Multiphoton Excited Fluorescence Microscopy:Principles, Instrumentation and Some Applications
Outline
1) MPE Photophysics
2) Instrumentation
3) Some Applications of MPE Fluorescence
4) MPE Photodamage Issues
5) 1,2,3 Photon Resolution
Interaction of Light with Matter
...3)3(2)2(1)1( EEEP
P = induced polarization,(n) = nth order non-linear susceptibilityE = electric field
Linear Processes · Simple Absorption/Reflection · Rayleigh Scattering
(3) << (2)<< (1) (5-7 orders of magnitude per term)
Second Order Processes
· Second Harmonic Generation*
· Sum-Frequency Generation
Third Order Processes
· Multi-Photon Absorption*
· Stimulated Raman Scattering
· Optical Kerr Effect
· White Light Generation
One and two photon absorption physics
Requires high power:Absorption onlyIn focal plane
Greatly Reduces out of plane bleaching
Simultaneous absorptionVirtual State:Very short lifetime ~10-17 s
Goeppart-Mayer, ~1936
e.g. fluorescein
2-photon excitation of fluorescein: 3D confinement
Absorption, Fluorescence only in middle at focal point
Compare 1 and 2-pAbsorption1-p excites throughout
Advantages of Non-linear Optical Excitation in Imaging · Intrinsic 3-Dimensionality (no pinhole)
· Little Near Infrared and IR Absorption of Biomolecules Greatly Minimizes Out-of-Plane Photo-Bleaching/Damage
· Comparable Lateral (X,Y) and Axial (Z) Resolution to confocal
· Large Depth of Penetration (scattering decreases in NIR)*5-10 fold
· Enhanced Contrast and Sensitivity* (non-descanned detection)
* enabling aspects for tissue imaging: brain, connective tissue, muscle
One Photon 2 photon
Absorptionprobability
AbsorptionCoefficient units
(50,000)
(10-16 cm2)
(10-50 cm4s)
10-50 cm4s=1 GM (Goppert-Mayer)
Power (photon)dependence
p P2 (gives rise to sectioning)
Laser Temporaldependence
none 1/
p p2 /
(virtual state)
One and 2-photon absorption characteristics
Cannot use cw lasers (Ar+)
22
12 ][
hc
NAPna
2-Photon Absorption Probability
Assume 10 GM cross section (fluorescein)100 femtosecond pulse1.4 NA800 nm80 MHz
Saturation (na≈1) occurs ~50 mW average power
Can propagate same form for three-photon absorption:Need at least 10 fold higher average power
Do most dyes have good enough two-photon cross sections for imaging?
Two-photon cross section measurement
Xu and Webb, 1996
Measure by fluorescence intensity, need quantum yield(same as 1 photon)
One photon absorption simply measured in UV-VIS Spectrometer, Beer’s Law A= εclneed new setup for two-photon absorption: need focused light
Measure wavelength
Measure pulse width
Epi geometry
Measure power
MeasureFluor.
22
12 ][
hc
NAPna
Control power
Two-photon spectrum of rhodamine B:Discrete points, not continuous like UV-VIS for 1-photon
Near-Infrared, rather than visible used for 1-p
Cro
ss
se
ctio
n G
M
Max 820 nmnot 1050 nm
Xu and Webb, 1996
Right slope of 2 atAll wavelengths:2-photon process
Fluorescein and rhodamine
Power Dependence to determine photon number: log-log plots
Same emission spectrumfor 1-p, 2-p excitationShould be, from Quantum MechanicsRelaxation is independent ofMode of excitation
Same emission spectrumFor different 2-p wavelengths:750 and 800 nmJust like 1-photon emission
Xu and Webb, 1996
Verify emission spectrum
Pulse Width Dependence
Correct pulse width dependence
Xu and Webb, 1996
22
12 ][
hc
NAPna
Slope of 1
fluorescein Rhodamine (best one)
Indo-1
Good 2-p properties
• Big conjugation
• Donor/Acceptor Pair(push-pull)
• Heteroatom substitution
2-P δ Rhodamine 5x>fluorescein because D/A pairAlso relaxes selection rules over fluorescein
medium
WeakUV absorb Small conj
strong
Conclusion: most dyes used for confocal work forTwo-photon excitation (some better than others)
Some Generalities about multi-photon absorption
1) Emission spectrum is the same as 1-p
2) Emission quantum yield is the same
3) Fluorescence lifetime is the same
4) Spectral positions nominally scale for the same transition: 2-p is twice 1-p wavelength for
5) Selection rules are often different, especially for xanthenes(fluorescein, rhodamine, and derivatives, (calcium green, fluos))
10 SS Nominally forbidden in 2-p
20 SS Nominally forbidden in 1-p: missing in fluoresceinAllowed and stronger in 2-p
10 SS 20 SS
Rhodamine Photophysics
10-12 s
1000 nm TPE
500 nm OPE
800 nm TPE
400 nm OPE
10-9 s
S0
S1
S2
800 nm stronger than 1000 nm band
Reverse of 1-photonFor all xanthenes:Fluorescein,rhodamines
All max ~830 nmNot ~1000 nm
1 and 2-photon bands
mediumstrong
Xu and Webb, 1996
FluoresceinRhodamine B
Fluorescein, RhodamineSimilar band structure
Cro
ss
se
ctio
n G
M
Cro
ss
se
ctio
n G
M
So et al
Simultaneous imaging not possible in one photon absorption :
Different transitions, need multiple lasers:390, 490, 540 nm
Multi-color imaging in tissue culture cells:Nucleus (blue), mitochondria (red), actin (green)
Image all 3 simultaneously via 2-photonwith ti:sapphire laser 780 nm: different selection rules for MPE
Confocal detection must be descanned through pinholeto achieve axial discrimination, eliminate out of focus light
Very inefficientOptical budget
Limited depth ofPenetration ~30 microns(based on one scatteringLength in most tissues)
Non-decannedDetection greatlyIncreases theSensitivity
3-5 foldVery significantFewer opticsNo pinhole, Detect scatteredlight
Open pinhole,Not necessaryBut stillConfocal, lossy
2-photon does not Need pinhole at allOr descanning toEliminate out ofFocus light(intrinsic sectioning)
Ti:sapphire100 femtosecond80 MHZ
“The Campagnola”
Very simple beampathRelative to confocal
Lasers for multiphoton excitation
Typical confocals use fixed laser lines,Can be limiting for multi-color imaging
**
*
Lasers for confocal microscopy
All cw: no peak powerNone useful for multiphoton excitationNot tunable or red enoughto match 2-photon bands
Tunable Lasers
Gain Medium with broad emission spectrum gives tunabilityto excite any fluorophoreStill monochromatic when lasing
Organic Dyes (e.g. rhodamines, styryls), titanium sapphire)
Ti:sapphire is almost universally used for 2-photon excitationTi: Al2O3
Titanium Sapphire Ti3+: Al2O3
E 3/2 excited
state
2T2 ground
state
Spin forbidden: long emission lifetime 300 µs
ε=14000, strong for spin-forbidden transition
Huge Stokes shift, Broad emission spectrum
Revolutionized laser industry in ability to make short pulses
Pumped with 532 nm Tunable 700-1000 nm 100 fs, 80 MHz
Modelocked ti:sapphire laser for 2-photon microscope2-photon would not have taken off without this laser
Tuning Range and Power of Ti:Sapphire
100 femtosecond pulses10 nm FWHM bandwidth
Longer wavelengths Less damaging
900 nm is often goodCompromise betweenPower and viability
Gets dodgy after 900Except with big laser,Alignment critical
Excite essentially every dye,Fluorescent proteinWith this wavelength range
Tissue Imaging: MPM enablingDepth and sensitivity
White, Biophys J, 1998
2-photon depth much improved:Reduced scattering 1047 vs 532 nm
1-p 2-p Contrast stretched
2-p descanned here
White, Biophys J, 1998
Non-descanned (direct) detection provides greater sensitivity
Confocal (1-p)<2-p descanned< 2-p direct2-p direct collects ballistic and scattered photons
Still fairlyConfocal
Photons areScattered, misspinhole
Sensitivity (best signals)
X-Zprojection
MFP~20-50 microns
Non-descanned (direct) detection provides greater sensitivity
White, Biophys J, 1998
Important at increasing depth
direct
X-Z
Make or break experiment with Highly scattering tissue
2-photon imaging of retina (salamander)
Fluorescein labeled
X-Z projection
Denk, PNAS, 1999
Too thick for 1-pGood contrast throughout
Svoboda, Neuron, 2006, vol 50, 823(review of 2-P for neuroscience)
Depth capability using 2-photon absorption
Autofluorescence of endogenous species in tissues
Need multi-photon excitation, non-descanned detectionFor enough sensitivity: small cross sections and quantum yields
Autofluorescence in Tumors
Mitochondria:NADH, Flavins
NAD not fluorescentNADH emission toMonitor respiration
NADH good diagnosticOf cell metabolism
Small cross sectionQuantum yield ~10%Small delta ~0.1 GMHigh concentration
Need non-descannedDetection to be viable
Imaging Muscle (NADH)With TPE Fluorescence
Low cross section butHigh concentration
Balaban et al
So et alAnn. Rev. BME2000
2-photon Tissue Imaging (Mouse Ear)
keratinocytesBasal cells
Collagen/elastinfibers
cartilage
Autofluorescence good for many layers
Strata corneum
Keratinocytes
Dermal layer(elastin, collagen)fibers
Human Skin Two-photon imaging
So et alAnn. Rev. BME2000 More versatile than dyes (but weaker)
MPM enabling, very weak in confocal
Endogenous only way for in vivo clinicalApplications.
Cannot use dyes (toxicity) cannot penetrate tissues or GFP expressions
Trend is multimodal: fluorescence + scattering fluorescence + CT fluorescence + PET
Needs multiphoton for depth of penetration and Sensitivity due to weak signals
Multiphoton for FRAP, uncaging, photoactivation
Multiphoton bleaching
Need 3D treatment, both radial, axial PSF
Calcein in RBL cells
Calcein In solution
2-photon FRAP in cells, solution
Webb et al
2-P photobleaching and fluorescence recoveryIn starfish oocyte
10 kD dye- DextranCross nuclear envelope
70 kD does not cross
Better Cell viability than 1-p due to confinement
Line scan bleach, page scan recovery
2-photon uncaging glutamate
Svoboda, Neuron, 2006
Fluo-5 calcium sensitiveAlexa Ca insensitive
Need 2-p localizationFor this
2-photon photoactivation of GFP
Svoboda, Neuron, 2006
Uncaging cross sections very smallFraction of 1 GMRequires high power, short wavelengthsPA FP can be more efficient
Measure of diffusion
Choice of Excitation Wavelength · Redder is always better for cell viability, imaging depth in tissue · Selection Rules different for One and Two-photon Excitation for
many dyes: check published literature (some not right)• Fluorescein 2-p maximum is 820 nm, but that band is invisible in 1-p excitation
•2-P absorption coefficients do not always scale with 1-p absorption Fluorescein 5x weaker 2-p absorber than Rhodamine All flourescein like dyes are somewhat weak in 2-P:
Calcium Green, Calcein, Fluos
· Ti:Sapphire is tunable 700-1000 nm, but only one wavelength at a time:
Optimize if multiple fluorophores, can usually do thisNot usually possible by 1-photon
Optical Considerations
Objective Lenses · Throughput: most lenses were designed for Visible Excitation, not near- Infrared (but changing) · Highly corrected lenses have losses of 2-4 fold, depending on wavelength (worse to the red) and square dependence on NLO= big losses · Neofluars worse transmission than Fluars · Some New lenses are available optimized for near- infrared Registration Issues
· Focus of White light vs Laser often different by 10-20 microns (dispersion) · Overlapping visible and near-infrared lasers difficult for uncaging
Practical Considerations of Multi-photon Excited Fluorescence Microscopy
MPE good for long term imaging because of sectioning but there are drawbacks · Still bleaches in plane just like one photon · New problems can arise from high peak power giving rise to unwanted non-linear effects Plasma formation leading to cell destruction (makes holes) Accidental 3 photon absorption of proteins and nucleic acids (700-800 nm) (abnormal cell division) damage thresholds highly wavelength dependentdetermined by cell division or live cell/dead cell membrane assay 700-800 nm ~10 mW at 1.4 NA is good limit at sample(Scales for lower NA) >850 nm little damage (still bleaches in-plane)
Damage can arise from Higher Order Absorption
Plasma formation:Very damaging
Bleaching, Free radical damageIn plane, same for 1,2 photonsFor first triplet
Piston, Biophys J. 2000
488 nm 1-photon
Slope=1.2
Bleaching of fluorescein dextran in droplets
710 nm 2-photon
Slope=1.9 (low power)
Like absorption probability
Two-photon Bleaching as function of average power, exposure
Piston,Biophys J.2000
Bleaching highlyNonlinear:16.5 mW>>8.3Much higher rateThan quadraticScalingGoes to higher states
Nonlinearity would be decreased at longer wavelengths
Piston,2000
NADH=3.65
Coumarin=5.1
Indo-1=3.5
Highly nonlinear:Higher order processesExcitation to higher states
Non-linear bleaching (ctd)
For same transition 2-pDoes not bleach moreThan 1-p!
Photodamage and Average Power in live cells (Neher, 2002)
Two criteria yieldSlope=2.4 (log-log)
Some 3-p contribution
visual
Fura 2-AM
Neher and Photodamage continued
Photodamage~-1.5
Also highly nonlinear
Linear with pixel dwell time:No change in peak powerConsistent with highly nonlineardamage
Pulse width dependence Pixel Dwell time dependence
RAYLEIGH CRITERION for Resolution for 2 Objects
Not resolved Barely resolved Completely Resolved
Determination of Point Spread Function of Microscope
NAd
2
22.1min
Abbe` Limit
175 nm fluorescent BeadSub-resolution
Volume is EllipsoidAxial ~NA2
PSF is measured size of infinitely smallPoint source of light
One and Two-photon axial discrimination
widefield
1-p confocal
2-photonLonger wavelength
2-photon confocal
1-p and 2-p Confocal geometries about the same resolution1-p confocal better than 2-P nonconfocal
Radial PSF
Axial PSF
Comparison of 1 and 2 photon PSFs2P wavelength twice that of one photon
One photon is better becauseShorter wavelength, but 2-p Better than Abbe limit(not twice 1-photon)NLO not diffraction limitedCooperative effect
Smaller PSF=Better resolution
Optical Resolution of the Campagnola Microscope
Imaging sub-resolution 100 nm fluorescent beads
Both agree well with theory
Is 2-photon excitation really necessary or advantageous for the experiment?
Weakly Fluorescent (autofluorescence) samples:
yes, sensitivity non-descanned detection can make or break experiment
Thick and turbid samples: yes, sensitivity, detection Reduced scattering
FRAP, FRET, Uncaging: yes, 3-D
routine Cell Imaging: NO!!