Fluorescence Lifetime Imaging Microscopy

Top left: A lifetime image a nuclear envelop that is double immunolabelled with 5-LO (donor) and FLAP (acceptor). FLAP is a scaffold protein for 5-LO and the interaction is important in the inflammation pathway. Top right: A histogram of the lifetimes shows three distributions; the longest lifetime of about 1800ps is non interacting donor, the slightly shorter

Autofluorescence in aged brain tissue

Proteins like lipofuscin accumulate and aggregate in all mammalian tissue over time. Since these proteins are autofluorescent, they contribute to the fluorescence signal and cause errors when measuring fluorescence lifetime.

The figures below show the lifetime image of a neuron from an aged mouse that was transfected with yellow fluorescent protein.

The lifetime of YFP is around 2000 ps. As you can see from the false color image on the left and the histogram of lifetimes on the right, there is a peak at just above 2000 and a much broader peak at about 650ps, with a low level of intermediate values between the two. The short lived peak is a piece of lipofuscin AF, with the intermediate values being pixels in which the two fluorophores are mixed.

It is difficult to remove this systematic error from the data for two reasons. 1) the decay profiles of AF are highly variable and non-exponential in nature, making fitting the data and treating the AF as a decay component almost impossible. 2) In many cases the AF is not as well localized as it is in the example above, making lifetime gating, as I have shown above, impossible. To make matters worse, when performing FRET experiments, as described in the previous section, autofluorescence lifetimes are often similar to the quenched FRET component of the decay, making it difficult to distinguish between a protein interaction and the presence of lipofuscin.

Lifetime (~1600ps) is indicative of a weak interaction with the shortest lifetime (~570ps) being a strong, close-knit interaction. Bottom: Using a segmentation algorithm, the locations of the three populations are mapped, showing that the non-interacting protein is inside the nuclear envelope with the weak and strong interactions occurring on the inner and outer walls of the nuclear envelope.

Solution: Multi-spectral subtraction

Coming soon... (August 2008)