the effects of natural

formsof activity in coupled neurons on the strengthof the electrical synapses between them.

We recorded from pairs of gap junction–coupled TRN neurons (Fig. 1A) within con-

ventional thalamocortical brain slices (34). Tomeasure electrical synaptic strength, we delivered

hyperpolarizing current injections into one neu-ron (cell 1) while recording voltage (V ) responses

in both neurons, which were maintained at a base-line Vm = –65 mV (Fig. 1B). Using these deflec-

tions, we determined the coupling coefficient cc12 =DVcell 2/DVcell 1, and from injecting current into cell

2, similarly determined cc21 = DVcell 1/DVcell 2.Wealso calculated coupling conductance GC (34) in

each direction. From a total of 313 paired record-ings of coupled TRN neurons, we found an av-

erage cc of 0.12 T 0.08 andGC of 0.80 T 0.63 nS(mean T SD) (Fig. 1C), which is in line with the

values for previous reports in TRN (17, 18, 33)

1Children’s Hospital, Department of Neurology, Harvard Uni-versity, 300 Longwood Avenue, Boston, MA 02115, USA.2Center for Brain Science, Harvard University, 52 Oxford Street,Cambridge, MA 02138, USA.

*To whom correspondence should be addressed. E-mail:julie.haas@gmail.com (J.S.H.); carole.landisman@hms.harvard.edu (C.E.L.)

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 389

REPORTS

and of similar size to cc values reported in thecortex and other areas (35–37). Other parameters,

such as intraneuronal distance and probability ofcoupling, were similar to previous reports (18).

Asymmetry of electrical synapses has beenobserved experimentally (36, 37). In our data,

coupling was rarely symmetrical, which is shownby the spread of values when plotting cc21 against

cc12 for each pair (Fig. 1D). We quantified asym-metry by the ratio of directional ccs (cc21/cc12)

for each pair; for all pairs, the mean ratio of ccswas 1.6 T 0.6 (n = 313 pairs) (Fig. 1E). Some of

Fig. 1. (A) Magnification 60× infrared image from patchrecordings of a coupled pair of TRN neurons. (B) Currentinjection into one cell (I1) of a coupled pair drives a directresponse in that cell (V1) and a gap junction–relayedresponse in the second cell (V2); cc12 = DV2/DV1. Scale bars,5 mV, 0.1 s. (C) Mean electrical synaptic conductance (GC)plotted against mean cc (dots). Open circles are binnedaverages, with a slope of 7.9 [bin width, 0.02; coefficient ofdetermination (r2) = 0.77]. (D) Directional cc (purple, scaledby 10) and GC (orange) for each pair; 1→2 representscoupling measured by current injection into cell 1, as in (B).(E) Coupling asymmetry was quantified by distribution ofratios (cc12/cc21 and G12/G21, larger value/smaller; binwidth, 0.05). (F) Spikes driven by current injection into onecell (gray) caused spikes in the unstimulated coupled cell(black), as shown for three pairs with cc between 0.2 and 0.4maintained at baseline Vm≈–65mV. Scale bar, 25mV, 0.1 s.(G) Wide-field image of TRN cells loaded with OGB-Bapta1AM (Invitrogen, Carlsbad, California). (H) Stimulation of apatched cell (gray) drove bursting and strong calcium re-sponses in that cell and in several neighboring cells (scalebars, 1% DF/F, 50 ms and 25 mV, 50 ms for bottom trace).Traces are from the cells labeled by color and number in (G).

I1

V2

V1

0 2 40

2

4

1→2

2→1

10∗cc G

C0 0.15 0.3

0

1

2

3

mean cc

mea

n G

C (

nS)

1 2 30

0.1

0.2

ratio

dens

ity

ccG

C

A B C D E

F

G H

1

2

3

4

5

6

7

8

Fig. 2. (A) Paired bursting driven by simultaneouscurrent injections into both cells of coupled pairs.Scale bars, 20 mV, 50 ms. (Inset) Close-up of pairedburst event. (B) Mean cc and GC before and afterpaired bursting (gray bar). (C) Average normalizedinput resistance (Rin) and membrane potential (Vm) forthe neurons summarized in (B). (D) Example pairedresponses before and after activity pairing as in (A).Scale bars, 100 ms, 2.5 mV (coupled response, inblack), 5 mV (direct response, in gray). (E) Burstingdriven by current injections into one cell of a coupledpair (gray trace) while the other neuron was quiescent(black trace). Scale bars, 20 mV, 50 ms. (Inset) Close-up of burst in cell 1 and burstlet in cell 2. (F) Mean ccand GC before and after single-cell bursting (gray bar).(G) Average normalized input resistance (Rin) andmembrane potential (Vm) for the neurons summarizedin (F). (H) Example paired responses before and afteractivity pairing as in (E). Scale bars, 100 ms, 2.5 mV(coupled response, in black), 5 mV (direct response,in gray).

−10 0 10 20

0.8

1

1.2

Elapsed Time (min)

Rin

, Vm

(no

rm)

−10 0 10 20

0.8

0.9

1

1.1

GC, c

c (n

orm

)

A

B

C

D

−10 0 10 20

0.8

1

1.2

Elapsed Time (min)

Rin

, Vm

(no

rm)

−10 0 10 20

0.8

0.9

1

1.1

GC, c

c (n

orm

)

E

F

G

H

control post−activity

control post−activity

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org390

REPORTS

the observed asymmetry in cc is due to mis-matches in input resistance; however, ratios of

directionally measured GC (G21/G12), which areindependent of input resistance (34), had a mean

of 1.2 T 0.27 (Fig. 1E).Like many thalamic neurons, TRN neurons

spike in two modes: conventional fast sodium-based tonic spikes and slower low-threshold cal-

cium spikes (LTS), known as bursts, that arecrowned by a barrage of fast sodium spikes. In

many pairs in the current study, bursts elicitedby positive current injection into one neuron were

sufficient to drive bursts in its coupled neighbor(Fig. 1F). Imaging experiments revealed that

bursting activity driven in a patched cell propa-

gated through a network of coupled cells (Fig. 1,G and H).

To determine the effects of bursting in cou-pled cells on electrical synaptic strength, we

tested coupling strength before and after 5 min ofsynchronous evoked bursting in pairs of coupled

neurons. Bursting was driven by simultaneous cur-rent injections of 100 to 300 pA for 50ms at 2 Hz

through the recording electrodes of both neu-rons, which weremaintained at membrane poten-

tials between –65 and –70 mV by means ofsteady-state current injection (Fig. 2A). After paired

bursting, ccwas reduced by 12.0 T 3.6%, andGC

was depressed by 13.2 T 1.8% (P < 0.05, two-

tailed unpaired t test, n = 7 pairs) (Fig. 2B). Thislong-term depression (LTD) persisted for the

length of recordings (for at least 30 min afterpaired bursting), with no apparent signs of di-

minishing (Fig. 2B). There were no significantchanges in input resistance or membrane resting

potential (Fig. 2C), ruling out the possibility thatthe observed changes in electrical synaptic strength

reflected changes in the intrinsic properties of theneurons at the whole-cell level. Although path-

ological changes in internal calcium concentra-tion are known to affect gap-junctional strength

(32, 38), our estimates of the calcium influx fromthe slow rate of bursting used here are much

smaller. Bursting rates in vivo are often faster

than those used here (19, 26).To determine whether bursting in one neuron

alone is sufficient to induce LTD, we repeated theactivity paradigm, this time only stimulating burst-

ing in a single neuron of a pair (Fig. 2E) whileholding the coupled cell at ~–70 mV so as to

prevent it from bursting. After single-cell bursting,cc was reduced by 15.0 T 3.4%, and GC was re-

duced by 13.0 T 2.3% (P < 0.05, n = 11 pairs)(Fig. 2F). The magnitude of LTD was not signif-

icantly different for the single-cell burst paradigmfrom the paired-bursting paradigm (unpaired t test).

To determine the contribution of sodiumspikes to LTD, we repeated the bursting para-

digm in both cells using a bath application of1 mM tetrodotoxin (TTX), which completely and

reversibly blocks the quick barrage of sodium-mediated action potentials crowning the calcium-

mediated bursts (Fig. 3A). After paired burstingin TTX, cc decreased by 12.3 T 3.2%, and GC

decreased by 11.7 T 2.6% (P < 0.05, n = 9 pairs)(Fig. 3B). We also repeated the bursting para-

digm in one cell alone in TTX (Fig. 3D). Aftersingle-cell activity in TTX, cc decreased by 6.5 T

2.3%, and GC was reduced by 6.0 T 2.0% (P <0.05, n = 11 pairs) (Fig. 3E).When depolarized to

rest just below spiking threshold (~–40 mV) andstimulated to spike with 50-ms pulses repeated at

2 Hz, in order to emulate spiking during bursting

without activating the LTS, coupling decreasedby a smaller and delayed amount (DGC = –7.2 T

2.0%, Dcc = –7.0 T 2.8%; P = 0.03, n = 8 pairs;spike frequency during this paradigmwas twice as

slow as during LTS bursts) (fig. S2). Of theseactivity paradigms, the amount of depression

from single-cell bursting in TTXwas significantlysmaller than others [P < 0.05, analysis of variance

(ANOVA)] (Fig. 3H).Activity paradigms in which only one cell

was active allowed us to characterize the timecourse of changes in electrical synaptic strength

by measuring the amplitude of the postsynapticburstlet in the coupled cell during the 5 min of

Fig. 3. (A) Paired bursting driven by simultaneouscurrent injections into both cells of coupled pairs, in thepresence of 1 mM TTX. Scale bars, 10 mV, 50 ms. (Inset)Close-up of paired burst events. (B) Mean cc and GC

before and after paired bursting in TTX (gray bar). (C)Average normalized input resistance (Rin) and mem-brane potential (Vm) for the neurons summarized in (B).(D) Bursting driven by injections of current into one cellof a coupled pair (gray trace) while the other neuronwas quiescent (black trace), also in TTX. Scale bars, 10mV, 50 ms. (Inset) Close-up of burst event and burstlet.(E) Mean cc andGC before and after single-cell bursting inTTX (gray bar). (F) Average normalized input resistance(Rin) and membrane potential (Vm) for the neuronssummarized in (E). (G) Burstlet amplitudes (from Fig.2E) during single-cell activity plotted against elapsed timeand normalized to final values. (H) Summary of changesinGC for the four paradigms: paired bursting (2B), single-cell bursting (1B), paired bursting in TTX (2B + T), andsingle-cell bursting in TTX (1B + T). Asterisk indicatessignificance (P < 0.05, ANOVA).

−10 0 10 20

0.8

1

1.2

Elapsed Time (min)

Rin

, Vm

(no

rm)

−10 0 10 20

0.8

0.9

1

1.1

GC, c

c (n

orm

)

A

B

C

G

−10 0 10 20

0.8

1

1.2

Elapsed Time (min)

Rin

, Vm

(no

rm)

−10 0 10 20

0.8

0.9

1

1.1

GC, c

c (n

orm

)

D

E

F

H

0 1 2 3 4

1

1.1

1.2

Elapsed time during pairing (min)

Bur

stle

tA

mpl

itude

(nor

m)

2B 1B 2B+T 1B+T−20

−10

0

Δ G

C (

%)

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 391

REPORTS

activity. For both single-cell bursting and single-cell bursting in TTX, changes in synaptic strength

(burstlet amplitude) reached their steady-statereduced values within 2 min of activity (Fig. 3G).

In two of our activity paradigms, the activityof the coupled pair, and thus the use of the syn-

apse, was also asymmetrical (Figs. 2E and 3D)—that is, one neuron was active while the other was

quiescent, resulting in largely unidirectional cur-rent flow across the gap junction channels during

activity. These asymmetrical stimuli allowed us toinvestigate whether the LTD was also expressed

asymmetrically. First, we quantified the effects ofactivity on each direction of coupling, with re-

spect to the active cell. Coupling measured with

current injection into cell 1 (the active cell duringpairing), or outbound coupling, we denote as cc12,

whereas couplingmeasured with current injectioninto the quiet cell 2 and relayed by the gap

junction back to the active cell 1, or inboundcoupling, is cc21 (Fig. 4A). For full bursting in

one neuron (Fig. 4B), the inbound coupling cc21

decreased by 16.0 T 3.4%, whereas outbound

coupling, cc12, decreased by 8.6 T 3.7% (P < 0.05for both directions; two-tailed, paired t test, n = 11

pairs) (Fig. 4, C and D). The change in cc21 wassignificantly larger than in cc12 (P < 0.05). Di-

rectional conductances decreased similarly; G21

decreased by 10.8 T 3.2%, andG12 decreased by6.8 T 3.2% (P < 0.05). For single-cell LTS burst-

ing in TTX (Fig. 4D), inbound coupling, cc21, de-creased by 10.0 T 3.0% (P < 0.05, n = 10 pairs),

whereas the change in outbound coupling, cc12,was not significant (–5.5 T 2.7%,P= 0.07, n = 10

pairs) (Fig. 4G and H). In TTX, outbound G12

decreased by 7.5 T 2.0% (P = 0.04), and G21

decreased by 6.6 T 2.5% (P = 0.09).In principle, asymmetrical use of a gap junc-

tion could potentially act to either decrease, in-crease, or preserve the pre-activity asymmetry of

coupling in any given pair. To examine the sys-tematic effects of unidirectional synapse use on

asymmetry, we plotted the ratios of directional

ccs andGCs (cc21/cc12 andG21/G12) for each pairafter unidirectional activity (Fig. 4, E and I, y

axis) against the initial values (Fig. 4, E and I, xaxis). The identity line corresponds to coupling

asymmetry that was unaffected by asymmetricaluse of the synapse. For full bursts in one cell,

ratios of ccs increased on average by 9.1 T 2.4%after activity (P < 0.01, n = 11 pairs) (Fig. 4E);

this shift represents a systematic trend of greaterchange in the coupling of inbound communica-

tion, cc21, relative to outbound communication,cc12. Ratios of GC also increased, by 5.0 T 2.2%

(P < 0.05). Changes in asymmetry were not due

to coordinated shifts in input resistance; R1/R2

decreased by 3.1 T 2.6% (P = 0.25). For LTS

bursts without sodium spikes in one cell, ratios ofccs and GCs fell along the identity line after ac-

tivity, with an insignificant change in rectificationfrom initial values (ratio of ccs: 6.0 T 4.2%, P =

0.6; ratio ofGCs: –0.6 T 2.0%, P = 0.76; R1/R2:–5.3 T 3.4%, P = 0.07; n = 10 pairs) (Fig. 4I),

indicating that the changes in rectification maybe due to sodium spikes. As expected, ratios of

coupling coefficients also did not change sig-nificantly for symmetrical synaptic use (paired

bursting).Although activity-dependent changes have

been extensively described and characterized at

chemical synapses, long-term modification ofelectrical synapses by precise patterns of activity

of coupled cells themselves has not yet been de-scribed. The changes we measured, ~15%, are

small as compared with some changes measuredat chemical synapses. Neurons receive thousands

of individual chemical synaptic inputs, which areeach very small, often distant from the soma, and

of short, stereotyped time courses. Chemical syn-aptic inputs are orders of magnitude smaller than

are electrical synaptic inputs and typically in-effective as single voices in driving a cell to spike.

The average coupling measured here (cc = 0.12)

Fig. 4. (A) For activity in cell 1, cc12 (blue) represents the “outbound”coupling measured with current injection into cell 1, and cc21 (green) rep-resents “inbound” coupling. (B) Single-cell bursting in cell 1 (gray) withpostsynaptic burstlets in cell 2. Scale bars, 15 mV, 25 ms. (C) Inbound cc21before and after full bursts in cell 1. (D) Outbound cc12 before and after fullbursts in cell 1. (E) Ratios of directional cc [black solid circles; division of thechanges in (C) divided by the changes in (D) for each pair] and GC (opencircles, P < 0.05 for both cc and GC) after full bursts in cell 1, plotted against

initial values. (F) Bursts in cell 1 (gray) in 1 mM TTX. Scale bars, 10 mV, 25 ms.(G) Inbound cc21 before and after bursts in cell 1 in TTX. (H) Outbound cc12before and after bursts in cell 1 in TTX. (I) Ratios of directional cc (red solidsquares; P = 0.6) and GC (open squares; P = 0.76) after bursts in cell 1 in TTX,plotted against initial values. (J ) Model of an asymmetrical gap junction astwo parallel branches. RC represents the minimum conductance (maximumresistance) common to both sides of the gap junction, and RD representsadditional, asymmetrical conductance in one direction.

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org392

REPORTS

applied to an average presynaptic burst (~50mV)yields a ~6-mV burstlet in a coupled cell, which

persists for the entire ~50 ms of the burst (Figs.1F, 2E, and 3D); a single burstlet is often large

enough to drive bursts directly in a coupled neigh-bor (Fig. 1). A reduction by <15% is considerable

for these already strong synapses and is sufficientto prevent a cell’s burstlet from driving its neigh-

bor to burst (fig. S1).By preferentially diminishing coupling inbound

to bursting cells, activity-dependent LTD couldunplug single bursting cells from an overly active

or synchronous neighbor or network or adjustinput preference between intra-TRN electrical

input and input from corticothalamic or thalamo-

cortical fibers. The effects of activity-dependentchanges may be more complex in vivo because

of multiple electrical synapses and/or recurrentsynapses between neurons.

What cellular processes might underlie the ob-served LTD of electrical synapses? Gap junctions

are plaques comprising hundreds to thousandsof individual channels. Insertion and deletion of

gap junction channels is a normal component ofcellular function and a candidate mechanism for

changing synaptic strength. In addition, con-nexin36 (Cx36) proteins have multiple phos-

phorylation sites (39, 40). Phosphorylation-relatedchanges in coupling mediated by either protein

kinaseA (41, 42) or CamKII (40) aswell as hemi-channel conductance changes at Cx35 channels

(43) have been described.Our experiments indicate that electrical syn-

aptic strength is asymmetrical at baseline and isfurther adjustable in a use-directional manner. Re-

sults in mice in which Cx36 has been knockedout indicate that synapses composed of non-

Cx36 proteins are more asymmetrical than thosein wild-type (44); thus, one possible source of

asymmetry is inclusion of non-Cx36 proteinsand/or pores at the synapse. Our results further

indicate that coupling asymmetry can be shiftedby activity; neurons can fine-tune the relative

proportion of signals they send or receive to orfrom coupled neighbors, respectively. Increased

expression or activation of non-Cx36 proteinscould account for this increase in asymmetry.

Despite evidence of gap-junctional rectifica-tion in mammalian systems, the canonical sym-

bol for those electrical synapses has remained thesimple linear resistor (RC). Our observations of

baseline asymmetry and activity-dependent shiftsin asymmetry (Fig. 4E) led us to reconsider the

standard model because a linear resistor cannotaccount for asymmetry or increases in asym-

metry. Diodes have been used to model heavilyrectifying invertebrate gap junctions (45) but have

not yet been considered for mammalian gap-junctional synapses. We suggest a model of a

mammalian gap-junctional synapse as two branchesin parallel (Fig. 4J): One branch carries the com-

mon resistance (RC) or the maximum resistance

(minimum of conductance GC) measured fromboth directions. A parallel branch consists of a

resistor (RD) in series with a diode, representingthe increase in conductance (or decreased re-

sistance) observed as asymmetry.

References and Notes1. C. R. Houser, J. E. Vaughn, R. P. Barber, E. Roberts,

Brain Res. 200, 341 (1980).

2. P. T. Ohara, A. R. Lieberman, J. Neurocytol. 14, 365 (1985).

3. D. Pinault, M. Deschênes, J. Comp. Neurol. 391, 180

(1998).

4. F. Crick, Proc. Natl. Acad. Sci. U.S.A. 81, 4586 (1984).

5. K. McAlonan, J. Cavanaugh, R. H. Wurtz, J. Neurosci. 26,

4444 (2006).

6. D. L. Rowe et al., Int. J. Neurosci. 115, 1273 (2005).

7. M. Steriade, D. A. McCormick, T. J. Sejnowski, Science

262, 679 (1993).

8. M. Deschênes, M. Paradis, J. P. Roy, M. Steriade,

J. Neurophysiol. 51, 1196 (1984).

9. M. Steriade, Trends Neurosci. 28, 317 (2005).

10. M. Inoue, J. Duysens, J. M. Vossen, A. M. Coenen,

Brain Res. 612, 35 (1993).

11. M. V. Bennett, R. S. Zukin, Neuron 41, 495 (2004).

12. B. W. Connors, M. A. Long, Annu. Rev. Neurosci. 27,

393 (2004).

13. M. Galarreta, S. Hestrin, Nat. Rev. Neurosci. 2, 425

(2001).

14. J. R. Gibson, M. Beierlein, B. W. Connors, Nature 402,

75 (1999).

15. Z. Chu, M. Galarreta, S. Hestrin, J. Neurosci. 23, 96

(2003).

16. M. Beierlein, J. R. Gibson, B. W. Connors, Nat. Neurosci.

3, 904 (2000).

17. C. E. Landisman et al., J. Neurosci. 22, 1002 (2002).

18. M. A. Long, C. E. Landisman, B. W. Connors, J. Neurosci.

24, 341 (2004).

19. P. Fuentealba, M. Steriade, Prog. Neurobiol. 75, 125

(2005).

20. M. Blatow et al., Neuron 38, 805 (2003).

21. G. Tamás, E. H. Buhl, A. Lörincz, P. Somogyi,

Nat. Neurosci. 3, 366 (2000).

22. K. Vervaeke et al., Neuron 67, 435 (2010).

23. C. C. Chow, N. Kopell, Neural Comput. 12, 1643

(2000).

24. T. J. Lewis, J. Rinzel, J. Comput. Neurosci. 14, 283

(2003).

25. K. L. Blethyn, S. W. Hughes, V. Crunelli,

Thalamus Relat. Syst. 4, 13 (2008).

26. M. Steriade, L. Domich, G. Oakson, M. Deschênes,

J. Neurophysiol. 57, 260 (1987).

27. R. C. Malenka, M. F. Bear, Neuron 44, 5 (2004).

28. R. C. Froemke, D. Debanne, G.-Q. Bi, Front. Syn. Neurosci.

2, 19 (2010).

29. K. P. Lamsa, D. M. Kullmann, M. A. Woodin,

Front. Syn. Neurosci. 2, 8 (2010).

30. F. S. Nugent, J. A. Kauer, J. Physiol. 586, 1487 (2008).

31. J. S. Haas, T. Nowotny, H. D. Abarbanel, J. Neurophysiol.

96, 3305 (2006).

32. A. E. Pereda, D. S. Faber, J. Neurosci. 16, 983 (1996).

33. C. E. Landisman, B. W. Connors, Science 310, 1809

(2005).

34. Materials and methods are available as supporting

material on Science Online.

35. J. R. Gibson, M. Beierlein, B. W. Connors, J. Neurophysiol.

93, 467 (2005).

36. A. Devor, Y. Yarom, J. Neurophysiol. 87, 3048 (2002).

37. P. Mann-Metzer, Y. Yarom, J. Neurosci. 19, 3298

(1999).

38. B. Rörig, B. Sutor, Mol. Neurobiol. 12, 225 (1996).

39. W. W. Kothmann, X. Li, G. S. Burr, J. O’Brien,

Vis. Neurosci. 24, 363 (2007).

40. C. Alev et al., Proc. Natl. Acad. Sci. U.S.A. 105, 20964

(2008).

41. S. Urschel et al., J. Biol. Chem. 281, 33163 (2006).

42. W. W. Kothmann, S. C. Massey, J. O’Brien, J. Neurosci.

29, 14903 (2009).

43. G. Mitropoulou, R. Bruzzone, J. Neurosci. Res. 72, 147

(2003).

44. T. Zolnik, B. W. Connors, presented at the Society for

Neuroscience Conference, Prog. 42.14. San Diego, CA, 2010.

45. E. J. Furshpan, D. D. Potter, J. Physiol. 145, 289

(1959).

Acknowledgments: We thank D. Feldman and K. Bender

for valuable feedback on previous version of the

manuscript. This work was supported by the Milton

Fund.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/334/6054/389/DC1

Materials and Methods

Figs. S1 and S2

26 April 2011; accepted 16 August 2011

10.1126/science.1207502

www.sciencemag.org SCIENCE VOL 334 21 OCTOBER 2011 393

REPORTS

Weekly Cell Signaling Journal fromScience

ScienceSignaling

As a AAAS member, addScience Signaling access for over60% off regular price. Subscribe in any of these ways:• Go to sciencesignaling.org• Call +1 202-326-6417

• Mail or fax this form with your payment to +1 202-842-1065

1200 New York Ave., NW • Washington, DC 20005 • +1 202-326-6417 • sciencesignaling.org

� AAAS member price – US$52 online only;

US$225 print and online (add US$100 for non-US delivery)

AAAS membership number required ______________________________

� Nonmember price – US$156 online only;US$465 print and online (add US$100 for non-US delivery)

Name ___________________________________________

Address _________________________________________

_________________________________________________

City _____________________________________________

State/Province____________________________________

Zip/Postal Code ___________________________________

Country __________________________________________

E-mail____________________________________________

Phone____________________________________________

Online resources on cell signaling

Payment

� Check (payable to AAAS – Science Signaling)Mail check and this form to:AAASAttn:Membership Department1200NewYork Avenue, NWWashington, DC 20005 USA

� Chargemy:

� VISA � MasterCard � American Express

Card Number_________________________________________

Expiration Date_______________________________________

Signature____________________________________________

Date ________________________________________________

If paying by credit card, youmay FAST FAX your order to+1 202-842-1065

Subscribe now for 1 year’s access toScience Signaling

(required for subscription activation)

Gain insight into the fast-paced research in cell signaling withoriginal reviews and perspectives by leading researchers.

Stay abreast of the latest developments with summaries oftheweek’s hottest research and the nightly updatedVirtualJournal, with primary research articles from 49 publishers.

Give your own research a boost with detailed protocols thatguide you through the latest techniques.

Learn about the relationships controlling cell behavior fromthe ConnectionsMaps pathways, a graphical interface into theDatabase of Cell Signaling, with information provided by cellsignaling experts.

Stay ahead in this rapidly advancing, multidisciplinary fieldwith custom alerts and personalization tools.

Sitewide access available for your institutiontoday. Contact sciencesignaling@aaas.org

or call +1 866-265-4152

ScienceSignaling.org

Now availablein print edition!

Prices valid until December 31, 2011.

NNoiinn

395www.sciencemag.org/products SCIENCE VOL 334 21 OCTOBER 2011

Newly offered instrumentation, apparatus, and laboratory materials of interest to researchers in all disciplines in academic, industrial, and governmental organizations are

featured in this space. Emphasis is given to purpose, chief characteristics, and availability of products and materials. Endorsement by Science or AAAS of any products or

materials mentioned is not implied. Additional information may be obtained from the manufacturer or supplier.

Electronically submit your new product description or product literature information! Go to www.sciencemag.org/products/newproducts.dtl for more information.

LIFE SCIENCE TECHNOLOGIES

NEW PRODUCTS: CELL/TISSUE CULTURE

Produced by the Science/AAAS Custom Publishing Office

AUTOMATED CELL CULTURE SYSTEM The CompacT SC is an automated T-fl ask cell culture system designed

to fully control expansion, differentiation, and maintenance of stem

cells. The new CompacT SC, which can be used to aseptically culture

multiple different stem cell lines in parallel using T75 and T175 fl asks,

has a number of features that makes controlling the growth and

differentiation of stem cells an easy, convenient task. The system

includes automated cell imaging and analysis, allowing scientists to

program the CompacT SC to monitor cell morphology and growth

curves. It also permits researchers to set the system to respond to

any pre-programmed changes by automatically passaging cells and

adding new media, reagents, or vectors at appropriate times. This

means scientists can consistently maintain cell lines in the state they

need them, either as pluripotent cells or during cell differentiation

and transfection. Processes can be performed at any time of the day

or night, without researchers having to be in the laboratory.

TAP Biosystems

For info: +44-(0)-1763 227200 www.tapbiosystems.com

MANUAL COLONY COUNTER

The aCOLade is an affordable, manual colony counter with automatic

result recording, ideal for use in any microbiology laboratory. The

new aCOLade colony counter consists of a compact white light

box with a pressure controlled back plate and a pen attached to an

integrated digital counter. Operating aCOLade is quick and simple

because as soon as microbiologists touch the Petri dish’s lid with

the pen to count a colony, a bleeping sound occurs and the count is

automatically added and displayed on the digital counter. To record

results, aCOLade can be connected to any computer and comes

with software, which allows scientists to automatically transfer their

counts to a spreadsheet. This fl exible system has a well-designed

adjustable plate ring that is used to hold plates as small as 50 mm

diameter and can be easily altered to accommodate 90 mm and

120 mm diameter plates, making it suitable for a range of counting

applications.

Synbiosis

For info: +44-(0)-1223-727125 www.synbiosis.com

ENZYME BLOCKING SOLUTION

In immunohistochemical staining, background staining can often

be traced to the existence of active endogenous enzyme, such as

peroxidase and alkaline phosphatase, in the tissue preparation. This

type of interference can generate false positive results or obscure

the specifi c staining of the target antigen. To assist the pathologist,

histologists, and other research scientists in obtaining meaningful

immunohistochemical data, BLOXALL Blocking Solution has been

designed as a dual blocking solution for endogenous alkaline phos-

phatase and peroxidase in tissues. BLOXALL inactivates all endog-

enous peroxidase, pseudoperoxidase, and alkaline phosphatase in

formalin-fi xed, paraffi n-embedded tissue sections, frozen tissue

sections, and cell preparations. BLOXALL Blocking Solution is a

simple, single step, ready-to-use reagent provided in a convenient

dropper bottle.

Vector Laboratories

For info: 650-697-3600 www.vectorlabs.com

CELL HEALTH/GENE EXPRESSION ASSAY

The ONE-Glo + Tox Assay is designed for improved analysis of re-

porter gene expression in the context of cell health. The assay com-

bines luciferase chemistry with a cell viability marker in a two-step,

addition-only process, to enable measurement of cell viability and

gene expression in a single well of a plate, negating the need to

run parallel assays. By analyzing reporter activity in relation to cell

viability, false results due to toxicity or changes in cell number can

be uncovered. The fi rst part of the assay is a non-lytic fl uorescence

assay (CellTiter-Fluor Cell Viability Assay) that measures the relative

number of the live cells in a culture population after experimental

manipulation. The second part of the assay uses the ONE-Glo Lu-

cifease Assay System to quantify fi refl y luciferase reporter gene ex-

pression from cells made to express this reporter enzyme. Ideally

suited for high and ultrahigh throughput applications, the volumes of

each assay component can be scaled to meet throughput needs and

is amenable to automation up to 1,536-well format.

Promega

For info: 800-356-9526 www.promega.com

INCUBATORS

The CB series incubators provide excellent growth conditions for mammalian

cell and tissue cultures, offering precise, reproducible control of oxygen

concentrations as low as 0.2% and quick recovery times, important factors

in conducting research under hypoxic conditions. Models CB 53, 150, and

210 allow sample access via optional small internal doors that, in addition to

maintaining steady CO2 and O

2 levels and accelerating recovery, minimize

the risk of contamination. The incubators feature as standard a 180°C hot-air

cycle that provides true sterilization. Interior walls stay dry through controlled

humidifi cation with a defi ned condensation point, and chamber preheating

with the VENTAIR air jacket provides homogeneous temperature distribution.

A seamless, deep-drawn interior features an integrated shelf-support system

for easy cleaning. Single-beam infrared sensor measures and controls CO2

in real-time, independent of other gasses and humidity, and provides stable

pH values for optimum, uniform cell growth throughout the entire chamber.

Binder

For info: 866-885-9794 www.binder-world.com

POSITIONS OPEN

TENURE TRACK FACULTY POSITIONS inStructural Biology and Dynamics with a focus on

Membrane Proteins

The Departments of Biochemistry & Molecular Bi-ology (website: http://www.bmb.msu.edu) andChemistry (website: http://www.chemistry.msu.edu) at Michigan State University (MSU) seek out-standing candidates for two tenure-track ASSIST-ANT PROFESSORSHIPS in the fields of structuralbiology and dynamics with an emphasis on mem-brane proteins and biomembranes. Successful candi-dates will develop a vigorous, externally funded researchprogram using and/or developing modern molecu-lar biophysical approaches to examine the structureand dynamics of membrane proteins involved in im-portant biological functions including but not limitedto, host pathogen interactions, signaling, ion or mo-lecular transport, bioenergetics, and membrane bio-genesis. These individuals will have the opportunity touse state-of-the-art facilities, which include a 900 MHzNMR, mass spectrometry facility, and dedicated syn-chrotron beamlines at Argonne National Laboratory.There are also opportunities to interact with a diversegroup of MSU faculty working in structural biologyand dynamics. In addition, the candidates will contrib-ute to teaching undergraduate, graduate, and/or med-ical students in their home department.

Review of application materials will begin onNovember 15, 2011 and will continue until suitablecandidates are identified. The Departments are highlycommitted to a diverse faculty and encourage applica-tions from individuals in groups that have beentraditionally underrepresented in science faculties in-cluding women, persons of color, veterans, and personswith disabilities. Application materials should includea single file containing a cover letter, curriculum vitae,statement of research interests and future directions,and the names and e-mail addresses of three refer-ences. All application materials should be uploadedelectronically at website: http://www.jobs.msu.edu for Position #5297. Questions regarding thisposition may be directed to Dr. Shelagh Ferguson-Miller (e-mail: fergus20@msu.edu) in Biochem-istry & Molecular Biology and/or Dr. David Weliky(e-mail: weliky@chemistry.msu.edu) in Chemistry.

ASSISTANT OR ASSOCIATE PROFESSORUCSD Department of Neurosciences

Website: http://neurosciences.ucsd.edu/

Faculty Position for Neuroscientist who studiesgenomic or epigenomic control of basic neural pro-cesses, and/or the influence of genomic variationon disease mechanism or therapy. Candidate shouldbe familiar with neuroscience, and have significantexpertise with bioinformatics tools, including high-resolution analyses of genomic variation and expres-sion. Ability to integrate information across platformsis highly desirable, together with willingness to in-teract in a highly collaborative environment while pur-suing independent research interests. Background inbioinformatics is desirable. Assistant or Associate lev-el. The La Jolla neuroscience community is extremelystrong and diverse, providing an outstanding oppor-tunity to develop an independent research program.Position will be within newly established TranslationalNeuroscience Institute in the School of Medicine.

Competitive salary and startup. Commensurate withqualifications and based on University of California payscales. Review of applications will begin November 30,2011 and continue until the position is filled.

Application materials should be submitted via UCSDAP On-Line RECRUIT (website: https://apol-recruit.ucsd.edu/), an electronic job application system. SeePosition Posting 10-165 in RECRUIT. Please beprepared to provide curriculum vitae, a two-pagestatement of research interests, and three letters ofrecommendation.

University of California, San Diego is an Affirmative Action/Equal Opportunity Employer with a strong institutional com-mitment to excellence through diversity.

POSITIONS OPEN

FACULTY POSITIONYale University School of Medicine

Department of NeurobiologyNew Haven, CT 06520-8001

Website: http://medicine.yale.edu/neurobiology/index.aspx

The Department of Neurobiology at YaleUniversity School of Medicine is seeking to hirea scientist using genetic, molecular, or cellularapproaches to examine nervous system develop-ment and function. Although the emphasis willbe placed on recruiting an ASSISTANT PRO-FESSOR, extraordinary applicants at the AS-SOCIATE PROFESSOR or PROFESSORlevels will also be considered. We seek excep-tional candidates with a track record of creativ-ity and productivity, demonstrated potential foroutstanding future achievements, and a desireto participate in a dynamic and growing Neuro-science community at Yale. Candidates are ex-pected to mount a productive and innovativeresearch program, to obtain outside funding, andto participate actively in graduate and medicaleducation.

Candidates must hold a Ph.D., M.D., or equiv-alent degree. Please send curriculum vitae, se-lected reprints, a research plan, and the names ofat least three references. All application mate-rials should be sent electronically to the followinge-mail: neuro.search@yale.edu. Applicationswill be reviewed as they are received, but mustbe received before December 1, 2011.

Yale University is an Affirmative Action/Equal Op-portunity Employer. Yale values diversity in its faculty,students, and staff and especially welcomes applicationsfrom women and underrepresented minorities.

The Department of Biology at the University ofMinnesota Duluth (UMD) invites applications for atenure-track ASSISTANT PROFESSOR position inGenetics, broadly defined, beginning August 2012.We seek a person with research experience in genetics,or closely related field, who will instruct lecture andlaboratory courses in genetics, and develop at least oneadvanced course in their area of specialization. Thearea of specialization is open; applications from can-didates with expertise ranging from population ge-netics to genomics will be equally considered. Thesuccessful candidate will establish an independent, ex-ternally funded research program involving undergrad-uates, and M.S. and Ph.D. students. Service to thedepartment, college, and University is also expected.Opportunities exist for collaboration with researchersat UMD_s Natural Resources Research Institute, LargeLakes Observatory, College of Pharmacy, School ofMedicine, and the EPA Mid-Continent Ecology Di-vision. State-of-the-art research and instruction facil-ities and competitive startup funding are available.Essential qualifications include a Ph.D. or terminal de-gree in the biological sciences, evidence of potentialfor achievement in teaching appropriate for appoint-ment at the Assistant Professor level, peer-reviewedpublications, and strong oral and written communi-cation skills. The University of Minnesota requiresthat you apply online for this position. For a com-plete position description and information on howto apply online, visit website: http://employment.umn.edu/, and search for Job Requisition 174713.Complete applications will be reviewed beginningNovember 22, 2011 and continue until the positionis filled. The University of Minnesota is an Equal OpportunityEducator and Employer.

online

@sciencecareers.org

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencecareers.org396

www.sciencecareers.org 397

UPCOMING FEATURES

Neuroscience: Emerging Fields—November 4

Focus on China—December 9

BS/MS Scientists (online only)—January 13

This summer, the European Commission announced a €7 billion (US$9.6 billion) investment in science, its largest funding package ever for research. This boost in science capital, up 9 percent from last year, is part of the European Union’s (EU) Seventh

Framework Programme for research and technology develop-ment (FP7), and will fund research that will “tackle the biggest societal challenges facing Europe and the world.” It is expect-ed to create approximately 174,000 jobs in the short-term and 450,000 jobs total, and nearly €80 billion (US$113 billion) in gross domestic product (GDP) growth over 15 years.

This is good news for scientists looking for work in Europe. “An integral part of excellence is integration or mobility of researchers from many different nations,” says Michael

Jennings, spokesperson for Máire Geoghegan-Quinn, European Commissioner of Research, Innovation and Science. “We are investing in research and innovation in Europe and creating the conditions and priorities to better attract and retain scientists.”

For academics who desire employment on this diverse and captivating continent, it is important to remember that while many aspects of European science are the same no matter which country you are in (for example, all EU scholars can apply for funds from the European Research Council [ERC]), there are also numerous differences that exist between the members states and their higher education systems. The postdoc appointment is one such illustration: “Although the concept of postdoctoral researchers is well understood,

there is considerable variability in job titles and in the practical organization of this stage in Europe,” according to the 2010 European Science Foundation (ESF) Report, “Research Careers in Europe: Landscape and Horizons” (http://scim.ag/qrFZHZ). But one thing is consistent across borders: Opportunities for employment abound, as long as you know how to navigate the member states’ systems.

GERMANY: STRONG AND OPEN FOR BUSINESSGermany, with a population of approximately 82 million, seems to be faring better than many of its EU neighbors continued »

CR

ED

ITS:

(LE

FT) ©

ISTO

CK

PHO

TO.C

OM

/TR

OTS

CH

E; (

SEIP

KE

) CO

RIN

NE

APP

IA-A

YM

E; (

KO

TAK

OSK

I) E

ER

O H

OLM

STR

ÖM

FOCUS ON CAREERS

FOCUS ON EUROPE

Produced by the Science/AAAS Custom Publishing Office

“You need to have something to offer—such as

collaborations, expertise on topics which would

complement theirs, or teaching experience—in

order to be selected for a permanent position.”

—Jani Kotakoski

Academic Opportunities in European Science

Norwegian University of Science and Technology

Bjørg Elisabeth Kilavik Ryan F. Seipke

Scientists who wish to pursue academic careers in Europe have much with which to contend, especially now amidst an uncertain fi scal landscape. With much of the continent still at risk for a recession, there is legitimate reason to be anxious over the future of science funding in the European Union. However, buttressed by the European Commission, several member states, including Germany, the United Kingdom, and France, as well as Scandinavia, have plans in place to bolster scientifi c research and innovation, and make those regions attractive destinations for early and mid-career professionals looking for academic positions. By Alaina G. Levine

Jani Kotakoski

www.sciencecareers.org398

during this economic tumult. According to the Economist, it is in a “comfortable” fi scal situation, “aided by a strongly growing economy will help to reduce the defi cit without tough public spending cuts or tax rises.” Cathleen Fisher, president of the American Friends of the Alexander Von Humboldt Foundation, which promotes and supports the activities of the parent Hum-boldt Foundation, which provides exchanges between German and American scientists, echoes this sentiment. “Germany is in a relatively good position, having weathered the 2008 fi nan-cial crisis quite well,” she states.

The nation is advancing in the midst of fi nancial fragility for a variety of reasons, says Andreas Pinkwart, former minister for research in North-Rhine-Westphalia, and currently dean of HHL–Leipzig Graduate School of Management. “The German economy has learned to be very export-oriented and global-oriented.” For example, in the last fi ve years, the government developed a ranking system for its universities in an effort to make them more internationally competitive, he adds.

This impetus to improve the higher learning institutes is one slice of a pie composed of three federal initiatives, which will support research and development, student funding, and uni-versity infrastructure. Each enterprise has its own time frame and budget, although the total amount exceeds €20.7 billion (US$28.4 billion), and “guarantees annual budget increases of at least 5 percent for the largest science funding and science performance organizations”, explains Max Vögler, director of the North America Offi ce of the German Research Foundation (DFG, Deutsche Forschungsgemeinschaft).

Jani Kotakoski, currently an adjunct professor of physics at the University of Helsinki, completed his postdoc at Technische Universität Darmstadt. His three-year contract was funded by DFG and focused on materials science and high-pressure phys-ics. “I was interested in a permanent position in Germany,” he says, “but I was extremely unlikely to obtain one with the expe-rience I had at the time.” By that point, Kotakoski explains, he had completed only one postdoc, which is not usually enough experience to be considered a top candidate—at least in a fi eld like materials science where the competition is very tough.

He returned to his native Finland and rejoined the group from

which he had received his Doctorate. Now Kotakoski is on his way to a new position at the University of Vienna, “which is likely to greatly improve my research profi le,” he says. “My view is that after two to three years there, I will be able to get a good position elsewhere, like Austria, Germany, Finland, or maybe other Nordic countries.”

For scientists interested in relocating here, Kotakoski advises interested parties to obtain contacts in advance and pursue a contract position. “One thing to also keep in mind is that the people who make the decision are the other faculty mem-bers,” he affi rms. “Hence, you need to have something to of-fer—such as collaborations, expertise on topics which would complement theirs, or teaching experience—in order to be se-lected for a permanent position.”

With a known defi ciency in tenure-track positions, and a distinctive tradition of requiring academics to complete a second thesis (called Habilitation) to even qualify for tenured employment, “Germany has tried to introduce new paths” to-ward landing these treasured jobs, says Fisher. “The addition of junior professorships as an alternative to the fulfi llment of a Habilitation is one such route, depending on the institution and fi eld.”

Furthermore, it is signifi cant to note that in Germany, the Ph.D. is considered to be an acceptable entry point for industrial jobs. Vögler cites that with 2.4 percent of the German workforce holding Doctorates, compared to an estimated 1.4 percent in the United States, “there is no general expectation to go into academia,” he says. Pinkwart comments that approximately half the Doctorates go to work in industry versus academia.

UK: GETTING BETTER ALL THE TIMEThe United Kingdom is the third largest economy in Europe, after Germany and France, and is a scientifi c powerhouse that cannot be doubted: “With just 1 percent of the world’s popula-tion, the UK receives over 12 percent of citations to published papers…, and receives 10 percent of internationally recognized prizes each year,” according to its embassy’s website.

But only a decade ago, the government commissioned a study in which various problems that negatively affect the supply chain of scientifi c academic jobs were identifi ed. “It appears that this is not an attractive career path for many of the brightest Ph.D. graduates. This is both harming the UK’s research base and causing recruitment and retention diffi cul-ties for universities,” according to the report led by Sir Gareth

Roberts, a Welsh physicist. Among the challenges concerning postdoctoral and other contract research staff noted were: Un-certain career prospects associated with work on a short-term contractual basis, unsatisfactory training in the skills required in an academic career, and increasingly uncompetitive salaries.

Today, the UK system still has its troubles. But things seem to be getting better, and the nation continues to attract sci-entists from abroad to its noted institutions. Case in point: Raymond E. Goldstein, the Schlumberger Professor of Com-plex Physical Systems in the Department of Applied Mathe-matics and Theoretical Physics at the University of Cambridge. An American who arrived in the United Kingdom

FOCUS ON CAREERS

FOCUS ON EUROPE

Produced by the Science/AAAS Custom Publishing Office

continued » CR

ED

IT: H

HL-

LEIP

ZIG

GR

AD

UA

TE S

CH

OO

L O

F M

AN

AG

EM

EN

T

Andreas Pinkwart

The Alexander von Humboldt Professorships are Germany’s way of creating a beacon effect

and energising its research landscape. Every year, the Alexander von Humboldt Foundation is

offering ten of the world’s leading researchers up to five million EUR each to create new or

consolidate existing internationally visible research focus areas at German universities.

Shine your light in GermanyProfit from excellent conditions for research

Academics of all disciplines are eligible for an Alexan-

der von Humboldt Professorship, provided that they

are established abroad and recognised internation-

ally as top-class researchers. They will be nominated

by German universities – where appropriate in co-

operation with non-university research institutions.

Each Alexander von Humboldt Professorship will be

sponsored for a period of five years on the premise

that the university presents a convincing strategy

to sustain the position once the funding period has

come to an end. Accordingly, universities are asked to

submit an implementation plan as part of the fund-

ing application. This will allow new, long-term re-

search groups to be established, conducting cutting-

edge (international) research. The GermanMinistry of

Education and Research is supporting this programme.

Visit www.humboldt-foundation.de/ahp for more

information. Next closing dates for applications:

15 November 2011 and 15 May 2012

Alexander von Humboldt FoundationJean-Paul-Str. 1253173 BonnGermanyE-Mail: info@avh.de

www.humboldt-foundation.de

Exzellenz verbindet –be part of a worldwide network.

on

lin

e @

scie

nce

care

ers

.org

focu

s o

n e

uro

pe

www.sciencecareers.org400

of higher education and re-search,” says a French offi -cial involved in the manage-ment of bilateral cooperation programs. Initiatives include the creation of a new fund-ing agency in 2005, the National Research Agency (ANR, L’Agence nationale de

la recherché), a decentraliza-tion of universities to give them more independence, and fi nancial incentives to internationalize laboratories, which is part of the “govern-ment push to attract faculty

from abroad,” he describes. One element of France’s higher education is seemingly

matchless in the EU: prior to the last fi ve years, academia offered “only tenured positions” explains the French offi cial. “Now we have contract positions also.” This evolution, com-bined with the fl exibility granted to the institutions to recruit and negotiate salaries for their research faculty, has helped the nation become more attractive to top scientifi c talent. But the completion of an in-country contract position is still the pre-ferred route to obtaining a professorship.

Bjørg Elisabeth Kilavik can attest to this manner of hiring. After completing a postdoc in Marseille, she will start a perma-nent job as a principal investigator/research scientist at the Na-tional Center for Scientifi c Research (CNRS, Centre National de la Recherche Scientifi que). As an employee of the state, Kilavik will be subject to a mandatory retirement age, but she will not have teaching responsibilities, although this could change. “At these types of institutions, we are very privileged,” she notes.

Kilavik realizes how attractive France is for academics seek-ing tenure. In Germany, where she completed her Ph.D., “I could maybe have gotten a fi ve year contract as a junior pro-fessor, but there are few possibilities of getting a tenure-track position,” she says. In France, “at least the permanent posi-tions exist,…which gives stability. More people in Europe are realizing…[France] is close to the only place where these posi-tions exist.”

SCANDINAVIA: FAIRING WELL AND GROWINGUnited by similar cultures and language, the Nordic countries of Norway, Sweden, and Denmark support internationalism and an ease of movement of scholars from one nation to another. Almost all universities are state funded, and grants are gener-ally bestowed by the individual countries’ research councils, either directly to a PI or through the PI’s institution, depending upon the country and type of grant. While many other coun-tries demonstrate fi scal distress, Scandinavia’s economies are resilient. Norway has a budgetary surplus, and Sweden is debt-free. All three countries have annually increased spending on research and innovation for the last few years.

“We are so fortunate in Scandinavia, we are not

FOCUS ON CAREERS

FOCUS ON EUROPE

continued »

Produced by the Science/AAAS Custom Publishing Office

“Germany is in

a relatively good

position, having

weathered the

2008 fi nancial

crisis quite well.”

—Cathleen Fisher

fi ve years ago, Goldstein was a professor at the University of Arizona when he was ap-pointed to the chair.

He admires the process to hire him. “Their interest in me was not the money I could bring in, but rather the science,” he says. Goldstein appreciates that faculty are paid full-year salaries, as op-posed to nine month salaries as seen in the States. “This is recognition that research is a fundamental part of your job,” he adds. There is a man-datory retirement age (67, which may be changing), but many retirees keep their labs, supported by their pension.

For younger scientists eyeing careers in British academia, Goldstein suggests that the best way to get a job may be to do an early postdoc here. Indeed, this is something that Ryan

Seipke, a senior research associate (essentially a postdoc) in the School of Biological Sciences at the University of East An-glia, is currently undertaking. Originally from the United States, he chose the United Kingdom for his training in part because he wanted to gain international experience. When his three-year contract ends in 2012, he can apply for another.

As he starts his job search process, Seipke notes a number of ways in which one could potentially join the faculty of an English university. One is to secure a postdoc-to-professor transition fellowship. These highly competitive, prestigious opportunities are offered by several organizations (private and governmental), including the Royal Society, Biotechnology and Biological Sciences Research Council, Leverhulme Trust, and Medical Research Council. “They are an unoffi cial stamp of approval to the university that you might be a star and successful in securing more grant money,” opines Seipke. At the conclusion of the fellowship, he says, you are often offered a permanent faculty job.

FRANCE: PUTTING SCIENCE BACK INTO THE HEART OF THE NATION The French Republic is the 20th largest country in the world by demography, but the fi fth biggest scientifi c power with over 210,000 public and private researchers and, in total, almost 800,000 engineers and scientists nationwide, according to a report issued by its Ministry of Higher Education and Research. With a modern history of domestic scientifi c achievement dating back to the end of World War II, France has capitalized on its scientifi c assets with the adoption, in 2009, of a National Research and Innovation Strategy with a very specifi c and worthwhile goal: “To put back research and innovation at the heart of French society and economy.”

Indeed, as the country transitions from an economy that incorporates more governmental control to one that is more autonomous, “France has strived to reshape the landscape

London Research InstituteResearch Group Leaders

Registeredcharityno1089464

The London Research Institute (LRI) is Cancer Research UK's flagship researchInstitute, focusing on the analysis of fundamental biological processes involved in cancer.The Institute’s international staff work in 50 research groups, housed in well-supportedlaboratories at Lincoln's Inn Fields in central London, and at Clare Hall in Hertfordshire.

LRI encourages pursuit of ambitious and longer-term research programmes at the highestlevel.The London Research Institute is core-funded by Cancer Research UK. LRI GroupLeaders receive generous Institute core funding for personnel (research fellows, graduatestudents and technical support), laboratory consumables, access to the Institute'scomprehensive core technology facilities, backed by a substantial laboratory spaceand equipment package and competitive employment terms.

For 2011 recruitment, we are interested in outstanding Scientists seeking to establishindependent innovative research programmes to address fundamental questions in

Cancer Biology including but not limited to:• Tumour Biology: tumour-host interactions, cancer models, human cancer genomics

•Chromosome Biology:DNA damage, Cell Cycle regulation

•Computational Biology: Bioinformatics, biological networks, image processing.

Informal enquiries may be made by e-mail to Julian.Downward@cancer.org.uk orJohn.Diffley@cancer.org.uk

Applications by Clinically qualified candidates wishing to hold Senior Clinical ResearchFellowships at the Institute are welcomed. For information about the London ResearchInstitute, its staff, and their research interests visit www.london-research-institute.co.uk

In 2015 the LRI will become part of the new Francis Crick Institute, based in astate-of-the-art laboratory building at St Pancras in central London. Research at the newInstitute will focus on interdisciplinary approaches to the biology of human health anddisease.The Crick will be home to 1,250 researchers from the LRI, the MRC NationalInstitute for Medical Research, and the London Universities UCL, KCL and Imperial College.

Read aboutThe Francis Crick Institute at www.crick.ac.uk

To apply, please visit https://lrigroupleader.cancerresearchuk.org and submit a CV, publication list,past and future research plans (approx 2,000 words) and the contact details of at least threeacademic referees in a .pdf format. Please note that referees will be instructed to submit lettersof recommendation online at the time your application is received.

Closing date: 27th November 2011.

on

lin

e @

scie

nce

care

ers

.org

focu

s o

n e

uro

pe

www.sciencecareers.org402

Kristoffer Meinander

DOI: 10.1126/science.opms.r1100110

Alaina G. Levine is a science writer based in Tucson, Arizona, USA.

FOCUS ON CAREERS

FOCUS ON EUROPE

in an economic crisis such as in central and south Europe,” says Mikael Lindgren, a Swede who is currently a professor of optical sciences at the Norwegian University of Science and Technology. “There appears to be no lack of money in Norway for research and development in sectors such as energy, health, offshore and construction, only a lack of skilled labor.” Permanent academic jobs can be found throughout the region, although they are scarce in Denmark and Sweden because there are few universities in those countries. “In Norway it can be diffi cult to recruit ‘local’ Ph.D. students and postdocs because of the competition from the industrial and health sectors,” he remarks.

Ylva Hellsten, a professor of exercise and sports science at the University of Copenhagen, has been in Denmark for 15 years. She acknowledges the diffi culty for foreigners in secur-ing permanent academic positions in this former seat of Viking power, particularly those from outside of Scandinavian coun-tries, who may be unfamiliar with the language. Furthermore, there are only eight universities, and “Ph.D. students typically stay within the Danish system for academic employment,” she says. However, the tide may be changing. In 2007, 10 percent of researchers at Danish universities were of foreign descent, states Christian Lundager, assistant to the director general of the Danish Agency for Science, Technology and Innovation (Forsknings- og Innovationsstyrelsen), but from 2007–2009, one-third of all new appointments at the assistant professor level or above were from abroad.

Peter Byass, a professor of global health at Umeå Centre for Global Health Research within Umeå University in Sweden, recently considered how the country might become more at-tractive to international researchers. He cites that while the na-tion of 9 million people boasts a positive working environment which emphasizes academic autonomy, public appreciation for science, and universal healthcare, there are still several hur-dles to recruiting foreign scholars. For example, the academic

review process for senior posts can seem “unbelievably slow and complex to outsiders,” and it can be “hard for outsiders to understand expectations put on researchers,” especially given unfamiliar management styles, he says.

But with world-class research institutions and the ability to submit grant proposals in English, Scandinavia is an attrac-tive option for international scientists. Kristoffer Meinander,

a Swedish-speaking Finn who is a postdoc at the University of Aarhus in Denmark, says that foreigners may be surprised to learn the extent of scientifi c knowhow and funding for research infrastructure that exists here. “I wouldn’t have expected to fi nd this high level in Scandinavia.”

Indeed, throughout Europe, international scholars are discovering its scientifi c assets are continuing to prosper even amidst an ambiguous economic landscape. And now with support from the European Union’s Seventh Framework Programme and related national initiatives, the continent seems poised to advance even further, creating more attractive opportunities for foreign scholars looking to contribute to its research endeavors.

Produced by the Science/AAAS Custom Publishing Office

FEATURED PARTICIPANTS

Norwegian University of Science and Technologywww.ntnu.edu

Seventh Framework Programme for Research and Technology Developmenteuropa.eu/legislation_summaries/energy/european_energy_policy/i23022_en.htm

Umeå Centre for Global Health Researchwww.globalhealthresearch.net

University of Cambridgewww.cam.ac.uk

University of Copenhagenwww.ku.dk/english

University of East Angliawww.uea.ac.uk

University of Helsinkiwww.helsinki.fi /university

Aarhus Universitywww.au.dk/en

American Friends of the Alexander von Humboldt Foundationwww.americanfriends-of-avh.org

Danish Agency for Science, Technology and Innovationen.fi .dk

Delegation of the European Union to the USAwww.eurunion.org/eu

Embassy of France in the U.S.ambafrance-us.org

HHL-Leipzig Graduate School of Managementwww.hhl.de

National Center for Scientifi c Research (CNRS)www.cnrs.fr/index.php

North America Offi ce of the German Research Foundation (DFG)www.dfg.de/en/dfg_profi le/head_offi ce/dfg_abroad/north_america/

For further information and access to the online application material, please consult:

www.ist.ac.at/professor-applications

Deadline for receiving Assistant Professor applications: January 15, 2012

IST Austria values diversity and is committed to equality. Female researchers are encouraged to apply.

ISTAUSTRIA IS LOOKINGFOR

Professors andAssistant ProfessorsIST Austria (Institute of Science and Technology Austria) invites applications

for Professors and Assistant Professors in all fields of the natural and mathe-

matical sciences and related disciplines. Outstanding scientists in physics,

chemistry, and mathematics are especially encouraged to apply.

The Institute, which is situated on the outskirts of Vienna, was established by the Austrian

government with a focus on basic research. The campus opened in 2009 and is expected

to grow to 45 research groups and over 500 employees by 2016. IST Austria is entitled

to award PhD degrees and includes an English-language graduate school. It aims to

achieve an international mix of scientists and chooses them solely on the basis of their

individual excellence and potential contribution to research.

The Institute recruits tenured and tenure-track leaders of independent research groups.

The successful candidates will receive a substantial annual research budget but are

expected to also apply for external research grants.

For further information and access to the online application please consult www.ist.ac.at/gradschool.

For inquiries, please contact gradschool@ist.ac.at. For students wishing to enter the program in the

fall of 2012, the deadline for applications is January 15, 2012.

IST Austria values diversity and is committed to equality. Female students are encouraged to apply.

CALL FORPhDSTUDENTSThe Graduate School at IST Austria invites applicants from all countries to its PhD program. IST Austria isa new institution located on the outskirts of Vienna dedicated to cutting-edge basic research in the natural sciencesand related disciplines. The language at the Institute and the Graduate School is English.

The PhD program combines advanced coursework and research, with a focus on Biology, Computer Science,Neuroscience, and interdisciplinary areas. IST Austria offers internationally competitive PhD salaries supporting 4-5 yearsof study. Applicants must hold either a BS or MS degree or equivalent.

The Institute offers PhD students positions with the following faculty:

Additional faculty members will be announced on the IST website www.ist.ac.at.

Nick Barton Evolutionary and Mathematical BiologyJonathan P. Bollback Evolutionary BiologySylvia Cremer Evolutionary and Behavioral BiologyCaroline Uhler StatisticsTobias Bollenbach Biophysics and Systems BiologyCalin C. Guet Systems and Synthetic BiologyCarl-Philipp Heisenberg Cell and Developmental BiologyHarald Janovjak Molecular and Cellular BiophysicsDaria Siekhaus Cell and Developmental BiologyMichael Sixt Cell Biology and Immunology

Jozsef Csicsvari Systems NeurosciencePeter Jonas NeuroscienceGašper Tkacik Theoretical Biophysics and NeuroscienceKrishnendu Chatterjee Game Theory and Software Systems TheoryHerbert Edelsbrunner Algorithms, Geometry, and TopologyThomas A. Henzinger Software Systems TheoryVladimir Kolmogorov Computer Vision and Graph AlgorithmsChristoph Lampert Computer Vision and Machine LearningKrzysztof Pietrzak CryptographyChris Wojtan Computer Graphics

CAMPUS

VISIT DAY

November 26,

2011

on

lin

e @

scie

nce

care

ers

.org

focu

s o

n e

uro

pe

Vertex creates new possibilities in medicine. Our team discovers,

develops and commercializes innovative therapies so people with

serious diseases can lead better lives. Vertex scientists and our

collaborators are working on new medicines to cure or significantly

advance the treatment of hepatitis C, cystic fibrosis, epilepsy and

other life-threatening diseases. Founded more than 20 years ago in

Cambridge, MA, we now have ongoing worldwide research programs

and sites in the United States, United Kingdom and Canada. Vertex has

consistently been recognized as one of the industry’s top workplaces by

leading publications such as the Boston Globe, Boston Business Journal,

San Diego Business Journal and The Scientist, and most recently was

named the top employer in Science magazine’s 2011 annual survey.

As the UK-based subsidiary, Vertex Pharmaceuticals (Europe) Ltd brings

together scientists from various disciplines and backgrounds to form a

highly creative and productive team that works on finding medicines

for the treatment of cancer and inflammatory diseases. Our new

research facility in Oxfordshire is equipped with the latest technology

that enables us to tackle all aspects of early phase small molecule drug

discovery and, with the help of the wider organization, take on some of

the most ambitious goals in the industry.

Structure based drug design, has enabled all our major achievements

so far. It is our intention that this technology will continue to play a

central role in the way we do drug discovery. The close proximity of the

Diamond Light Source makes the UK site particularly important in this

field of Vertex research. The now vacant job of Head of Crystallography

is therefore a wonderful career opportunity for someone with skills in

structural biology and a desire to make important new drugs.

You are most likely to have a PhD and extensive experience in protein

crystallography. Strong leadership skills are required and excellent

written and oral publication record is a must. Experience with

pharma-biased methodologies such as fragment-based lead generation

and the ability to rapidly assess and introduce novel technologies

would be an advantage. The Head of Crystallography will be expected

to influence Vertex research on both sides of the Atlantic within the

structural biology groups and also by providing ideas and impetus for

new projects. In order to do this, communication skills and the ability

to work in collaboration must be first rate. An interest in biological and

chemical science beyond structural biology would be most useful since

this would allow the candidate to take career opportunities in multi

disciplinary project leadership.

If you are innovative and focussed on achieving success within a

collaborative environment, please visit our website www.vrtx.com to

find out more information and apply.

Closing date for applications is 7th November 2011.

Head of Crystallography, Oxford

Graduate Program in Computational

Engineering ScienceThe Aachen Institute for Advanced Study in Computational Engineering

Science (AICES) is a graduate school established within the frame of the

German Excellence Initiative. AICES focuses on computational engineer-

ing science including such diverse fields as modeling and simulation;

optimization; inverse problems and high-performance computing.

There is a limited number of openings in the AICES graduate program for

exceptionally qualified students who hold a bachelor´s or master´s degree

in engineering, natural science, mathematics, or computer science.

AICES offers a five-year path to the doctorate for students with bachelor’s

degrees, or a three-year path with master’s degree, due to a novel advis-

ing and training concept. Admitted candidates receive tax-free stipends

of 2000 € per month for the doctoral phase and 500 € per month for the

master´s phase.

Graduate School AICESRWTH AachenGermanywww.aices.rwth-aachen.de

www.aices.rwth-aachen.de/admission

Photo:

Peter Winandy

on

lin

e @

scie

nce

care

ers

.org

focu

s o

n e

uro

pe

GRADUATE PROGRAM

These newly established Fellowships in Experimental or Theoretical Quantum

Sciencewill be awarded both on the PhD- and the postdoc-level on the basis of

an international competition. The appointments are for a three-year duration.

Postdoctoral Fellowships carry a competitive annual salary, and offer an annual

research expense fund. PhD Fellowshipswill participate in the Vienna graduate

program CoQuS.

TheViennaQuantumFellowship programhasbeenestablishedwith the support

from the Austrian Ministry of Science and Research to offer young scientists

the best possible opportunity to develop their talents in the environment of

the Vienna Center for Quantum Science and Technology (VCQ).

The VCQ faculty provides a broad variety of research opportunities in the

areas of Experimental and Theoretical Quantum Science (see http://vcq.

quantum.at):

• Matter-wave interferometry and quantum atom optics

• Micro- and nanoscale quantum optics and quantum optomechanics

• Microoptics and novel quantum states of light

• Cold atoms and degenerate quantum gases

• Many-body quantum physics and quantum simulations

• Entanglement-based quantum communication on Earth and via satellites

• Quantum information and foundations of physics

Applicationmaterial should be sent to vcq@quantum.atDetailed and additional

information regarding application can be obtained from http://vcq.quantum.

at/fellowships

Deadline for the application is December 1, 2011. Fellowship candidates will

automatically be considered for other available postdoctoral positions in their

fields of interest.

The Vienna Center for Quantum Science and Technology (VCQ) invites

applications for the

Vienna Quantum Fellowships

More information about ATTRACT and PEARL as well as the other funding opportunitiesoffered by the National Research Fund Luxembourg can be found on the FNR’s website.Go and see what’s behind on www.fnr.lu/pearl and www.fnr.lu/attract

For an overview on research in Luxembourg, have a look at www.publicresearch.lu

ATTRACTLUXEMBOURG’S RESEARCH PROGRAMME FOR OUTSTANDINGYOUNG RESEARCHERS FROM ALL OVER THE WORLD

If you are an outstanding young researcher, ourresearch programme ATTRACT allows you to set up anindependent research team within a research institutionin Luxembourg. Through your innovation, dynamism andcreativity as well as the high scientific quality of yourresearch project, you will help to enhance Luxembourg’sposition in the international world of research. Projectsselected under ATTRACT have a lifespan of five yearsand the financial contribution will be up to 1.5M EUR.The 6th ATTRACT Call will be launched in December 2011.

PEARLLUXEMBOURG’S RESEARCH PROGRAMMEFOR INTERNATIONALLY RECOGNISED SENIOR RESEARCHERS

If you are an internationally recognised seniorresearcher, our research programme PEARL gives youthe opportunity to establish a high-profile researchprogramme in a research institution in Luxembourg(University of Luxembourg and other Research Centres)and thus to accelerate the development of andto strengthen the country’s research priorities.The financial contribution will be up to 5M EUR overa duration of five years.

Research Opportunities in Luxembourg.See what’s behind.

INVESTIGATING FUTURE CHALLENGES

challenges love solutions

We think there’s a solution for every problem. That’s why BASF researchers from all disciplines

always work on innovations with passion. Help us, in a modern environment, to find not just

products but comprehensive solutions for tomorrow’s challenges. That’s how we create

chemistry. At BASF. Find out more now and send your application to: www.basf.com/career

on

lin

e @

scie

nce

care

ers

.org

focu

s o

n e

uro

pe

The Leipzig School of Human Origins- An International Max Planck Research School -

by

The University of Leipzig

and

The Max Planck Institute for Evolutionary Anthropology

The Leipzig School of Human Origins offers a unique interdisciplinarygraduate program to study the evolutionary history of humans and greatapes. Graduate students are accepted into one of the following areas,but are encouraged to take part in courses and seminars from all threedisciplines:

Comparative and Molecular Primatology

Evolutionary and Functional Genomics, Ancient DNA,Molecular Anthropology and Genome Bioinformatics

Human Paleontology, Prehistoric Archaeologyand Archaeological Science

The language of the school is English. Visit www.leipzig.de for informati-on on living in Leipzig, Germany, in the center of Europe.

For project and application details go to www.leipzig-school.eva.mpg.deor contact us at:

e-mail: leipzig-school@eva.mpg.dephone: ++49 (0) 341 3550-0fax ++49 (0) 341 3550-119Application deadline: January 31, 2012

The Norwegian University of Science and Technology (NTNU) in Trondheimrepresents academic eminence in technology and the natural sciences as well asin other academic disciplines ranging from the social sciences, the arts, medicine,architecture to fine art. Cross-disciplinary cooperation results in innovative break-throughs and creative solutions with far-reaching social and economic impact.

Jobbnnorg

e.n

o

Faculty of Medicine

Professorship/Qualification Fellowshipin Systems NeuroscienceThe Kavli Institute for Systems Neuroscience of the Faculty ofMedicine at the Norwegian University of Science and Technology(NTNU) invites applications for a faculty position in systemsneuroscience. The new position is part of NTNU’s strategic effort inthe field of neuroscience.

We seek applicants with experience and interest in using state-of-the-art molecular and cellular technologies to understand neuralnetworks and behaviour. The successful candidate holds a Ph.D. andhas experience leading a research team and attracting internationalfunding. The candidate has a track record in both molecular-cellularand systems neuroscience, with outstanding publications, andshe/he demonstrates ability to develop an internationally competitiveresearch programme. Participation in teaching activities at master’sand PhD level is required.

The applicant will benefit from the strong infrastructure at the KavliInstitute for Systems Neuroscience at NTNU (www.ntnu.no/cbm).Start-up funding, including scientific equipment and PhD/postdoctoral fellows, is negotiable. The position is advertised at therank of Professor but may alternatively be defined as a qualificationfellowship for a period of no longer than 3 years in case of sufficientfuture potential.Young applicants not yet qualified for fullprofessorship are thus encouraged to apply.

Applicants should submit a cover letter, a CV with a completepublication list, copies of 5 selected papers, a summary of researchachievements, a research plan, and 3 letters of reference (refereesshould send letters directly to Edvard Moser, Director of the KavliInstitute, edvard.moser@ntnu.no).

The complete advertisement is available at

www.jobbnorge.no.

Submit applications through www.jobbnorge.no

within 1 December, 2011.

For further information about the position contact Edvard Moser,edvard.moser@ntnu.no, tel. +47 73598278; information about theapplication process contact Brit Løbeck Fladvad, HR- section,Faculty of Medicine, brit.fladvad@ntnu.no.

See also http://www.medisin.ntnu.no/eng/

The Molecular Biology Program of the Sloan-Kettering Institute, Memorial

Sloan-Kettering Cancer Center (www.ski.edu), has initiated a faculty search

at the Assistant Member level (equivalent to Assistant Professor). We are

interested in outstanding individuals who have demonstrated records of

significant accomplishment and the potential to make substantial contributions

to the biological sciences as independent investigators. Successful applicants

will have research interests that move the Program into exciting new areas that

complement and expand our existing strengths in the areas of maintenance

of genomic integrity, regulation of the cell cycle, and regulation of gene

expression. Faculty will be eligible to hold appointments in the Gerstner

Sloan-Kettering Graduate School of Biomedical Sciences, the Weill Cornell

Graduate School of Medical Sciences, as well as the Tri-Institutional MD/PhD

Training Program.

The deadline for applications is November 1, 2011. Interested candidates

should visit http://facultysearch.ski.edu to apply via the on-line faculty

application. Please visit the site as soon as possible, as it contains important

information on the required application materials, including deadlines for

submission of letters of reference.

Informal inquiries may be sent to Julie Kwan at kwanj@mskcc.org

or to Dr. Kenneth Marians, Chair, Molecular Biology Program at

kmarians@sloankettering.edu. MSKCC is an equal opportunity and affirmative

action employer committed to diversity and inclusion in all aspects of recruiting

and employment. All qualified individuals are encouraged to apply.

www.mskcc.org

Faculty PositionMolecular Biology

Sloan-Kettering Institute

on

lin

e @

scie

nce

care

ers

.org

FOCUS ON EUROPE GRADUATE PROGRAM

POSITIONS OPEN

Senior Level Research Position

WEILL CORNELL MEDICAL COLLEGE and

NEW YORK-PRESBYTERIAN HOSPITAL

Thrombosis research with particular reference to diagnosis, treatmentand prevention of occlusive vascular diseases.

Since 1958, the laboratories have been located at the New York VAMedical Center and Weill Cornell Medical College.

The major focus is on the etiology and pathogenesis of increased,therapeutically resistant blood cell and vessel wall reactivity.

The main emphasis of the research is on the vascular and molecularbiology of all blood cell and vessel wall components. Metabolic effectsof cell-cell interactions as initiated with multiple agonists are understudy for identification of transcellular metabolism, both in vitro and exvivo. The guiding premise is that thrombosis is strictly a multicellularprocess and must be treated by newer methodologies.

Requirements: The applicant should have a Ph.D. or an M.D.-Ph.D.with a minimum of five years of post-doctoral experience. It would beimportant to have a publication record-especially in high impactjournals. The incumbent should also have a proven ability to obtainindependent funding previously. The ability to develop a researchproject and work independently on the main theme of the laboratory isdesirable. The ability to interact with colleagues and synergize with themto advance the research of the laboratory is highly important. The abilityto write and co-author research and clinical publications and also todevelop specific aims for a research grant would be expected.

Qualified candidates should submit their complete curriculum vitae andbibliography in addition to three reference letters to the Department ofHuman Resources, WEILL CORNELL MEDICAL COLLEGE,1300 York Avenue, Room C610 (Box 113), New York, NY 10065;ajmarcus@med.cornell.edu.

EOE/M/F/D/V

Climate Change & Sea Level Rise InitiativeOld Dominion University has created an exciting new Climate Change and SeaLevel Rise Initiative and seeks a senior level faculty member in any disciplinerelated to climate change and sea level rise to help lead this initiative. The goal ofthe Initiative is to foster research, education and outreach on the impact that climatechange and sea level rise may have on metropolitan communities that are situatedat or close to sea level. Old Dominion University, a state-assisted Carnegiedoctoral/research-extensive institution that serves almost 25,000 students includingmore than 6,000 graduate students, is itself located in the city of Norfolk in themetropolitan Hampton Roads region of coastal Virginia. A broad range of facultyfrom across the entire university, ranging from science and engineering to the socialsciences, education, business and health sciences, are currently involved in thisinitiative (see http://www.odu.edu/ao/research/ccslri/). The successful candidatewill be part of the leadership team, will be an effective leader and advocate for theInitiative, and will be expected to contribute to the Initiative through research,teaching and service activities.

Applicants should possess an appropriate terminal degree and an academic recordthat merits a tenured appointment at the rank of associate or full professor in one ofthe academic departments within the University. A successful record in research andgrant writing is required, as is evidence of leadership and the ability to interact andcommunicate clearly with internal and external communities.

Applications should include a letter of interest that addresses the Initiative’s goals, acurriculum vitae, and contact information for three professional references includingemail addresses and phone numbers. Review of applications will begin November11, 2011 and the position will remain open until an appointment is made.Applications and nominations should be submitted electronically to: CCSLR SearchCommittee, attention Judy Bowman (jbowman@odu.edu), 222 Koch Hall,Norfolk, VA 23529.

Old Dominion University is an affirmative action, equal opportunity institution andrequires compliance with the Immigration Reform and Control Act of 1986.

on

lin

e @

scie

nce

care

ers

.org

FOCUS ON EUROPE

Tenure/Tenure Track Faculty Positionin Biological Sciences

The Department of Biological Sciences at Wayne State University(http://www.clas.wayne.edu/biology/) anticipates hiring a tenure-trackprofessor with research expertise in systems or computational biology.Preference will be given to candidates working in areas complementingthe department’s existing strengths in transcription and gene regula-tion, organismal and evolutionary development, intra- and intercellularsignaling, genomics, and community and landscape ecology. Rank willbe dependent upon qualifications.Wayne State University is a large, comprehensive, nationally ranked

research institution that offers state-of-the-art research facilities andhighly competitive start-up packages. The metropolitan Detroit areaoffers a rich cultural and educational environment, an excellent standardof living, and easy proximity toMichigan’s lakes, forests and recreationalsites.Applicants must have a Ph.D. degree, postdoctoral experience andan outstanding record of research achievement. Successful applicantsare expected to establish and maintain vigorous, externally fundedresearch programs and to participate in graduate and undergraduateeducation.All positions are posted on-line at jobs.wayne.edu. In additionto an online application that includes cover letter and curriculum vitae,applicants must submit a 2-page statement of their research plans andhave three letters of reference sent to the Faculty Search Committee:ad5348@wayne.edu. Please apply by November 30, 2011 for fullconsideration. Applications will be considered only when all materialshave been received.

Wayne State University is an affirmative action/equal opportunityemployer. Women and members of minority groups are especially

encouraged to apply.

Tenure-Track Position in Cell BiologyMcGill University

The Department of Biology at McGill University invites applications

for a tenure-track position in cell biology. We are seeking an energetic,

interactive individual whowill complement the Department’s recognized

strengths, across a wide range of model organisms, in developmental

genetics, neuroscience, biophysics, and cell biology. The ideal applicant

will employ advanced modern techniques to address questions of broad

biological significance. The candidatewill have convenient access to state-

of-the-art imaging facilities and other major core services as a member of

theMcGill Life Sciences Complex.Applicants should possess a Ph.D. or

equivalent degree in Biology or a related discipline, postdoctoral experi-

ence, and a significant track record of research excellence. The successful

applicant will be expected to conduct a vigorous program of independent,

externally funded research and to contribute to teaching at both the under-

graduate and graduate levels.We anticipate that this position will be filled

at the ASSISTANT PROFESSOR (tenure track) level, but applications

from more established candidates will be considered for recruitment at

the ASSOCIATE or FULL PROFESSOR rank. Competitive startup

and equipment funding packages are available. Persons wishing to be

considered for this position should forward via e-mail: a curriculum vitae,

a statement of research interests, a statement of teaching interests, PDF

files of major publications, and arrange to have three letters of reference

submitted directly by e-mail to: recruit.biology@mcgill.ca In the subject

line, please enter Biology Faculty Search and your name.Acceptable file

formats are Microsoft Word and PDF.

The application deadline is December 2nd, 2011.

All qualified applicants are encouraged to apply; however, Canadian and

permanent residents will be given priority.McGill University is committed

to diversity and equity in employment. It welcomes applications from

indigenous peoples, visible minorities, ethnic minorities, persons with

disabilities, women, persons of minority sexual orientations and gender

identities and others who may contribute to further diversification.

Assistant Professor of Cell Biology and Neuroscience

The Department of Cell Biology and Neuroscience at Rutgers, TheState University of New Jersey, Piscataway, seeks to fill two tenure-track positions at the Assistant Professor level. Individuals workingon all aspects of cell biology or neurobiology are invited to apply;those with research programs that integrate with and complement cur-rent faculty research will be given highest priority. The new facultymay be part of the Brain Health Institute initiative. The Departmentis located on the Rutgers Busch Campus and is part of the Divisionof Life Sciences, a group of Departments and Institutes that serves toprovide opportunities for interdisciplinary research. Current collabora-tions within the DLS range from biomaterials and nanotechnologyto human genetics and stem cells. The Campus is located near theUMDNJ-Robert Wood Johnson Medical School and is less than onehour away from New York City and Philadelphia.Applicants must have a Ph.D. and/or M.D. with a minimum of

three years postdoctoral experience. The successful candidate willbe expected to establish an independent research program sup-ported by external funding and to contribute to undergraduate andgraduate education. The Department offers excellent facilities andcompetitive start-up packages. Interested individuals are encouragedto apply online through the CBN website (http://cbn.rutgers.edu)with a curriculum vitae, a brief statement of research plans, and thenames, addresses, and contact information of three individuals whowill provide a letter of reference.Applications received after December 1, 2011 will be considered

if positions remain available.

Rutgers University is an Equal Opportunity/Affirmative ActionEmployer.

The School of Life Sciences and The Biodesign Institute at Arizona StateUniversity invite applications for a tenure-track faculty position at the levelof Assistant Professor whose research is at the interface between Genomicsand Life Sciences. Research methods may include theoretical, computational,and empirical approaches in population genetics, functional and comparativegenomics, and bioinformatics. The successful candidate will be expected todevelop an innovative, extramurally-funded, research program, teach at theundergraduate and graduate levels, and have a commitment to outreach andservice. The successful candidate will be expected to mentor undergraduateand graduate students as well as postdoctoral fellows. A competitive start-uppackage and teaching load compatible with high research productivity will beprovided.Arizona State University has made a commitment to accelerating the

translation of basic discoveries into practical benefits for society throughthe construction of state-of-the-art research facilities and the recruitmentof world-class faculty members. The successful candidate will participatein university-wide health and/or sustainability initiatives supported by corefacilities for functional genomics and next generation sequencing, functionalproteomics, high throughput cellular screen, bioinformatics, high performancecomputing, and imaging. More information on genomic research opportunitiesat the Biodesign Institute and the School of Life Sciences at ASU can be foundat http://genomics.asu.edu.Candidates must have a Ph.D. (or equivalent) in an appropriate field, and

a minimum of 2 years of postdoctoral training is preferred. Demonstratedteaching and research excellence is preferred.To apply, send cover letter, your curriculum vitae, three representative

publications, separate statements of future research plans and teachingphilosophy and interests, and contact information for three references to be sentto Alan Rawls, Chair, Genomics Faculty Search Committee, School of LifeSciences, PO Box 874501, Tempe, AZ 85287-4501. Electronic applicationssent as PDF files to solsfacultysearch@asu.edu are preferred. The initialclosing date for receipt of applications is November 11, 2011; applicationswill be reviewed weekly thereafter until the search is closed. For additionalinformation on this position and the School of Life Sciences, please visit http://sols.asu.edu/jobs. A background check is required for employment. ArizonaState University is an equal opportunity/affirmative action employer committedto excellence through diversity. Women and minorities are encouraged to apply.

Faculty Positionin Genomics

on

lin

e @

scie

nce

care

ers

.org

The University of Michigan seeks outstanding applicants for tenure-track and tenured clinical andbiomedical informatics faculty positions. UM hosts a Clinical and Translational Sciences Award (CTSA)Biomedical Informatics Program, 48 current Bioinformatics PhD students and NIGMS bioinformatics andNCI proteome informatics training grants. We have active national outreach for minority candidates.We are currently recruiting up to 3 senior and 5 junior faculty members to establish independent

individual and team-based research programs. Informatics research teams are welcome to apply as aunit. After multiple recent recruitments in the area of basic bioinformatics, our emphasis this cycle is inbiomedical/clinical informatics. Specific joint appointments may be considered, for appropriate candidates,with the Dept. of Human Genetics, Electrical Engineering and Computer Science, School of Information,School of Public Health, School of Nursing or other appropriate units. In addition, affiliation with Centers,including the Comprehensive Cancer Center, Depression Center, Cardiovascular Center and Metabolomicsand Obesity Center will be encouraged. There are extensive computational and information infrastructuralresources are available to individual recruits and teams. For appropriate individuals, opportunities exist forfaculty leadership roles in Research Information Technology management and operations and to influenceinstitutional priorities in clinical and biomedical informatics.Successful candidates will have a PhD and/or MD degree, or equivalent, with post-doctoral training,

on topics such as biomedical data mining and machine learning; multi-scale integrative analysis; naturallanguage processing (NLP) and ontologies applied to biomedicine; informatics related to healthcare deliveryand personalized medicine (e.g. clinical decision support or pharmacogenomics). Publications and funding,as evidence of research productivity, a detailed research plan as well as evidence for an interest in in graduateand post-doctoral education will be essential components of the application. The rank of selected candidateswill depend upon experience and qualifications.Applicants should send a letter of interest with Curriculum Vitae, Research Plan, and a list of three or

more references with current contact information to: Search Committee, Department of ComputationalMedicine and Bioinformatics, Job Code 200. The University of Michigan, 2017 Palmer Commons,100 Washtenaw Ave, Ann Arbor, MI. 48109-2218, email: ccmbrecruit@umich.edu. For appropriatelyqualified candidates, simultaneous application to theUMBiological Sciences Scholars Program, BSSP, isstrongly encouraged (for more information see: http://www.med.umich.edu/medschool/research/bssp/).Applications will be reviewed through March 2012 beginning October 2011.

Ann Arbor is a remarkable cultural and living environment. The University of Michigan is responsive tothe needs of dual career families and is an Equal Opportunity Affirmative Action Employer committed to

diverse faculty, staff and student body.

http://ccmb.med.umich.edu

University of Michigan Medical School

Department of Computational Medicine

and Bioinformatics

Assistant, Associate and Full Professors

FACULTY POSITIONDepartment of Chemistry and Biochemistry

Texas State University -San Marcos

The Department of Chemistry and Biochemistry at Texas State University-San Marcosseeks to fill a new faculty position in any area of Chemistry or Biochemistry, beginning fall2012.Applicants at all rankswill be considered. The successful candidate at theAssistant Profes-sor rank must have a Ph.D. in Chemistry, Biochemistry, Chemical Education or a closely-relatedfield; postdoctoral research experience; a track record substantiating the potential to establish anexternally-funded research program that involves undergraduate and/or graduate students; and thecapability to teach courses in chemistry, biochemistry, and/or chemical education. The successfulcandidate at the Associate Professor or Professor rank must meet all qualifications for the Assis-tant Professor rank, in addition to the following: five or more years of experience as a universityfaculty member; and a record consistent with (1) success in obtaining externally-funded researchgrants, (2) quality peer-reviewed publications, and (3) significant professional service activities.At all ranks, preference will be given to candidates whose areas of teaching and research expertisecomplement the department’s goals; and to candidates who can effectively mentor students.Texas State University- San Marcos is located in the burgeoning Austin-San Antonio corridor

at the edge of the hill country, and is the 5th largest campus in Texas with more than 34,100students. The Department of Chemistry and Biochemistry currently has 20 faculty, 30 M.S.graduate students, and 350 undergraduate majors, most of whom participate in research. TheDepartment also is a participant in a proposed interdisciplinary Ph.D. program in MaterialsScience, Engineering, and Commercialization. For additional information, please visit http://www.txstate.edu/chemistry.

All application materials should be emailed to Chem_Search@TxState.edu. A completeapplication consists of the following: (1) a cover letter identifying the rank for which you areapplying, the area(s) of your teaching and research interests, and a list of the three individuals(with contact information) who will be submitting letters of reference; (2) a CV; (3) a one-pagesummary of teaching philosophy and interests; (4) an outline of research plans (3 pages or less);(5) undergraduate and graduate transcripts; and (6) three letters of reference, emailed directlyfrom each referee. To receive full consideration, application submissions should be completeby December 4, 2011. Review of complete applications will begin December 5 and will continueuntil the position has been filled.

Texas State University - San Marcos is an equal opportunity employer; women and members ofunderrepresented minorities and individuals with disabilities are encouraged to apply.

Executive DirectorDivision of Earth and Ecosystem SciencesPosition #30-001

on

lin

e @

scie

nce

care

ers

.org

ScienceCareers.org

An exceptional career requires insightful planning and management.That’s where Science Careers comes in. From job search to careerenhancement, Science Careers has the tools and resources to helpyou achieve your goals. Get yourself on the right track today and get areal career plan that works. Visit ScienceCareers.org.

Get a Career Plan that Works.

TheDepartment of Civil and Environmental Engineering at Northwest-

ern University invites applications for a tenure-track or tenured faculty

position in Environmental Engineering and Science. We seek outstand-

ing applicants in any area of environmental research who have a strong

fundamental knowledge base, perform cutting-edge cross-disciplinary

research, and have the ability to contribute substantially to our core teach-

ing efforts environmental science and engineering. Areas of particular

interest include: environmental microbiology, biotechnology, andmicro-

bial ecology, innovative approaches to sustainable water use, design and

development of advanced treatment technologies, and linkages between

water, ecosystems, biogeochemical cycles, and global change.

Review of applications will begin in October 2011, and the search will

proceed until the position is filled. Preferencewill be given to applications

submitted by November 30, 2011.

Applications should be submitted electronically as a PDF document

containing a cover letter, curriculum vitae, a two-to-three-page descrip-

tion of research accomplishments and plans for future work, a one-to-two

page description of teaching interests, and a list of at least three persons

who can provide letters of recommendation.Applicationmaterials should

be submitted to the Environmental Search Committee Chair via the web

interface at http://facultysearch.mccormick.northwestern.edu/apply/

index/NDI

It is anticipated that this position will be filled at the junior level but

outstanding senior candidates will also be considered.

Northwestern University is an Affirmative Action, Equal Opportunity

Employer. Women and individuals in underrepresented groups in science

and engineering are encouraged to apply. Hiring is contingent upon

eligibility to work in the United States.

Faculty Tenure Track Positions

Faculty Positionin Environmental Engineering

and Scienceat Northwestern University

We invite applicantswith aPhD,MDorequivalent, anda recordof outstanding

promise and achievements, for our faculty positions in cardiovascular and

metabolic disorders. The Duke-NUS Cardiovascular andMetabolic Disorders

(CVMD) Signature Research Programwill bring together top-flight researchers

investigating the interactionsbetweenmetabolic disorders and cardiovascular

disease, focusing on translational discoveries that can impact clinical care.

Investigators will address diverse aspects of the problem from diabetes and

dyslipidemia to hypertension and vascular disease.

Positions include full salary aswell as generous start-up and 5 years of annual

research funding to assure a stable base of support to be supplemented by

competitive awards. Candidates should submit a cover letter, curriculum

vitae, summary of research accomplishment and an outline of future plans

by February 1, 2012 to:

Thomas Coffman, MD, Director

Cardiovascular and Metabolic Disorders Program

Duke-NUS Graduate Medical School Singapore

8 College Road, Singapore, 169857

E-mail cvmd.recruit@duke-nus.edu.sg

ABOUTDUKE-NATIONALUNIVERSITYOF SINGAPOREGRADUATEMEDICAL

SCHOOL

Duke–NUS brings post-baccalaureate, research-intensivemedical education

and research to Asia, and represents a truly global partnership betweenDuke

University in the U.S. and National University of Singapore (NUS). Duke-NUS

shares amodern campuswith Singapore’s largest hospital, SingaporeGeneral

Hospital, and several national centers, including the National Heart Centre.

The facilities in nearby Biopolis also provide a unique array of complementary

resources. In addition, the CVMD program has close associations with the

Duke Cardiovascular Research Center and Duke Global Health Institute.

www.duke-nus.edu.sg

Faculty Positions in Cardiovascular & Metabolic Disorders

Chair,Biomolecular Chemistry

University of Wisconsin School of Medicineand Public Health, Madison, Wisconsin

The UWSMPH invites applications and nomina-

tions for the position of Chair of the Department

of Biomolecular Chemistry (BMC), to succeed

Dr. Robert Fillingame, who will be stepping down

after ten years as Chair. The BMC Department has

thirteen faculty members, each with an active

research program in a variety of areas ranging from

biophysical chemistry to cellular and developmental

biology. The Department’s graduate program, which

is collaboratively directed by faculty in BMC and the

Biochemistry Department in the College of Agricul-

ture and Life Sciences attracts, on average, 25 very

high quality students annually. The department also

attracts graduate students from other multidisci-

plinary programs on campus and currently houses

40 - 45 students in the various research laboratories in

addition to 30 post-doctoral trainees and other

research staff. The department currently participates

in several multidisciplinary training grants and

has an impressive portfolio of extramural grants

to the faculty.

We seek an accomplished scientist with a strong

record of academic and administrative leadership.

Special emphasis will be given to experience in both

graduate student and medical student education, the

mentoring of junior faculty, and a sustained record

of extramural research funding. The successful

candidate will have an exciting vision for the future

of biochemical research, education, and training in a

leading academic medical center. Candidates must

have a PhD degree or equivalent and must have

academic credentials for a tenured faculty appoint-

ment at the University of Wisconsin-Madison.

Send a letter of application or nomination, with

curriculum vitae, to: K. Craig Kent, M.D.,

and Rod Welch, Ph.D., BMC Chair Search

Committee, c/o Jamie Edge, 4150 HSLC,

750 Highland Avenue, Madison, WI,

53705-2111, jledge@wisc.edu.

To receive full consideration, applicationsshould arrive by February 1, 2012.

Unless confidentiality is requested in writing, information regarding

applicants must be released upon request. Finalists cannot be

guaranteed confidentiality. Wisconsin Caregiver Law applies.

The University of Wisconsin is an equal opportunity, affirmative

action employer. For more information: www.med.wisc.edu,

http://www.bmolchem.wisc.edu/index.html

on

lin

e @

scie

nce

care

ers

.org

Heal the sick, advance the science, share the knowledge.

The Division of Pulmonary and Critical Care Medicine, in

conjunction with the Mayo Clinic Cancer Center in Rochester,

MN, is seeking a highly productive Ph.D. and/or M.D. scientist

to lead a developing Program in Lung Cancer. The ideal

candidate will be an established investigator at the Associate or

Full Professor level with a strong record of extramural funding,

exceptional productivity in cancer research and leadership

experience. Academic appointment is available in any of the

Mayo Graduate School of Medicine basic science departments.

The Mayo Clinic has over 200 biomedical research laboratories,

institutionally supported state-of-the-art animal, molecular and

microscopic core facilities, a number of unique human disease

tissue banks and opportunities to readily translate discovery to

the bedside. Very competitive start-up and sustained intramural

funding will be provided.

Interested applicants should submit curriculum vitae,

description of research plans and names and addresses of three

references by December 1, 2011 online at: www.mayoclinic.

org/physician-jobs (job #6442BR), or email to:

Ms. Trish Iverson

Search Committee Secretary

Email: iverson.patricia@mayo.edu

Mayo Foundation is an affirmative action and equal opportunity employer andeducator. Post-offer/pre-employment drug screening is required.

Faculty Position in Cell/MolecularBiology of Lung Cancer

Yourcareer isour cause.

• Job Postings

• Job Alerts

• Resume/CVDatabase

• Career Advice

• Career Forum

www.sciencecareers.org

Get help

from the

experts.

on

lin

e @

scie

nce

care

ers

.org

Why not

Postdoctoral Position inAstronomy/AstrochemistryNY Center for Astrobiology

Rensselaer Polytechnic Institute

The successful applicant will join a team seekingto identify major chemical pathways that leadfrom simple molecules in interstellar clouds tocomplex organic molecules in protoplanetarydisks around new stars. The appointee will partic-ipate in acquisition and analysis of new and exist-ing astronomical observations, modeling, andpublication of results. A Ph.D. (or foreign equiva-lent) in astrophysics or a related field is required;prior relevant research experience is advanta-geous. The position will be for one year withexpectation of renewal.

Apply by December 15, 2011 to PostdoctoralResearch Associate, NY Center forAstrobiology, search number 20110001 at:https://rpijobs.rpi.edu/applicants/jsp/shared/frameset/Frameset.jsp?time=1313683304474

Vitae, publications, research interests and namesof 3 professional references may also be mailedto: AM Strack, Astrobiology, Cogswell, RPI,110 8th Street, Troy, NY 12180 or by email tostraca@rpi.edu

Rensselaer is an AA/EOE

change the world?

The University invites applications for two tenure track Assistant Professor positions in RNA scienceand technology.

CELL/DEVELOPMENTAL BIOLOGIST, Department of Biological Sciences: conducting researchon the role of RNA, including but not exclusive to non-coding or microRNA molecules, in post-transcrip-tional gene regulation or other cellular and/or developmental processes.

CHEMIST/BIOCHEMIST, Department of Chemistry: conducting research in RNA science and itsapplications in areas such as, but not limited to, modified nucleosides, synthesis, imaging, and analyticalchemistries as it pertains to RNA structure/function relationships, including interactions with proteins andother RNAs.

Both positions will be affiliated with the RNA Institute (http://www.albany.edu/rna) with state-of-the-art laboratories housed in the Life Sciences Research Building (http://www.albany.edu/lifesciences).The Institute brings together more than 35 investigators from the College of Arts & Sciences, the Collegeof Nanoscale Science and Engineering, the School of Public Health, and regional institutions includingthe Wadsworth Center, Rensselaer Polytechnic Institute, and Albany Medical College. This creates an out-standing environment for research collaborations.

Instructional responsibilities will be consistent with the position and those of the faculty in the homedepartment, and the interests of the candidate.

Submit applications for CELL/DEVELOPMENTAL BIOLOGY at:http://albany.interviewexchange.com/jobofferdetails.jsp?JOBID=27938

Submit applications for CHEMIST/BIOCHEMIST athttp://albany.interviewexchange.com/jobofferdetails.jsp?JOBID=27907

Applications must include a CV with publications cited in detail and any present or past grant funding,statement of research interests, statement of teaching interests, and a minimum of three references withcontact information.

The successful candidates for both positions will be offered a competitive salary, start-up package, andresearch space in the Life Sciences Research Building.

Qualifications for both Candidates: Ph.D. from a college or university accredited by the U.S. Departmentof Education or an internationally recognized accrediting organization and a strong publication recordreflecting significant scientific accomplishments. Applicants must address in their applications their abili-ty to work with and instruct a culturally diverse population. Preferred qualifications include productivepost-doctoral training and the potential or demonstrated ability, to obtain independent extramural funding.Review of applications will begin on November 15, 2011 and continue until the positions are filled.

The University at Albany is an EEO/AA/IRCA/ADA employer.

Two Faculty Positions in RNA Research

The Biology Department at City College of the City University of New York invites applications for a tenure-track or tenured positionin Molecular Neurobiology at the level of either Assistant or Associate Professor to begin Fall 2012. We are searching for anoutstanding molecular neurobiologist performing cutting-edge research in fundamental cellular or developmental processes in thenervous system. Candidates should have demonstrated research excellence and collaborative skills to interact with a vibrant,expanding neurobiology group. The candidate’s research program should complement the current departmental research inmolecular neurobiology, systems neurobiology, and behavior. The successful candidate will be expected to teach in bothundergraduate and doctoral programs and work collaboratively within the City University of New York. For areas of departmentalstrengths, see www.sci.ccny.cuny.edu/biology.

QUALIFICATIONS

Junior candidates should have a Ph.D., postdoctoral experience, and a strong record of publications; senior candidates should havea strong history of federal funding, research productivity, and teaching at the undergraduate and graduate level.

COMPENSATION

Commensurate with qualifications and experience. Competitive start-up package available.

HOW TO APPLY

If you are viewing this job posting in CUNYFirst, please click on "Apply Now" on the bottom of this page and follow the instructions.If you are viewing this job posting externally, please apply as follows:- Go to www.cuny.edu and click on "Employment"- Click "Search job listings"- Click on "More options to search for CUNY jobs”- Search by Job Opening ID number- Click on the "Apply Now" button and follow the instructions.

To be considered for this position, you must include a curriculum vitae (CV), summary of past research accomplishments and futureresearch plans, and a statement of teaching and mentoring experience in one document in any of the following formats: doc, .docx,.pdf, .rtf, or text format.

Letters of recommendation from at least three referees should be sent directly to the search committee at:Molecular Neurobiology Search Committee - Department of Biology, J526The City College of New York, 160 Convent Avenue, New York, NY 10031biosearches@sci.ccny.cuny.edufax: 212 650-8585

EQUAL EMPLOYMENT OPPORTUNITY

We are committed to enhancing our diverse academic community by actively encouraging people with disabilities, minorities,veterans, and women to apply. We take pride in our pluralistic community and continue to seek excellence through diversity andinclusion. EO/AA Employer.

Assistant or Associate ProfessorMolecular Neurobiologist

Job Opening ID Number: 4678

Closing Date: Until Filled; Applicationswill be reviewed starting December 1, 2011

The City College, host site of the New York StructuralBiology Center, home campus of The City University ofNew York Macromolecular Assemblies Institute, and aCUNY flagship campus, invites applications for a tenure-track faculty position in Structural Biology with emphasison Electron Microscopy and/or X-ray crystallography. Weare seeking a candidate who will develop a world-class,externally funded research program that aims toelucidate structure/function correlations in biomolecules.This Chemistry Department position may be at the rankof Assistant or Associate Professor. The new facultymember will be expected to teach both undergraduateand graduate level courses and mentor undergraduateand graduate students. The successful candidate shouldhave a Ph.D. in biophysics, chemistry, or biochemistryand substantial postdoctoral experience in structuralbiology or related fields and ability to build a productiveexternally funded research program.

HOW TO APPLY

Please apply as follows:- Go to www.cuny.edu and click on "Employment"- Click "Search job listings"- Click on "More search options"- Search by Job Opening ID number- Click on the "Apply Now" button and follow theinstructions.

To be considered for this position, a curriculum vitae(CV), list of publications, a description of current andfuture research plans, a description of teachingphilosophy and 2-3 representative journal articles shouldbe combined and uploaded as a single PDF. The 3 lettersof reference should be sent directly by email to Ms.Denise Addison (dma@sci.ccny.cuny.edu).

Assistant or Associate ProfessorStructural Biology –

Tenure Track

Job Opening ID Number: 4543

Closing Date: November 15, 2011

on

lin

e @

scie

nce

care

ers

.org

POSITIONS OPEN

Outstanding POSTDOCTORAL FELLOWS aresought to conduct research into the pathogenesis ofAIDS. Using the nonhuman primate model of AIDSwe are examining host and viral determinants andmechanisms that underlie host control of and protec-tion from infection.

The successful candidate will have a DVM, M.D., orPh.D. in the life sciences along with evidence of firstauthor papers published or accepted in peer-reviewedjournals. The candidate will also have excellent writtenand verbal communication skills and analytical capa-bilities and a solid understanding of immunology andvirology and techniques such as polychromatic flowcytometry, immunohistochemistry, and in situ hybrid-ization. Molecular biology skills will also be useful. Toapply, send a cover letter and curriculum vitae and thenames of three individuals who may be contacted forreferences to Andrew A. Lackner, DVM, Ph.D.; Pro-fessor of Microbiology, Immunology and Pathology;Director, Tulane National Primate Research Center ate-mail: rita@tulane.edu.

Tulane University and Louisiana State University are Affir-mative Action/Equal Opportunity Educators and Employers. In-dividuals from under-represented minorities are strongly encouragedto apply.

FACULTY POSITION inMicrobiology

San Jose State University

San Jose State University (SJSU) invites applicationsfor a tenure-trackASSISTANT PROFESSOR positionin microbiology to begin August 20, 2012 (website:http://www.sjsu.edu/facultyaffairs/Unit_3/Tenure_Track/Employment/index.htm). Preferencewill be given to applicants in such areas as eukaryoticmicrobiology, virology, microbial genomics, microbialproteomics, and host-microbial areas will also be con-sidered. For full consideration send a letter of appli-cation, curriculum vitae, university undergraduate andgraduate transcripts, statement of teaching interests/philosophy and research interests, and at least three of-ficial letters of reference with contact information to:Cleber Ouverney, Department of Biological Sciences,San Jose State University, San Jose, CA 95192-0100.E-mail: cleber.ouverney@sjsu.edu. For full considera-tion, apply by November 4, 2011. Please include JobOpening Identification (JOID) 14215 on all corre-spondence. SJSU is an Equal Opportunity/Affirmative ActionEmployer committed to the core values of inclusion, civility, andrespect for each individual.

FACULTY POSITION inNeurophysiology

San Jose State University

San Jose State University (SJSU) invites applicationsfor a tenure-track ASSISTANT PROFESSOR posi-tion in neurophysiology to begin August 20, 2012(website: http://www.sjsu.edu/facultyaffairs/Unit_3/Tenure_Track/Employment/index.htm).Preference will be given to applicants whose back-grounds complement and augment existing faculty andprograms. Experience in vertebrate neurophysiology isessential. For full consideration send a letter of appli-cation, curriculum vitae, university undergraduate andgraduate transcripts, statement of teaching interests/philosophy and research interests, and at least threeofficial letters of reference with contact informationto: Michael Sneary, Chair, Department of Biolog-ical Sciences, San Jose State University, San Jose, CA95192-0100. E-mail: michael.sneary@sjsu.edu. Forfull consideration, apply by November 4, 2011. Pleaseinclude Job Opening Identification (JOID) 14216 onall correspondence. SJSU is an Equal Opportunity/AffirmativeAction Employer committed to the core values of inclusion, civility,and respect for each individual.

POSITIONS OPEN

ASSISTANT PROFESSORof Genome Biology

The School of Biological Sciences at IllinoisState University (website: http://www.bio.illinoisstate.edu) invites applications for atenure-track faculty position at the level of As-sistant Professor in the area of Genome Biol-ogy. The successful applicant should be engagedin research that blends bioinformatics with mo-lecular experimentation to address fundamentalquestions in broad areas of genome structure,function, evolution, or related area. We seek ap-plicants with a Ph.D., or equivalent, and post-doctoral experience, a demonstrated potential tosecure external funding and an interest in work-ing within a diverse intellectual community. Thesuccessful applicant will be expected to devel-op an independent externally funded researchprogram, and be involved in both graduate andundergraduate teaching and mentoring. TheSchool of Biological Sciences at Illinois StateUniversity is currently home to 85 M.S. andPh.D. candidates and over 500 B.S. majors. Toapply, send descriptive cover letter, curriculumvitae, a one- to two-page statement of future re-search goals, and contact information for threereferences as a single PDF file to Dr. WadeNichols, c/o Sally Little via e-mail: salitt2@ilstu.edu. Review of applications will begin onNovember 1, 2011 and continue until the po-sition is filled. Intended start date August 16,2012. Illinois State University is an Equal OpportunityUniversity encouraging diversity.

ASSISTANT/ASSOCIATE PROFESSORDepartment of Neuroscience & Physiology

State University of New YorkUpstate Medical University

We seek applications to fill two tenure-track posi-tions at either the Assistant or Associate Professorlevel from individuals studying any area of Neurosci-ence. The successful applicants will be expected to de-velop well-funded research programs and to contributeto graduate and medical teaching. We offer a highlycompetitive startup package and salary. Appointmentat the Associate Professor level will require demonstra-tion of outstanding achievement and current extramuralfunding. Further information about the Departmentcan be found at website: http://www.upstate.edu/neuroscience/.

Candidates should have a Ph.D. or equivalent, post-doctoral experience, and a strong publication record.Applicants should e-mail a PDF file containing cur-riculum vitae, summary of research accomplishments,and future research plans to e-mail: neurosci@upstate.edu. In addition, three letters of reference should beaddressed to: Dr. Barry E. Knox, Chair, Departmentof Neuroscience & Physiology, WH 3223, 750 EastAdams Street, Syracuse, New York 13210.

Review of applications will begin November 1, 2011,and continue until the positions are filled.

SUNY Upstate Medical University is an Affirmative Action/Equal Opportunity Employer engaging excellence through diver-sity. Women and minorities are strongly encouraged to apply.

MEDICAL DIRECTOR

Physician, biomedical researcher, or other medical/bioscience professional sought by Manhattan familyto research and coordinate family medical and healthcare issues. This person will manage a small team ofprofessionals and interface with physicians, medical re-searchers, and consultants (in academia and otherwise)to ensure delivery of highest-quality medical care tofamily members. Considerable weight will be givento unusual academic distinction and other intellectu-al achievements. Excellent communication skills are amust, a Ph.D. or M.D. is strongly preferred, and clini-cal experience is a plus. This is a full-time positionwith a highly attractive compensation package and sig-nificant upside potential. Resume to e-mail: mdrecr@gmail.com.

POSITIONS OPEN

The Department of Biological Sciences, Uni-versity of Denver (DU), invites applications foran Ecologist in a tenure-track position at theASSISTANT PROFESSOR level to begin Sep-tember 1, 2012. We are particularly interestedin candidates in the fields of population ecology,community ecology, plant-animal interactions,evolutionary ecology, and physiological ecolo-gy. Field-oriented ecologist studying whole or-ganisms; potential for research in the RockyMountain region is especially desirable. The suc-cessful candidate will have a Ph.D. and post-doctoral experience in appropriate fields, willdevelop an extramurally funded research program,will supervise undergraduate research projectsand M.S. and Ph.D. students, and will teachundergraduate and graduate courses in ecologyand specific areas of specialty. All candidatesmust submit their application through website:https://www.dujobs.org. The online applica-tion should include: curriculum vitae, statementsof teaching philosophy and research interestsand two recent publications. Under a separatecover, send three letters of recommendationto: Dr. Shannon Murphy, Chair, EcologistSearch Committee, Department of Biolog-ical Sciences, University of Denver, Denver,CO 80208. Review of applications will beginDecember 1, 2011. The University of Denver is com-mitted to enhancing the diversity of its faculty and staff andencourages applications from women, minorities, peoplewith disabilities and veterans. DU is an Equal Employ-ment Opportunity/Affirmative Action Employer. Pleasesee our extensive benefit package at website: http://www.du.edu/hr/benefits.

ASSISTANT PROFESSORInstitute of Bioinformatics and theDepartment of Computer Science

University of Georgia

The Institute of Bioinformatics and the Depart-ment of Computer Science at the University of Georgiainvite applications for a joint tenure-track faculty po-sition at the Assistant Professor level, to start August2012. The tenure home is in Computer Science. We areinterested in strengthening and complementing ourexisting broad program in Bioinformatics at the Uni-versity of Georgia. The candidate should have a Ph.D.or equivalent degree in Computer Science, Bioinfor-matics or a closely related field and a strong researchtrack record at the interface of computer science andbiology. The candidate will be expected to maintain arigorous, externally funded research program and tocontribute to undergraduate and graduate teachingin both Computer Science and Bioinformatics degreeprograms. For information about the breadth of theunits, see websites: http://iob.uga.edu and http://cs.uga.edu.

Applications should be uploaded to website:http://recruitment.franklin.uga.edu and should in-clude a cover letter, curriculum vitae, and brief state-ments of research and teaching interests. Three lettersof recommendation should be uploaded separatelyto the same website. The committee will begin re-viewing applications on November 15, 2011 until theposition is filled.

The Institute of Bioinformatics, the Department of ComputerScience, the Franklin College of Arts and Sciences, its many units,and the University of Georgia are committed to increasing thediversity of its faculty and students, and to sustaining a work andlearning environment that is inclusive. Women, minorities andpeople with disabilities are encouraged to apply. The University isan Equal Employment Opportunity/Affirmative Action Institution.

Translational Neuroscience: The University of Ari-zona seeks to hire a tenure-eligible physician-scientistat the ASSISTANT or ASSOCIATE level in Neu-roscience. Visit website: https://www.uacareertrack.com (Job Number 48123) or contact Search Com-mittee Chair, Carol A. Barnes, Ph.D. at e-mail: carol@nsma.arizona.edu. Affirmative Action/Equal OpportunityEmployer.

online

@sciencecareers.org

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencecareers.org414

Read inspiring stories of women working

in “Green Science” who are blending

a unique combination of enthusiasm for

science and concern for others

to make the world a better place.

This booklet is brought to you by the

AAAS/Science Business Office in partnership

with the L’Oreal Foundation

Download this free bookletScienceCareers.org/LOrealWiS

WOMEN IN SCIENCE

forging new pathwaysin green science

Download your free copy today at

ScienceCareers.org/booklets

Brought to you by theAAAS/Science Business Office

on

lin

e @

scie

nce

care

ers

.org

POSITIONS OPEN

The Department of Biological Sciences (website:http://biology.fiu.edu) at Florida International Uni-versity (FIU) is seeking applicants for a full-time,OPEN RANK FACULTY POSITION in molecu-lar biology to begin Fall 2012. The position is part ofthe strategic initiative of the School of Integrated Scienceand Humanity (SISH) to support the planned Insti-tute of Biomolecular and Biomedical Sciences (IBBS).The area of molecular biology is broadly defined and in-cludes research that will build the department_s strengthin genomics, proteomics, and biomedical applications.The successful candidate will have a strong record ofscholarly activity and external funding. Rank is open,but preference will be given to senior applicants whoare attracted to the challenges of playing a leadershiprole in building the IBBS and fostering its multidis-ciplinary environment. Advanced junior applicants withstrong research credentials and demonstrated poten-tial for obtaining external funding will also be consid-ered. A Ph.D. in the appropriate discipline is required.Interested candidates should apply online at website:http://www.fiujobs.org and reference Position Num-ber 35539. Please include curriculum vitae, list ofpublications, contact information for three potentialreferences, and a letter describing research interests. In-quiries may be addressed to Dr. Walter Van Hamme,Chair, Biomolecular Science Search Committee (e-mail:vanhamme@fiu.edu). Review of applications will be-gin on December 1, 2011, and continue until the po-sition is filled.

FIU is a multi-campus public research university lo-cated in Miami, a vibrant, international city.It has anewly established medical school that emphasizes med-ical research and community based medicine. FIU of-fers more than 180 baccalaureate, masters, professional,and doctoral degree programs to over 42,000 students.

FIU is a member of the State University System of Floridaand is an Equal Opportunity/Equal Access/Affirmative ActionEmployer.

TWO NEUROBIOLOGY POSITIONS: Aspart of a University-wide expansion in neuroscienceresearch, West Virginia University (WVU) invites ap-plications for two tenure-track positions at the AS-SISTANT PROFESSOR level in the Departmentof Biology, to begin August 2012. These individualswill join a rapidly-growing neuroscience communityat WVU, which numbers almost 50 full-time faculty(website: http://www.hsc.wvu.edu/wvucn/). Weseek a systems-level neurophysiologist and a compu-tational neuroscientist. The neurophysiologist will usea multidisciplinary approach in non-mammalian orinvertebrate systems to ask integrative questions in eithersensory or motor neurobiology. For the computa-tional position, we are especially interested in someoneworking at the level of cells, circuits, and/or systems.For both positions, we seek individuals with a broadbiology background and strong interdisciplinary skills.They will develop externally funded, independent re-search programs, and contribute to the undergraduateand graduate teaching missions of the department.Two years of postdoctoral experience, excellent writ-ten and oral communication skills, and the potentialto secure external funding are required. Qualified ap-plicants should submit research and teaching state-ments, curriculum vitae, and arrange for three lettersof reference to be sent to e-mail: wvubiology@mail.wvu.edu. Review of applications will commence onNovember 15 and continue until the position is filled.For more information about the position, contactJim Belanger, e-mail: jim.belanger@mail.wvu.edu.For more information about WVU and the city ofMorgantown, West Virginia website: http://www.as.wvu.edu/biology/faculty/positions.htm.

Women, minorities, and persons with disabilities are stronglyencouraged to apply, and the university is supportive of the needsof dual career couples. West Virginia University is an Equal Op-portunity/Affirmative Action Employer and the recipient of anNSF ADVANCE award for gender equity.

POSITIONS OPEN

MARINE GENETICIST

The Florida International University (FIU) is seek-ing applicants for a Marine Geneticist (rank open, seniorapplicants are encouraged) for a tenure-track positionin the Department of Biological Sciences (website:http://casgroup.fiu.edu/Biology/). He or she willparticipate in the Marine Science Program (website:http://casgroup.fiu.edu/marine/), a new and grow-ing interdisciplinary initiative emphasizing researchand teaching in coastal marine science. The MarineScience Program is housed in a recently completedbuilding in FIU Biscayne Bay Campus in the city ofNorth Miami, and currently is home to 12 researchlaboratories, two teaching laboratories, wet labs, amesocosm facility, and running sea water systems.The successful candidate will be expected to maintainan externally funded research program, supervise grad-uate students in our Ph.D. program, as well as teachundergraduate courses including General Geneticsand other courses in their areas of expertise. Areas ofprospective research foci may include but are not lim-ited to: genomics/phylogenetics, metagenomics, popu-lation genetics, ecological genetics, and environmentalmicrobiology. To ensure full consideration, applicationsshould be received by November 21, 2011. Screen-ing of applications will begin on that date and continueuntil a suitable candidate is selected. Applications willonly be accepted electronically as PDF files. Applica-tions should include a cover letter, curriculum vitae, asummary of research interests and teaching goals.Please submit applications to e-mail: trexlerj@fiu.edu.In addition, applicants should arrange for three let-ters of reference to be sent directly to the same e-mailaddress. Interested applicants are also required tosubmit their applications online via website: http://www.fiujobs.org reference Position Number 35538.FIU is a member of the State University System of Floridaand is an Equal Opportunity/Equal Access/Affirmative ActionEmployer.

RANK-OPEN FACULTY POSITIONin Bioinformatics

Florida International University

The Department of Biological Sciences at FloridaInternational University (FIU) (website: http://casgroup.fiu.edu/Biology/) is seeking applicantsfor an open-rank tenure-track position in Bioinfor-matics. The successful candidate will be expected tomaintain an externally funded research program, su-pervise graduate students in our Ph.D. program, aswell as teach undergraduate courses including genet-ics and other courses in their areas of expertise. Areasof prospective research foci may include but are notlimited to: genomics, metagenomics, proteomics, com-putational biology, systems biology, genome-wide as-sociation mapping, and phylogenetics. Candidatesutilizing theoretical approaches are encouraged to ap-ply. To ensure full consideration, applications shouldbe received by November 21, 2011. Screening of ap-plications will begin on that date and continue until asuitable candidate is selected. Applications will onlybe accepted electronically as PDF files. Interested ap-plicants should submit a (1) cover letter, (2) curricu-lum vitae, (3) statement of research interests, teachinggoals, and service interests, and (4) arrange to havethree or more references sent independently to Ericvon Wettberg at e-mail: eric.vonwettberg@fiu.edu.Interested applicants are also required to submit theirapplications online via website: http://www.fiujobs.org, reference Position Number 35537. FIU is a mem-ber of the State University System of Florida and is an EqualOpportunity/Equal Access/Affirmative Action Employer.

We delivercustomized job alerts.

www.ScienceCareers.org

POSITIONS OPEN

FACULTY POSITION – BIOCHEMISTRY

The Department of Chemistry and Biochemistry atAuburn University invites applications for a tenure-track, nine-month position in biochemistry at the levelof ASSISTANT PROFESSOR. The candidate isexpected to develop a vigorous, externally funded re-search program. Although we are particularly inter-ested in persons with expertise in biomacromolecularX-ray crystallography, we welcome applications fromindividuals with research interests in any area of bio-chemistry. Duties also include teaching at the under-graduate and graduate levels. A Ph.D. in chemistry orbiochemistry and at least one year of postdoctoral ex-perience are required. The candidate selected for thisposition must meet eligibility requirements to work in theUnited States on the date the appointment is scheduled tobegin (August 2012) and continue working legally forthe proposed term of employment. Excellent communica-tion skills required. Minorities and women are encour-aged to apply. Applicants should submit curriculumvitae, statements of research plans and teaching phi-losophy, and have three letters of reference sent elec-tronically to e-mail: bchesearch@auburn.edu or bymail to: Biochemistry Search Committee, Depart-ment of Chemistry and Biochemistry, AuburnUniversity, AL 36849-5312. Review of applicationswill begin November 28, 2011, and continue until theposition is filled. Auburn University is an Affirmative Action/Equal Opportunity Employer.

RANK-OPEN FACULTY POSITIONin Comparative Immunology

Florida International University

The Department of Biological Sciences at FloridaInternational University (FIU) is seeking a Compara-tive Immunologist (open rank) to fill a tenure-trackposition. All candidates must have postdoctoral expe-rience and strong publication and funding records.Senior candidates should have a demonstrated his-tory of leadership and consistent extramural fundingin the field of comparative immunology. The depart-ment has a broad range of faculty for potential col-laborative interactions (website: http://casgroup.fiu.edu/Biology/). Teaching expectations include ageneral undergraduate course in immunobiology andgraduate specialty courses. All application materialsshould be submitted electronically as PDF files to:Dr. Laurie Richardson, Chair, Comparative Immu-nology Search Committee, e-mail: laurie.richardson@fiu.edu and received by November 21, 2011. Eachapplication should include a cover letter, curriculumvitae, and a summary of professional and teaching in-terests. In addition, applicants should arrange forthree to four letters of reference to be sent directlyto the same e-mail address. Interested applicants arealso required to submit their applications online viawebsite: http://www.fiujobs.org reference Posi-tion Number 34952. FIU is a member of the State Uni-versity System of Florida and is an Equal Opportunity/EqualAccess/Affirmative Action Employer.

POSTDOCTORAL FELLOWSHIPS

The Geophysical Laboratory, Carnegie Institutionof Washington, invites applications for postdoctoralfellowships. The Geophysical Laboratory emphasizesinterdisciplinary experimental and theoretical researchin fields spanning geoscience, microbiology, chemistry,and physics. The Laboratory supports world-class fa-cilities in high-pressure research; organic, stable isotopeand biogeochemistry; mineral physics and petrology;and astrobiology.

Please see website: http://www.gl.ciw.edu/employment/Postdoctoral_Positions for informa-tion on the application process. Also, see website:http://www.gl.ciw.edu/ for a listing of personnel,current research interests, major facilities, and appli-cation information.

Completed applications for Carnegie fellowshipsshould be submitted by December 31, 2011, to:Russell J. Hemley, Director, Attn: Danielle Appleby,Geophysical Laboratory, 5251 Broad Branch Road,NW, Washington, DC 20015-1305, USA (e-mail:dappleby@ciw.edu). E-mail is preferred.

The Geophysical Laboratory is an Equal OpportunityEmployer.

on

lin

e @

scien

ceca

ree

rs.o

rg

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencecareers.org416

Visit our

ENHANCED

WEBSITE!

ScienceCareers.org

Science Careersis the forum that

answers questions.

Science Careers is dedicated to

opening new doors and answering

questions on career topics that

matter to you. We’re the go-to

career site for connecting with top

employers, industry experts, and

your peers. We’re the source for

the latest and most relevant career

information across the globe.

With community feedback and

a professional atmosphere, our

careers forum allows you to

connect with colleagues and

associates to get the advice

and guidance you seek.

Science Careers Forum:

» Relevant Career Topics

» Advice and Answers

» Community, Connections,and More!

Visit the forum and get your

questions answered today!

Your Future

Awaits.

Come work with us!

http://web.mit.edu

Faculty SearchesDepartment of Earth,Atmospheric and PlanetarySciencesApplications for all of the following positionsare being accepted at Academic JobsOnline (https://academicjobsonline.org/ajo).Please do not ask your referees to uploadletters at the time of application; letters willbe requested directly by MIT. To receiveconsideration, a completed applicationmust be received.

Faculty Positions – Climate-related FieldsThe MIT Department of Earth, Atmosphericand Planetary Sciences announces a majorexpansion of its activities in climate scienceand seeks applicants for up to three facultypositions in climate-related fields. Preferencewill be given to junior appointments at theassistant professor level, but a senior appointmentcan be considered for an individual withexceptional qualifications. Areas of specificinterest include observations, models andtheory of the atmosphere, ocean and cryosphere,and climates, biogeochemical cycles, andecology. Successful candidates must havea strong record of accomplishment in theirdiscipline, a strong commitment to teachingand student advising, a keen interest in relatingtheir work to complementary research in theDepartment and in the MIT/Woods Hole JointProgram in Oceanography. Joint appointmentswith other MIT departments are also potentiallynegotiable where appropriate.

A completed application will include acurriculum vitae, a statement of research andteaching objectives, and the names of fivepotential referees. More information about thisposition can be obtained by writing ProfessorKerry A. Emanuel at emanuel@mit.edu.

Tenure-Track Junior Faculty Position –ExoplanetsWe seek an outstanding scientist with interestin and potential for innovation and leadershipin teaching and research. The search is in thebroad area of exoplanets, including theory,observation, and instrumentation. However,we are especially interested in individualswhose research complements existing MITexpertise. The Department encourages stronginteraction with research programs within theEarth, Atmospheric, and Planetary Sciences.

A completed application will include acurriculum vitae, a one-page descriptionof research plans, and the names of threepotential referees. The deadline for thisopening will be December 31, 2011; allapplications received by this date will receivefull consideration. More information about thisposition can be obtained by writing ProfessorSara Seager at seager@mit.edu.

Junior Faculty Position –Sedimentary GeologyWe seek an individual with a field-basedobservational program, broad research interests,and a commitment to interdisciplinary studies.

Applicants should submit a curriculum vitae,one-page descriptions of research and teachingplans, and the names, email addresses, andphone numbers of three professional referees.Review of applications will take place beginningOctober 1, 2011, but will not continue pastMarch 1, 2012. Questions may be addressedto Prof. Leigh Royden, Search CommitteeChair, at lhroyden@mit.edu.

Search contact for all of the above-listedpositions: Mr. Michael Richard, HRAdministrator, EAPS, Massachusetts Instituteof Technology, 54-926, 77 MassachusettsAvenue, Cambridge, MA 02139-7307;mjr@mit.edu; 617-253-5184; 617-253-8298 (fax).

MIT is an Equal Opportunity/Affirmative Actionemployer. Applications from women andunderrepresented minority candidates areencouraged.

on

lin

e @

scie

nce

care

ers

.org

POSITIONS OPEN

FACULTY POSITION IN GENETICSDepartment of Biological SciencesThe George Washington University

The Department of Biological Sciences at the GeorgeWashington University is accepting applications for atenure-track faculty position at the rank of ASSIST-ANT PROFESSOR with expertise in the field of Ge-netics. We are searching for candidates who use geneticor genomic experimental methods to address funda-mental questions about gene expression or developmen-tal biology or who are able to integrate gene functionand regulation of complex traits from the genome tothe phenotype level. Research activities should beconducted within the context of comparative and/orevolutionary biology and complement the research fo-cus of the department. The successful candidate willbe expected to establish and maintain an externallyfunded research program that involves undergraduateand graduate students. Teaching responsibilities willinclude an undergraduate introductory course in ge-netics that includes a lab. Basic Qualifications: a Ph.D.in an appropriate discipline, postdoctoral experience,ability to teach basic genetics, and accomplishmentin biological research in genetics as demonstrated bypublications in peer-reviewed journals. Application Pro-cedure: to be considered please send electronically aletter of application, a complete curriculum vitae, briefdescriptions of teaching and research plans, three pub-lications, and the names and addresses of three peoplewho will be willing to submit letters of recommenda-tion. Applications should be sent to the search chairat e-mail: gwgeneticssearch@gmail.com. Only com-plete applications will be considered. Review of appli-cations will begin on November 18 and will continueuntil the position is filled.

The George Washington University is an Equal Opportunity/Affirmative Action Employer. The University Search Committeeseeks to attract an active, culturally, and academically diverse fac-ulty of the highest caliber.

FACULTY POSITION IN CELL BIOLOGYDepartment of Biological SciencesThe George Washington University

The Department of Biological Sciences at theGeorge Washington University is accepting appli-cations for a tenure-track position at the rank ofASSISTANT PROFESSOR in Cell Biology. Weseek broadly trained candidates working on problemsof cell-cell interactions including, but not limited to,host-pathogen interactions, symbiosis, developmentalprocesses, or neurobiology. Research activities shouldbe conducted within the context of comparative and/or evolutionary biology and complement the researchfocus of the department. Preference will be given tocandidates who integrate cellular analyses with modernimaging approaches. The successful candidate willbe expected to establish and maintain an externallyfunded research program that involves undergraduateand graduate students. Teaching responsibilities willinclude an undergraduate, introductory course in cellbiology and a second course within the candidate_sarea of expertise. Basic Qualifications: a Ph.D. in anappropriate discipline, postdoctoral experience, abilityto teach cell biology, and accomplishments in cell bi-ology as demonstrated by publications in peer-reviewedjournals. Application Procedure: to be considered pleasesend electronically a letter of application, a completecurriculum vitae, brief descriptions of teaching and re-search plans, three publications, and the names andaddresses of three people who will be willing to submitletters of recommendation. Applications should besent to the search chair at e-mail: gwcellbiolsearch@gmail.com. Only complete applications can be eval-uated. Assessment of applications will begin onNovember 18, 2011, and will be ongoing until theposition is filled.

The George Washington University is an Equal Opportunity/Affirmative Action Employer. The University Search Committeeseeks to attract an active, culturally, and academically diverse fa-culty of the highest caliber.

POSITIONS OPEN

FACULTY POSITIONin Animal Behavior (Invertebrate)Department of Biological SciencesThe George Washington University

The Department of Biological Sciences of the GeorgeWashington University is accepting applications for atenure-track faculty position at the rank of ASSIST-ANT PROFESSOR with expertise in the field ofAnimal Behavior, focusing on invertebrates. We aresearching for candidates who use comparative and/orexperimental approaches to investigate any aspect ofinvertebrate behavior, including but not limited to mat-ing, foraging, feeding, communication, learning, ag-gression, and cooperation. Applicants who examinethe genetic and/or neurological basis of behavior arealso encouraged to apply. Teaching responsibilitiesinclude an undergraduate course with laboratory inAnimal Behavior and a second course in their area ofexpertise. The successful candidate is expected to es-tablish and maintain a vigorous research programcapable of attracting external funding that involvesgraduate and undergraduate students and will add toour growing departmental strength in environmentalbiology and ecology. Basic Qualifications: a Ph.D. inan appropriate discipline, postdoctoral experience, abil-ity to teach animal behavior, and accomplishments inbiological research in animal behavior demonstratedby publications in peer-reviewed journals. Applica-tion Procedure: to be considered please send electron-ically a complete curriculum vitae, brief descriptionsof teaching and research plans, three publications,and the names and contact information for three ref-erences to the search chair, Dr. John Lill, at e-mail:gwbehaviorsearch@gmail.com.

Only complete applications will be considered. Re-view of applications will begin on November 18, 2011,and will continue until the position is filled.

The George Washington University is an Equal Opportunity/Affirmative Action Employer. The University Search Committeeseeks to attract an active, culturally, and academically diverse fac-ulty of the highest caliber.

UNIVERSITY OF CALIFORNIA, IRVINE-FACULTY POSITION IN ATMOSPHERICCHEMISTRY. The Department of Chemistry invitesapplications for a tenure-track position as ASSIST-ANT PROFESSOR in the area of Atmospheric Chem-istry. We are seeking an outstanding Ph.D. who willestablish a vigorous research program in areas, whichwill complement existing strong programs in atmo-spheric chemistry within the Department of Chem-istry, and preferably also take advantage of a breadthof related programs and collaborations on campus(seewebsite: http://airuci.uci.edu/). Typical (but notexclusive) areas include the development of new an-alytical methods, studies of the fundamental chemistryand photochemistry of organic or inorganic atmo-spheric systems, and molecular level studies of inter-actions of outdoor or indoor atmospheric species withbiological or energy-related systems. The ability toeffectively teach chemistry at the undergraduate andgraduate levels is required. Applications must be sub-mitted electronically via the Internet atwebsite: https://recruit.ap.uci.edu. Applicants should upload a coverletter, curriculum vitae (including publication list), aconcise statement of research plans, and the namesand contact information for at least three references.Applications and supporting materials should bereceived by November 30, 2011 for full consideration.The University of California, Irvine is an Equal Opportunity/Affirmative Action Employer committed to excellence through di-versity and strongly encourages applications from all qualified ap-plicants, including women and minorities. UC Irvine has an activeADVANCE Gender Equity Program.

↓Find your future here.

www.ScienceCareers.org

POSITIONS OPEN

COMMUNITY ECOLOGIST

The Department of Biological Sciences at BowlingGreen State University (BGSU) seeks applicants for anASSISTANT PROFESSOR level, tenure-trackCommunity Ecology position. Applicants with post-doctoral experience and research expertise that com-plements existing strengths in evolution, population,and conservation ecology are preferred. Successful can-didates are expected to develop a highly productive,externally funded research program, as well as contrib-ute to the teaching missions of our Ph.D./M.S. pro-gram (È80 students) and undergraduate program,which includes a Specialization in Ecology and Con-servation Biology. To apply, electronically submit cov-er letter, curriculum vitae, statements of research plansand teaching philosophy/experience, representativepublications, and three reference letters to e-mail:dmclean@bgsu.edu or by mail to: Community Ecol-ogist Search Committee, Department of BiologicalSciences, BGSU, Bowling Green, OH 43403-0208by November 30, 2011. Contact Helen Michaels ate-mail: hmichae@bgsu.edu for additional informa-tion. Information about our department can be foundat website: http://www.bgsu.edu/departments/biology. BGSU is an Affirmative Action/Equal OpportunityEmployer/Educator and encourages applications from women,minorities, veterans, and persons with disabilities.

TENURE-TRACK POSITION inTheoretical or Computational Biophysics

Department of BiophysicsJohns Hopkins University

The Thomas C. Jenkins Department of Biophysicsseeks candidates for a tenure-track, ASSISTANT PRO-FESSOR position in theory and simulation of bio-logical systems. Areas of special interest include, butare not restricted to the application of theory andcomputation to biological macromolecules and as-semblies, cell dynamics, cellular physiology, or applica-tions of statistical thermodynamics, polymer physics,and physical chemistry in biology, systems biology,and biological networks. Applicants should electron-ically send a single PDF file with a cover letter, cur-riculum vitae, and a brief description of research plansto e-mail: biophysicssearch@jhu.edu. Three lettersof recommendation should be sent to this e-mail ad-dress or to: Faculty Search Committee, T. C. JenkinsDepartment of Biophysics, Johns Hopkins Univer-sity, 3400 N. Charles Street, Baltimore, M.D. 21218-2685; telephone: 410-516-7245. All materials shouldbe received by November 30, 2011.

Johns Hopkins University is an Affirmative Action/EqualOpportunity Employer.

More scientists agree—we are the most useful website.

√

www.ScienceCareers.org

MARKET PLACE

Promab Biotechnologies Inc.

Custom MonoclonalAntibody $4,200

>3,000 CLONESWILL BE SCREENED

1-866-339-0871

www.promab.com info@promab.com

on

lin

e @

scie

nce

care

ers

.org

21 OCTOBER 2011 VOL 334 SCIENCE www.sciencecareers.org418

1200 New York Ave., NW • Washington, DC 20005 • +1 202-326-6417 • ScienceTranslationalMedicine.org

AAAS member price:

US$52 online only

US$225 print and online (add US$100 for non-US delivery)

AAAS membership number required ______________________________

Nonmember price:US$156 online onlyUS$465 print and online (add US$100 for non-US delivery)

Name ___________________________________________

Address _________________________________________

_________________________________________________

City _____________________________________________

State/Province ____________________________________

Zip/Postal Code ___________________________________

Country __________________________________________

E-mail ________________________________________

Phone ___________________________________________

Payment

Check (payable to AAAS – Science Translational Medicine)

Mail check and this form to:

AAAS

Attn: Membership Department

1200New York Avenue, NW

Washington, DC 20005 USA

Charge my:

VISA MasterCard American Express

Card Number________________________________________

Expiration Date______________________________________

Signature___________________________________________

Date _______________________________________________

If paying by credit card, youmay FAST FAX your order to+1 202-842-1065

Subscribe now for 1 year’s access to Science Translational Medicine

(required for subscription activation)

Science Translational Medicine, the newest weekly journal from AAAS,is focused on applications of basic research knowledge to improvehuman health.

The goal of Science Translational Medicine is simple: help the scientificcommunity harness decades of progress in research at the basic leveland translate these biological discoveries into medical advances.

Science Translational Medicine publishes:• Peer-reviewed primary research papers

• Perspectives and reviews on research from basic scienceand clinical viewpoints

• Survey of recent literature and findings in other journals

• Commentary on policy, funding, regulatory issues, and more.

As a AAAS member, add Science Translational Medicineaccess for over 60% off of the regular price. Subscribein any of these ways:

• Go to ScienceTranslationalMedicine.org

• Call +1 202-326-6417

• Mail or fax this form with yourpayment to +1 202-842-1065

ScienceTranslationalMedicine

ScienceTranslationalMedicine.org

Weekly journal integrating science and clinical medicine

Prices valid until December 31, 2011.

Sitewide access available foryour institution. Recommend asubscription to your library today.ScienceOnline.org/recommend

Our stock is growing.With 670 sandwich ELISA kits now available covering10 species, you havemore choices under one roof.

More kits, more choices! Raybiotech now carries ELISA kits for human, rhesus macaque, mouse,

rat, cotton rat, pig, cow, horse, cat, and dog target proteins.

• Pre-coated strip plates

• Reproducible and reliable

• Ideal for validation of antibody array data

• Extensively validated in publications

• Money back guarantee

ELISA testing service available!

Discover more. Publish faster.

90-day money-back guarantee. Applicableto U.S. customers only. For details, visit:www.raybiotech.com/risk_free_sci.asp

Try any RayBio® ELISA kit

RISK-FREE!