Luminescent Quantum Dots

Investigating Biological Processes at the Single Molecule LevelUsing Luminescent Quantum Dots

THOMAS PONS1 and HEDI MATTOUSSI2

1Laboratoire Photons et Matie`re, CNRS UPR5, ESPCI, 10, rue Vauquelin, 75005 Paris, France; and 2Division of OpticalSciences, US Naval Research Laboratory, Washington, DC 20375, USA

(Received 16 September 2008; accepted 12 May 2009; published online 12 June 2009)

AbstractIn this report we summarize the progress made inthe past several years on the use of luminescent QDs to probebiological processes at the single molecule level. We start byproviding a quick overview of the basic properties ofsemiconductor nanocrystals, including synthetic routes, sur-face-functionalization strategies, along with the main attri-butes of QDs that are of direct relevance to single moleculestudies based on uorescence detection. We then detail somevaluable insights into specic biological processes gainedusing single QDs. These include progress made in probingbiomolecular interactions, tracking of protein receptors bothin vitro and in live cells, and single particle resonance energytransfer. We will also discuss the advantages offered andlimitations encountered by single QD uorescence as aninvestigative tool in biology.

KeywordsSemiconductor, Nanocrystal, Quantum dot,

Single molecule, Fluorescence, FRET, Biosensing, Hybrid-

ization, Molecular motors.

INTRODUCTION

Semiconductor nanocrystals, or quantum dots(QDs), have recently emerged as a new set of uoro-phores that can enhance biological assays, uorescencedetection and imaging.11,45,53,64,65,67,69,93,95,104 This isdue to some of their unique optical and spectroscopicproperties, which are often not matched by mostconventional uorophores, such as organic dyes anduorescent proteins. QDs exhibit high uorescencequantum yields, high photobleaching thresholds, and apronounced resistance to photo- and chemical-degra-dation.45,67,69,95,104 They also exhibit narrow and tun-able emission bands along with broad excitationspectra and high extinction coefcient (~10100 timeslarger than most dyes). These properties allow the

exibility to excite them efciently far from theiremission peaks. There has been a tremendous interestin using them to develop a variety of biological assaysand for sensor design. These include imaging of liveand xed cells, imaging of tissue, immunoassays andenergy transfer based assays. Energy transfer basedsensing has in particular expanded in the past fewyears, with sensors developed for the detection of smalland large molecule targets, hybridization interactions,and enzyme digestion.11,53,64,65,67,93,95

Ensemble measurements are macroscopic in natureand they provide information about average propertiesof samples such as size, conformation or orientation ofproteins. In comparison, single molecule measure-ments are able to resolve molecular scale heterogene-ities and the fate of individual molecules. Fluorescencedetection applied to single molecules can allow accessto valuable molecular scale information, and it hasbecome one of the most commonly used single mole-cule techniques in biological research.18 Furthermore,recent progress in optical instrumentation and thedevelopment of highly sensitive detection tools (such asAPD and CCD) allowed easier detection and resolu-tion of uorescence from single uorophores. This hasin turn allowed the development of a variety of singlemolecule assays to study ligand-receptor binding,changes in the conformation of macromolecules (e.g.,proteins, short and large oligonucleotides), and singlemolecule diffusion and transport.18,106 Outside bio-physical research, single molecule uorescence wassuccessfully applied to individual colloidal QDs. Itallowed access to truly unique and remarkable infor-mation about the uorescence of single QDs that werenot accessible from ensemble measurements. In par-ticular, two unique properties distinguish luminescentQDs from most conventional uorophores: (1) singleQD exhibit very narrow PL spectra compared to thoseaveraged for macroscopic samples (FWHM ~ 15 nmvs. 3040 nm for ensemble spectra at room tempera-ture) and (2) the uorescence emission of isolated

Address correspondence to Hedi Mattoussi, Division of Optical

Sciences, US Naval Research Laboratory, Washington, DC 20375,

USA. Electronic mail: [email protected]

Annals of Biomedical Engineering, Vol. 37, No. 10, October 2009 ( 2009) pp. 19341959DOI: 10.1007/s10439-009-9715-0

0090-6964/09/1000-1934/0 2009 US Naval Research Laboratory

1934

nanoparticles under continuous excitation is intermit-tent in nature (blinking).24,26,49,50,74

The unique optical properties of QDs can be par-ticularly benecial for single molecule uorescencemeasurements. Due to their large extinction coecient,narrow emission spectra, and their resistance to photo-degradation, QDs may be individually detected withhigh signal-to-noise ratios and for extended periods oftime (several minutes under sustained irradiation), afeature that is not available to traditional organic dyes.This can permit easy discrimination of dierent singleQD colors. This also makes them particularly suitablefor multiplexed single molecule uorescence imaging.As such, single molecule uorescence has emerged asone of the major aspects of employing these uoro-phores in biology.16,56

In this report, we start with a brief description of themost eective synthetic procedures to prepare highquality colloidal nanocrystals, outline the commonlyused surface functionalization techniques, and describethe most pertinent photo-physical properties of use tobiology. We then review the progress that use of singleQD uorescence has allowed in biology, with a focuson a few representative examples where unique andvaluable insights into specic biological processes havebeen gained. These include progress made in probingbiomolecular interactions, tracking of protein recep-tors in vitro and in live cells and single particle energytransfer. We will also discuss the advantages offeredand limitations encountered by single QD uorescenceas an investigative tool in biology.

SYNTHESIS, SURFACE-FUNCTIONALIZATIONAND PHYSICAL PROPERTIES

OF COLLOIDAL QDS

Synthesis of Dispersible and Highly Luminescent QDs

Luminescent QDs that have found most use inbiology are colloidal in nature, and they are preparedusing solution phase reactions. The rst solution-phasegrowth of the nanoparticles was realized within inversemicelles.38,90 However, the major advance in solution-phase synthesis took place in 1993, when Bawendi andco-workers showed that pyrolysis of organometallicprecursors can provide high quality CdSe nanocrystalsthat have crystalline cores, narrow size distribution(~10% or less) and exhibit relatively high photoemis-sion quantum yields.71 The rst demonstration of thisreaction scheme employed the rapid injection of dim-ethylcadmium (CdMe2) and trioctylphosphine selenide(TOP:Se) mixed with trioctylphosphine (TOP) into ahot (280300 C) coordinating solution of trioctyl-phosphine oxide (TOPO). One of the key aspects of

this synthesis route was that colloidal QDs couldreproducibly be made to exhibit narrow emission withlow defect contributions and relatively high roomtemperature quantum yields (QY ~ 510%). This hasallowed performance of several viable photophysicaland structural characterizations (using for exampleabsorption, uorescence spectroscopy and scatteringtechniques).61,71,75 It has further raised the potentialfor technological applications based on QDs (e.g.,LEDs and photovoltaic devices).13,32,43,63,70

In subsequent studies, Peng and co-workers furtherrened this reaction scheme and showed that severalprecursors which are less pyrophoric and less volatilethan CdMe2 could potentially be used for preparinghigh quality colloidal nanocrystals of CdS, CdSe, andCdTe.80,86 Other groups quickly followed up on Pengswork, and those efforts combined have outlined theimportance of impurities (usually acids coordinating tothe metal precursors) in the reaction progress. In thisroute, high purity TOPO and controlled amounts ofmetal coordinating ligands and metal precursors, suchas CdO, Cd carboxylates, phosphonates, or acetyl-acetonates, were used to synthesize various Cd-basednanocrystals, as schematically represented inFig. 1.80,86,89 The high temperature synthetic route wasalso applied to making near-IR emitting PbSe QDsinitially by Murray and co-workers, and then by othergroups, using oleic acid and Lead(II) acetate trihydrateor lead oxide.19,72,103,110

To further take advantage of the progress made inimproving the quality of QDs achieved using hightemperature synthesis and to improve the luminescencequantum yield, researchers applied the concept of bandgap engineering developed for semiconducting quan-tum wells to the growth of colloidal nanocrystals (seeFig. 1).48 It was rst demonstrated by Guyot-Sionnnest and co-workers that overcoating CdSe QDswith a layer of ZnS could improve the PL quantumyields to values of 30%.39 Following that two com-prehensive reports by Dabbousi et al.15 and Penget al.79 detailed the complete reaction conditions forpreparing a series of CdSeZnS and CdSeCdS coreshell nanoparticles that are strongly uorescent andstable. We should add that air-stable precursors havealso been used for the overcoating reactions, followingthe reports of synthesis of CdS, CdSe nanocrystals.54,89

One of the issues associated with overcoating QDs withZnS shell to improve the PL yield is the crystal latticemismatch between core and shell materials, which canproduce non-homogenous shell structure due to strain-induced defects. To potentially address this problem afew groups have grown multilayer shells on the nativecore (e.g., CdSeCdSZnS)105; this progressivelyadapts the crystalline lattice parameters of the ov-ercoating shell to those of the core. We should also

Investigating Biological Processes at the Single Molecule Level Using Luminescent QDs 1935

emphasize that QD cores have been synthesized withvarious materials combinations, including II-VI (e.g.,CdSe, CdS, CdTe, ZnSe) and III-V (e.g., InP,InAs) semiconductors and their alloys.77 However,thus far only CdSe-based QDs have provided both lowsize dispersity and high QYs; these have been used inall single QD studies presented in this review.

Water-Solubilization Strategies

Since publication of the rst reports on developingcolloidal QDs as biological labels, several strategiesaimed at developing stable water dispersions of lumi-nescent QDs have been developed. These strategies canbe loosely divided in two main categories: ligand-exchange and encapsulation within block copolymersor phospholipid micelles (see Fig. 2).

Ligand Exchange

Water-solubilization via ligand exchange involvesreplacing the native hydrophobic surface ligands(mainly TOP/TOPO) with bi- and/or multi-functionalhydrophilic ligands. The hydrophilic ligands are madeof metal-coordinating anchor group(s) (often thiol-based groups) at one end for binding to the nano-crystal surface. At the other end the ligands presenthydrophilic groups (such as carboxyl or poly(ethyleneglycol)) that promote anity to aqueous solutions.

Water-transfer via cap-exchange with small ligandshas been attempted by several groups in the past dec-ade, because of its ease of implementation and thecommercial availability of several of these ligands. Abenet of this strategy is that the resulting hydrophilicQDs are compact in size,85 a property that can beextremely important for applications such as intracel-lular transport.67 However, the stability of the resultingQDs in buffer solutions is often poor. For instance,thiol-alkyl-carboxyl ligands (including mercaptoaceticacid MAA, and mercapto-undecanoic acid, MUA)produce nanocrystals that tend to progressivelyaggregate in physiological conditions within a rela-tively short period of time. A concomitant decrease inthe uorescence quantum yield is also measured. Thisis most likely due to desorption of the ligands from theQD surface. Peng and co-workers attempted to resolvethis problem by preparing hydroxyl-terminated den-dron ligands that were also end-functionalized withthiol groups to promote anchoring on the QD surface,then later used amine-terminated dendrimers for capexchange; the latter allowed cross-linking of theligands to form dendron-boxed QDs.33,101 To addressthe long-term stability of QDs prepared via cap-exchange our group relied on the use of bidentate thiolligands such as dihydrolipoic acid (DHLA).62

However, most of the commercially available thiol-terminated ligands (MAA, MPA, MUA as well asDHLA) rely on the deprotonation of the carboxyl

Selenium precursor: TOP:Se

TOPOC16H33NH2

CdSeCadmium Precursors:Cd(acac)2Cd(acetate) 340~350 C

HDDO

TOP / TOPO

CdO

140~180 C

ZnEt2 + TMS2S

ZnS

CdSe

(b)(c)

(d)

(a)

FIGURE 1. (a and b) Schematics of the stepwise synthesis route based on high temperature reaction of organometallic precur-sors to prepare CdSeZnS coreshell QDs. (c) Representative absorption and emission spectra for a set of CdSe core only QDsprepared using the high temperature method. (d) A typical high resolution TEM image of CdSe QDs having a diameter of ~40 A.

T. PONS AND H. MATTOUSSI1936

groups terminal groups to achieve dispersion in water,and this has limited ones ability to prepare homoge-neous QD dispersions at acidic pHs.62 We haverecently developed a new set of DHLA ligand deriva-tives that can potentially allow one to overcome theseconstraints/limitations. This relied on appendingpoly(ethylene glycol) (PEG) segments with variablechain-lengths onto dihydrolipoic acid.68,99 QDs cappedwith DHLAPEG ligands are well dispersed in aque-ous media and stable over an extended period of timeand over a relatively broad pH range. Subsequently,we demonstrated that inserting reactive groups at thelateral end of the DHLAPEG ligand (such carboxylor amine) could permit easy implementation of knowncoupling reaction to attach QDs to proteins andpeptides.58,96,97 With the ligand exchange strategy, thenature of the anchoring group to the QD surface(e.g., monodentate vs. multidentate) can seriouslyaffect the long-term stability of the hydrophilic QDs.

In a somewhat dierent rationale, Weiss andco-workers have used phytochelatin-related peptidesterminated with a few cysteine groups for anchoring onthe QDs.81 This clever strategy also relies on the multi-dentate thiol interactions with the inorganic surface ofthe nanocrystals, which improves the stability of the

ligand/QD system. Moreover, using peptides as theactual ligands can potentially allow direct biologicaltargeting. Rather large amounts of peptide ligands areneeded to carry out an effective cap exchange, becausethis process naturally requires a large excess of ligands.

In a few recent studies, amine-rich polymers havebeen employed to functionalize the QD surfaces, and ithas been proposed that for these polymers, watertransfer to aqueous media is realized via cap exchangeof the native ligands. Systems used include amphiphilichyperbranched polyethyleneimine,73 poly(ethyleneglycol)-grafted polyethylenimine,20 and poly(ethyleneglycol-b-2-N,N-dimethylaminoethyl methacrylate)(PEG-b-PDMA) diblock copolymer.100 Cap exchangewith thiol-terminated silane ligands, followed by fur-ther cross-linking to form a silica shell coating was alsoutilized as a means to promote hydrophilicity ofCdSeCdS(ZnS) coreshell nanocrystals.6,30

Encapsulation within Block Copolymers and Phospho-lipid Micelles

This strategy relies on the use of bifunctionalamphiphilic polymers having hydrophobic segments,which selectively interact (and interdigitate) with the

ZnS

CdSe

ZnS

CdSeCdSe CdSe

Hydrophilic groups

Hydrophobic S S

OO O OH

n

groups

O O

OOH

n

O

OH

O O

OOH

n

O

OH

SS

O O

S S

OO O OH

nS

SOSO

O

HO

S

OOn

ZnSS S

O O

S S

OO O OH

nS

SOSO

O

HO

S

OOn

ZnS ZnSO

OO

OHn

SS

OO

SSSS

OOOHO

n

SS

OO

n

SS CdSe O

OO

OHn

SS

OO

SSSS

OOOHO

n

SS

OO

n

SS CdSe CdSe

O

O

HOn

OO

OHO n

On O

O

HOn

OO

OHO n

On

FIGURE 2. Schematics of the two main strategies used for surface-functionalization and transfer to water of colloidal QDs usedto-date. (Left) Representative example of the cap exchange strategy using polyethylene glycol-appended dihydrolipoic acid(DHLAPEG). (Right) Encapsulation within an amphiphilic block copolymer.


native TOP/TOPO shell, and hydrophilic units thatpromote QD dispersion in aqueous media.28,57,78,104,109

In general achieving a controlled balance of hydropho-bic and hydrophilic blocks within the polymers is nec-essary for its ability to promote water solubility of theQDs. In one of the early uses of this strategy, Wu et al.used a block copolymer shell consisting of 40% octyl-amine-modied polyacrylic acid.104 Pellegrino et al.used poly(maleic anhydride alt-1-tetradecene) for QDencapsulation.78 Following adsorption on the TOP/TOPO QDs, further addition of bis(6-aminohexyl)amine formed cross-linked polymers on the QDs, andsubsequent hydrolysis of the unreacted anhydride unitswas shown to make the QDs water soluble.

The second encapsulation method employedphospholipid derivatives which formed micelles. Inparticular, the rst demonstration of this strategy wasreported by Dubertret and co-workers who usedphospholipids containing 40% 1,2-dipalmitoyl-sn-gly-cero-3-phosphoethanolamine-N-[methoxy(polyethyleneglycol)-2000] (mPEG-2000 PE) and 60% of 1,2-dipalmitoylglycero-3-phosphocholine (DPPC) to carryout encapsulation of CdSeZnS QDs and their transferto buffer solutions.21,27 This technique is simple toimplement as most phospholipids are commerciallyavailable, and like polymer encapsulation can yieldwater-soluble QDs with QYs higher than thoseobtained via ligand exchange. This strategy has someof the stability issues at very low concentrationsencountered by micelles in general. In addition, it is

possible that more than a single nanocrystal could becontained within each micelle; nonetheless, it has beenrecently shown that these can be removed usingultracentrifugation or high pressure liquid chroma-tography.7 Overall one of the constraints of usingencapsulation is that the overall hydrodynamic size ofthe nanoparticles becomes inevitably large.

Photo-Physical Properties of Colloidal QDs

Semiconductor QDs are neither wholly atomic norbulk semiconductors; instead they are nanoscaleassemblies of atoms (1001000 s depending on the size).The spatial connement of intrinsic electron and holecarriers to the physical dimensions of the nanocrystal(quantum connement eects) imparts on these mate-rials novel electronic properties not exhibited by theirbulk parent materials or individual atoms/molecules.These connement eects manifest when the nanocrys-tal size becomes comparable to, or smaller than, the bulkBohr exciton radius.29,47 The best known andunderstood connement effect is the widening of theenergy band gap with decrease of nanocrystal size. Thismanifests itself as a blue shift of the rst absorption peakand the photoluminescence maximum with decreasingparticle size, along with the appearance of discrete en-ergy states in both the valence and conduction bands. Italso manifests in the appearance of tunable (by size andcomposition) photoemission properties, as shown inFig. 1 for CdSe-based QDs and Fig. 3 for other QD

)

CdTeCdS PbSe

banc

e (a.

u.

ance

(a.u.

)

Abso

rb

Abso

rba an

ce (a

.u.)

Abso

rba

500 600 700 1000 2000 3000Wavelength (nm) Wavelength (nm) Wavelength (nm)

300 400 500

UV< 250 2500 >IR400 700Visible (nm)

ZnSeCdS

CdTe PbSe/TeCdSe/Te CdHgTe Alloys

ZnS CdSe PbS

FIGURE 3. (Top) (Top) Representative absorption spectra for CdS, CdTe, and PbSe QDs prepared using high temperature solutionreaction. The shift in the position of the first absorption peak reflects increase in the nanocrystal size. (Bottom) PL emissionwindows for various IIVI, IIIV and hybrid IIIVI core materials. Figures partially reproduced from Medintz et al.,67 Peng and Peng,80

and Yu et al.,110 with permission from the American Chemical Society and NPG.


materials. Resolution in the excited state energy levelsand separation between the valence and conductionband levels depends on the width of the nanocrystal sizedistribution and the type of semiconducting materialsused.

In addition to the size and composition dependentoptical and spectroscopic properties, colloidal lumi-nescent QDs have another characteristic that distin-guishes them from organic uorophores: intermittentphotoemission (blinking) of individual nanocrystals.The photoluminescence of isolated QDs displays analternation of on (emitting) and o (dark) peri-ods, which durations follow a heavy-tail power lawdistribution with time.49,50,74 This implies that indi-vidual QDs can experience dark periods, where noemission can be collected. This feature is especiallyimportant for single molecule detection, as a QD canperiodically disappear from the observation eld.PL blinking of single QDs has attracted much inter-est, because its underlying mechanisms are not fullyunderstood.50 Two recent studies have shown thatCdSe QDs coated with thick CdS shells can exhibit adrastic reduction in the blinking rate.10,59 One mayexpect that these improvements will soon produceQD probes that are better suited for biologicallyrelevant single molecule studies. These propertiescombined with the high quantum yields realized forcoreshell QDs and the very high extinction coef-cient constitute the most important properties for usein biology.

Finally, the size of a QD varies from ~10 to~50 nm, depending on the type of solubilizationtechnique used.67,85 It is larger than organic uoro-phores but comparable to that of uorescent proteins.The relatively large size allows the ability to conjugateseveral biomolecules on the surface of a single QD,which can be advantageous in certain instances.Enhanced target afnity and higher rates of energytransfer can be realized with multifunctionalQD-conjugates. However, the large size of hydrophilicQDs and their conjugates can inuence their diffusionin biological media and may affect their ability toaccess certain cellular compartments. In addition,multivalency may be a limitation in instances when astrict one-to-one QDbiomolecule conjugation isrequired.

It should be noted that despite the progress made indesigning reproducible techniques to achieve theeective transfer of luminescent QDs and an array ofother inorganic nanoprobes to buer media, there arestill several issues associated with the use of QDs inbiology. These issues stem from problems such aslimited temporal stability to changes in pH for severalnanocrystals prepared using cap exchange with mono-thiol-alkyl-COOH ligands and some of the polymer

encapsulation approaches developed in the past dec-ade. The limited pH stability can often producenanocrystals that experience aggregation build up inlive cells and in blood. pH limitation in general stemsfrom sensitivity of the capping/encapsulating layerused to promote hydrophilicity of the nal nano-crystals. Insertion of PEG oligomers have alleviatedsome of these issues in some of the latest stud-ies.58,96,97,109 The wide range of size variation ofhydrophilic QDs has also been a serious problem forapplications requiring compact nanocrystals and theirconjugates. Batch to batch variation of commerciallyavailable QDs have also been reported in variousstudies. There have been sustained efforts by severalgroups to address these problems and eventuallyproduce nanocrystals that can allow one to takeadvantage of some the unique properties of theseprobes.

SUMMARY OF SOME OF THE PROGRESSMADE USING QDS FOR ENSEMBLE

MEASUREMENTS

There has been an array of biological assaysdeveloped in the past 8 years where researchers havetried to exploit the unique properties of luminescentQDs. These developments can essentially be sum-marized in three areas: use for live cells imagingwhere easy to implement multiplexing capabilities,slow degradation and high photo-bleaching thresholdwere demonstrated. These range from multicolorlabeling of xed cells using QDantibody conjugates,investigating specic biological processes that requireextended observation, to probing membrane specicprocesses such as T cell stimulation.2,21,45,87,104 Thevery high two-photon action cross-section of QDshas been effectively used to realize deep tissueimaging with reduced IR excitation and very weakauto-uorescence contribution.51 In somewhat simi-lar rationale, one-photon uorescence of NIR emit-ting QDs was also used to carry out real timemapping of sentinel lymph nodes in live animalsduring surgery, where nodes as deep inside the tissue(~100 mm1 cm deep) were imaged in real time usingrelatively low excitation intensity.46 Use of QDs forin vitro sensor development based on uorescencedetection and energy transfer has also witnessed asubstantial development. They have been used forsingle and multiplexed immuno-sensing of solubletoxins.31 Energy transfer based assays have been usedto detect DNA hybridization, recognition of fullenzymes by QDaptamer conjugates,53 enzymecleavage of peptides,65,93 and glycosylation levels inproteins.76


USE OF SINGLE QDS FOR THE IN VITRODETECTION OF BIOMOLECULAR

INTERACTIONS

Because of the very narrow PL spectrum of indi-vidual QDs, the proportion of photons that arerejected by the narrow band detection lters issubstantially reduced, and this simplies the spectraldeconvolution of signals from dierent QD colors.This could enhance the signal-to-noise ratio and can bevery benecial to single molecule assays, where thenumber of detected photons is inherently low. Theability to resolve single QD emission for extendedperiods of time will undoubtedly allow higher detec-tion sensitivity and data collection over long timeintervals.

Co-Localization of Distinct QD Probes

This conguration has been used to develop multi-plexed assays, where interactions (and binding) of twoQD-bioconjugates emitting at dierent spectral win-dows to dierent sites on the same target molecule canbe simultaneously detected. A typical example caninvolve two distinct emission QDs conjugated to twodierent DNA probes, mixed in a solution with a

target oligonucleotide that contain complementarysequences for both probes. The target DNA hybridizessimultaneously with the two QDDNA conjugates(dual hybridization). If immobilized on a substrate,this dual hybridization could then be detected opticallyusing wide eld uorescence microscopy, where thesignals from the two QD probes are simultaneouslyimaged.

Wang and co-workers40 applied this strategy to carryout a genetic analysis of anthrax pathogenicity. Ingeneral, a positive identication of this bacteriumrequires the simultaneous detection of three distinctgenes, namely rpoB, pagA, and capC in the samesample. Three pairs of target-specic DNA probes wereconjugated (combinatorially) to three distinct colorQDs emitting at 525, 605, and 705 nm, respectively. Astargets, they used three synthetic oligonucleotides, eachderived from conserved sequences from each of thethree anthrax-related genes; these were used as simu-lated targets (in a background of E. coli genomic DNA)for analysis. They rst used the simultaneous uores-cence signals in the combined pseudocolors of thespots, namely, indigo, magenta, and orange, as anindication of the presence of rpoB, pagA, and capC,respectively (Fig. 4a). They then carried out four sets ofexperiments, where the specicity of the assay was

FIGURE 4. Simulated multiplexed analysis of anthrax-related genetic targets: (a) color set for the three pairs of target-specific QDnanoprobes and their resulting co-localized fluorescent images upon hybridization. (b) Four samples containing different com-binations of the three targets, rpoB, capC, and pagA. Checks represent the existence of certain target sequences. Sample IV is thenegative control. (c) Fluorescent images I, II, III, and IV correlate with samples I, II, III, and IV, respectively (Bar dimension is 1 lm).Figure partially reproduced from Ho et al.,40 with permission from the American Chemical Society.


essentially dened by its capacity to correctly identifyand quantify the individual target sequences in a com-plex mixture of various target combinations (Fig. 4b).In the rst step, uorescent spots in all the three com-bined pseudocolors were detected, which indicated thepresence of the three targets in the solution (Fig. 4c, I).In the next two experiments, they found that the pres-ence of two (Fig. 4c, II) or only one (Fig. 4c, III) targetcould also be unambiguously determined (i.e., withoutfalse positive or negative). In the fourth negative con-trol sample, only uorescent spots stemming fromunhybridized blue, green, and red QD probes wereobserved (Fig. 4c, IV); in this sample no specic targetwas present. A similar assay has been reported usingQDantibody conjugates binding different sites of thesame target protein.1

Alternatively, the conjugate dispersion could beinjected into a small glass capillary and the QD signalscould be detected sequentially as the conjugates passthrough the observation volume. In both cases, thepresence of the target could be inferred from theco-localization of two distinct single QD emissionsignals, because the size of the total targetDNAprobeQD complex is smaller than the optical resolu-tion of the detection system. This solution-phasemethod is fast (solution vs. surface immobilizedreagents) and does not require separation of unboundprobes; these are usually not co-localized. It allowsidentication of samples without prior polymerasechain reactions (PCR) amplications, which are notquantitative and can introduce non-negligible errors. Italso allows detection of low abundance targets and hasa higher sensitivity than microbead assays in whichweak target signals may be dicult to discriminateagainst the strong microbead coding signal. The sen-sitivity of this assay format has been demonstratedeven in the presence of a strong nonspecic back-ground (i.e., samples mixed with cell extracts).1

Other reported assays using this format includedQD-conjugate binding two extremities of individualDNA molecules that were stretched on a glass surface,and the identication of single large entities such asviruses or bacteria from dierent strains in a macro-scopic mixture.14,23,36 These assays often use QDs withspectrally distinct emissions conjugated to differentantibodies. In the virus constructs, the collected uo-rescence signal (being a mix of signals emitted by dif-ferent QDantibody conjugates bound onto a singlescaffold) provided a spectral bar code that accountedfor the nature and number of specic proteins displayedon the virus surface. This has for example allowed arapid identication of the virus strain.1

We should emphasize that these assay formats, suchas the one used for virus identication could beperformed with organic uorophores. However,

comprehensive analysis requires the usual complexspectral deconvolution and tedious data analysis. Incomparison, QDs oer advantages such as highersensitivity and spectral unmixing (deconvolution),which enable the multiplexed detection of single smallbiomolecules carrying only a few uorescent markers.Nonetheless, the limitation of assays based on QDuorophores, derives from the intermittent emission ofsingle QDs. This can cause a fraction of the co-local-ized QD probes to occasionally appear as isolatedQDs, which may introduce false negative counts.

Co-Localization Studies Using QDDye Pairs

The same concept was applied by Webb andco-workers to investigate the relationship betweenensemble uorescence quantum yields measured formacroscopic samples and the blinking properties ofindividual QDs.107 The authors coupled Streptavidin-functionalized 525-nm emitting QDs with biotin-functionalized Alexa Fluor 594 dye and used theuorescence collected from dot and dye to distinguishbetween dark and bright freely diffusing single nano-crystals. This presents a different conguration fromthat of surface-immobilized QDs (e.g., dispersed inpolymeric lms). The bright fractions of QDs weremeasured by uorescence coincidence analysis andtwo-photon uorescence correlation spectroscopy. Inparticular, they reported that the bright fraction ofQDs was proportional to the measured macroscopicquantum yields of the samples. However, brightness ofindividual QDs was constant across samples with dif-ferent quantum yields. They then applied the samemeasurements to solutions with much higher viscosity,where nanocrystal residence time in the illuminatedvolume was substantially increased. Their ndingsclearly indicated that increasing the residence time byas much as 10-fold did not change the fraction ofapparently dark QDs, which they attributed to thepresence of two populations of QDs in a sample: onedark non-emitting and one emitting; only the emittingfraction contributed to the measured macroscopicuorescence yields. These results may be very infor-mative, as they can potentially provide correlationbetween ensemble quantum yield and single moleculeuorescence; additional work is still needed to betterunderstand the correlation between the uorescenceproperties of single and ensemble QDs.

SINGLE PARTICLE ENERGY TRANSFER

Fluorescence Resonance Energy Transfer (FRET)involves the non-radiative transfer of excitation energyfrom a uorescent donor to a ground state proximal


acceptor. It is governed by the dipoledipole interac-tion (Forster) mechanism, which results in a strongdependence of the transfer eciency, E, on the donoracceptor separation distance, r (E ~ 1/r6) and on thespectral overlap between the donor emission and theacceptor absorption. Because of the strong dependenceof the transfer efciency on separation distance, thistechnique provides a unique tool to probe intermo-lecular association and intra-molecular conformationalchanges in the 110 nm range.17,34 FRET provides apowerful means for studying biomolecular interactionsinvolved in drug screening or disease diagnosis. It hasbeen widely used for studying changes in protein andoligonucleotide conformation in vitro and in live cellsusing dye donordye acceptor FRET pairs. It has beenshown in the past 5 years that QDs offer a few uniqueadvantages (compared to organic dyes) for useas FRET donors.11,12,66,84,94 As demonstrated forensemble measurements, some of the size-tunableabsorption and emission properties can be very bene-cial to implementing single particle energy transfer.The broad absorption spectra of QDs allow excitationof the donor far from the acceptor absorption peak,which substantially reduces direct excitation contri-bution to the acceptor signal. In addition, the narrowemission of single QDs can be tuned to match theacceptor absorption spectrum.12 This narrow emissionalso allows easy deconvolution of the donor andacceptor emissions. Finally, because a QD can beconjugated to several acceptors, the overall FRET ratecan be substantially increased, a clear advantageespecially with longer separation distances. Followingthe success in using QDs as energy transfer donors inensemble assays, there has been a natural desire totranslating these assays to the single molecule level.This can allow access to single molecule information,reduces materials consumption, while potentiallyallowing higher sensitivity to be reached.

Single particle/molecule (sp- or sm-) FRET can useeither surface-immobilized or freely diusing QDs andQD-bioconjugates. Probing immobilized QD-assem-blies can allow one to collect and resolve time-depen-dent conformational changes at the single moleculelevel. In this conguration one plots time-traces ofdonor and acceptor emission proles, and changes ofthese proles are used to account for alterations in theFRET interactions. However, spFRET using surface-immobilized assay remains dicult to implement, dueamong others to weak FRET eciency (in particular,for one QDone dye pairs), a feature associated withsome of the commercial nanocrystals; these oftenrequire multi-layer functionalization and additionalconjugation steps.67 Moreover, the intermittent natureof single QD emission prevents one from quantitativelymonitoring variations in the FRET efciency with time.

Finally, interaction of the QD-bioconjugate with thesurface may alter the biomolecule conformation and itsbehavior.

Probing freely diusing QD-bioconjugates in solu-tion samples can allow one to circumvent some of theissues encountered with immobilized conjugates, by forexample limiting the specimen excitation to the resi-dence time within the illuminated volume. It alsoreduces issues associated with intermittent emissionand data collection and analysis. Solution-phase singleparticle FRET conguration utilizes confocal micros-copy combined with a highly focused laser beam forspecimen excitation, two dichroic mirrors and twohighly sensitive detection systems (Fig. 5a). The col-lected signal consists of simultaneous bursts of PLintensities from the donor (QD), collected on thedonor channel, ID, and the acceptor (dye), collected onthe acceptor channel, IA, corresponding to Browniandiffusion of single QD conjugates in and out of theconfocal volume. In practice, the spFRET signatureconsists of plotting the population fraction of events(or simply the number of events) vs. emission ratio, g,dened as g = IA/(IA + ID); g is also be referred to asthe spFRET efciency, E. For a control solutioncontaining QD donors only, the population fractionvs. g (or number of events vs. E) is reduced to a peakcentered at g = 0 (or E = 0), due to absence of FRET.For samples containing QDdye pairs, non-zero con-tribution from the acceptors due to energy transferbroadens the distribution peak and shifts it to higher gvalues (Fig. 5).84

In what follows we will describe a few representativeexamples where single particle FRET applied to QDsused to provide valuable and unique informationabout the dynamics of Holliday junction, spectrallyresolved FRET, the detection of target DNA sequenceand the heterogeneity in QDprotein bioconjugates. Inthese examples, either surface-immobilized or solution-phase congurations were employed.

spFRET of Immobilized QDOligonucleotideand QDProtein Conjugates

We describe two recent examples. In the rst, singleimmobilized QDoligonucleotide conjugates were usedto probe the dynamics of DNA Holliday junctions.41

One of the characteristics of this four DNA strandassembly is that its conformation uctuates betweentwo different forms depending on salt concentration.For example, in the absence of multivalent ions and atlow salt concentration, the junction has an open con-guration. At high salt concentration, effects of elec-trostatic repulsions are reduced and the junction foldsinto one of two-stacked conformers. The authors usedstreptavidinQDs which were rst immobilized on a


glass substrate using biotinylated BSA. First, theyexposed the immobilized QDs to solution of doublestranded DNA sequence having one end labeled withbiotin while the other end carries a Cy5 dye to promotethe formation of surface-tethered QDCy5 pairs. Thisexperiment was used to demonstrate that blinking ofindividual QDs is indeed temporally correlated withthat of the dye only when surface-immobilizedQDstreptavidinDNACy5 conjugates were formed,i.e., FRET interactions were present. Through analternation between green laser excitation (able toexcite the QDdye pair) and red laser excitation (whichexcites the dye only), they showed that FRET inter-actions could be eliminated by bleaching of the dyeunder sustained red light excitation, which renderedthe dye acceptor inactive and unable to interact withthe QDs; after dye recovery correlation between theblinking of QD and Cy5 emissions could again bedetected.41 They then applied this technique to explorethe dynamics that characterize DNA Holliday junc-tion, which was assembled from four single strandedDNA molecules, two of them were unlabeled, one pre-labeled with biotin and the other labeled with Cy5acceptor (see schematics in Fig. 6). Experiments were

carried out at pH 7.5 in Tris:HCl buffer where con-formational transition between the two states are slowenough to allow resolution on a CCD camera(~100 ms). They found that a small fraction of the QDuorescence showed anti-correlated two-state uc-tuations between the QD signal and that of the Cy5dye, with both states exhibiting signicant acceptorcontributions. The behavior was different from the oneobserved when unconjugated QDs are probed (in thesame sample), where no alternation in the blinkingpattern was measured. They conrmed that this char-acteristic behavior is similar to the case where aCy3Cy5 pair was used instead (see Fig. 6 and Hohngand Ha41). They further conrmed their observationby carrying out dwell time analysis of high and lowFRET states using QDCy5 and Cy3Cy5 pairs andfound that there is a good agreement in dwell time ofthe high and low FRET sates collected from bothpairs, even though the system using QDs allowed amuch larger total number of transitions to be observed.These nding clearly conrm that conformationalchanges in the Holliday junction induce uctuations inthe FRET efciency measured between the QD anddye at the single molecule level.

(b)(a)

Sample

60

80

100

detection threshold

Sample

Obj

320 321 322 323 324 3250

20

40

counts

detection thresholdDichroic 1

Di h i 2

exc=488nm

Notch filter

time (s)APDD

APDA

Dichroic 2LP filter

0.6

0.8no RR0.1 RR0.2 RR0.5 RR1 RR2 RR

fract

ion

(c)

0.2

0.4 4 RR8 RR

Popu

latio

n

-20 0 20 40 60 80 100 120 1400.0

IA / (IA + ID) (%)

(d)

FIGURE 5. (a) Confocal microcopy set up used for implementing solution-phase spFRET. (b) A typical example of superimposeddonor (green) and acceptor (red) time traces. Only bursts with the sum of both signals exceeding the threshold level (indicated byarrows) are used for analysis. (c) Emission ratio distributions obtained from spFRET measurements as a function of the acceptor-to-QD ratio. (d) Emission events vs. efficiency E reported in Zhang et al.114 Figure partially reproduced from Pons et al.84 and Zhanget al.,114 with permissions from the American Chemical Society and NPG.


In the second example, our group applied spFRETto immobilized QDproteindye conjugates to conrmthe presence of spectrally resolved energy transferobserved for ensemble measurements.83 Spectrallyresolved FRET is a unique property of QDs and is dueto a combination of broad dye absorption and narrowsingle QD emission. Indeed, an ensemble PL spectrumis composed of the sum of very narrow single QDspectra. This produces distinct single QD spectraloverlap with the acceptor absorption spectrum,resulting in an apparent wavelength-dependent FRETefciency. We measured the intensities and emissionwavelengths of single QDs from a 575-nm centered

population, conjugated to 20 proteins that were eitherunlabeled or labeled with a QXL520 (quencher) andimmobilized on a glass cover slip. The QD PL coin-cides with the red shoulder of the dye absorption. Todetermine the position of each QD emission, the uo-rescence image was split into two windows using adichroic mirror centered at 570 nm. Three QD frac-tions could be detected, one emitting in the red win-dow, one emitting in the blue window of the dichroic,while a third having emission around 570 nm leakedinto both channels (Fig. 6). The QD signal from eachwindow identied as blue IB and red IR components,while an emission ratio r = IB/(IB + IR), was used to

FIGURE 6. (Left panel) Single-pair FRET between 585-nm QDs and Cy5 bound via a Holliday junction. (a) Schematics of thejunction assembled on the immobilized streptavidinQD via biotinylated BSA. (b) Time-traces collected from some of theassemblies showing anticorrelated fluctuation in the signals of QD and Cy5. (c) Similar traces for Cy3Cy5 pair used for the samejunction. (Right panel) (a) Simultaneous PL images on the blue B and red R sides of the dichroic showing a few 575 nm QDMBPconjugates. IR black and IB gray intensity traces of a QD exhibiting r ~ 0.45. (b) Population distribution of QDs conjugated to MBP(squares) and QXL520MBP (open circles) vs. r. (c) Average intensities of single QDs conjugated to MBP (squares) or to MPBQXL520 (filled circles) vs. emission ratio r. The normalized intensity distribution I/Imax for QDMBPQXL520 is shown for com-parison (open circles). (d) Direct comparison between the wavelength-dependent FRET rates from steady-state (solid line) andsingle QD measurements (circles) for the QDQXL520 pair. A total of 220 single QD traces for the QD20 MBP and 310 single QDtraces for the QD20 MBPQXL520 were used. Data reproduced from Hohng et al.41 and Pons et al.,83 with permission from WileyVCH and AIP.


determine the QD central emission, e.g., r = 0.5 cor-responds to a 570-nm QD whereas higher and lowervalues indicate bluer and redder QDs, respectively.Two histograms were constructed, one for the popu-lation distribution as a function of the ratio r, N(r), bycounting the fraction of QDs exhibiting a particular rvalue, the other for the average intensity I = IB + IRvs. r. Figure 6 shows that the population distributionN(r) was the same for QDs conjugated to labeled orunlabeled MBP, with an average emission ratio of 0.50.There was, however, a signicant change in the shapeof the intensity distribution I vs. r, with bluer QDsexhibiting higher quenching than the redder counter-parts (Fig. 6, right panel c). This reects weakerspectral overlap of the redder QDs with the dye thantheir bluer counterparts. A FRET rate spectrum wasextracted from the ratios between the two intensitydistributions in the absence and presence of theacceptors, and compared to the one derived from thesteady-state experiments. These data also showed that

the single QD and solution ensemble experiments wereconsistent (see Fig. 6). This heterogeneity shouldalways been taken into account when analyzing singleQD FRET data.83

Solution Phase spFRET Applied to DNA Hybridization

In the rst demonstration of solution-phasespFRET with QDs Zhang et al. applied the techniqueto DNA hybridization and showed that it could allowdetection of a target sequence with high sensitivity(Fig. 7).114 Their reagents included a reporter probelabeled with an organic dye acceptor, a biotinylatedcapture probe and streptavidinQDs (SAQDsfrom Invitrogen). The probes were designed to havecomplementary sequences to two different regions ofthe target sequence; thus simultaneously hybridizationof the two probes with the target DNA can take placeif the latter is present. They rst mixed the two probeswith the DNA target, which resulted in the formation

FIGURE 7. Schematic of single-QD-based DNA assemblies. (a) Conceptual scheme showing the formation of a sensing assemblyin the presence of targets. (b) Fluorescence emission from Cy5 on illumination on QD caused by FRET between Cy5 acceptors andthe QD donor the assembly. (c) Sensing responsitivity at different target concentrations for nanosensors and molecular beacons.Reprinted with permission from NPG114.


of a sandwich hybrid between the capture probe, thereporter probe, and the target sequence (as shown inFig. 7). When mixed with SAQDs the preformedsandwich structures bound to the QD surface, bring-ing several Cy5 in close proximity to the donor centerin the conjugate. The solution was then introduced ina glass capillary and owed through a small obser-vation volume where uorescence signals from indi-vidual bioconjugates could be detected (Fig. 7). Thepresence of the target was identied by the simulta-neous detection of the emissions from QD (ID) andCy5 (IA), indicative of the QDDNA complex for-mation. This assay showed high detection sensitivity(in terms of minimal detectable target concentration)compared to molecular beacon assays employingorganic dyes (Fig. 7c).

Zhang and Johnson further showed that using asimilar conguration of QDs surrounded by a few

hybridized oligonucleotides an increase in the owvelocity in the capillary could substantially increasethe measured spFRET eciency.111,112 Such effect wasattributed to DNA deformation in the capillary underapplied Poiseuille ow eld, which effectivelydeformed the conjugate structure and reduced thecenter-to-center separation distance between dye andQD. In particular, they used a 50-base target oligo-nucleotide sandwich-hybridized with one biotinylated25-base capture probe and one 25-base reporterlabeled with Alexa Fluor 647, similar to the congu-ration described above (see Fig. 8a).112 When thesandwich hybrid oligonucleotides are immobilized onthe surface of 605 nm emitting streptavidinQDs, anaverage separation distance of 2225 nm is expectedfrom the conjugate structure together with experi-mental parameters under relaxed conditions (Fig. 8b).Such distance is far beyond the range of Forster

200

400 605QD

Time / s

Phot

on c

ount

s / m

s-1

0

100

200 Alexa Fluor 647

0 0

200

400

605QD

0 1 2 3 4 5 0 1 2 3 4 5

0

200

400

Time / s

Phot

on c

ount

s / m

s-1

Alexa Fluor 647

(a)

(c)

(d)

550 600 650 7000

30

60

90

120

Wavelength / nm

Fluo

resc

ence

0 6/1 12/1 18/1 24/1

605QD

QD emission

Excitation (488nm)Alexa Fluor 647

Alexa Fluor 647emission

QDFRETdsDNA

QD emission

Excitation (488nm)Alexa Fluor 647

Alexa Fluor 647emission

QDQDFRETdsDNA

(b)

FIGURE 8. (a) Schematic depiction of the conjugate assembly using dsDNA bridge with 605-nm emitting QDs and 647 Alexa Fluoracceptor. (b) A schematic representation of the conjugate with the expected separation distance in the QDdsDNAAlexa Fluor 647.(c) Evolution of the ensemble fluorescence spectra as a function of the increasing DNA/QD ratio; no change was measured.(d) Representative traces of fluorescence bursts from QDdsDNA/Alexa Fluor 647 complexes in a diffusion state (left, QD con-centration: 0.1 nM and concentration of Alexa Fluor 647 labeled 25-mer dsDNA was 1.8 nM), and in a microfluidic flow with a flow rateof 4.0 lL/min (right, QD concentration = 0.02 nM and concentration of Alexa Fluor 647-labeled 25-mer dsDNA was 0.36 nM. Figurereproduced from Zhang and Johnson,112 with permission from Wiley-VCH Verlag GmbH & Co. KGaA., and kindly provided byDr. Zhang.


allowed interactions (r> 2R0, R0 being the Forsterradius), and this was indeed conrmed by both sp-and ensemble FRET measurements where no QD PLquenching was measured (see Fig. 8c). However,under microuidic ow, uorescence bursts wereclearly collected on donor and accepter channels(Fig. 8d). Furthermore, the population of events(similar to g in Fig. 5a) was found to progressivelyshift toward higher g (reported here as FRET ef-ciency) when the target concentration was increased.The latter essentially increases the number of sand-wich hybrids arrayed on the nanocrystal surface, thusenhancing the FRET interactions per individual con-jugate. This nding is quite interesting, as it combinesthe benets of solution-phase spFRET and micro-uidics to both extend the range of accessible energytransfer but also allow enhanced sensitivity to detectrather large oligonucleotides.

In a separate study Zhang and Johnson appliedspFRET as a means to quantify the interactionsbetween a specic peptide on the Rev protein (Rev-peptide) and the Rev responsive element within theRNA gene (RRE-RNA) and to identify and charac-terize potential inhibitors.113 Interactions between theregulatory protein Rev and a portion of env genewithin the RNA gene (RRE-RNA) are critical to HIV-1 replication. Rev peptide is a sequence from the basicregion of Rev with reported high afnity to RRE.The authors demonstrated that the stoichiometry of

Rev-peptide binding to RRE-RNA sequence can beaccurately determined using such FRET-based singleQDRRE-RNA sensing assemblies (Fig. 9). Theyfurther used this single particle conguration topotentially quantify the inhibitory effects of proavinon the afnity between Rev-peptide and RRE. Inparticular, spFRET permitted them to quantifythese inhibitory effects even in the presence ofsubstantial levels of interference uorescence fromhigh-concentration proavin; the latter is a blue uo-rescent compound that often prevents the discrimina-tion of FRET signals in ensemble measurements.

In all these solution phase studies, the main factorsenabling ecient FRET from a QD to acceptorslocated at the other extremities of long double-strandedDNA and eective single molecule discriminationcould be attributed to two specic features: (1) EachQD-conjugate had several sandwich hybrids arrayedaround the nanocrystal surface. This increased thelocal acceptor concentration around the QD andresulted in high FRET eciency as well as brightacceptor uorescence bursts. (2) Direct excitationcontribution to the acceptor signal was extremelyweak, since the system could be excited in a region ofminimal acceptor absorbance, which reduced thedetection background to a minimal level. These twofactors combined produced enhanced detection sensi-tivity. In the study of Zhang et al.,114 for example,these QD-assemblies were successfully applied in

FIGURE 9. (a) Secondary structure of biotin-functionalized RRE IIB RNA. The nucleotides with strong specific affinity for Rev arehighlighted in red. (b) Cy5-labeled Rev peptide used. (c) Proflavin chemical structure. (d) Schematic representation of the single-QD FRET-based sensing assembly utilized for evaluating Rev-peptideRRE interaction along with the inhibitory efficacy of pro-flavin; commercial 605QD was used. Figure reproduced from Zhang and Johnson,113 with permission from the American ChemicalSociety, and kindly provided by Dr. Zhang.


conjunction with an oligonucleotide ligation assay toreach single nucleotide mutation detection.

Determining the Heterogeneity of QD-BioconjugatesUsing spFRET

Heterogeneity in QD-conjugates is a natural processand is due to the presence of multi-reactive sites on thesurface of a QD. Thus when a surface-functionalizednanocrystal is exposed to target proteins and peptides,a distribution in the conjugate valence naturally resultsduring the conjugate formation. We used single-QDFRET to characterize the heterogeneity of self-assem-bled QD-bioconjugates and to gain information aboutthe distribution in conjugate valence.84 These conju-gates were formed by attaching (via metal-afnitydriven self-assembly) average numbers of maltosebinding protein (MBPs) labeled with rhodamine red(RR) onto CdSeZnS QDs capped with dihydrolipoicacid (DHLA) ligands. The inuence of acceptor-to-donor ratio was examined by varying the fraction of

labeled-to-unlabeled proteins while keeping the sametotal number of proteins per QD-conjugate. This studyshowed that within a macroscopic homogeneoussample, heterogeneity in conjugate valence is a keyfeature of self-assembled QDprotein conjugates.More importantly, it was shown that the number ofacceptors per QD followed a Poisson distribution,where the probability of nding a conjugate havingexactly n acceptors for sample with a nominal numberof proteindye per QD, N, obeys the relation84:

pN; n Nn expN=n! 1In particular, using the distribution plots shown inFig. 5c, spFRET allowed us to determine the fractionof QD-conjugates having zero valence (i.e., QDs thatare not conjugated to any MBPdye), p(N,0), for aseries of macroscopic samples with increasing nominalvalence, N. This was extracted from a very narrowwindow in the distribution of population fractioncentered at g = 0 (Figs. 10a and 10b). Data clearlyshowed that the zero valence population fraction

1.00.8

no RR

(a)

(c)

(b)

0.6

0.8

0.4

0.6

no RR0.1 RR0.2 RR0.5 RR1 RR2 RR4 RR8 RR

0.0

0.2

0.4

0.0

0.2Popu

latio

n fra

ctio

n

Popu

latio

n fra

ctio

n

Popu

latio

n fra

ctio

n

N0 2 4 6 8 10-20 0 20 40 60 80 100 120 140

IA / (IA + ID) (%)

0.16

0.08

0.10

0.12

0.14exp. P()n=0n=1n=2n=3n=4n 5

0.02

0.04

0.06n=5n=6n=7theor. P()

-20 0 20 40 60 80 100 1200.00

IA/(IA+ID) (%)FIGURE 10. (a) spFRET population fraction vs. emission ratio g with increasing QD-conjugate valence. (b) Fraction of QDs withoutany acceptors or zero valence (g < 10%; squares) and that engaged in FRET (g > 10%; triangles) as a function of N, the averagenumber of RR acceptors per QD, obtained from spFRET measurements shown in (a). The fits corresponds to the Poisson distributionp(N,0) ~ exp(2N) and 1 2 p(N,0). (c) Comparison between predictions using Poisson function and experimental populationdistributions for N = 4. Figure partially reproduced from Pons et al.,84 with permission from the American Chemical Society.


(p(N,0)) decreased exponentially with increasing N,which agrees well with the prediction of Eq. (1). Theagreement between measured population distributionand the Poisson statistics was further veried for largervalues of N. We used the separation distance, directdye excitation and quantum yield ratios extracted fromensemble measurements in combination with Eq. (1) toconstruct a theoretical distribution for conjugateswith exactly n dyes using Eq. (1). A composite distri-bution of all these sub-populations (n = 0, 1, 2) wasthen constructed using these individual distributions.A side-by-side comparison between predictions andexperimental distribution for N = 4 is shown inFig. 10c. There is clearly a good agreement betweenthe Poisson distribution function and the experimentaldata. Additional details on applying these ndings toQDprotein sensors can be found in Pons et al.84

IN VITRO AND IN VIVO TRACKINGOF PROTEIN USING SINGLE QDS

In Vitro Detection of Kinesin and Myosin MotorMovement

Use of single molecule tracking has elucidated thecharacteristics of motor proteins such as kinesin,myosin and dynein that are responsible for activeintracellular transport of vesicles, organelles and pro-tein complexes (as cargos) along microtubules or actinlaments.42,91,98 Polymer beads of a few hun-dreds nanometer in size labeled (or loaded) with uo-rescent dye or simply individual organic dyes were usedto gain unique insights into the structural propertiesand movement of individual motors, determine theirvelocity, show that their movement is not necessarilycontinuous, and determine the size and rate of theirsteps.42,91,98 These experiments allow observation foronly a few seconds to a few tens of seconds dependingon the experimental conditions, nonetheless. Use ofQD tags, instead, could open up new possibilities tostudy the dynamic properties of these motor proteinsby extending the observation times.

In this aspect, we discuss a few representativestudies where use of QDs provided useful and uniqueinformation on the motion of motor proteins. In therst study, Surrey and co-worker used streptavidinQDs to label individual biotinylated kinesin motorsand track their motion characteristics with precision,including velocity, step length, and dwell time underdierent conditions.92 The authors utilized QD-con-jugates bearing single or multiple kinesin motors, andexplored motility in congurations of crowdedmicrotubules and in the presence of obstacles. Inparticular, they demonstrated that in the presence of

obstacles or in crowded conditions, kinesin motorscan wait on the microtubule in a strongly bound stateuntil the obstacle(s) unbinds, allowing the motor tomove to freed site in the next step. They postulatedthat this behavior could explain pauses in the motionof kinesin-transported cargos observed in living cells.In another study, Warshaw et al. labeled the heads ofindividual myosin V motors expressing a C-terminalbiotin with two different color QDs in a mixture.102

Using two-color single QD tracking, the authors wereable to locate the positions of the heads with ~6 nmprecision for several minutes, follow the motion ofindividual myosin molecules and probe changes intheir intramolecular conformation. In particular, theyfound that the heads on the same V motor movealternatively along the actin lament by ~74 nm steps,while maintaining an average separation distance of~36 nm during pauses in motion. These ndingsconrm the predictions of the hand-over-hand modelof myosin mode of motor proteins along an actinlament (Fig. 11).108

Conversely, QDs were used to label single fragmentsof the actin or microtubule, the substrate on whichmotors (e.g., kinesin) move. In these sliding assays,motors are adsorbed on a surface, and then exposed tosubstrate fragments. The immobilized motors are ableto bind these fragments and induce their transportacross the surface. A rst proof-of-principle studyby Mansson et al. demonstrated the use of QDs tolabel actin laments sliding on a surface of adsorbedmyosin motors, where movement of actin lamentswas tracked for several seconds.60 A more renedstudy published by Leduc et al. used single QDs toinvestigate the gliding motility of microtubules over avarying number of kinesin-1 motors attached to a glasssurface. This allowed the investigation of cooperativeinteractions between kinesin motors with the samemicrotubule (MT).52 For this, kinesin-1 motors werecombined with uorescent proteins (GFP-kinesin) thenimmobilized on a glass surface using 5-His antibodies,so that their locations could be determined by uo-rescence imaging; additional casein allowed passiv-ation of the rest of the surface (Fig. 12). QD-MTassemblies were then deposited on the substrate andtheir motion was tracked by detecting the QD emissionwith a ~3 nm precision (Fig. 12 bottom). By analyzingthe motion, velocity and step size of microtubulesinteracting with one, two or higher number of kinesin-1 motors, the authors identied distinct forward andbackward jumps on the order of 10 nm, which essen-tially indicates the existence of fractional motion stepsand absence of synchronization between severalmotors interacting with the same microtubule. Thesendings led researchers to envision that the streptavi-dinQDs used to label the actin laments could also be


used as anchor points (or a scaffold) to bind andtransport large biotinylated cargos.

Tracking of Protein Receptors in Live Cells

Transposing the achievements in single QD conju-gate tracking described above from in vitro conditionsto living cells has the potential to unravel complexbiomolecular interactions underlying specic cellularactivities. Spatial organization of lipids, proteins, andcarbohydrate on the cellular membrane plays a criticalrole in their interactions with the extracellular envi-ronment and surrounding tissue and in the cell sig-naling. Understanding when and how the cellreorganizes these membrane components is therefore akey challenge in the study of cellular activity. Forinstance, trafcking of neurotransmitting receptorswithin membrane compartments and between mem-brane and intracellular space plays an important rolein the regulation of neuronal activity. Nonetheless,single QD tracking of receptor proteins in cell mediahave been successfully reported only for easily acces-sible targets. These include labeling of transmembraneproteins to investigate membrane-bound receptor dif-fusion or endosomal trafcking. Indeed, tracking ofindividual proteins in live cells with QDs encountersdifculties that are often shared with other nanopar-ticle tags and uorophore. The main difculty iscaused by nonspecic interactions of QD-conjugates

with their environment, either due to steric hindrance(crowding effects, limited access to particular cellu-lar compartments) or to simple non-specicadsorption of endogenous molecules on the QD sur-face. Other hurdles include difculties of transfectingisolated QD-conjugates across the plasma membraneinto the cytoplasm.

There has been a few promising demonstrationswhere some of the key properties of QDs have pro-vided researchers the opportunity to track single trans-membrane receptors for long periods of time and led toa better understanding of receptor tracking in neu-rons. In a representative study, Dahan et al. reportedthe rst use of QD for single receptor tracking in cul-tured neurons in 2003.16 The authors labeled individ-ual glycine receptors with single QDstreptavidinconjugates via primary and secondary antibodies.Single QDs were identied and tracked by theirblinking properties (Fig. 13). They observed a consid-erably higher uorescence signal compared to standarduorophores, resulting in a spatial resolution reaching510 nm, compared with 40 nm with an organic dye.The ability to follow the trajectory of single QD-labeled receptors for more than 20 min (compared to~510 s for standard organic dyes), allowed the visu-alization of receptor trajectories on a time scale thatwas not previously accessible and revealed new diffu-sion dynamics. In particular, the authors showed theexistence of several membrane domains corresponding

FIGURE 11. Progressive motion of Myosin V motors labeled with red and green emitting QDs along an actin filament. Green andred open circles are the two different QD positions determined by Gaussian fits. The green arrow identifies a sub-step. Upper leftare averaged QD images for steps labeled AD. The yellow lines connect QD centers emphasizing alternating relative headpositions. Histograms of inter-head spacing and step size are shown in the bottom right. Figure partially reproduced fromWarshaw et al.,102 with permission from the Biophysical Society.


to synaptic, perisynaptic and extrasynaptic regionswith different receptor diffusion behaviors. A fewstudies have since used QDs to track trajectories ofindividual membrane receptors, in neurons or othercell types.35,8,9,25,35,37 Single QD tracking will con-tinue to gain more attraction in the future for studyingtrans-membrane protein dynamics, accessible to bothspecialists and non-specialists.

Recently, Tsien and co-workers used uctuations insingle QD intensities as a means to probe of theirchemical environment, in addition to theirlocalization.115 The authors were able to measurerelevant neuro-physiological parameters that werepreviously inaccessible, such as kiss-and-run and full-collapse fusion intervals and occurrences, fusion poreopen time and kinetics of re-acidication of the vesicle

FIGURE 12. (Top) (a) Principle of an in vitro gliding motility assay. The fluorescent beads were used as references to correct forspatial drift. (b) Images of gliding motility, where Rhodamine-was used to visualize MTs. The positions of the GFP-kinesins (laser-GFP) and the QDs (laser-QD655) were obtained by TIRF microscopy of the same field of view. The image on the right (merged)corresponds to the overlay of the three colors. The arrowheads show the kinesin positions in the laser-GFP image. The asteriskindicates the bead position in all colors (Scale bar 5 2 lm). (Bottom) Distinct jumps in the MT displacement in the three-motorcase. (a and b) Projected walked distances of two QDs (red curves) and the corresponding sideways motion (blue curves). Arrowsindicate the presence of jumps in the forward (solid line) and the backward (dashed line) direction. Figure partially reproducedfrom Leduc et al.,52 with permission from the National Academy of Sciences.


interior. They incubated hippocampal neurons withQDs so that they can be loaded into intracellular ves-icles at synapses. Electrical stimulation of the neuronselicited vesicle exocytosis, which occurred followingeither full-collapse fusion (FCF) or kiss-and-run(K&R). Whereas in FCF, the vesicle fully fuses with

the cell membrane and loses its identity, in K&R thevesicle fuses transiently with the cell membrane torelease its content into the extracellular medium througha 15 nm pore and is then rapidly recycled/reformed(see Fig. 14). FCF of a QD-containing vesicle is easilydetected by diffusion of the QD out of the vesicle into

FIGURE 13. Visualization of multiple exchanges between extrasynaptic and synaptic domains, in which an individual Glycenereceptor (GlyR) alternated between free and confined diffusion states. Images extracted from a sequence of 850 frames (acquisitiontime: 75 ms). (A1) to (A8) correspond to frames 6, 118, 150, 267, 333, 515, 629, and 850, respectively. QD fluorescence spots (green)and FM4-64-labeled synaptic boutons (red). One QD (arrow), first located at bouton b1, diffuses in the extrasynaptic membrane[(A2) to (A5)] and associates with bouton b2 [(A6) to (A8)]. Scale bar, 2 lm. Figure reproduced from Dahan et al.,16 with permissionfrom AAS.

FIGURE 14. Experimental traces illustrating typical emission patterns under three conditions, which are classified as K&R, FCF,or K&R + FCF. Also shown is a schematic depiction of the hypothesized effect of acute or chronic block of vesicular H+-adenosinetriphosphatase with bafilomycin A1 (Baf). Without Baf (normal), QD emission is reduced (maroon) by acidic luminal pH (gray).Vesicle fusion deacidifies the environment producing QD brightening (red). Acute application of Baf (acutely blocked) preventsreacidification after vesicle retrieval; chronic Baf (chronically blocked) removes all pH gradients. Figure reproduced from Zhanget al.,115 with permission from AAAS.


the extracellular medium. In comparison, duringK&R, the QD stays inside the vesicle. The internal pHof normal vesicles is ~5.5, but during K&R, this pH isequilibrated (through the fusion pore) with that of theexternal medium, and increases to ~7. When the porecloses and the vesicle recovers its integrity, its internalmedium is re-acidied to pH 5.5. To detect K&Revents, the authors relied on the pH-dependence of QDPL (~15% change in ensemble PL between pH 5.5 andpH 7.4). This pH change translated into a ~15%increase single QD PL while the fusion pore was open,and a decrease back to its initial level when the vesiclerecovers its initial pH. The exact mechanism respon-sible for the pH dependence of the QD PL was notelucidated, however. In addition, measuring repro-ducible small changes in single QD PL due to small pHchanges is still an open question; the authors showedhomogeneous single QD emission with rather lowsignal noise levels beyond what has often been reportedin the literature. Issues related to reproducibility andstability of hydrophilic QDs we discussed previouslywere simply not been discussed in this report. Futurestudies will require control over (and understanding of)the QD sensitivity to pH changes in order to obtainreproducible, specic QD PL responses to differentchemical environments. The results reported by Zhanget al.,115 nonetheless, constitute a step forward in usingsingle QD spectroscopy.

QDs have also given access to new informationregarding endosomal tracking of trans-membranereceptors. A study by Lidke et al. in 2004 demon-strated that QDs conjugated to the epidermal growth

factor (EGF) using streptavidinbiotin binding couldbe used to monitor the binding of EGF to specicmembrane receptors and track their intracellular fate(Fig. 15).56 EGFQDs were shown to correctly bindand activate the targeted erbB1 receptor, followed byendocytotic uptake of the EGFQDreceptor com-plex. The authors were then able to track the trajec-tories of individual EGFQDreceptor and show theexistence of a new retrograde transport mechanism inlopodia from the cell surface to the cell body. In asubsequent study, the authors used single EGFQDtracking to examine the nature of this retrogradetransport in detail.55 In particular, analysis of singleQD trajectories and their mean square displacementsshowed a behavior consistent with active transport andnot a Brownian motion. The authors showed that thistransport was supported by the actin network, asindicated by analysis of transport velocities and theeffects of pharmacological agents. They next exposedthe cells to two different colors of EGFQD conjugatesand observed that the onset of active transport corre-lated with merging of two EGFQD conjugates. Thisshowed that dimerization of EGF-activated receptorswas necessary to trigger the active transport. Finally,they examined whether retrograde transport occurredbefore or after endocytosis of the EGFQDreceptorconjugates using FRET between QD donors andorganic dyes. Membrane-impermeable biotinylated dyeswere able to bind the remaining biotin-binding sites onQDEGF conjugates, resulting in efcient FRET, evenafter the onset of retrograde transport. This demon-strated that these conjugates were still available to the

FIGURE 15. Retrograde transport of EGF-QDs (red) on filipodia. (a) A431 cell expressing receptor erbB3-mCitrine (green); max-imum intensity projection of four 0.5 lm confocal sections as a function of time. (b) Magnified image of filipodium indicated in thelast panel of (a), showing the uniform migration of the EGFQDs toward the cell body with a velocity of ~10 nm/s. All scale bars,5 lm. Figures reproduced from Lidke et al.,56 with permission from the Nature Publishing Group.


extracellular medium, and that receptor endocytosisonly occurred after retrograde transport along the l-opodia, when reaching the cell body. In another study,Vu and co-worker used antibody-conjugated QDs totarget specic membrane receptors to the nerve growthfactor.88 The authors showed that QDs allowed mon-itoring not only of the receptor cellular uptake, butalso of its intracellular fate. QD-labeled receptor pro-cessing by the cells was consistent with that of unla-beled receptors, including active shuttling of QDreceptors conjugates to new neural processes.

In a dierent study, Weiss and co-workers used theprocess of avidin polypeptide chain fusion to the gly-cosylphosphatidylinositol (GPI)anchoring sequenceof human CD14 receptors as a means to target avidinreceptors expressed in the cytoplasmic membrane ofHeLa cells. For this they used the direct peptide capexchange approach on the QDs they recently devel-oped and employed biotin-peptide-coated CdSeZnSQDs to label the avidin receptors expressed in thecytoplasmic membrane of these HeLa cells, and fol-lowed their motion using single molecule resolution.They in particular tracked the bound QDpeptideconjugates for several minutes as they diused in the

membrane of live cells and tracked in the cytosol.69,82

This allowed them to probe the relationship betweenGPI-anchored receptors and lipid rafts in the cellmembrane.

Using the same peptide surface coating strategyWeiss and co-workers further developed a generalplatform to use QDs for labeling and tracking of cellmembrane proteins at the single molecule level. Theystarted by inserting a small fraction of hapten-modiedpeptides (along with PEG-appended peptide) duringthe hydrophilic transfer of their QDs, producing hap-ten-functionalized nanocrystals with reduced non-spe-cic interactions. The haptens were then used to targetcell surface fusion protein containing single chainantibody fragments (scFVs) specic to haptens. In thisdesign, a scFV and the corresponding hapten recog-nition pair are separately attached to the target proteinand the QD, respectively. They then applied thisstrategy to probe the specic anity between 4M5.3scFv developed against uorescein. In particular theydemonstrated targeting of 4M5.3 scFv displayed onthe membrane of yeast and murine neuronal cells.44

Figure 16 shows a test of these probes on a mobiletarget protein mouse prion protein (PrP) expressed on

FIGURE 16. (a) Cartoon representation of a scFvPrP anchored in the membrane of an N2a cell via its GPI-anchor with a boundfluorescein-peptide-coated QD (FL-pc-QD). (bd) Single-particle tracking of scFvPrP in live neuroblastoma cells immobilized on aglass coverslip. (b) DIC image with overlaid QD fluorescence (white spots). (c) Thousand frames single-QD trajectory of the QDindicated by the white square in (b). (d) Intensity time trace exhibiting a blinking pattern typical of a single QD. Scale bars are 10 lmin (b) and 1 lm in (c). Figure partially reproduced from Iyer et al.,44 with permission from the American Chemical Society.


neuroblastoma cell line N2a, where the scFv was fusedto the N-terminus of PrP. The trajectory isolated for asingle QDhapten conjugate indicates that there aretwo distinct diffusion regimes for the protein on theimmobilized cell membrane; single protein movementwas identied by the intermittent (blinking) uores-cence signal typical of individual QD emission.

In order to circumvent some of the limitationsassociated with eective uptake of QD-bioconjugatesacross the cell membrane, Courty and co-workers usedosmotic lysis of pinosomes to deliver individual QDkinesin-motor conjugates inside the cell cytoplasm.22

This mode of delivery involves exposure of the cells torst a hyper-osmotic solution to induce QD pinocy-tosis, then to a hypo-osmotic solution to disrupt thepinosomes and release their content. The authorsreported that conjugates delivered via this mechanismcould be tracked individually over time. They foundthat the majority of QD-conjugates showed trajectoriestypical of freely diffusing Brownian motion, but asmaller fraction (~10%) of conjugates exhibited arather linear directed motion. This directed motionwas not observed when using unconjugated QDs orwhen inhibiting kinesin motor transport by a chemi-cally induced depolymerization of the actin laments.When analyzing the directed motion of this fraction ofQD-conjugates, the authors observed a velocity con-sistent with in vitro observations, which show thatkinesin movement was not affected by its QD cargo.This study is particularly encouraging for futuredevelopments using QDs to track individual biomole-cules in the cytoplasm. Nonetheless, additionalsystematic studies using a variety of cell types will stillbe needed to validate this type of cellular delivery andto ensure the absence of non-specic interactions of theQDs with the complex intracellular environment.

The few studies described in this report clearlyindicate that QDs have the potential to oer valuableinsights into intracellular tracking of membranereceptors and can reveal previously inaccessibletransport mechanisms. The studies exploited the highphoto-resistance of QDs to track individual emittersfor long period of time and report the tracking of theconjugated biomolecules. In general, the main limita-tions of these techniques originate from the still rela-tively large size of the QD conjugates, which mayinuence the QD-labeled protein diusion in connedspaces like synaptic clefts, or from non-specic inter-actions with endogenous molecules. The size limitationadds constraints even to probing extracellular trans-membrane protein tracking (e.g., due to the presence oflarge transmembrane carbohydrates). Blinking of QDuorescence though often causes occasional interrup-tion of the reconstructed trajectories, has not seriously

limited the use of QDs in single protein tracking aslong as the density of labeled receptors is low enoughthat the same QD can be identied again during thenext emitting period. Blinking has indeed been used asmeans to distinguish single QD-conjugate from thoseof aggregates, and it has allowed unequivocal identi-cation and tracking of single molecules in macro-scopic samples. However, it may be necessary tocorrect quantitative transport measurements from QDblinking eect in certain applications.22

CONCLUSION

By allowing measurements with high signal-to-noiseratios and extended observation times, QDs haveclearly opened up new possibilities for studying avariety of biological processes at the single moleculelevel. For example, they allowed tracking of singleprotein receptor trajectories with high spatial resolu-tion and over several tens of minutes. The set of singlemolecule studies summarized in this review constituteonly a small fraction of what use of QD uorophorespromises in biology. In addition to improving existingsingle biomolecule assays these materials have thepotential to allow access to previously inaccessibleinformation about biomolecular interactions, as hasbeen demonstrated in a few preliminary in vitro andlive cell studies. One should expect that optical tech-niques for single molecule tracking, in particular, willtremendously benet from further development in QDfunctionalization and conjugation techniques. Furtherprogress combined with extended observation timeswill bring even better spatial and temporal resolutionof reconstructed trajectories inside live cells.

However, further progress in several domains willbe required in order to take advantage of the fullpotential of these materials and realize the potentials ofsingle QDs studies. In particular, better and soundcharacterization of the eects of surface coating andconjugation strategies on the long-term stability ofQDs and QD-bioconjugates in buers and in intra-cellular environments is needed. Intracellular com-partments are complex and usually rich in ionicmaterials, and non-specic interactions often alter QDmovements in these media. One can expect that asbetter and more controllable surface functionalizationstrategies are developed and improved QD stability inbiological media and easy to implement conjugationstrategies become available, single molecule assaysusing QDs will ultimately transition from proof-of-principles performed by specialists to general use bythe broader biological and biophysical community.


ACKNOWLEDGMENTS

We acknowledge NRL, the Oce of NavalResearch, Army Research Oce and DTRA fornancial support. We also thank Dorothy Farrell atNRL for the assistance with TEM image shown inFig. 1, and Xavier Michalet at UCLA from the fruitfuldiscussions.

REFERENCES

1Agrawal, A., C. Y. Zhang, T. Byassee, R. A. Tripp, andS. Nie. Counting single native biomolecules and intactviruses with color-coded nanoparticles. Anal. Chem.78:10611070, 2006. doi:10.1021/ac051801t.2Anikeeva, N., T. Lebedeva, A. R. Clapp, E. R. Goldman,M. L. Dustin, H. Mattoussi, and Y. Sykulev. Quantumdotpeptide-MHC biosensors reveal strong CD8-depen-dent cooperation between self and viral antigens thataugment the T cell response. Proc. Natl. Acad. Sci.103:1684616851, 2006. doi:10.1073/pnas.0607771103.3Bats, C., L. Groc, and D. Choquet. The interactionbetween stargazin and PSD-95 regulates AMPA receptorsurface trafcking. Neuron 53:719734, 2007. doi:10.1016/j.neuron.2007.01.030.4Bouzigues, C., and M. Dahan. Transient directed motionsof GABA(A) receptors in growth cones detected by aspeed correlation index. Biophys. J. 92:654660, 2007.doi:10.1529/biophysj.106.094524.5Bouzigues, C., M. Morel, A. Triller, et al. Asymmetricredistribution of GABA receptors during GABA gradientsensing by nerve growth cones analyzed by single quan-tum dot imaging. Proceed. Nat. Acad. Sci. 104:1125111256, 2007. doi:10.1073/pnas.0702536104.6Bruchez, M., Jr., M. Moronne, P. Gin, S. Weiss, and A. P.Alivisatos. Semiconductor nanocrystals as uorescentbiological labels. Science 281:20132016, 1998. doi:10.1126/science.281.5385.2013.7Carion, O., B. Mahler, T. Pons, and B. Dubertret. Syn-thesis, encapsulation, purication and coupling of singlequantum dots in phospholipid micelles for their use incellular and in vivo imaging. Nat. Protocols 2:23832390,2007. doi:10.1038/nprot.2007.351.8Charrier, C., M. V. Ehrensperger, M. Dahan, et al.Cytoskeleton regulation of glycine receptor number atsynapses and diffusion in the plasma membrane. J. Neu-rosci. 26:85028511, 2006. doi:10.1523/JNEUROSCI.1758-06.2006.9Chen, H. F., I. Titushkin, M. Stroscio, et al. Alteredmembrane dynamics of quantum dot-conjugated integrinsduring osteogenic differentiation of human bone marrowderived progenitor cells. Biophys. J. 92:13991408, 2007.doi:10.1529/biophysj.106.094896.

10Chen, Y. F., J. Vela, H. Htoon, J. L. Casson, D. J. Werder,D. A. Bussian, V. I. Klimov, and J. A. Hollingsworth.Giant multishell CdSe nanocrystal quantum dots withsuppressed blinking. J. Am. Chem. Soc. 130:50265027,2008. doi:10.1021/ja711379k.

11Clapp, A. R., I. L. Medintz, and H. Mattoussi. Forsterresonance energy transfer investigations using quantum-dot uorophores. ChemPhysChem 7:4757, 2006. doi:10.1002/cphc.200500217.

12Clapp, A. R., I. L. Medintz, J. M. Mauro, B. R. Fisher,M. G. Bawendi, and H. Mattoussi. Fluorescence reso-nance energy transfer between quantum dot donors anddye-labeled protein acceptors. J. Am. Chem. Soc. 126:301310, 2004. doi:10.1021/ja037088b.

13Coe, S., W.-K. Woo, M. Bawendi, and V. Bulovic. Elec-troluminescence from single monolayers of nanocrystalsin molecular organic devices. Nature 420:800803, 2002.doi:10.1038/nature01217.

14Crut, A., B. Geron-Landre, I. Bonnet, S. Bonneau,P. Desbiolles, and C. Escude. Detection of single DNAmolecules by multicolor quantum-dot end-labeling.Nucleic Acids Res. 33:e98, 2005. doi:10.1093/nar/gni097.

15Dabbousi, B. O., J. Rodriguez-Viejo, F. V. Mikulec, J. R.Heine, H. Mattoussi, R. Ober, K. F. Jensen, and M. G.Bawendi. (CdSe)ZnS coreshell quantum dots: synthesisand characterization of a size series of highly luminescentnanocrystallites. J. Phys. Chem. B 101:94639475, 1997.doi:10.1021/jp971091y.

16Dahan, M., S. Levi, C. Luccardini, P. Rostaing, B. Riveau,and A. Triller. Diffusion dynamics of glycine receptorsrevealed by single-quantum dot tracking. Science 302:442445, 2003. doi:10.1126/science.1088525.

17Deniz, A. A., M. Dahan, J. R. Grunwell, et al. Single-pairuorescence resonance energy transfer on freely diffusingmolecules: Observation of Forster distance dependenceand subpopulations. Proc. Natl. Acad. Sci. 96:36703675,1999. doi:10.1073/pnas.96.7.3670.

18Deniz, A. A., T. A. Laurence, M. Dahan, et al. Ratio-metric single-molecule studies of freely diffusing biomol-ecules. Ann. Rev. Phys. Chem. 52:233253, 2001. doi:10.1146/annurev.physchem.52.1.233.

19Du, H., C. Chen, R. Krishnan, T. D. Krauss, J. M.Harbold, F. W. Wise, M. G. Thomas, and J. Silcox.Optical properties of colloidal PbSe nanocrystals. NanoLett. 2:13211324, 2002. doi:10.1021/nl025785g.

20Duan, H., and S. Nie. Cell-penetrating quantum dotsbased on multivalent and endosome-disrupting surfacecoatings. J. Am. Chem. Soc. 129:33333338, 2007. doi:10.1021/ja068158s.

21Dubertret, B., P. Skourides, D. J. Norris, V. Noireaux,A. H. Brivanlou, and A. Libchaber. In vivo imaging ofquantum dots encapsulated in phospholipid micelles.Science 298:17591762, 2002. doi:10.1126/science.1077194.

22Durisic, N., A. I. Bachir, D. L. Kolin, et al. Detection andcorrection of blinking bias in image correlation transportmeasurements of quantum dot tagged macromolecules.Biophys. J. 93:13381346, 2007. doi:10.1529/biophysj.107.106864.

23Edgar, R., M. McKinstry, J. Hwang, et al. High-sensi-tivity bacterial detection using biotin-tagged phage andquantum-dot nanocomplexes. Proc. Natl. Acad. Sci.103:48414845, 2006. doi:10.1073/pnas.0601211103.

24Efros,

Luminescent Quantum Dots

Documents

Transcript of Luminescent Quantum Dots