Multiphoton and Spectral Imaging

Multiphoton microscopy

• Predicted by Maria Göppert-Mayer in 1931

• Implemented by Denk in early 1990’s

• Principle: Instead of raising a molecule to an excited state with a single energetic photon, it cam be raised to an excited state by the quasi-simulatneous absorption of two (2-photon) or 3 (3-photon) less energetic photons

Multiphoton-photon Jablonski diagram

Multiphoton

• In multiphoton microscopy, the intermediate state is not a defined state, and so is “quantum forbidden”

• However, in quantum mechanics, forbidden is not absolute

• Therefore, the requirement for quasi-simultaneity• Practically, it means within ~10-18 seconds• In single photon, probablility of excitation is

proportional to I; in two-photon, it is proportional to I2

Excitation volume

http://www.loci.wisc.edu/multiphoton/mp.html

Advantages of multiphoton microscopy

• Fluorescence excitation is confined to a femtoliter volume – less photobleaching

• Excitation wavelengts are not absorbef by fluorophore above plane of focus

• Longer excitation wavelengths penetrate more deeply into biological tissue

• Inherent optical sectioning

Increased contrast in multiphoton

Centonze,V.E and J.G.White. (1998) Biophysical J. 75:2015-2024

Light sources

• Light flux necessary for multiphoton microscopy can be achieved by femtosecond pulsed IR lasers

• Ti-Sapphire lasers tunable from 700-900 nm

• http://micro.magnet.fsu.edu/primer/java/lasers/tsunami/index.html

Spectra Physics Mai Tai, Coherent Chameleon

Tuning Ranges 680-1080 nm

Sealed box units; no adjustments necessary

Computer controlled tuning

Stable pointing as you scan spectrum

Dyes for multiphoton microscopy

• Multiphoton excitation spectra for dyes is an active field of exploration

• Generally, 2PE peaks are broad

• General rule: start a little more energetic than λmax for single photon

• For example: EGFP: λmax for single photon = 488; λmax for two photon ≈ 900 nm

2PE Spectra

Detector configuration for multiphoton

Molecular Expressions web site

Note, in particular the descanned detector and the “Whole Area PMT Detector” = Nondescanned detector.

Descanned detector

• Uses same scan mirror to descan beam as was used to scan it.

• Better alignment with confocal• However, only collects the amount of light

represented by the projection of the mirror onto the specimen: less sensitivity

• Do not forget to open up the confocal pinhole, because the nature of multiphoton restricts excitation to a femtoliter volume

Nondescanned detector

• Because our excitation volume is restricted to a femtoliter volume, and is automatically an optical section, we do not need to descan

• Cone projected onto specimen is much wider, so much more sensitivity

• However, also much more sensitive to stray light

Confocal spectral imaging

• In many case, the spectra of dyes overlap either in their excitation spectrum, their emission spectrum, or both.

• What can we do?• Excitation overlap – for instance, tetramethylrhodamine

excitation spectrum overlaps that of fluorescein, so if we use the 488 and 543 lines simulatanously, we see overlap

• Solution: – Choose different dyey (fluorescein and Texas red)– Multitracking (sequential scanning) – excite at 488 while the

fluorescein image is being collected and at 543 while the rhodamine is.

What about emission?

Molecular Probes

Choose different dyes

Sometimes you can’t avoid overlap

• Autofluorescence frequently overlaps fluorescein emission

• NADH/Flavoprotein: on 2-P excitation at 800 nm, the 450 nm NADH emission is clean, but the 550 nm flavoprotein emission band has about 30% NADH emission

• Fluorescent proteins

Example: Lambda stack of cells expressing either CFP or GFP on

chromatin

What do we do?

• Acquire a Lambda stack of our image

• Acquire a lambda stack of our reference dyes, or, alternatively, identify areas in the image that will be pure.

• Mathematicall, through linear unmixing, apply linear algebra to separate the individual dye spectra from the multispectral image

Linear Unmixing

• Different amounts of pink and blue generate different spectra

Pairwise comparison of dyes that can or cannot be unmixed

Note that for pairs that cannot be unmixed (ie, DiO and eGFP), the shape of the spectra are very similar

Unmixing: fluorescein phalloidin and Sytox green

Problems with linear unmixing

• It takes a lot longer to acquire lambda stacks than single images

• The software – at least on the Leica – is not transparent to use

Solutions

Zeiss META Both use a prism to separate

Nikon CSI the spectrum to multiple

channels

Both have software that is easier to use