FRET and FRET-FLIM microscopy imaging of localised protein interactions in living cell nucleus

FRET microscopy imaging is widely used to detect protein-protein interactions inside living cells. This application note describes the use of one and two-photon FRET and in characterising the dimerisation of C/EBPa protein expressed in living mouse pituitary GHFT1-5 cells. We also describe the use of FRET using FLIM (time-domain).

Within the living cell, interacting proteins are assembled into molecular machines that function to control cellular homeostasis. These protein assemblies are traditionally studied using biophysical or biochemical methods such as affinity chromatography or co-immunoprecipitation.

Recently, two-hybrid and phage-display methods have been used for detecting protein-protein interactions. These in vitro screening methods have the advantage of providing direct access to the genetic information encoding unknown protein partners. These techniques, however, do not allow direct access to interactions of these protein partners in their natural environment inside the living cell. But using the approach of fluorescence resonance energy transfer (FRET) microscopy, this information can be obtained from single living cells with nanometer resolution.

In this study we implemented FRET microscopy to characterise intranuclear dimer formation of the transcription factor C/EBPa in living pituitary cells. Members of the C/EBP family of transcription factors are critical determinants of cell differentiation. C/EBPa controls the transcription of genes involved in energy, including those encoding anterior pituitary growth hormone (GH) and prolactin (PRL).

C/EBPa is a basic region-leucine zipper (b-zip) transcription factor that forms dimers through contacts in the leucine zipper and binds to specific DNA elements via the basic region. We recently showed that GFP-tagged C/EBPa expressed in mouse pituitary GHFT1-5 cells was localised to subnuclear sites associated with pericentromeric heterochromatin, and this pattern was identical to that for the endogenous protein in differentiated mouse adipocytes. Our studies indicated that the b-zip region of C/EBPa (AA 244-358) fused to GFP was sufficient for subnuclear targeting of the fusion protein in pituitary GHFT1-5 cells. Since this region contains the dimerisation domain, we sought to determine whether the expressed fusion proteins were associated as dimers in these subnuclear sites using the techniques of Confocal and Multi-photon FRET microscopy.

What is FRET?

FRET is a quantum mechanical process involving the radiationless transfer of energy from a donor fluorophore to an appropriately positioned acceptor fluorophore over a very limited distance with subsequent fluorescence emission by the acceptor fluorophore. Four conditions must be fulfilled for FRET to occur.

First, the donor emission spectrum must significantly overlap (>70%) with the acceptor absorption spectrum.

Second, the distance between the donor and acceptor fluorophores must fall within the range of 1 to 10 nm.

Third, the donor emission dipole moment, the acceptor absorption dipole moment and their separation vectors must be in favourable mutual orientation.

Finally, the donor should have a high quantum yield. Examples of widely used FRET pairs include FITC-Rhodamine, Alexa 488-Cy3, Cy3-Cy5, BFP-GFP, BFP-YFP, CFP-YFP, CFP-dsRED and more pairs can be found in the literature.

FRET is mainly used to provide spatial information between two interacting proteins, at a resolution higher than conventional microscopy. Many biological questions including extent of colocalisation of proteins, dimerisation of proteins, transcription mechanisms, transduction pathways, molecular motor motions and protein folding can be addressed using FRET.

Fluorescence imaging microscopy suffers from various drawbacks like autofluorescence, detector noise, optical noise and photobleaching. In addition to these, spectral bleed-through or cross talk is the major problem in FRET microscopy. There are typically two components of bleed through: the donor bleed through (Figure 1 - shaded), which is part of the emission spectrum of the donor overlapping the emission spectrum of the acceptor and the second component is due to part of the excitation spectrum of the acceptor overlapping with the excitation spectrum of the donor (shaded). Due to these effects, the observed FRET signal is higher than the actual signal. We developed dedicated software in an approach to correct all the above-mentioned contamination from the FRET signal.

Methods

For the studies described here the sequence encoding the DNA binding and dimerisation domain of the transcription factor C/EBPa was fused in-frame to the commercially available CFP or YFP colour variants (www.clontech.com) to generate CFP-C/EBPa and YFP-C/EBPa. Mouse pituitary GHFT1-5 cells were harvested and transfected with the indicated plasmid DNA(s) by electroporation.

For imaging, the cells were inoculated drop wise onto a sterile cover glass in 35 mm culture dishes and allowed to attach prior to gently flooding the culture dish with media. They were maintained for 18 to 36 hours prior to imaging. The cover glass with attached cells was inserted into a chamber containing the appropriate medium and the chamber was then placed on the microscope stage.

Confocal and multi-photon imaging

Imaging was done using a Bio-Rad Radiance2100 confocal and multi-photon system. Confocal excitation wavelengths used for the donor and acceptor were 457 and 514 nm, respectively. Filters used for donor and acceptor emission were 485/30 and 545/50, respectively. For multi-photon excitation, a Coherent Ti:sapphire laser system was used at 820 and 920 nm to excite the donor and acceptor, respectively. Emission filters were the same as above.

Imaging for FRET

A number of different images must be collected to process the FRET data. These include donor only slide, acceptor only slide and a slide with both the donor and acceptor fluorophores. All three specimens must be viewed with both donor and acceptor excitations for each experiment. To verify the occurrence of FRET, a negative control must be included. One must also optimise the fluorophore concentrations. Generally, the acceptor concentration is greater than that of the donor. This is required to increase the probability of dipole moment orientation of the donor and acceptor molecules.

The fluorescence intensity in both channels were measured for donor only, acceptor only, and donor and acceptor samples. Imaging parameters were kept constant. Controls were imaged to determine bleed-through which was eliminated from the FRET data. An algorithm developed by us was used to remove cross talk to obtain the True FRET data. This algorithm not only removes spectral bleed-through but also corrects the fluorophore expression level variation in living cells and tissues.

Table 1: Summary of images required for FRET processing. D=donor, A=acceptor, a-g=images.

Single label donor slide

Select a well-expressed cell. Adjust the gain and black level of the donor and acceptor channels. The acceptor channel will show spectral bleed-through from the donor emission channel using donor excitation wavelength.

Single label acceptor slide

The acceptor channel image will show acceptor bleed-through when excited by donor excitation wavelength.

Double labelled donor and acceptor slide

The gain and black level adjustment for each channel should be kept constant for all the other image acquisitions. FRET signal is obtained along with both donor and acceptor bleed-through signals in the acceptor channel with the donor excitation wavelength.

Results

Removal of bleed-through

We applied our new algorithm to remove both the donor and acceptor spectral bleed-through problems and correct the variation in fluorophore expression level (FEL). The algorithm allows us to calculate both the energy transfer efficiency (E%), and to estimate the distance between donor and acceptor molecules. The details of the algorithm are explained in the literature.

Figure 2 describes confocal FRET imaging using a Radiance2100 confocal system. The donor was excited at 457 nm, and the donor image (D) and the unprocessed FRET (uFRET) image of proteins localised in the nucleus of a single living cell are shown. PFRET images (processed FRET) are obtained after application of FRET data analysis algorithm. The histogram below each image shows the respective signal strength. The occurrence of FRET indicates the formation of CEBPa dimers in the nucleus of live pituitary cells.

Figure 3 shows the same cell imaged by multi-photon microscopy.

The same algorithm was applied to obtain the true or processed FRET (PFRET) image. Confocal microscopy is limited by the availability of standard laser lines. Multi-photon (MP-FRET) microscopy provides a wide tuning range (700-1000 nm) allowing excitation of a wide variety of fluorophores. Multi-photon (MP-FRET) microscopy is also less damaging to living cells, thus limiting problems associated with fluorophore photobleaching, photodamage and intrinsic fluorescence of cellular components. An alternative to excite CFP would be the 405 nm line, which is available as an option on Radiance systems.

Figures 2 and 3: Localisation of CFP- and YFP-C/EBPa protein expressed in mouse GHFT1-5 cells using confocal and multi-photon FRET. The doubly expressed cells (CFP-YFP-C/EBPa) were excited by donor excitation wavelength and the donor (D) and unprocessed FRET (uFRET) images of proteins localised in the nucleus of a single living cell were acquired. FRET data analysis algorithm was used to obtain the processed PFRET (PFRET) image (5). The respective histograms below the images indicate the decrease in signal level after the correction (in the ROI) of the PFRET image. Note that the increase in signal in the MP-FRET compared to the confocal-FRET microscopy.

Discussion

Limitations and cautions for FRET

The failure to detect FRET from a pair of labelled proteins may not translate to the absence of interaction between them. There are several reasons why interacting proteins may not produce a FRET signal. Because the detection of FRET signal relies on the efficiency of FRET, it is important to optimise the efficiency by choosing a donor fluorophore that has a significant overlap with the acceptor absorption. Since FRET is most efficient with a stoichiometry favouring the interaction of proteins fused to the donor fluorophore with the proteins fused to the acceptor fluorophore, increasing the concentration of acceptor to donor will be helpful.

As with any approach using expression vectors in living cells, artifacts that arise from over expression must be taken into account. Control experiments with labelled proteins that colocalise, but do not interact physically, can be used to assess non-specific interactions that may contribute to the FRET signal. FRET measurements are also limited by the accuracy of quantifying fluorescence intensity, which is more prone to artifacts at weaker energy transfer signals. It is important to quantify lower intensities accurately. In these cases, fluorescence lifetime imaging (FLIM) of the donor fluorophore may offer significant improvement in sensitivity for determining the physical interactions between molecules in living cells.

FRET and acceptor photobleaching

When FRET occurs, there is quenching of the donor fluorescence signal since some of the donor excitation energy is transferred to the acceptor. Both quenching of the donor fluorescence and the acceptor emission should be measured to confirm the occurrence of FRET and this is made possible by the technique of acceptor photobleaching.

The selective photobleaching of the acceptor abolishes FRET and in regions where FRET occurred, there will be an increase in donor emission because of dequenching. FRET efficiency can be improved by increasing the overlap between the donor emission and acceptor absorption. The trade-off for this improved efficiency will be an increase in the background signal resulting from bleed-through. While this approach has the advantage of using a single double-labelled specimen, it cannot be used for living cells, since the exposure to extended laser energy is believed to cause ill effects.

Multi-photon and FRET

The excitation process in multi-photon FRET microscopy differs from that in a one-photon FRET, but the emission process is the same. MP-FRET eliminates out-of-focus light altogether by limiting the fluorophore excitation to the focal plane. This minimises photobleaching of the donor and reduces absorbance and light scattering from the sample.

In addition, excitation is in the infrared region, which causes significantly less damage to living cells. While confocal microscopy limits the use of fluorophores to standard laser lines, most of the fluorophore pairs are excitable using the MP-FRET microscopy. Multi-photon also helps to detect FRET signals deeper within tissues.

FRET-FLIM microscopy

Factors that limit the detection of weak FRET signals include autofluorescence, photobleaching and light scattering. The latter two can be overcome by detecting FRET through the lifetime of the donor fluorophore. The FLIM technique detects the nanosecond decay of the fluorophores and provides a spatial lifetime map of these probes within a cell.The fluorescence lifetime of a fluorophore is critically dependent on the local environment surrounding the probe. As biological interactions occur over time, monitoring localised changes in probe lifetime can provide an enormous advantage for imaging dynamic cellular events.

An advantage of FLIM is that it is independent of change in probe concentration, photobleaching and other facts that limit intensity based steady state measurements. Instrumental methods for measuring fluorescence lifetimes are divided into two major categories, frequency-domain and time-domain.

Frequency-domain fluorometers excite the fluorescence with light, which is sinusoidal and modulated at radio frequencies (for nanosecond decays), and then measure the phase shift and amplitude attenuation of the fluorescence emission relative to the phase and amplitude of the exciting light. Thus, each lifetime value will cause a specific phase shift and attenuation at a given frequency. In time-domain methods, pulsed light is used as the excitation source, and fluorescence lifetimes are measured from the fluorescence signal directly or by using photon counting. We are now investigating FRET using time-correlated single photon counting .

Table 2: Two-photon excitation FRET-FLIM imaging of GHFT1-5 cells using Radiance2100 confocal/multi-photon microscopy. CFPex 820 nm and YFPex920 nm. The intensity and lifetime images of donor-alone, acceptor-alone and double-labeled donor channel images are shown. The mean lifetime values and the lifetime decay curve for these images are also displayed.

FRET from donor to acceptor molecules significantly influences the donor lifetime. Combining FRET and FLIM, only the donor lifetime needs to be monitored to detect interactions between the labelled proteins, making it unnecessary to consider spectral bleed-through correction in FRET-FLIM images (4). The technique is independent of acceptor emission and only the acceptor molecules close enough to receive donor energy are detected.

Conclusion

FRET methodology provides higher spatial resolution beyond the limits of conventional microscopy. In this application note, we describe the confocal and multi-photon FRET imaging of transcription factor C/EBPa in the nucleus of living mouse pituitary GHFT1-5 cells. We used this approach to quantify the dimerisation of the transcription factor in living cells.

* Dr Ammasi Periasamy is Director of the W.M. Keck Center for Cellular Imaging at the University of Virginia, USA.