Förster Resonance Energy Transfer (FRET) between two fluorophore molecules can be used for studying protein interactions in the living cell. Here, a protocol is provided as to how to measure FRET in live cells by detecting sensitized emission of the acceptor and quenching of the donor molecule using confocal laser scanning microscopy.
Förster Resonance Energy Transfer (FRET) is the radiationless transfer of energy from an excited donor to an acceptor molecule and depends upon the distance and orientation of the molecules as well as the extent of overlap between the donor emission and acceptor absorption spectra. FRET permits to study the interaction of proteins in the living cell over time and in different subcellular compartments. Different intensity-based algorithms to measure FRET using microscopy have been described in the literature. Here, a protocol and an algorithm are provided to quantify FRET efficiency based on measuring both the sensitized emission of the acceptor and quenching of the donor molecule. The quantification of ratiometric FRET in the living cell not only requires the determination of the crosstalk (spectral spill-over, or bleed-through) of the fluorescent proteins but also the detection efficiency of the microscopic setup. The protocol provided here details how to assess these critical parameters.
Microscopy-based analysis of Förster Resonance Energy Transfer (FRET) permits assessment of interactions between proteins in live cells. It provides spatial and temporal information, including information on where in the cell and in which subcellular compartment the interaction takes place and if this interaction changes over time.
Theodor Förster laid the theoretical foundation of FRET in 19481. FRET is a radiationless transfer of energy from an excited donor to an acceptor molecule and depends upon the distance of the molecules and the relative orientation of their transition dipoles as well as the overlap between the donor emission and acceptor absorption spectra. The rate of energy transfer is inversely proportional to the sixth power of the donor-acceptor distance. Thus, FRET can be used to measure molecular proximity in the range of 1-10 nm.
FRET competes with other de-excitation processes of the donor molecule and results in the so-called donor-quenching and sensitized emission of the acceptor. Donor-quenching is a reduction of the number of emitted donor photons, while sensitized emission is an increase in emitted acceptor photons. Many microscopic FRET analyses use fluorescence intensity measurements, including acceptor photobleaching2, donor photobleaching2, or FRET-sensitized photobleaching of the acceptor3.
Here, a step-by-step experimental protocol and mathematical algorithm are presented to quantify FRET using donor quenching and acceptor sensitized emission4,5, a method often referred to as ratiometric FRET. Many protocols on how to approximate sensitized emission have been published, few have quantified the absolute FRET efficiency6,7,8,9. The quantification of FRET efficiencies in the living cell requires determining (i) the crosstalk (spectral spill-over, or bleed-through) of the fluorescent proteins and, also (ii) the detection efficiency of the microscopic setup. While crosstalk can be assessed by imaging cells expressing only one of the fluorophores, the assessment of the relative detection efficiency of the donor and acceptor fluorescence is more complicated. It requires the knowledge of at least the ratio of the number of donor and acceptor molecules giving rise to the measured signals. The number of fluorophores expressed in live cells varies, however, from cell to cell and is unknown. The so-called α factor characterizes the relative signal strengths from a single excited donor and acceptor molecule. Knowledge of the factor is a prerequisite for quantitative ratiometric FRET measurements in samples with variable acceptor-to-donor molecule ratios as encountered during live-cell imaging with fluorescent proteins. Using a 1-to-1 donor-acceptor fusion protein as a calibration probe permits the determination of the α factor and also serves as a positive control. This genetically coupled probe is expressed by cells in unknown total amounts but in a fixed and known relative amount of one-to-one. The following protocol lays out how to construct the 1-to-1 probe and how to use it for quantification of FRET efficiency. A spreadsheet that includes all formulae can be found in the supplement and can be used by the readers to enter their own measurements in the respective columns as outlined below.
While the protocol uses the GFP-Cherry donor/ acceptor pair, the presented approach can be performed with any other FRET pair. The Supplementary File 1 provides details on cyan-yellow pairs.
1. Plasmid construction
2. Cell culture and transfection
3. FRET Imaging
4. Image analysis for detecting absolute FRET efficiencies using donor quenching and sensitized emission
NOTE: Here, a practical step-by-step guide as to how to determine FRET efficiency with the use of the attached spreadsheet (Supplementary File 2) is provided. Theory and derivation of the presented equations can be found in detail in previous publications4,15,16,17. With the described settings, the following fluorescence intensities are collected.
Figure 1 shows the images obtained in the donor channel, channel 1 (488, 505-530 nm), the transfer channel, channel 2 (488, >585 nm), and the acceptor channel, channel 3 (561, >585 nm), respectively. Representative images of cells expressing GFP only, Cherry only, co-expressing GFP and Cherry, and expressing the GFP-Cherry fusion protein. The mean cellular FRET efficiencies calculated in NRK cells expressing GFP-Cherry fusion protein (positive control, Figure 2A) and those co-expressing GFP-Cherry (negative control, Figure 2B) are plotted versus the acceptor-to-donor ratio intensity ratio (Q) or molecular ratio NA/ND in each cell. Figure 2C illustrates an example on how to outline a region of interest and avoid perinuclear vesicles with high autofluorescence.
The presented algorithm can be used to quantify FRET efficiency in any region of interest including the quantification in every pixel of the image in the transfer channel. Figure 2D shows normalized pixel-by-pixel FRET images of cells expressing the GFP-Cherry fusion protein, co-expressing GFP and Cherry as negative control, and expressing receptor subunits of the Ashwell-Morell receptor. The rat variant of this receptor, the rat hepatic lectin (RHL1 and RHL2), is a two-subunit receptor system that is known to hetero-oligomerize. All FRET efficiencies were normalized to that of the GFP-Cherry fusion protein. We labeled RHL1 and 2 with GFP and Cherry on the cytoplasmic side of the plasma membrane. The image shows distinct FRET values at the plasma membrane compared to intracellular vesicles. Deleting the stalk domain of RHL1 (GFP-RHL1Δstalk) which is thought to mediate the tight interaction between the two subunits decreases the detected FRET efficiencies. In Figure 2E, mean cellular FRET efficiencies of cells expressing GFP-RHL1 and Cherry-RHL2, and GFP-RHL1Δstalk and Cherry-RHL2 are plotted versus the acceptor-to-donor molecular ratio (NA/ND). For further FRET analyses of this receptor system the reader may refer to a previous publication5.
Figure 1: Representative images of cells expressing fluorescent proteins. Cells expressing GFP only (A), Cherry only (B), GFP and Cherry co-expression (C), and GFP-Cherry fusion protein (D). Images in channel 1 (488, 505-530 nm), channel 2 (488, >585 nm), and channel 3 (561, >585 nm), respectively. Images were obtained with 63x oil objective and zoom set to 3x. Scale bar: 10 μm. Please click here to view a larger version of this figure.
Figure 2: Quantification of FRET images. Mean cellular FRET efficiencies for cells expressing the GFP-Cherry fusion protein plotted versus the acceptor-to-donor intensity ratio (Q) (A) and versus the acceptor-to-donor molecular ratio (NA/ND) for cells co-expressing GFP and Cherry (B). Open circle thick line represents single cell expressing GFP-Cherry fusion protein. Open circle thin line represents a cell co-expressing GFP and Cherry. Note that at higher NA/ND ratios the fraction of the useful signal, the sensitized emission (IDEα) in the transfer channel becomes smaller and smaller relative to the direct excitation of the acceptor (IAS2). This results in a larger error in the determination of E. Pixel-by-pixel FRET images calculated using the presented algorithm (C). This panel illustrates a cell co-expressing GFP and Cherry, the negative control, in the donor channel, the transfer channel, and the acceptor channel. It also shows a possible outline of a region of interest avoiding perinuclear vesicles with high autofluorescence which may negatively impact the precision of the FRET calculation. (D) All FRET efficiencies were normalized to that of the GFP-Cherry fusion protein. From left to right, GFP-Cherry fusion protein (positive control), GFP Cherry co-expression (negative control), GFP-RHL1 and Cherry-RHL2, and GFP-RHL1Δstalk and Cherry-RHL2. Scale bar: 10 μm. Color-coded scale bar: normalized mean FRET efficiency (normalized to the mean value of the positive control, the GFP-Cherry fusion protein). (E) This panel shows mean cellular FRET efficiencies for cells expressing the GFP-RHL1 and Cherry-RHL2 as well as GFP-RHL1Δstalk and Cherry-RHL2 plotted versus the acceptor-to-donor molecular ratio (NA/ND). Closed grey circle represents a single cell expressing GFP-RHL1 and Cherry-RHL2. Closed black circle represents a cell expressing GFP-RHL1Δstalk and Cherry-RHL2. Please click here to view a larger version of this figure.
Supplementary File 1: Algorithm for other donor-acceptor pairs. Algorithm for other donor-acceptor pairs such as different versions of cyan (ECFP, CyPet, mTFP1, Cerulean, mTurquoise2) and yellow (EYFP, Citrine, Venus, SYFP2, YPet) fluorescent proteins. Please click here to download this File.
Supplementary File 2: Spreadsheet with the presented FRET algorithm and use of the GFP–Cherry fusion protein to quantify FRET by sensitized emission of the acceptor and donor-quenching. Cells A2, B2, C2: Mean background signals (B1, B2, B3) in channels 1, 2, and 3, respectively. Cells D2 and E2: Mean values for cross-talk factors S1, and S2. Cell G2: Value for extinction coefficient ratio . Cell I1, and I2: Extinction coefficients of GFP and Cherry at 488-nm laser light. Cell J2: Mean value for factor. Column C (C5 and up): measured fluorescence intensity of a cell expressing GFP in channel 1. Column D (D5 and up): measured fluorescence intensity of a cell expressing GFP in channel 2. Column E (E5 and up): measured fluorescence intensity of a cell expressing GFP in channel 3. Columns F, G, and H (F5, G5, H5 and up, respectively): measured fluorescence intensities subtracted by mean background intensities in all 3 channels. Column I (I5 and up): Calculated cross-talk factor S1. Column M (M5 and up): measured fluorescence intensity of a cell expressing Cherry in channel 1. Column N (N5 and up): measured fluorescence intensity of a cell expressing Cherry in channel 2. Column O (O5 and up): measured fluorescence intensity of a cell expressing Cherry in channel 3. Columns P, Q, and R (P5, Q5, R5 and up, respectively): measured fluorescence intensities subtracted by mean background intensities in all 3 channels. Column S (S5 and up): Calculated cross-talk factor S2. Column W (W5 and up): measured fluorescence intensity of a non- or mock-transfected cell in channel 1. Column X (X5 and up): measured fluorescence intensity of a non- or mock-transfected cell in channel 2. Column Y (Y5 and up): measured fluorescence intensity of a non- or mock-transfected cell in channel 3.Column AD (AD5 and up): measured fluorescence intensity of a cell expressing the GFP-Cherry fusion protein (or co-expressing GFP and Cherry, or any protein pair of interest) in channel 1. Column AE (AE5 and up): measured fluorescence intensity of a cell expressing the GFP-Cherry fusion protein (or co-expressing GFP and Cherry, or any protein pair of interest) in channel 2. Column AF (AF5 and up): measured fluorescence intensity of a cell expressing the GFP-Cherry fusion protein (or co-expressing GFP and Cherry, or any protein pair of interest) in channel 3. Columns AG, AH, and AI (AG5, AH5, AI5 and up, respectively): measured fluorescence intensities subtracted by mean background intensities in all 3 channels. Column AJ (AJ5 and up): Calculated α factor. Column AK (AK 5 and up): Calculated mean FRET efficiency E. Column AL (AL5 and up): calculated corrected acceptor-to-donor intensity ratio (Q). Examples for calculated parameters from a FRET experiment as expressed as mean and standard deviation: S1 = 0.2232 ± 0.0060. S2 = 0.2039 ± 0.0074. α = 1.9463 ± 0.1409. E = 0.2713 ± 0.0220. Please click here to download this File.
Supplementary Figure 1: GFP and Cherry absorption and emission spectra. Normalized absorption and fluorescence emission spectra of eGFP and mCherry. The excitation laser lines (488 and 543 nm) and filter transmissions used for the donor channel (ch1: 505-530 nm) and the transfer/acceptor channels (ch2 & 3: >585 nm) in the confocal microscope are marked by shading. For excitation of the acceptor, 561 or 590-nm laser lines can also be used. Source fluorescent protein data base (fpbase.org). Please click here to download this File.
The presented protocol details the use of the genetically coupled one-to-one fluorescent protein calibration probe for quantifying FRET using the detection of sensitized emission of the acceptor and quenching of the donor molecule by confocal microscopy. This method can be applied to assess protein interactions in the physiological context of the living cell in different subcellular compartments. Spatial resolution can be further improved by applying the presented algorithm to calculate FRET efficiencies in each pixel of an image (pixel-by-pixel FRET). Intensity-based determination of absolute FRET efficiencies requires the determination of the cross-talk, quantified here with the S factors, and the detection efficiency of donor and acceptor molecules by the given microscopic set-up, quantified by the α factor. Here, a protocol is provided that permits quantification of both cross-talk and detection efficiency. Direct coupling of the genetic information of fluorescent protein guarantees equimolar expression in live cells and thereby makes the determination of the α factor possible. Knowledge of the α factor in turn is a prerequisite for the quantification of the FRET efficiency in live cells. Critical steps in the provided protocol are the proper cloning of the directly coupled fluorescent protein chimera, sufficient time after transfection to allow for fluorescent protein maturation, use of fluorescent proteins in a similar cellular microenvironment, e.g., fluorescent proteins in cytoplasm and on cytoplasmic side of plasma membrane, and a stable microscope set-up.
Experimental set-up and algorithm presented here are designed for the GFP-Cherry FRET pair. We provide in the Supplementary File 1 the algorithm for different versions of cyan (ECFP, CyPet, mTFP1, Cerulean, mTurquoise2) and yellow fluorescent proteins (EYFP, Citrine, Venus, SYFP2, YPet). The advantage of these fluorescent protein pairs is a greater spectral overlap between the emission spectrum of the cyan and the absorption spectrum of the yellow protein (as compared to GFP-Cherry) resulting in somewhat higher R0 values and larger FRET efficiencies. However, for the same reason, the number of non-negligible crosstalk factors and their magnitudes are also larger (see Supplementary Text).
Limitations of quantitative microscopic ratiometric FRET in live cells need to be considered and is discussed. Prerequisite of using FRET for detecting molecular interactions is the application of fluorescent tags. Our protocol uses genetically encoded fluorescent proteins. Since these fluorescent proteins may be comparable in size (27 kDa) to the tagged protein of interest, they may change localization and function of the protein of interest. Therefore, both localization and functionality of any tagged protein of interest should be tested and compared with those of the endogenous unlabeled protein. Another critical point to keep in mind is the endogenous unlabeled pool of the protein of interest. The interactions of the labeled with unlabeled endogenous proteins will decrease the FRET signal. Ideally, all proteins of interest are labeled. This can be achieved by using cells which do not have the protein of interest endogenously (such as in the example given in Figure 2C and 2E), using cells derived from knock-in mice, using CRISPR modified cells, etc. Even with the use of pixel-by-pixel FRET, the signal of donor and acceptor molecules will be averaged over a diffraction-limited spot, the resolution limit of a confocal microscope. Therefore, it is impossible to resolve different donor populations within a diffraction-limited spot. There might be donors without an acceptor, or donors with multiple acceptors contributing to the average FRET signal. Dissecting different molecular subpopulations by FRET requires fluorescence lifetime imaging22. Another problem, especially at high expression levels is the so-called random FRET between fluorophores in close proximity without underlying interaction of the proteins of interest23. This random FRET can be significant in the plasma membrane given the 2-D confinement of the membrane proteins compared to freely diffusive cytoplasmic proteins. Therefore, control experiments should always be done such as deleting domains that mediate the presumed interaction between molecules and test for reduction in the detected FRET signal.
The uncertainty of exact stoichiometries of interacting proteins in live cells limits the use of FRET as a spectroscopic ruler to assess molecular distances between proteins in live cells. Even in the case of a strict one-to-one interaction, (i) the flexibility of the linker attaching the fluorescent protein to the protein of interest as well as (ii) the lack of knowledge of the relative orientation of the dipole moments of the dyes (determining the so-called κ2 factor) may confound exact distance measurements. Studies on acceptor fusion constructs with stiff linkers of different lengths separating the two fluorophores and displaying different FRET efficiencies have been reported elsewhere12. Conversely, assessing stoichiometries with FRET measurements in live cells is complex, however possible using the presented ratiometric FRET quantification. The inclined reader may be referred to previous work detailing a feasible approach5.
In summary, the quantitative FRET approach presented here permits the detection of (i) protein interactions in the physiological context of the living cell, (ii) changes in protein interactions over time and (iii) differences in interactions in different subcellular compartments down to the pixel-by-pixel level of a confocal image, and (iv) the dependence of the detected FRET signal upon the molecular acceptor-to-donor ratio expressed in a live cell.
The authors have nothing to disclose.
We would like to thank the Neuroscience Imaging Service at Stanford University School of Medicine for providing equipment and space for this project. This research was supported by intramural funding of the Stanford Cancer Institute and the Gynecologic Oncology Division Stanford as well as GINOP-2.3.2-15-2016-00026, GINOP-2.3.3-15-2016-00030, NN129371, ANN135107 from the National Research, Development and Innovation Office, Hungary.
0.5% Trypsin-EDTA without phenol red (10x) | Thermo Fisher Scientific | 15400054 | |
Clontech mCherry N1 vector | Addgene | 3553 | |
DMEM without phenol red | Thermo Fisher Scientific | 11054020 | |
Fugene 6 | Promega | E2691 | |
HEPES | Thermo Fisher Scientific | 15630080 | |
LabTek 8-well chambers #1.0 | Thermo Fisher Scientific | 12565470 | |
L-Glutamine (200 mM) | Thermo Fisher Scientific | 25030081 |