Critique (everyone)Monitoring dynamic protein interactions with photoquenching FRET. Nature Methods, vol 3, 519-24 (2006)

“Two-photon fluorescence correlation microscopy reveals the two-phase nature of transport in tumor” Jain et al, Nature Medicine, vol 10 203-7 (2004)

Suggested presentation

1. Campagnola, P.J., et al., High resolution non-linear optical microscopy of living cells by second harmonic generation. Biophys. J., 1999. 77: p. 3341-3349.

2. Campagnola, P.J., et al., 3-Dimesional High-Resolution Second Harmonic Generation Imaging of Endogenous Structural Proteins in Biological Tissues. Biophys. J., 2002. 82: p. 493-508.

3. Moreaux, L., O. Sandre, and J. Mertz, Membrane imaging by second-harmonic generation microscopy. J. Opt. Soc. Am. B, 2000. 17: p. 1685-1694.

Assigned Reading for Next Week

Outline:

1) Fluorescence Lifetime Imaging (FLIM)

2) Fluorescence Resonance Energy Transfer (FRET)

3) FRET/FLIM

4) FCS

Fluorescence Lifetime motivation

1) Sensitive to environment: pH, ions, potentialSNARF, Calcium Green, CameleonsPerform in vitro calibrations

1) Results Not sensitive to bleaching artifacts

2) Not sensitive to uneven staining (unless self-quenched)

3) Not sensitive to alignment (intensity artifacts)

iscf

f

kk

k

nonradiscf

f

kkk

k

iscf kk 1

Fluorescence Quantum Yield φ: important for dyesRatio of emitted to absorbed photons

Measured lifetime is sum ofRates of natural lifetime and non radiative decay paths

(k is rate,(k is rate,Inverse of time)Inverse of time)

Quantum Yield:Quantum Yield:

fk 1

0 Natural lifetime

Very fast1-10 ps

Einstein A coefficient A21=1/τOscillator strength, f, and fluorescence lifetime τ

2122

3

21 8A

e

mcf

o

For band centered at 500 nm, Fully allowed transition has lifetime of 4 ns(for one electron)

Dyes has several valence electrons, larger fLifetimes between 1-4 ns

Fluorescent lifetime depends on environment:Fluorescent lifetime depends on environment:Used in microscopy as contrastUsed in microscopy as contrast

υυ=light frequency, m=mass of electron,=light frequency, m=mass of electron,c=speed of light, c=speed of light, ee= electron charge= electron charge

Unquenched emission:Normal QY, lifetime

Quenched emissionDecreased QY, lifetimee.g. metals, aggregation

Unquenched and Quenched Emission

iscf

f

kk

k

nonradiscf

f

kkk

k

Gold somewhat quenchesthe fluorescence

2 general approaches: time domain and frequency domain

Short pulse laser modulate CW laser

Frequency Domain Methods for Lifetime Measurements:Modulate laser and measure phase change of fluorescence

Use cw laser (e.g. argon ion)Modulate at rate nearInverse of emission lifetime10-100 MHz

Measure phase changewith Lockin amplifier

Modulation Methods in Frequency Domain

Modulate laser and ICCD(intensified CCD camera)Better S/N

Modulate laser

ICCD Detectors for Lifetime measurements:Frequency domain and some time-domain

Needs to be gated rapidlyWidefield imaging (no sectioning)High quantum yieldVery expensive $80KRegular CCDs:10-20K

Historically Most common

Microchannel platesAmplify signal ~10 fold

Time-domain Widefield Lifetime imaging with ICCD

Variable delayed gate or many gatesis scannedTo sample exponential decay:Many frames (for each delay)

ICCD has no time intrinsic response: slow readoutGated gain

Two-photon has short pulse laser for time-gating

French

Ti:sapphire

Higher viscosityShorter lifetimeBetter chance forNon-radiative decay

Time domain methods for lifetime measurementsWith gated electronics and fast detectors (not gain modulated)Best for point detection, PMT on laser scanning

Synchronized Gating done by pulsed laser (e.g. ti:sapphire laser)

Collect data from multiple gates (windows)At the same time, fit to exponential

PMT Detectors for Lifetime measurements

~300 picosecond resolutionBetter with deconvolutionCost ~$500

~30 picosecond resolutionNo dispersionCost ~$15000fragile

PMTS have low quantum yield(10-20%), MCP worse ~5%

Microchannel plate photomultiplier: full of holes, kick off electrons

Dispersion in time of flightacross 14 dynodesLimits time response

300-500 picosecond resolutionVery small area (200 sq microns)Not good for scanningHigh quantum yield (up to 70% at 700 nm)Low count rate (~10 MHz)$5KExtremely fragile!!

Avalanche Photodiode (APD)

Time gating measurements of fluorescence decayTemporal Resolution defined by IRF (laser, detector, electronics)

IRF=instrument response function,Must be (much) shorter than fluorescence lifetime (delta function)to avoid convolution

Measure IRF with reflectionor known short lifetimee.g. Rose Bengal (90 ps)

Ideal IRF Real IRF

Gate away from IRF (laser pulse, PMT response)Lose photons

Practical determinations of theInstrument Response Function

1) Laser modern lasers: ti:sapphires 100 femtosecondLifetimes: nanosecondsNot a factorWas 20-30 years ago before modelocked lasers

2) DetectorsAPD or PMT response ~200 picoseconds: can beMCP-PMT 30 ps: not typical limitation

3) TCSPC or gating Electronics20-50 ps (depending on sophistication) Can be convolved with MCP-PMT response

Time-correlated single photon counting:•most flexibility, most accurate,•samples whole decay•Best time response

Measures time of flight of photonsAfter excitation pulse

Bins data at each time intervalRather than gating

Collect enough photons to approximate exponential:

Slower than gating butBetter measurement, Can separate biexponentials: Multiple components

Principles of time-correlated single photon counting

TAC or TDC measures time of flight, bins photons

Been aroundFor decades

Time-Correlated Single Photon Counting electronicsOn laser scanning microscope (recent)

TCSPC electronics synchronized with laser scanning electronics:Pixel, line, frame synchHistorically very hard: mostly homebuilt (e.g. Gerritsen)

Becker &Hickel addon to Zeiss Laser scanning confocal

Electronics all in one PCI board, ~50K addon

Intensity vs fluorescence lifetime image

Same dye, different lifetime because of environment

Quenched close to Nucleus due toHigher concentrationLower lifetime

Intensity and lifetime measurements

CFP-YFP linked by short peptide chainEnergy is transferred from CFP to YFPLifetime reveals info intensity does not

Duncan, J. Microscopy 2004

TCSPC FLIM using ECFP

2 distinct lifetimes: meaning?

Performance of Frequency and time domain methods

TCSPC best for efficiency, S/NBut more expensive (ti:sapphire laser)But already have if have 2-photon microscope

Long Acquisition Times for TCSPC FLIM:Need enough data to approximate decay

Bright stains106/s

Dim stains104/s

May bleach before done imagingDetection with 2-4 gates may be better ifShort on photons

need100-100000Photons/pixel

GFP lifetime increasesWith increasing viscosity

Different lifetime for B cells at immuno Junction with natural killer (NK) cellEGFP::MHC

Autofluorescence of Rat Ear

Contains collagen, elastin :Single exponential not sufficient for multiple components

Fits to two discrete components noisy

Continuous lifetime distributionBetter for multiple components

Mean tauFor pixels

Width, h, of distributionFor pixel

Unless know componentsStretched exp is betterRepresentative of physiologyand provides more data

FLIM as Diagnostic of Joint Disorder

H&E staining

Widefieldfluorescence

WidefieldFLIM

Little info

Detail revealed by FLIM

Fixed, thin sections(few microns)

FLIM as Cancer Diagnostic

Benign

Carcinoma

H&E staining Widefield FLIM

FLIM shows morphology like H&E histologyCan optically section and no staining with FLIMWith 2-p can do thick tissues (few hundred microns)

More contrastThan H&E

ProbablyNADH, FAD

Widefieldfluorescence

WidefieldFLIM

FLIM Diagnostics of arterial plaque

Clear lifetimeDifference in Normal and plaque:Not visible by Fluorescence intensity

FLIM via endoscope as clinical toolWorks like through microscope

Lifetime of NADH, FAD changes from normal To cancer

White PNAS2007

DonorExcitation

Donor Emission

Donor Excitation

AcceptorEmission

Fluorescence Resonance Energy Transfer (FRET)

Donor emission overlaps with Acceptor Absorption:Highly distance dependent

FRET probes conformational changes

Different conformation givesDifferent FRET signature

FRET increasesIn both cases

Inter and IntramolecularForms of FRET withProteins

CFP-YFP good combo

Protein-Protein InteractionsIn cytoplasm andmembranes

No FRET forNo overlap of donor emission,acceptor absorption

When FRET Occurs

No FRET forOrthogonal dipoleorientation

No FRET for moleculesmore than 10 nm apartR0=distance where FRET=0.5

Typical Values of Ro

Donor Acceptor Ro (Å)

Fluorescein Tetramethylrhodamine 55

IAEDANS Fluorescein 46

EDANS DABCYL 33

Fluorescein Fluorescein 44

BODIPY FL BODIPY FL 57

Fluorescein QSY 7 dye 61

Cy3 Cy5 53

CFP YFP 50

green red

GFPs and other colored “FPs have transformed FRET microscopy

Before had to label proteins, then introduce

Number of FRET Publications since 1989

60

6

60

RR

RFRET

Fluorescence Resonance Energy Transfer -Detection of Probe Proximity

0

0

0

0

0

0

D

DD

AMax

A

AA

D

DD

FF

FF

F

FFFRET

R0 typically 40-50 Angstroms50% transfer

Practical Challenges to FRET Quantitation

• Emission from A contaminates D channel (filters)• Emission from D contaminates A channel• Unknown labeling levels for D and A• Signal variation due to bleaching

– Complicates kinetic studies– Bleaching rate of D can actually be slowed by FRET

Solutions:• Separately labeled D and A controls to define bleedthrough• Acceptor destruction by photobleaching to establish • Dual wavelength ratio imaging to normalize away variations in label levels and bleaching effects

0DF

Fluorescent Proteins as D-A PairsIssue of Spectral Overlap

Better overlap,FRETBut more bleedthrough

Poor Spectral overlap,But less bleedthrough

Survey of FRET-Based Assays

• Protease activity

• Calcium Ion measurements

• cAMP

• Protein tyrosine kinase activity

• Phospholipase C activity

• Protein kinase C activity

• Membrane potential

Principle of Operation of Chameleon Calcium Indicators

FRET Increases when CaM binds Calcium ionsConformation changes, CFP-YFP closer together

Potential Sensor Based on FRET

Mechanism and Single CellsGonzalez and Tsien, Biophys J., 1995

Demonstration on Leech GanglionKleinfeld, et al., Neuron, 1999

Improved indicatorsGonzalez JE, Tsien RY. 1997.

Chemistry and Biology 4:269-277.

Donor= Di4-ANEPPSFast voltage sensor

Acceptor=OxonolSlow voltage sensorFRET pair more sensitive

Lifetime and FRET

Large change in lifetime for quenched donor upon FRET

0

0

0

0

0

0

D

DD

AMax

A

AA

D

DD

FF

FF

F

FFFRET

FRET should have bi-exponential decay, quenched and unquenched:Long and short lifetime components

CFP and YFP FRET by Lifetime Imaging

Channel changes conformation, distance changes, Donor quenching occurs due to FRET

Short lifetime is FRET from DonorFor given pixel Ratio of fast to slow decay coefficientsis estimate of FRET efficiency

Duncan, J. Microscopy 2004

CFP and YFP tethers FRET by Lifetime Imaging

Donor Lifetime goes up post acceptor bleaching

FRET Outcomes

Donor decreases

Acceptorincreases

Donor lifetimedecreases

Donor fluorescenceAnisotropy increasesAcceptor decreases

FRET pair anisotropy

Donor Anisotropy Increases: shorterLifetime, less likely to Rotate before emission

Extent of depolContains relativeorientation

Emission dipole usuallyParallel to excitation dipole:FRET to other orientationDepolarizes acceptor emissionNot constrained by laser

Much better dynamic rangeThan lifetime based changes ~10x

Anisotropy measurement more accurate

Piston, BJ2004

Fluctuation (fluorescence) Correlation Spectroscopy (FCS)

Fluctuations in excitation volumedue to Diffusion, reactions

Diffusion times for globular proteins

0

100

200

300

400

500

0 200000 400000 600000 800000 1000000

molecular weight (kD)diff

usi

on t

ime (

mic

rose

cond

s)

Spherical molecule

Diffusion coefficient

Diffusion Demonstration

http://physioweb.med.uvm.edu/diffusion/FrapPages1.htm

Normalized Autocorrelation Function

Compares probability of detecting photon at time t with some latter time t + τ

Autocorrelation the expected value of the product of a random variable or signal realization with a time-shifted version of itself

FCS: autocorrelation of the fluorescence photons of one and the same molecule.

FCS decay: diffusion, bindingMaximum at tau=0Small tau: large correlationLarge tau: tend to 0

Form for translational diffusion

N=concentration of molecules in focal volume

τD =diffusion time, R=ωz/ωxy of observation volume

Need to measure Point Spread FunctionTo determine observation volume

Determination of Point Spread Function of Microscope

NAd

2

22.1min

Abbe` Limit

175 nm BeadSub-resolution

Volume is EllipsoidAxial ~NA2

PSF and Beam Waist

Imaging sub-resolution 100 nm fluorescent beads

Use 1/e2 points to get ω beam waist (87%)

D =2/4D

8D for two-photon

webb

FCS of Rhodamine in Sucrose Solution

Higher concentrationsShorter correlation times

1/y=number of moleculesIn focal volume

webb

Molecular interactions

FCS is a method to study molecular interactions between – fluorescence-tagged ligands and target molecules such as receptors, – unidentified, untagged compounds and tagged ligands

in competitive binding – protein-protein interactions – individual events of signal transduction cascades within cells.

Concentration range

FCS allows molecular interactions to be characterized over a wide dynamic range-concentrations of labeled particles between 200 nM to 200 pM -can be used. Because of the ultra-lowmeasuring volume of 10-15 l required for FCS measurement, nanoliter to microliter sample volumes are sufficient.

Physical state

Differences in the physical state of the molecule of interest, such as bound vs. free, cleaved vs. intact,can be discerned by FCS provided that the variant forms differ sufficiently in their size-related diffusion – peptides bound to soluble receptors, – ligands bound to membrane-anchored receptors, – viruses bound to cells, – antibodies bound to cells, – primers bound to target nucleic acids, – regulatory proteins /protein-complexes in interaction

with target DNA or RNA – enzymatic products.

If the diffusion properties of the reactants are too similar, both reactants have to be labeled withfluorescent dyes with different excitation and emission spectra.

Mathematical model for autocorrelation

Two component autocorrelation curve

Diffusion of structures > 50 nm may be hindered in cytoplasmtau values > 1 ms may not represent free diffusion

Binding to immobile species,e.g. microtubules

Binding to mobile receptor Binding to immobile receptor Motility along microtubule

ConcentrationDiffusion of receptor

ConcentrationKd

On rate (M-1sec-1)Off rate (sec-1)

Mobile/immobileMean squared displacement

The slow component in living cells

The slow component for hnRNP A2 (tau = 21 msec) may represent granule movement on microtubules

Measuring dissociation constant by FCS

Model:

ACCA

ABBAKd

Kd

Where A is a protein with low mobility, B is a mobile fluorescently labeled protein, and Cis endogenous, unlabeled version of B. Both B and C bind to A with the same affinity.

When B binds to A, its mobility changes. This can be observed by FCS as twocomponents with diffusion times 1 and 2 and amplitudes y1 and y2.

Equations

21

2

*

*

yy

yR

AC

CAKd

AB

BAKd

ACCCt

ABBBt

ACABAAt

Where At, Bt, and Ct are total concentrations of proteins A, B, and C correspondingly.

The dissociation constant can be expressed from the equations as

RCtBtR

AtKd

1

tau2= off rate

Carson, JMB

Need to measureEndogenous, addedProtein concentrations

Determine taus,Mobile fractionsBy bleaching

fluorescentmolecules

Cross-correlation spectroscopy

3D Gaussianconfocal detection volume

~1 femtoliter

diffusiontrajectories

Individual fluorescent molecules are detected as single channel photon count fluctuations. Bound

molecules are detected as coincident dual channel fluctuations.

Cross-correlation analysis provides a measure of the

number and rate of diffusion of bound molecules.

Cross-correlation function Grg(t) = < Ig(t).Ir(t+t) >

1

1.02

1.04

1.06

1.08

1.1

10 100 1000 10000

microseconds

Alexa488 RNASyto61cross-correlation

Dual channel fluctuation

10000

15000

20000

25000

30000

35000

40000

45000

50000

0 1 2 3 4 5 6 7 8 9 10

seconds

Alexa488 RNASyto61

Count rate

Photon Counting Histogram

Same raw data as autocorrelation

Measure of probability of brightness at given time

Allows determination of dimerization or oligerimization

By diffusion, assuming hardspheresDimer would increase τD

By only 26%. Cannot measure by FCS

However brightness would increase by 2x

PCH can determine concentrations

Gratton, BJ

PCH can differentiate fluorophores based on brightness