FRET – the principle
Fluorescence resonance energy transfer (FRET) describes the non-radiative transfer of energy stored in an excited fluorescent molecule (the donor) to a non-excited different fluorescent molecule (the acceptor) in its vicinity. Three conditions must be fulfilled for FRET to take place:
- Overlap of donor emission spectrum with acceptor excitation spectrum (Figure 1)
- Molecules must be in close proximity on a nanometer (10–9 m) scale (Figures 2–4)
- Molecules must have the appropriate relative orientation
Due to its strong distance dependence with r–6 (Figure 4) FRET occurs on a spatial scale which is highly relevant for biochemical reactions, such as protein-protein or protein-DNA interactions. FRET can probe molecular interactions by a sensitive fluorescence read-out. This allows researchers to study molecular interactions both in vitro and in vivo. By linking two interaction partners of interest with suitable fluorescent labels it is possible to analyze bi-molecular interactions. Alternatively, FRET allows the construction of biological probes reporting concentrations of second messengers or ion strength by means of an intra-molecular FRET due to strong conformational change.
Not surprisingly, FRET has developed into a widely used tool in cell biology, biophysics and biomedical imaging.
There are different techniques to detect FRET in the context of microscopy. Commonly known are techniques based on fluorescence intensity of either the donor (acceptor-photobleaching, FRET AB) or the acceptor (sensitized emission, FRET SE).
Intensity-based FRET can be readily applied using standard confocal microscopes. However, it also has some drawbacks. FRET AB cannot be applied in time series experiments and is susceptible to reversible photobleaching or photoconversion of the donor molecules. FRET SE, on the other hand, suffers from spectral cross-talk inherent to all FRET pairs and requires careful calibration measurements as well as linear unmixing of resulting images. This application letter introduces a different approach to measuring FRET which is based on fluorescence lifetime imaging microscopy (FLIM).
The process of fluorescence is often understood in terms of energy transitions from the electronic ground state (S0) to its excited state (S1) in a molecule (Figure 5, left). Such transitions can be elicited by incident light with the appropriate energy (i.e. frequency or wavelength). The absorbed energy is stored by the fluorescent molecule for a short time before it can be emitted as fluorescence. The time a molecule spends in its excited state is known as the fluorescence lifetime. It is typically in the order of nano-seconds (10–9 s) for many organic dyes and fluorescent proteins.
Fluorescence lifetime and FRET
An alternative process to relax from the excited state is, for example, FRET. By FRET excitation energy is non-radiatively transferred to an acceptor molecule. The acceptor in turn can relax by fluorescence (Figure 5, right). Since donor fluorescence and energy transfer are competing processes the rate depleting the excited state increases in the presence of FRET. One might say, the longer the donor molecules spend in the excited state the more likely it is that FRET occurs. Only those photons from donor molecules which relax by fluorescence are observed. Energy transferred to acceptor molecules is not detected due to the longer wavelength of acceptor fluorescence. Therefore, FRET shortens the donor lifetime (Figure 6).
Fluorescence lifetime imaging (FLIM)
The Leica SP8 FALCON measures fluorescence lifetimes in the time domain using pulsed lasers and single photon counting detectors. The lifetime is determined by building up a histogram of detected fluorescence events. This reveals a single or multi-exponential fluorescence decay. Numerical curve fitting renders the fluorescence lifetime and the amplitude (i.e., number of detected photons).
Since FRET decreases the donor lifetime one can quantify the extent to which FRET occurs, provided the donor lifetime without FRET is known. This donor lifetime τ serves as an absolute reference against which the FRET sample is analyzed. Therefore, FLIM-FRET is internally calibrated – a property alleviating many of the shortcomings of intensity-based FRET measurements. Since its fluorescence lifetime is an inherent property of a dye it is widely invariant to otherwise detrimental effects such as photobleaching, image shading, varying concentrations or expression levels.
The major limitation using intensity based FRET measurement is the underlying assumption that all observable donor molecules undergo FRET. This is usually not the case. This varying "unbound" fraction of donor molecules introduces considerable uncertainty to the measured FRET efficiency, making comparisons between experiments impossible. FLIM-FRET overcomes this disadvantage.
Recording FLIM images using live cells
To this end τ must be known from a measurement using a sample which contains the donor only. It is important to exclude any emission from the acceptor. Using external detectors one must use a band pass filter. Internal detectors can be adjusted to record only donor emission. The same settings must be used for both the donoronly measurement as well as the measurement using the FRET sample.
CFP-YFP FRET in live cells
In this work we use cultured RBKB78 cells transiently transfected with a FRET construct consisting of a CFP-YFP fusion protein (Fig. 7). The two FPs are connected by a short linker of two amino acids . Such a donor-acceptor fusion can also serve as a good positive control for FRET in a real-world scenario. The “donor only” sample consists of the same cells transfected with CFP only (Figure 7A). As a first approximation the average lifetime was computed in fast FLIM mode, for both, the donor only and the FRET sample (Figure 7B). The lifetime distribution histograms indicate that the average lifetime τ of the donor is 2.1 ns (Figure 8). The donor lifetime of the FRET construct is 1.4 ns. One obtains a FRET efficiency E = 1 – (1.4/2.1) = 33 %.
Top row: Fig. 7: RBKB78 cells transfected with a CFP donor only (A) and CFP-YFP fusion (B). The detection band was set between 445–495 nm using spectral FLIM detectors. The colored region has been used for analysis. Colors represent intensity modulated fluorescence lifetimes. Courtesy of Prof. Gregory Harms, University of Würzburg, Germany. We acknowledge experimental support by Dr. Benedikt Krämer (Picoquant, Berlin), Jan-Hendrik Spille and Wiebke Buck.
Bottom row: Fig. 8: Fluorescence lifetime distribution of donor only (yellow) and FRET (green) samples using average lifetimes. There is a clear shift of 0.7 ns towards shorter lifetimes in the FRET sample.
Ratios of FRET vs. no-FRET
It is known for CFP to have at least two lifetime components in its own right [2, 3]. Also, it is a priori not known whether all molecules under study undergo FRET. In order to do justice to this complexity one needs to have information on more than the average lifetime. We can perform a two-component fit which will give us two lifetimes and two amplitudes. The latter allow us to estimate the relative proportions of one lifetime over the other. In particular, using the amplitudes we can estimate the relative proportions of the fraction exhibiting FRET (bound fraction) and the fraction not exhibiting FRET (unbound fraction). FPs with a weak second fraction, such as EGFP or Sapphire are ideal for this type of analysis.
- He L, Olson DP, Wu X, Karpova T, McNally JG, Lipsky PE: A flow cytometric method to detect protein-protein interaction in living cells by directly visualizing donor fluorophore quenching during CFP-YFP fluorescence resonance energy transfer (FRET). Cytometry Part A 55A (2003) 71–85.
- Al H, Henderson JN, Remington SJ, Campbell RE: Directed evolution of a monomeric, bright and photostable version of Clavularia cyan fluorescent protein: structural characterization and applications in fluorescence imaging. Biochem J 400 (2006) 531–40.
- Jose M, Nair DK, Reissner C, Hartig R, Zuschratter W: Photophysics of Clomeleon by FLIM: Discriminating Excited State Reactions along Neuronal Development. Biophys J 92 (2007) 2237–22.