SEARCH

SEARCH BY CITATION

Keywords:

  • fluorescence (Förster) resonance energy transfer (FRET);
  • flow cytometry;
  • green fluorescent protein;
  • cyan fluorescent protein;
  • yellow fluorescent protein;
  • confocal microscopy;
  • spectrofluorometry

Abstract

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. APPENDIX A
  8. APPENDIX B
  9. LITERATURE CITED

Background

Use of distinct green fluorescent protein (GFP) variants permits the study of protein–protein interactions and colocalization in viable transfected cells by fluorescence (Förster) resonance energy transfer (FRET). Flow cytometry is a sensitive method to detect FRET. However, the typical dual-laser methods used in flow cytometric FRET assays are not generally applicable because they require a specialized krypton ultraviolet (UV) laser. The purpose of this work was to develop a flow cytometric method to detect FRET between cyan fluorescent protein (CFP; donor) and yellow fluorescent protein (YFP; acceptor) by using the 458-nm excitation from a single tunable argon-ion laser.

Methods

FUSE-binding protein (FBP) interacting repressor (FIR) and FBP are c-myc transcription factors and are known to interact physically. To examine their interaction within viable cells, FIR and the binding motif of FBP, the FBP central domain (FBPcd), were fused with CFP and YFP, respectively, and this pair of fluorescently-tagged proteins was used to detect FRET in vivo. Cells transfected with expression plasmids encoding a CFP-FIR fusion protein and YFP as a negative control, a CFP-YFP fusion protein as a positive control, or CFP-FIR and YFP-FBPcd fusion proteins were examined for FRET after excitation with a 458-nm line from a tunable argon-ion laser. FRET was measured as the ratio of YFP:CFP emission or as YFP emission at 564–606 nm. Conventional FRET using the 413-nm UV line from a krypton laser was examined for comparison. Fluorescence signals were separated with a customized optical filter configuration using 530-nm shortpass, 500-nm longpass, and 560-nm shortpass dichroics in addition to 488/30 nm (CFP), 530/30 nm (YFP), and 585/42 nm (FRET) bandpass filters. Further, a laser-scanning confocal microscopic photobleach technique was used to document that FRET occurred by showing that the intensity of donor CFP fluorescence increased after its acceptor YFP was photobleached. Steady-state spectrofluorometry was used to confirm and validate the results detected by flow cytometry.

Results

Upon excitation with the 458-nm line of the argon-ion laser, the enhancement of the acceptor YFP signal and the decrease of the CFP signal were easily detected in cells transfected with the CFP-YFP construct or CFP-FIR and YFP-FBPcd. Similarly, FRET was detected under these conditions when the YFP emission was assessed at 564–606 nm. A strong correlation was observed between the increase in the YFP:CFP ratio and the YFP emission detected at 564–606 nm, consistent with the conclusion that FRET was detected comparably by both methods. A conventional flow cytometric krypton UV-laser technique was also used to confirm that FRET occurred with the CFP-YFP fusion protein and from CFP-FIR [RIGHTWARDS ARROW] YFP-FBPcd. FRET also was confirmed by a confocal photobleaching technique, in which donor CFP intensity was enhanced after its acceptor YFP was photobleached. The flow cytometric and confocal microscopic results were confirmed by spectrofluorometry.

Conclusion

These results demonstrated the feasibility of flow cytometric detection of FRET signals from CFP to YFP by excitation with the 458-nm line from the tunable argon-ion laser. The method was as efficient as excitation with the krypton UV laser and therefore should make FRET a more generally available flow cytometric technique. Cytometry Part A 53A:39–54, 2003. Published 2003 Wiley-Liss, Inc.

Fluorescence (Förster) resonance energy transfer (FRET) is a dipole–dipole interaction that occurs when two fluorophores are located within a few nanometers of each other and the emission spectrum of one fluorophore (donor) overlaps the excitation spectrum of the other (acceptor) (1–6). The energy of the donor fluorophore transfers and excites the acceptor fluorophore. FRET is widely used in biological investigations, including enzyme modification and screening (1, 7–10), protein–protein interaction (9, 11–21), single-nucleotide polymorphisms (22–24), and analysis of regulatory sequences (12, 25–29).

In vivo analyses of gene expression and subcellular localization of target gene products frequently use green fluorescence protein (GFP) and its variants, blue fluorescent protein, cyan fluorescent protein (CFP), yellow fluorescent protein (YFP), and the unrelated, red fluorescent protein (5, 9, 11, 30–35). With the availability of GFP variants, examination of protein–protein interactions using blue fluorescent protein and GFP or CFP and YFP pairs in viable cells is possible by FRET (11, 16, 30, 35). Detection of the FRET signal can be accomplished by various means, including fluorescence lifetime image microscopy and spectrofluorometry. Flow cytometry can also be used for the measurement of FRET signals because of its capacity to compensate electronically for overlapping signals from the individual fluorescent proteins. Recently, a pair of the GFP variants, CFP and YFP, was used to measure FRET from the donor CFP, which was excited at an ultraviolet (UV) wavelength of 407/413 nm by a krypton laser, to an acceptor YFP, which was not excited at this wavelength (4, 35). However, quantitation of the degree of FRET requires estimation of the amount of the YFP and CFP signals and the FRET signal. This requires two lasers, a UV laser to provide the 407/413-nm wavelength for CFP excitation and an argon-ion laser to generate the 488-nm wavelength for YFP excitation and detection. The availability of the krypton UV laser is limiting for some laboratories. Moreover, the argon-ion laser produces a range of emissions (e.g., 488 nm) that have considerable overlap with the emission spectrum of CFP, making its use problematic. Therefore, a means for single-laser excitation to detect FRET would enable more laboratories to use this powerful research tool. In the present study, we describe an approach that permits efficient detection of FRET from CFP to YFP by using excitation with the 458-nm wavelength line from a tunable argon-ion laser.

MATERIAL AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. APPENDIX A
  8. APPENDIX B
  9. LITERATURE CITED

Plasmid Construction

The full-length FUSE-binding protein (FBP) interacting repressor (FIR) cDNA was cut from its yeast hybrid plasmid by double digestion by using EcoRI and ApaI (36) and cloned into pECFP-C1 and pEYFP-C1 (Clontech, Palo Alto, CA). The c-myc transcription factor FBP DNA-binding domain was obtained with double digestion by EcoRI and BamHI of the FBP central domain (FBPcd) in pcDNA1-amp (Invitrogen, Carlsbad, CA), and the fragment was inserted into the same sites of the pEYFP-C1 vector. A fusion of pECFP and YFP with a two–amino acid linker was constructed by amplifying the YFP coding region by polymerase chain reaction using oligos 5′AAG TCC GGA ATG GTG AGC AAG GGC GAG GA3′ and 5′ TCA AGA TCT CTT GTA CAG CTC GTC CAT GCC 3′ and inserting the fragment into the BspEI and BglII sites of pECFP-C1. Authenticity of each construct was confirmed by DNA sequence analysis.

Tissue Culture and Transfection

Culture of HEK 293 cells was carried out in Dulbecco's Modified Eagle's Medium with 10% fetal bovine serum. Transfection of 500 ng of the appropriate GFP variant plasmid into log-phase growth HEK 293 cells was achieved with Lipofectamine (Life Technologist, Gaithersburg, MD), according to the manufacturer's directions. Cells were harvested 24 h after transfection by trypsinization and resuspended in phosphate buffered saline with 1% bovine serum albumin for flow cytometric analysis.

Flow Cytometry Analysis and FRET

All data presented were collected with a FACSVantage SE (Becton Dickson, San Jose, CA) flow cytometer and cell sorter. This instrument is equipped with three lasers, including an Enterprise tunable argon-ion laser (regularly used at a wavelength of 488 nm and tuned to 458 nm as indicated for these experiments), a 635-nm Spectrum laser, and a krypton UV laser, which is tunable to a wavelength of 407/413 nm. For single-laser analysis, the primary argon-ion laser was tuned to 458 nm for CFP and YFP stimulations. For dual-laser analysis, the argon-ion laser was tuned to 488 nm for YFP excitation and detection, and the krypton UV laser was tuned to 413 nm for CFP excitation. Laser output power was consistently set at 50 and 150 mW for the 458/488-nm argon-ion laser and at 150 mW for the krypton UV laser.

When an excitation wavelength of 458 nm was used, the optical settings were modified from a previously described method (37–40), as shown in Figure 1. Briefly, the enhanced CFP fluorescence signal in FL2 was separated from YFP fluorescence with a 530-nm shortpass (SP) dichroic filter, reflected by a 500-nm longpass dichroic filter, and collected with a 480/30-nm bandpass (BP) filter. The YFP signals were separated with a 560-nm SP dichroic filter and collected with a 530/30-nm BP filter. All optical filters were purchased from Omega Optical (Brattleboro, VT). At 458-nm excitation, FRET was calculated as the ratio of YFP to CFP intensity. In addition, FRET emission was detected first by reflecting the red YFP emission with a 560-nm SP dichroic filter and then collecting with a 584/42-nm BP filter. Analysis of the red YFP emission optimized detection of the FRET signal by maximizing the ratio of the FRET component of the YFP emission from the background YFP emission. This was accomplished by electronic compensation.

thumbnail image

Figure 1. Diagram of the configuration of the FACSVantage SE optical systems using an excitation of 458 nm from the argon-ion laser. The optical setting are shown for simultaneous measurement of cyan fluorescent protein (CFP), yellow fluorescent protein (YFP), and fluorescence resonance energy transfer (FRET) signals by 488/30-nm, 530/30-nm, and 585/42-nm bandpass filters, respectively. Three different dichroic filters (530 shortpass [SP], 500 longpass [LP], and 560 SP) were used to split the fluorescence signal. SSC, side scatter.

Download figure to PowerPoint

Dual-laser FRET analysis also was carried out with the krypton UV laser set at a wavelength of 413 nm (UV) for excitation of CFP, and the argon-ion laser was set at a wavelength of 488 nm for excitation of the YFP (11). Signals for CFP and FRET were detected from the krypton UV laser excitation with a 480/30-nm BP filter (into FL2) and a 585/42-nm BP filter (into FL6), respectively. YFP signal detection after excitation by the argon-ion laser used a 530/30-nm BP filter (FL1).

Steady-State Fluorometric Measurement

Preparation of cell extracts was done as described previously (41), and western blotting was performed to monitor GFP fusion protein expression. Steady-state excitation and emission spectra of cell extracts containing CFP, YFP, or the CFP-YFP fusion protein were collected with a SPEX Fluorolog-3 spectrofluorometer (JY Horiba, Edison, NJ) in photon-counting mode. The exciting light was vertically polarized, whereas fluorescence was viewed at right angles to excitation through an emission polarizer oriented at 55 degrees with respect to the vertical (the so-called “magic angle”). This ensured that there were no orientation effects and that the emission being viewed was proportional to the total fluorescence intensity. Sample integration time was set so that a minimum of 60,000 counts was obtained at the peak wavelength to ensure good signal-to-noise ratio. For emission spectra, the excitation bandwidth was 5 nm, the emission bandwidth was 2.5 nm, and data were collected every 2.5 nm. Conversely, for excitation spectra, the excitation bandwidth was 2.5 nm, the emission bandwidth was 5 nm, and data were collected every 2.5 nm. Sample path length was 3 mm. The exciting light intensity was monitored simultaneously with the sample fluorescence intensity to allow correction for variations in the exciting light intensity with time. Baselines were recorded with an extract from untransfected cells and then subtracted from the respective fluorescence signals. In virtually all cases, the baseline made very little (several percentages, at most) contribution to the total emission intensity. The only exceptions were for those samples that had very little fluorescence intensity. A minimum of two measurements was made for each spectrum by using separately prepared cell extracts. Except for those factors that depended on absolute protein concentration, the results for the different extracts were in excellent agreement.

Western Blot

FIR and CFP or YFP western blots were performed as previously described (41, 42). FIR and FBP antibodies were generously provided by Dr. David Levens of the National Cancer Institute, and GFP antibody, which reacts with CFP and YFP molecules, was purchased from Roche (Indianapolis, IN).

FRET by Confocal Microscopy

HEK 293 cells transfected with CFP and YFP fusion plasmids were examined routinely with a 100× objective. Confocal microscopic images were obtained with a Carl Zeiss laser-scanning microscope using LSM 510 software. An excitation wavelength of 458 nm and an emission wavelength of 480–500 nm were used for CFP, and an excitation wavelength of 514 nm and an emission wavelength of 515–545 nm were used for YFP.

FRET was assessed and quantitated according to an acceptor photobleaching method that was developed for laser-scanning confocal microscopy (43). The experimental protocol assessed the extent of FRET by measuring the donor fluorescence before (Da) and after (D) photobleaching of the acceptor. CFP and YFP images were assessed repetitively over a 38-s interval. After 12 s of observation, YFP was photobleached with the full power of the 514-nm line, and measurement of CFP and YFP emissions was monitored over an additional 12 s. The amount of energy transfer detected by confocal microscopy (FRETc) was calculated as the ratio of donor fluorescence in the presence or absence of the acceptor.

  • equation image

where Da is the fluorescence intensity of the donor in the presence of the acceptor (i.e., before photobleaching) and D is the fluorescence intensity of the donor alone (i.e., after photobleaching).

The ratio of D to Da is equal to or less than 1.0 in the absence of FRET. If D/Da is greater than 1.0, then FRET is considered to have occurred.

Statistical Analyses

The ratio D/Da was compared with the null hypothesis value of 1.0 by one-group t tests.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. APPENDIX A
  8. APPENDIX B
  9. LITERATURE CITED

FRET Detection by Spectrofluorometry

The goal of this study was to develop a means of detecting CFP [RIGHTWARDS ARROW] YFP FRET with the use of single-laser excitation at 458 nm. To document that this approach was feasible, cells were transfected with plasmids encoding CFP or YFP alone, cotransfected with CFP and YFP plasmids, or transfected with a CFP-YFP fusion plasmid. Comparable amounts of the expressed protein were detected (Fig. 2A).

thumbnail image

Figure 2. Expressions of cyan fluorescent protein (CFP), yellow fluorescent protein (YFP), and the CFP-YFP fusion protein and emission spectra after excitations at 514 and 458 nm. A: Green fluorescent protein immunoblot shows relatively equal amounts of fluorescent proteins in the spectrofluorometric assay. Sequential dilutions (5, 2.5, and 1.25 μl) of protein extracts were performed for each sample. B: Spectrofluorometric analysis shows the identical YFP emission spectra for YFP and CFP-YFP fusion proteins. Curves show the emission spectra of the CFP-YFP fusion protein and free CFP and YFP normalized for molar concentration, as described in Appendix A. C: Emission spectrum shows fluorescence resonance energy transfer (FRET) measurement at 458-nm excitation. Curves show the emission spectra of CFP-YFP and free CFP and YFP normalized for molar concentration (free CFP × 0.69, free YFP × 0.37). In addition, curves normalized for the amount of energy transfer (ET) are shown (1.26 × YFP, CFP + YFP + ET, CFP/0.68), as described in Appendix A. The emission regions of CFP, YFP, and FRET detected by flow cytometry are shown as FL2, FL3, and FL6, respectively.

Download figure to PowerPoint

Initial experiments used steady-state spectrofluorometry to examine the excitation and emission spectra of the expressed GFP variant proteins. Results from those measurements are presented in Figure 2. The initial experiments examined the emission spectrum of the CFP-YFP fusion protein after excitation at 514 nm to document that the YFP component of the fusion protein had similar properties to the native YFP protein. Excitation at 514 nm was chosen because CFP is not excited at this wavelength. Figure 2B shows the emission spectra of free CFP (black line), free YFP (red line), and the fusion protein (solid circles) over the wavelength range of 525–600 nm after excitation at 514 nm. The relative fluorescence intensity (photons per second) is plotted against the emission wavelength (nm), and all three spectra were corrected to have the same exciting light intensity. The effective molar ratios of free CFP and free YFP were normalized as described in Appendix A. As can be seen, native CFP (black line) or CFP in the fusion protein (dotted line) had relatively little, if any, emission intensity at this excitation wavelength. This finding was consistent with its published absorption spectrum in buffer (44) and the excitation spectrum we measured in cell extracts (Appendix A), which indicated that CFP has relatively low absorbance at 514 nm. It should be noted that we could not directly measure the absorption spectra of any of the three proteins because of the strong wavelength−4-dependent light scattering of the cell extracts. The excitation wavelength of 514 nm is essentially at the absorbance peak of YFP (Appendix A). The findings that the normalized emission spectrum of free YFP (red line) produces a nearly perfect fit to the fusion protein emission spectrum (dotted line) and that CFP has essentially no fluorescence emission under these conditions indicated that, for excitation at 514 nm, only the YFP moiety of the fusion protein emits and then only through its direct excitation. Importantly, these results indicated that the spectral properties of the YFP component of the fusion protein are similar to those of free YFP.

Figure 2C shows the emission spectrum of the proteins after excitation at 458 nm. Emission was monitored over the wavelength range of 475–600 nm, and the individual spectra were again corrected to have the same exciting light intensity. As was the case for excitation at 514 nm, the contribution that directly excited YFP makes to the fusion protein spectrum (red line) was normalized by multiplying the free YFP spectrum by the factor of 0.37 used in Figure 2B (Appendix A). The contribution of the CFP moiety (black line) to the fusion protein spectrum was normalized by multiplying the free CFP spectrum by a factor of 0.69 so that it fit the fusion protein spectrum at 475 nm, where YFP does not emit (Appendix A). Because YFP can be directly excited at 458 nm, the YFP moiety makes a significant contribution to the fusion protein's emission spectrum beginning at 500–525 nm. Direct excitation of YFP and CFP alone, however, cannot fully account for the fusion protein's spectrum. To model the putative contribution of YFP because of energy transfer from the CFP moiety, we again used the rescaled spectrum of the former, this time multiplied by a factor of 1.26 (Appendix A). The result is shown by the green line in the figure. The sum of these three components (i.e., emission from CFP with some loss of intensity because of energy transfer to YFP, and emission from YFP through its direct excitation and energy transfer from CFP) is shown by the blue line in the Figure 2B, which is seen to superimpose with the emission spectrum of the fusion protein. Also shown (dashed line) is the calculated level of CFP (donor) fluorescence in the absence of the 38% FRET, determined as shown in Appendix A.

The conclusions from these studies are that the fusion protein contained functionally intact and unaltered CFP and YFP moieties and that excitation at 458 nm could be used to detect FRET efficiently as a decrease in the CFP emission spectrum measured below 525 nm or an increase in the YFP emission spectrum above 525 nm. Importantly, the increase in the YFP emission could be observed at all wavelengths longer than 525 nm.

FRET Measurement by Flow Cytometry Using a Single-Laser Excitation Wavelength of 458 nm

The major goal of this study was to develop a means of detecting CFP [RIGHTWARDS ARROW] YFP FRET by flow cytometry with single-laser excitation at 458 nm. The interaction between the c-myc transcription activator FBPcd and its repressor FIR was chosen to examine the feasibility of using this approach as a means of detecting protein–protein interaction in viable cells. This interaction was chosen for study because it was documented to occur by in vitro immunoprecipation and the yeast two-hybrid system (36). The comparable expression of CFP-FIR and YFP-FBPcd in HEK 293 cells was demonstrated by GFP, FIR, and FBP western blots (data not shown).

Cells transfected with CFP-FIR or YFP-FBPcd alone produced the expected signal when stimulated with the 458-nm excitation of the argon-ion laser (Fig. 3A). Mock-transfected cells produced no signal, whereas there was no CFP signal from YFP-FBPcd–transfected cells and no YFP signal from CFP-FIR–transfected cells. The cells cotransfected with the CFP-FIR and YFP plasmids served as a negative control for FRET and a positive control for CFP and YFP double transfectants (Fig. 3A). Notably, when the cells were transfected with the positive control plasmid encoding the fusion protein, CFP-YFP, there was a dramatic augmentation of the FRET signal detected in FL6 (Fig. 3A). Although the spectrofluorometric results indicated that FRET could be detected effectively at any point in the YFP emission spectrum greater than 525 nm, the use of electronic compensation with flow cytometry made it possible to detect FRET clearly in FL6 (564–606 nm). This increase in the FL6 FRET signal was a clear indication of the interaction of CFP with YFP in vivo at the individual cell level.

thumbnail image

Figure 3. Flow cytometric bivariate profiles of yellow fluorescent protein (YFP; FL3) intensity versus cyan fluorescent protein (CFP; FL2) intensity or fluorescence resonance energy transfer (FRET; FL6) intensity for each sample using an excitation of 458 nm from the argon-ion laser. A: As controls, cells mock transfected or transfected with CFP plus FUSE-binding protein (FBP) interacting repressor (FIR) or YFP plus FBP central domain (FBPcd) alone did not show a FRET signal. In contrast, CFP-YFP or CFP-FIR + YFP-FBPcd transfected cells manifested FRET signals. In the positive control, CFP-YFP transfected cells, the regions for double positive, single YFP-positive, and double negative cells are indicated as R1, R2, and R3, respectively. B: Histograms of FRET intensity from each gated region shown in A. MFI, mean fluorescence intensity of emission for each gated subpopulation.

Download figure to PowerPoint

It was noteworthy that most cells transfected with the positive control fusion protein CFP-YFP were positive only for the YFP signal (Fig. 3A, bottom left panel), whereas most cells cotransfected with CFP-FIR and YFP emitted CFP and YFP signals. Similar to the cells transfected with the CFP-YFP fusion protein, cells cotransfected with CFP-FIR and YFP-FBPcd also exhibited a diminished CFP signal (Fig. 3A). To determine the reason the CFP:YFP ratio was skewed toward YFP-only cells, regions R1, R2, and R3 (Fig. 3A) were gated for double-positive, YFP-positive, and double-negative cells, respectively. The data are shown in Figure 3B as histograms of FRET intensity detected at FL6. Cells in R2 had greater FRET intensity than those in R1. Moreover, in both populations there was a positive correlation between YFP intensity and FRET intensity and a negative correlation between CFP intensity and FRET. These results showed that in some cells there was nearly complete FRET from CFP to YFP and, consequently, the nearly complete loss of CFP intensity in some of the transfected cells shown in R2. These results implied that the loss of the CFP signal and the strength of FRET signal correlate with the extent of protein–protein interaction.

Intensity of the FRET Signal From the Acceptor Fluorophore Is Augmented, Whereas the Signal of the Donor Is Decreased

These results were consistent with the conclusion drawn from flow cytometry that, when protein–protein interaction occurs, the intensity of the donor fluorophore emission is diminished, whereas the signal intensity of the acceptor fluorochrome emission is increased. To document this observation, the intensities of CFP, YFP, and FRET signals were analyzed in greater detail. As predicted, the relative donor CFP signal intensity detected in FL2 decreased in the cells transfected with the positive control CFP-YFP plasmid or the CFP-FIR and YFP-FBPcd plasmids as the acceptor YFP signal intensity in FL3 and the FRET signal in FL6 increased (Fig. 4A,B). To confirm that the signal detected in FL6 actually reflected FRET activity, FRET was also calculated from the ratio of YFP to CFP emission as previously described (9, 30). To calculate the YFP:CFP ratio, the YFP and CFP intensities were first normalized based on the intensity of emission from the mock-transfected control and then the YFP:CFP ratio was calculated for the negative control (CFP-FIR + YFP), the positive controls CFP-YFP, and the experimental samples (CFP-FIR + YFP-FBPcd). The ratios were 1.02, 6.96, and 2.52, respectively (Fig. 4B). These results confirmed that FRET and, by implication, protein–protein interaction occurred in these samples. These data were a clear indication of the interaction of FIR with FBPcd in vivo at the individual cell level, as previously shown at the population level by immunoprecipitation and the yeast two-hybrid system (36).

thumbnail image

Figure 4. Quantitation of donor cyan fluorescent protein (CFP), acceptor yellow fluorescent protein (YFP) and fluorescence resonance energy transfer (FRET) intensities. A: Flow cytometric analysis of CFP (FL2), YFP (FL3), and FRET (FL6) intensities in mock-transfected cells and cells transfected with CFP and FUSE-binding protein (FBP) interacting repressor (FIR) plus YFP, CFP-YFP, and CFP-FIR plus YFP and FBP central domain (FBPcd) after excitation with the 458-nm line from the argon-ion laser. B: Magnitude of changes in intensity of fluorescence for the cells shown in A and the emission ratio of YFP to CFP. Relative fluorescence intensity was calculated as the ratio of emission intensity in the samples to that in the mock control.

Download figure to PowerPoint

Importantly, these data also indicated that FRET can be detected by flow cytometry in two different ways: first, by analyzing the YFP:CFP ratio in FL3 and FL2, respectively, and, second, by assessing the emission at 564–606 nm (FL6). There was good correlation between the two methods, as shown by the observation that cells in R2 with the highest YFP:CFP ratio (36.3) had the highest emission in the FL6 FRET channel (mean fluorescence intensity of 57.4), whereas the cells in R1 with a moderate YFP:CFP ratio (4.8) had a lower emission in FL6 (mean fluorescence intensity of 26.2; Fig. 3A,B). These results indicate that, with nearly equal expression and appropriate electronic compensation, the emission detected in the red range as FL6 is a simple and direct method to detect FRET.

YFP Signal Intensity Remains Constant When Cells Transfected With CFP-YFP or Cotransfected With CFP + YFP Are Excited With the 488-nm Wavelength Laser

The data suggested that the acceptor fluorophore emission intensity increases after stimulation of the donor fluorophore because of FRET. However, this increased signal may have had other explanations. The greater YFP signal could have been related to its overexpression in the cells cotransfected with CFP-FIR and YFP. To rule out this possibility, the same samples previously tested with the argon-ion laser set at the 458-nm wavelength were analyzed with the argon-ion laser tuned to 488 nm. The 488-nm line excites YFP and not CFP. The mean fluorescent intensity of YFP was the same for CFP-FIR + YFP and CFP-YFP (Fig. 5). These results were consistent with those found by spectrofluorometry and indicated that the functional YFP signals are similar in both transfectants when directly stimulated. Moreover, the findings implied that the differences previously noted represent differential FRET activity and not differences in YFP expression or functional activity. To confirm that the loss of CFP emission in cells transfected with the CFP-YFP fusion plasmid did not result from misfolding of the CFP fusion protein or diminished expression of the protein, cells were excited with a krypton UV laser at 407/413 nm to excite CFP but not YFP. If CFP in the CFP-YFP fusion protein or in CFP-FIR were not properly folded, CFP emission and CFP-dependent FRET would not be detected. As shown in Figure 6, the CFP and FRET signals were detected in cells transfected with CFP-YFP or cotransfected with CFP-FIR and YFP-FBPcd, but not in cells cotransfected with CFP-FIR and YFP. Notably, the intensity of the CFP emission decreased in the CFP-positive cells transfected with CFP-YFP, but not with CFP-FIR + YFP, which likely reflected energy transfer to YFP. Importantly, the percentage of FRET-positive cells in the same sample detected after stimulation with the 458-nm wavelength of the argon-ion laser (at FL6) or the 407/413-nm wavelength of the krypton UV laser was not significantly different (Fig. 6B), implying that stimulation with 458 nm is sufficient to detect nearly all FRET activity by flow cytometry. In conclusion, the FRET signals detected after excitation with 458 nm resulted from FRET from CFP to YFP and not from alterations in CFP or YFP expression.

thumbnail image

Figure 5. Analysis of yellow fluorescent protein (YFP) fluorescence intensity (FL1) upon excitation with the 488-nm line of the argon-ion laser that excites YFP but not cyan fluorescent protein (CFP). YFP expression is shown from the following samples: CFP and FUSE-binding protein (FBP) interacting repressor (FIR) plus YFP (negative control), CFP-YFP (positive control), and CFP-FIR + YFP and FUSE-binding protein central domain (FBPcd).

Download figure to PowerPoint

thumbnail image

Figure 6. A: Fluorescence resonance energy transfer (FRET) analysis by conventional excitation with the 407/413-nm line from the krypton ultraviolet laser for the same samples shown in Figure 5. B: Comparison of the FRET signals after excitation with the 458-nm line of the argon-ion laser and the 407/413-nm line of the krypton ultraviolet laser. CFP, cyan fluorescent protein; FBPcd, FUSE-binding protein central domain; FIR, FUSE-binding protein interacting repressor; MFI, mean fluorescence intensity; YFP, yellow fluorescent protein.

Download figure to PowerPoint

FRET Detection by Confocal Microscopy

In the final experiments, confocal microscopy was used to document that the FRET detected by flow cytometry indeed reflected protein–protein interaction between CFP and YFP. The approach involved photobleaching (43, 45–47). After colocalization of CFP and YFP was demonstrated, FRET was documented by photobleaching the acceptor YFP and analyzing the intensity of CFP emission. If FRET had occurred, photobleaching of YFP would have increased CFP emission (Fig. 7A).

thumbnail image

Figure 7. Documentation by confocal microscopy that FRET occurs in cells transfected with the CFP-YFP fusion protein and those cotransfected with CFP-FIR and YFP-FBPcd. A: Diagram shows the principle underlying the detection of FRET by photobleaching and confocal microscopy. When CFP- and YFP-tagged proteins interact with each other, energy from 458-nm–excited CFP molecules are directly transferred to YFP molecules, thereby decreasing CFP emission detected at 480–500 nm and increasing YFP emission detected at 515–545 nm. However, when the acceptor YFP is photobleached, emission intensities of the donor CFP are increased because of the inability of the photobleached YFP to accept energy transfer from CFP. B: CFP and YFP images immediately before and after photobleaching in cells transfected with a plasmid encoding the CFP-YFP fusion protein. Three regions, R1, R2, and R3, were analyzed for CFP intensity, photobleached, and reanalyzed for CFP intensity. Pseudo-colored YFP and CFP images are shown. FRET efficiency as calculated from CFP emission before and after photobleaching is shown at the bottom of images, with the statistical significance. C: The changes for CFP and YFP intensities of R1, R2, and R3 shown in B before and after photobleaching. The time was divided into 10 intervals at which CFP and YFP images were assessed. The locations of upward and downward arrows indicate the beginning and end of photobleaching, respectively. D: Laser-scanning confocal microscopy of cells transfected with CFP-FIR and YFP-FIR as a negative control for FRET. Fluorescence was excited with 458 nm, and CFP and YFP emissions were detected at 480–500 nm and 515–545 nm, respectively. FIR was localized mostly in the nucleus. Eight spots in eight individual cells showing apparent colocalization of CFP and YFP fluorescences were photobleached at full power with the 514-nm line to inactivate YFP and subsequently analyzed for the intensity of CFP emission. FRET efficiency as calculated from the CFP emission before and after photobleaching is shown at the bottom of the images with the statistical significance. E: FRET between CFP-FIR and YFP-FBPcd as documented by laser-scanning confocal microscopy. CFP and YFP images before and after photobleaching in cells cotransfected with CFP-FIR and YFP-FBPcd are shown. FBPcd and FIR were confined largely to the nucleus. Three regions, R1, R2, and R3, were analyzed for CFP emission before and after photobleaching. FRET efficiency as described in Materials and Methods is shown at the bottom of the images with the statistical significance. CFP, cyan fluorescent protein; FBPcd, FUSE-binding protein central domain; FIR, FUSE-binding protein interacting repressor; FRET, fluorescence resonance energy transfer; FRETc, amount of energy transfer detected by confocal microscopy; MFI, mean fluorescence intensity; YFP, yellow fluorescent protein.

Download figure to PowerPoint

As a positive control, photobleaching of YFP documented significant FRET in cells transfected with the CFP-YFP fusion protein (Fig. 7B,C). Thus, photobleaching of the YFP component of the CFP-YFP fusion protein resulted in significant enhancement in the CFP emission. The magnitude of this effect over time for CFP and YFP for three separate regions within one cell (R1, R2, and R3) is shown in Figure 7C. In contrast, no FRET was observed in the cells cotransfected with YFP-FIR and CFP-FIR because photobleaching of YFP had no effect on the intensity of CFP emission even though there was colocalization of CFP and YFP (Fig. 7D). However, there was significant FRET in the cells cotransfected with the CFP-FIR and YFP-FBPcd plasmids, and there was significant enhancement of CFP intensity after its acceptor, YFP, was photobleached (Fig. 7E). The magnitude of FRET in these cells was comparable to that noted with cells transfected with the CFP-YFP fusion plasmid, documenting that protein–protein interaction between FIR and FBPcd occurred in these cells.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. APPENDIX A
  8. APPENDIX B
  9. LITERATURE CITED

Although the phenomenon of FRET was observed by Perrin at the turn of the previous century (2, 5) and its basis was described as long-range dipole–dipole interactions between fluorescent molecules more than 50 years ago (48), only recently has FRET become more widely used for a variety of applications. With regard to its applicability in biomedicine, FRET can be used to obtain in vivo structural information, analyze single-nucleotide polymorphisms and protein–protein interactions, and monitor the roles of regulatory sequences of gene promoters in gene transcription.

With the complete human genome sequence being revealed rapidly, it is not difficult to predict the function of new genes by searching for homology to the known functional genes (49–53). More challenging is to connect the functions of different gene products within the complete network of human physiologic complexity. Because of the sheer size of the genome, methods that can streamline or automate these processes are highly desirable. The FRET technique as measured by flow cytometry has the advantages of being able to measure protein–protein interactions in vivo rapidly and quantitatively in a vast number of cells, making it potentially a more practical tool than other methods more commonly used to analyze it.

A CFP-YFP fluorochrome pair has been shown to be an excellent FRET tool to detect protein–protein interactions in vivo without interference, regardless of the complexity and heterogeneity of the system (9, 11, 30, 35). The most important advantage of this pair is that CFP and YFP have sufficient separation in their excitation spectra so that CFP as the donor fluorophore can be excited by the 413- or 407-nm emission of the UV laser, but not the acceptor YFP. In addition, their emission spectra are sufficiently separated so that they can be independently monitored. However, achieving maximal efficiency is difficult and must be carried out in a laboratory that has access to a sophisticated flow cytometer equipped with specific lasers, including a krypton UV laser to excite the CFP. Moreover, the FRET method in such a setup occupies the detector signals from two lasers, one of which might otherwise be used for other signals. Therefore, a method using the commonly employed tunable argon laser in a single-laser system to detect in vivo protein–protein interactions by FRET is highly desirable.

The argon-ion laser tuned to a wavelength of 458 nm recently was reported to have appropriate energy to excite CFP, YFP, and GFP simultaneously with sufficient separation of their respective emission signals (37–40). However, YFP is not optimally stimulated by the 458-nm line of argon-ion laser in comparison with its excitation with the 488-nm line of the argon-ion laser. In theory, this makes it possible for YFP to be excited indirectly by energy transfer from a donor fluorophore such as CFP and for the resultant increase in emission to be detected as a FRET signal. Importantly, the fact that YFP is partly stimulated by the 458-nm line makes it possible to estimate the molar ratios of CFP and YFP and thereby estimate the relative degree of FRET.

To test this hypothesis, we used a model of protein–protein interaction previously documented by immunoprecipation and the yeast two-hybrid system. The model system involved the FIR protein that was isolated by its ability to interact with the DNA-binding motif, FBPcd (36), which positively regulates c-myc gene transcription. We constructed plasmids encoding CFP-FIR, YFP-FBPcd, and the positive control fusion protein CFP-YFP. As predicted, the cells transfected with CFP-YFP or CFP-FIR + YFP-FBPcd manifested FRET signals, which did not exist in the cells cotransfected with the CFP-FIR + YFP (negative control) or any individual fluorochrome alone. These data were consistent with a characteristic of FRET that the intensity of the detected emission signal of the donor decreases while the acceptor signal intensity increases, which leads to a significant increase in the YFP:CFP ratio.

The increase in the YFP fluorescence coupled with the decrease in the detected emission intensity of the CFP donor is consistent with the conclusion that authentic protein–protein interactions were detected. The changes in CFP and YFP emissions reflected FRET and not differences in levels of YFP expression, because the YFP signal intensity remained almost unchanged for the cells transfected with CFP-FIR + YFP, CFP-YFP, or CFP-FIR + YFP-FBPcd when using an excitation wavelength of 488 nm, where CFP is not excited. In addition, changes in the CFP emission intensity were not related to differences in CFP expression or CFP misfolding, as documented by spectrofluorometry and flow cytometry. FRET did not result from abnormalities in the YFP emission produced by creation of the CFP-YFP fusion protein because the YFP and CFP-YFP were identical after stimulation with 514 nm that did not excite CFP. Importantly, the FRET signals detected after excitation with the 458-nm line were confirmed by using excitation with the conventional 413-nm krypton UV laser.

Notably, upon excitation at 458 nm, most CFP-YFP and CFP-FIR + YFP-FBPcd transfected cells showed diminished expression of CFP fluorescence. In contrast, nearly all CFP-FIR + YFP cotransfected cells remained doubly positive for CFP and YFP. Analysis of the profiles of the FRET signal intensity versus the YFP and the CFP signal intensities showed that the YFP-positive cells but not the double-positive CFP and YFP cells manifested the greatest FRET signals, implying that a large fraction of the donor energy had been captured by the acceptor. It is interesting to note that, when CFP was stimulated more efficiently with the krypton UV laser, a lesser diminution of the CFP signal was noted even when FRET occurred within a comparable number of cells. These results implied that this method can be used not only to detect protein–protein interaction but also to estimate the stoichiometry of the interaction at an individual cell level.

The use of this method permits FRET to be detected in two ways, the previously reported ratio of YFP to CFP emission and the detection of YFP emission in the red range (FL6, 564–606 nm). As expected, cells with the greatest ratio of YFP:CFP emission also exhibited the greatest red emission, indicating a strong correlation between FRET detected as red emission and FRET detected as a YFP:CFP emission ratio. Therefore, YFP signals detected in the red range of emission (564–606 nm) with electronic compensation can be used as a simple means to estimate FRET.

Laser-scanning confocal microscopy also was used to confirm that FRET between CFP and YFP detected by flow cytometry actually reflects protein–protein interaction at close distances. Photobleaching was used to inactivate the acceptor YFP, and then CFP intensity was assessed to document that FRET had occurred as previously reported (43, 45–47). After YFP photobleaching of the YFP-CFP fusion or YFP-FBPcd, the intensity of CFP emission from the fusion protein or CFP-FIR significantly increased, confirming that FRET detected by flow cytometry reflects interactions between cellular proteins in close proximity.

Results from the steady-state spectrofluorometric measurements confirmed that FRET occurs when CFP is excited at 458 nm. Beyond the range of CFP absorbance at 514 nm, the fusion protein emission spectrum can be described entirely in terms of emission from YFP after its direct excitation. When excited at 407 nm, where YFP does not absorb, the fusion protein spectrum still contains a contribution from YFP caused by FRET from CFP. At intermediate wavelengths, such as 458 nm, where the CFP and YFP moieties absorb, emission from YFP because of FRET must be invoked to explain the results of the measurements. The results from the analysis by spectrofluorometry confirmed that excitation at 458 nm can be used to detect FRET between CFP and YFP.

It should be emphasized that the single-laser flow cytometric approach can detect the presence of FRET, but its use to measure FRET efficiency accurately is problematic. This relates to the inability to assess the molar ratio of the donor and acceptor molecules accurately after single-laser excitation. Although we were able to estimate the FRET efficiency by spectrofluorometry using a CFP-YFP fusion protein, this would not be practical with the flow cytometer. Moreover, using this approach to estimate FRET efficiency between individual protein–fluorophore constructs would not be feasible by either approach. As a result, the single-laser flow cytometric approach can be used to estimate whether FRET occurs between two independently fluorescent-tagged proteins but cannot be used to quantitate the interaction accurately. The simplicity of the method of measuring molecular interactions therefore is counterbalanced by the inability to quantitate the interaction fully.

Together, the data indicated that the flow cytometric method described in this study effectively detects protein–protein interactions within individual cells when using the YFP:CFP emission ratio change or the isolated red component of YFP emission assessed at 564–606 nm. Although the method depends on there being nearly equimolar quantities of donor and acceptor, the YFP:CFP emission ratio after 458-nm excitation results in a FRET efficiency equivalent to that for the conventional excitation at 407 nm. Notably, the successful determination of FRET when using only the red component of YFP emission is even more dependent on equimolar amounts of donor and acceptor, so that a measurement would be useful only as a qualitative indicator. Importantly, FRET determined by flow cytometry was confirmed by spectrofluorometry and a confocal photobleaching technique. This simple single-line dual-channel flow cytometric technique should be useful for detecting a variety of protein–protein interaction in vivo, especially until newly developed solid-state lasers emitting at 405 and 488 nm become more widely available.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. APPENDIX A
  8. APPENDIX B
  9. LITERATURE CITED

We thank Dr. David Levens of NCI for providing FBP and FIR cDNA and antibodies, and Dr. Kenneth Spring of NHLBI for many helpful discussions.

APPENDIX A

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. APPENDIX A
  8. APPENDIX B
  9. LITERATURE CITED

EMISSION AND EXCITATION SPECTRA

Experiments were carried out to measure the emission and excitation spectra of CFP, YFP, and CFP-YFP fusion protein in cell lysates. The upper panel of Figure 8A shows the emission spectra of free CFP (black line), free YFP (red line), and the CFP-YFP fusion protein (filled circles) over the range of 475–600 nm for excitation at 407 nm. YFP alone exhibited little fluorescence intensity because of its very low absorbance at this excitation wavelength. The absorbance was judged from the published absorption spectrum (44) and our own excitation spectral measurement shown in Figure 8. In Figure 8, we reduced the YFP intensity by the same factor of 0.37 used in Figure 2B to correct for the apparent difference in concentration. This was accomplished by scaling the YFP emission to that of the CFP-YFP fusion protein after stimulation at 514 nm, where CFP does not absorb or emit. The black line shows the emission spectrum of free CFP reduced by a factor of 0.69, so that it fits the fusion protein emission spectrum at the short-wavelength end of its emission spectrum (475 nm), where YFP does not emit. Unlike the case for YFP, however, the correction factor for CFP presumably is not just the ratio of the fusion protein to free CFP concentrations. Rather, it should be the product of their molar ratio and the fraction of absorbed light that is emitted directly and not lost through energy transfer to YFP (Eq. AB4). Given that YFP makes little contribution to the fusion protein spectrum through direct excitation (red line), it is clear from the graph that FRET generates YFP emission from the fusion protein. To account for the emission of the fusion protein at 525 nm, for example, we must assume a third component (green line), which we take to be emission from YFP caused by energy transfer from CFP. This was modeled by multiplying the free YFP spectrum at 458-nm excitation by a factor of 0.69 (Fig. 2C). The sum of the three contributions (i.e., emissions from CFP and from YFP through its direct excitation and energy transfer from CFP) is shown by the blue line in Figure 8A, which is seen to fit the fusion protein spectrum quite well.

thumbnail image

Figure 8. Results of a spectrofluorometric analysis of emission and excitation spectra for CFP, YFP, and the CFP-YFP fusion protein. A: Upper panel. Emission spectra of free CFP (black line), free YFP (red line), and the CFP-YFP fusion protein (solid circles) over a range of 475–600 nm after excitation at 407 nm. The emission spectra of CFP and YFP have been scaled to account for the differences in protein concentration. The calculated ET from CFP [RIGHTWARDS ARROW] YFP is represented by the green line. Lower panel. Emission spectra of the proteins after excitation at 458 nm, with the estimated ET. B: Upper panel. Excitation spectra of CFP, YFP, and the CFP-YFP fusion protein over the range of 350–450 nm and monitored at 500 nm. Lower panel. Excitation spectra of the proteins over a range of 350–550 nm and monitored at 575 nm. CFP, cyan fluorescent protein; ET, energy transfer; YFP, yellow fluorescent protein.

Download figure to PowerPoint

The lower panel of Figure 8A shows the emission spectrum of the proteins after excitation at 458 nm and is similar to that shown in Figure 2C. Emission was monitored over the wavelength range of 475–600 nm, and the individual spectra were again corrected to have the same exciting light intensity. As was the case for 407-nm excitation, the contribution that directly excited YFP makes to the fusion protein spectrum (red line) was modeled by multiplying the free YFP spectrum by the factor of 0.37 obtained for excitation at 514 nm (Fig. 2B, upper panel). The contribution of the CFP moiety (black line) to the fusion protein emission spectrum was again modeled by multiplying the free CFP spectrum by a factor of 0.69 so that it would fit the fusion protein emission at 475 nm, where YFP does not emit. Because YFP is directly excited at 458 nm, the YFP moiety makes a greater contribution to the fusion protein's spectrum near 525 nm than was the case for excitation at 407 nm. Direct excitation of YFP and CFP alone cannot fully account for the fusion protein's spectrum, however. To model the putative contribution of YFP because of energy transfer from the CFP moiety, we used the rescaled spectrum of YFP, this time multiplying it by a factor of 1.26. This was determined by scaling the YFP emission spectrum so that the sum of three components (emissions from CFP, YFP through direct excitation, and YFP because of FRET from CFP) equaled the emission spectrum of the fusion protein. The result is shown by the green line in the lower panel of Figure 8A. The sum of three components (i.e., emissions from CFP with some loss of intensity because of energy transfer to YFP and emissions from both YFP through its direct excitation and energy transfer from CFP) is shown by the blue line in the figure, which nearly superimposes with the emission spectrum of the fusion protein.

The upper and lower panels of Figure 8B show the excitation spectra of CFP, YFP, and the CFP-YFP fusion protein. In the upper panel, the samples were excited over the range of 350–475 nm, and the steady-state fluorescence intensity at 500 nm was monitored to favor absorption and emission by CFP. The spectra were corrected to have the same constant exciting light intensity across the range of excitation wavelengths. The red line shows the excitation spectrum of YFP, which is seen to have relatively little absorbance or emission intensity in this case. The black line shows the excitation spectrum of CFP, the shape of which matches the published absorption spectrum of the fluorophore (44). (Our spectrum has its peak near 435 nm, and the shoulder on its red edge is centered near 452 nm. The corresponding published values are 433 and 453 nm, respectively (44)). The solid circles indicate the excitation spectrum of the fusion protein, and the blue line shows the free CFP spectrum multiplied by a factor of 0.72 to normalize the curves for protein concentration. That the latter produced an almost perfect fit to the fusion protein spectrum and that YFP had little or no signal in this region indicated that the fusion protein spectrum is related mostly to absorption by CFP under these conditions, as expected. The agreement between the free CFP correction factor used here (0.72) and that used for the emission spectra (0.69) indicated that the results of our different measurements are quite consistent.

The lower panel in Figure 8B shows the excitation spectra of the three fluorophores over the range of 350–550 nm, with emission observed at 575 nm. As for the upper panel, the red and black lines show the free YFP and CFP spectra, and the solid circles show the excitation spectrum of the fusion protein, all of which were corrected to have the same constant exciting light intensity across the range of excitation wavelengths. Under these conditions, emission from YFP is favored, although both YFP and CFP absorb in this region. In this case we find that a sum of 0.37 times the free YFP spectrum plus 1.3 times the free CFP spectrum is required to model the fusion protein spectrum (blue line). The YFP correction factor of 0.37 is in perfect agreement with that found for the emission spectra and gives the contribution made to the fusion protein spectrum through direct excitation of YFP. The correction or scaling factor of 1.3 needed for CFP is significantly greater than that found for the emission spectra or the excitation spectrum above, however. This is related to the fact that photons absorbed by CFP are reappearing as emission from CFP or from YFP through energy transfer, both of which we are now monitoring at 575 nm. Because 0.72 times the free CFP spectrum accounts for direct emission (Fig. 8B, upper panel), whereas 1.3 times the free CFP spectrum accounts for direct emission and energy transfer to YFP, the difference (0.58) should be a measure of the FRET efficiency, as calculated in Appendix B, if we assume that each fusion protein molecule has properly matured and folded CFP and YFP moieties.

It is notable that the correction factors for CFP at 407- and 458-nm excitations (0.69) are the same. This is the anticipated result because the concentration ratio of fusion protein to free CFP is the same in both cases, as is the energy transfer efficiency and thus the fraction of absorbed light emitted by CFP directly. The latter is related to the fact that the shape of the donor emission spectrum and thus its overlap with the acceptor (YFP) absorption spectrum, which, in addition to the medium's index of refraction and orientation factor, determine the transfer efficiency (54), is the same at both excitation wavelengths. In addition, the relative YFP intensities at 407 and 458 nm caused by FRET from CFP are the same as the relative CFP intensities. (i.e., the intensity at 407-nm excitation of YFP caused by energy transfer is 0.69 times that for excitation at 458 nm). This is as would be anticipated, because the FRET acceptor intensity should be proportional to the size of the donor population, which is measured by the CFP fluorescence intensity. We would not expect this sort of consistency if the enhanced emissions at 525 nm were not caused by FRET.

APPENDIX B

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. APPENDIX A
  8. APPENDIX B
  9. LITERATURE CITED

CALCULATION OF FRET EFFICIENCY

The steady-state fluorescence intensity (I) of a fluorophore in solution is equal to the product of the amount of light it absorbs and the fraction of the absorbed light it subsequently emits as fluorescence. For dilute solutions and unit exciting light intensity, the first quantity is essentially the absorbance, which is given by the Beer-Lambert law, Abs = ϵ(λex)cd, where ϵ(λex) is the fluorophore's molar extinction coefficient at the wavelength of excitation λex, c is its molar concentration, and d is the optical path length. The second quantity is by definition the fluorescence quantum yield or efficiency, Φ. For a solution of several fluorophore species, the products are additive, so that in our case,

  • equation image(AB1)

The first term gives the intensity of CFP fluorescence, where E is the FRET efficiency, cfuse is the fusion protein concentration, ΦCFP is the quantum yield of CFP in the absence of the FRET acceptor, and SCFPem) is a “shape” function that gives the fraction of the total fluorescence emitted at λem. The second term gives the fluorescence intensity of YFP because of its direct excitation, whereas the third term gives the YFP intensity because of FRET from CFP and is equal to the light absorbed by CFP that is lost because of FRET times the quantum yield of YFP. Because the optical path length is the same for all three species, d does not appear in equation AB1.

We should point out that in writing this equation we have assumed that the CFP and YFP moieties of each fusion protein molecule have properly folded and matured so that there are no molecules for which one or the other is not functional. A more general expression that would take into account the possibility of partial folding would require additional information.

FRET efficiency is usually calculated from the donor fluorescence intensities in the absence and presence of the acceptor. Unfortunately, this was not possible because, unlike the case for YFP at 514-nm excitation, we could not fluorometrically determine the fusion protein-to-free CFP molar ratio. Instead, we can calculate the efficiency from the steady-state fluorescence intensity of the acceptor because we know its relative contributions from direct excitation and FRET. From the emission spectra for excitation at 458 nm, we know that the ratio of the third to the second terms in equation AB1 is equal to 1.26, or

  • equation image(AB2)

(This result also can be obtained from equation 3.14 of Cheung [54].) To arrive at values for the extinction coefficients at 458 nm, we applied published peak values (44) to our excitation spectra (Fig. 8C), which we corrected for constant exciting light intensity, to appropriately scale them. (We could not directly measure the absorbance because of the amount of scattered light.) Using the calculated extension coefficients, equation AB2 produces approximately 0.38, or 38%, as the energy transfer efficiency between the CFP and YFP moieties in the fusion protein ensemble.

To confirm this result, we also calculated the FRET efficiency from our excitation spectra. From the results for emission at 500 nm (upper panel of Fig. 8B), we know that

  • equation image(AB3)

or

  • equation image(AB4)

(i.e., the scaling factor of 0.72 for CFP is simply equal to the fraction of its fluorescence intensity that is not lost because of FRET times the molar ratio of fusion protein to free CFP). From the results for emission at 575 nm (lower panel of Fig. 8B), we have

  • equation image(AB5)

By combining equations AB4 and AB5, we obtain

  • equation image(AB6)

or

  • equation image(AB7)

which yields

  • equation image(AB8)

or

  • equation image(AB9)

By using a published value of 0.2 for ΦCFPYFP (55) and our own measured values for SYFP(575 nm) and SCFP(575 nm), we obtain 0.37, or 37%, for the energy transfer efficiency (E) between the CFP and YFP moieties in the fusion protein. This value is in excellent agreement with the 38% efficiency we calculated based on emission spectra.

LITERATURE CITED

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. APPENDIX A
  8. APPENDIX B
  9. LITERATURE CITED