Imaging protein interactions with bioluminescence resonance energy transfer (BRET) in plant and mammalian cells and tissues

Abstract

FRET is a well established method for cellular and subcellular imaging of protein interactions. However, FRET obligatorily necessitates fluorescence excitation with its concomitant problems of photobleaching, autofluorescence, phototoxicity, and undesirable stimulation of photobiological processes. A sister technique, bioluminescence resonance energy transfer (BRET), avoids these problems because it uses enzyme-catalyzed luminescence; however, BRET signals usually have been too dim to image effectively in the past. Using a new generation electron bombardment-charge-coupled device camera coupled to an image splitter, we demonstrate that BRET can be used to image protein interactions in plant and animal cells and in tissues; even subcellular imaging is possible. We have applied this technology to image two different protein interactions: (i) dimerization of the developmental regulator, COP1, in plant seedlings; and (ii) CCAAT/enhancer binding protein α (C/EBPα) in the mammalian nucleus. This advance heralds a host of applications for imaging without fluorescent excitation and its consequent limitations.

Interactions among proteins are key to the performance of their cellular activity. Identifying the partners with which a protein associates has been a major approach, in addition to genetics, toward discovering the key components of a biological pathway. Moreover, quantifying and localizing these protein interactions are crucial to the ultimate understanding of how proteins accomplish their raison d'etre. Many different methods are available for measuring protein interactions (1), but one of the most useful is based on Förster resonance energy transfer (2). In particular, FRET is a well established resonance energy technique for monitoring protein interactions (3). When two fluorophores (the “donor” and the “acceptor”) with overlapping emission/absorption spectra are within ≈50 Å of one another and other conditions are met, the donor fluorophore is able to transfer its excited-state energy to the acceptor fluorophore. FRET can act as a “molecular yardstick” when two fluorophores of overlapping emission/absorption spectra are within a radius of ≈50 Å and other conditions are met. Therefore, if appropriate fluorophores are linked to proteins that interact with each other, the proximity of those proteins can be detected by FRET in living cells in which the fusion proteins are produced endogenously (3). Thus, the presence or absence of FRET acts as a molecular yardstick.

However, because FRET demands that the donor fluorophore be excited by light, the practical usefulness of FRET can be limited by the concomitant consequences of that irradiation: photobleaching, autofluorescence, and direct excitation of the acceptor fluorophore. Furthermore, some tissues can be damaged by the excitation light or might be directly photoresponsive (e.g., retina and most plant tissues) so that a photobiologically regulated interaction can be disturbed by FRET. For these reasons, we developed a complementary technique, bioluminescence resonance energy transfer (BRET) (4), that takes advantage of the natural phenomenon in which GFP participates (5). In BRET, the donor fluorophore of the FRET technique is replaced by a luciferase. In the presence of a substrate, bioluminescence emanating from the luciferase can excite the acceptor fluorophore if the luciferase and fluorophore are within a radius of ≈50 Å. BRET thereby avoids the problems of direct excitation while retaining the advantages of FRET. BRET also has the potential to outshine FRET for deep penetration of animal tissues (6) and for high-throughput screening because an internal light source is not required in the screening instrument and because BRET does not photobleach the energy donor (7).

On the other hand, because the light emission from BRET generally is dim, it has been thought to be less suitable for imaging than is FRET (8, 9). In particular, BRET has been thought to be “not currently suitable for single-cell analyses and thus information on cellular distribution and spatial resolution cannot be provided” with BRET (7). Until now, that is. We have coupled a very sensitive charge-coupled device (CCD) camera with a Dual-View image splitter to display BRET localizations in highly autofluorescent plant tissues and cells. Here, we present the tissue localization for dimerization of the plant regulator of light signaling, constitutive photomorphogenesis 1 (COP1) protein. Furthermore, we are able to image the subcellular distributions of interactions of the CCAAT/enhancer binding protein α (C/EBPα) in single isolated mammalian cells. This advance augurs imaging applications that are not limited by problems associated with fluorescent excitation.

Results

Imaging of BRET Interactions in Whole Tobacco Seedlings.

Although low-resolution BRET imaging has been reported for whole-animal imaging (6), Fig. 1 demonstrates that BRET imaging at a higher level of resolution can be applied to plant seedlings by using a modified electron bombardment-CCD (EB-CCD) camera and a Dual-View microimager attached directly to a ×4 microscopic objective. The microimager allows us to collect two wavelength ranges simultaneously by including a dichroic that splits the image at 505 nm and short-pass/long-pass filters that refine the spectral distinction [see supporting information (SI) Text]. The spectra distinguished by the Dual-View are (i) wavelengths shorter than 505 nm (= B for blue light) versus (ii) wavelengths longer than 505 nm (= Y for yellow light; see SI Fig. 5). A “BRET ratio” then is derived over the image to quantify the amount of resonance transfer from luciferase to fluorophore, and these BRET ratios will be expressed herein as “Y÷B” or “Y:B.” Fig. 1 A–D depicts the BRET signal when Renilla luciferase (RLUC) is expressed alone in tobacco seedlings. RLUC luminescence is expressed strongly from root hairs as well as stems and weakly from cotyledons (Fig. 1 B and C). As expected, the luminescence image for RLUC in the blue window (Fig. 1B) is significantly brighter than that in the yellow window (Fig. 1C), resulting in a BRET ratio over the entire seedling that is mostly <1.0 (average Y:B over the entire image = 0.85 ± 0.08 SD; Fig. 1D). On the other hand, luminescence from the fusion protein RLUC·enhanced yellow fluorescent protein (EYFP), which exhibits BRET (4), shows a brighter luminescence in the yellow than in the blue (Fig. 1 F and G), resulting in a BRET ratio over the seedling predominantly >1.0 (Y:B = 1.31 ± 0.15 SD; Fig. 1H).

Fig. 1.

BRET macroimaging of tobacco seedlings using the light-tight box apparatus (see SI Text). (A–L) Tobacco seedlings ≈7 days after germination. Shown are seedlings transformed with (i) P_35S::Rluc (A–D), (ii) P_35S::Rluc·EYFP (E–H), and (iii) P_35S::Rluc·COP1 + P_35S::EYFP·COP1 (I–L). (A, E, and I) Bright-field images. (B, F, and J) Short-pass luminescence images (blue). (C, G, and K) Long-pass luminescence images (yellow). (D, H, and L) BRET ratios (Y÷B) over the entire image (pseudocolor scale is shown above D). Quantification of the average BRET ratios (Y÷B ± SD) over the entire images for these samples are as follows: for RLUC, Y÷B = 0.85 ± 0.08 SD (n = 4, including D); for RLUC·EYFP, Y÷B = 1.31 ± 0.15 SD (n = 3, including H); and for RLUC·COP1/EYFP·COP1, Y÷B = 1.26 ± 0.12 SD (n = 4, including L). A ×4 noninfinity-corrected microscopic objective was coupled directly to the Dual-View, which was connected in turn to the EB-CCD camera. Exposure time averaged 7.5 min. Coelenterazine concentration was 10 μM at 22°C.

We previously showed in a nonimaging study that BRET could be used to measure the dimerization of COP1 protein, a repressor of light-regulated development in plants (10, 11). Because that investigation was performed in onion epidermal cells, it could not demonstrate the localization of COP1 interaction among the tissues of seedlings/plants. Here, we use BRET to visualize the tissue localization of COP1 dimerization in tobacco seedlings with two independent COP1 fusion proteins; both fusion proteins are N-terminal fusions: RLUC·COP1 and EYFP·COP1 (10). These constructs were not expressed as strongly in the tobacco seedlings as were RLUC and RLUC·EYFP, but they clearly were expressed in the rootlet and cotyledons (Fig. 1 J–K). The BRET ratio was high (Y:B = 1.26 ± 0.12 SD) wherever the RLUC·COP1 expression was strong enough to obtain a detectable luminescence image (Fig. 1L). This result indicates that dimerization occurs for the photomorphogenesis-regulating protein COP1 throughout the seedling, even in tissues such as roots that do not normally receive strong illumination and therefore would not be expected to express pathways involved in light-regulated development.

BRET Imaging in Arabidopsis Tissues and Cells.

Fig. 1 clearly shows that macroimaging of plant tissue BRET is possible. We decided to zoom in with an inverted microscope coupled with our EB-CCD/Dual-View on specific tissues and cells of Arabidopsis, the most widely used plant for genetic research. Fig. 2 illustrates the application of BRET to the cotyledons of seedlings (Fig. 2 A–I) and to a single isolated plant cell (Fig. 2 J–N). Fig. 2A shows spectra of whole Arabidopsis seedlings expressing RLUC or RLUC·EYFP. Fig. 2 B and C shows the emission of RLUC-expressing seedlings in the blue and the yellow, respectively. Fig. 2D shows the BRET ratio over the cotyledons, which generally is <1.0 (Y:B = 0.58 ± 0.02 SD), which corresponds well with that expected from the whole-seedling spectrum shown in Fig. 2A. Fig. 2 E–I shows another cotyledon expressing the RLUC·EYFP fusion protein. By virtue of its EYFP moiety, this protein is fluorescent, and therefore the distribution of its fluorescence over the cotyledon can be visualized (Fig. 2F). Assay of luminescence demonstrates that the yellow emission (Fig. 2H) is brighter than the blue emission (Fig. 2G), resulting in a BRET ratio >1.0 (Y:B = 1.18 ± 0.05 SD; Fig. 2I), as expected from the whole-seedling BRET spectrum (Fig. 2A).

Fig. 2.

BRET microimaging of Arabidopsis seedlings using an inverted fluorescence microscope apparatus. (A) Spectra of RLUC and RLUC·YFP emission from whole Arabidopsis seedlings. Luminescence spectra were normalized to the emission at 480 nm. Shown are seedlings transformed with (i) P_35S::Rluc (B–D) (no bright-field image is available for this sample, and because RLUC is not fluorescent, there is no fluorescence image), and (ii) P_35S::Rluc·EYFP (E–I) (optics arrangement 1 with ×2 objective, 5-min exposure time). (J–N) A single isolated Arabidopsis cell expressing RLUC·EYFP from a suspension culture [optics arrangement 2 with ×40 objective, N.A. 1.30 (oil immersion), 7.5-min exposure time]. (E and J) Bright-field images. (F and K) RLUC·EYFP fusion protein's fluorescence images. (B, G, and L) Short-pass luminescence (blue) images. (C, H, and M) Long-pass luminescence images (yellow). (D, I, and N) BRET ratios (Y÷B) over the entire image (pseudocolor scale shown above D). Quantification of the average BRET ratios (Y÷B ± SD) over the entire images for these samples are as follows: for RLUC, Y÷B = 0.58 ± 0.02 SD (n = 6, including D); and for RLUC·EYFP, Y÷B = 1.18 ± 0.05 SD (n = 6, including I). Coelenterazine concentration was 10 μM at 22°C.

We then extended BRET imaging to single Arabidopsis cells isolated from a suspension culture line and expressing the RLUC·EYFP fusion protein (Fig. 2 J–N). As was the case for cotyledon tissue, single cells expressing RLUC·EYFP exhibit fluorescence (Fig. 2K) and luminescence that is brighter in the yellow (Fig. 2M) than in the blue (Fig. 2L), resulting in a BRET ratio >1.0 (Fig. 2N). RLUC·EYFP is localized in the cytoplasm (the outer ring) and the nucleus (the bulge at the 10 o'clock position of the ring) but is excluded from the large central vacuole that is characteristic of most plant cells.

Correction of BRET Signal in Differentially Absorbing Tissues.

In our measurements of RLUC·EYFP emission in Escherichia coli (4) and mammalian cells (see below), the emission peaks in the blue versus the yellow windows have been approximately equal (i.e., Y:B ≈ 1.0). Therefore, we initially were perplexed that the yellow peak was reproducibly higher when RLUC·EYFP was expressed in plant tissue (as in Fig. 2A). We considered the possibility that this was because of the fact that green plant tissue contains many pigments that might differentially absorb luminescence of different wavelengths. This hypothesis is easily tested because plants can be germinated and grown in darkness, resulting in “etiolated” or essentially nonpigmented tissue. When we measured the whole-plant emission spectrum of RLUC·EYFP from etiolated tobacco seedlings, we indeed found an emission spectrum that was closer to 1:1 for Y:B (Fig. 3A). Data from imaging of “green” versus “etiolated” seedlings bore out this hypothesis: for RLUC-expressing seedlings, Y:B = 0.85 ± 0.08 SD for green seedlings, but Y:B = 0.54 ± 0.14 SD for etiolated seedlings; for RLUC·EYFP-expressing seedlings, Y:B = 1.31 ± 0.15 SD for green seedlings, but Y:B = 0.98 ± 0.04 for etiolated seedlings; and for RLUC·COP1/EYFP·COP1-expressing seedlings, Y:B = 1.26 ± 0.12 SD for green seedlings, but Y:B = 0.96 ± 0.01 SD for etiolated seedlings (BRET images for etiolated seedlings are not shown).

Fig. 3.

Correction of BRET images from plants for differential absorption of luminescence. (A) RLUC·EYFP emission spectra for light-grown (green) and dark-grown (etiolated) tobacco seedlings, normalized to the emission at 480 nm. (B) RLUC·EYFP emission spectra for green and etiolated tobacco seedlings, normalized to the emission at 530 nm. (C) Absorption spectra of an ethanol extraction of pigments from green and etiolated tobacco seedlings. (D and E) D shows the BRET ratio image from Fig. 1D, shown with a red box encasing the pigmented (cotyledon) portion that is corrected in E. (F and G) F shows the BRET ratio image from Fig. 1H, shown with a red box encasing the cotyledon portion that is corrected in G. Correction factor for boxed regions of E and G was 1.27 (see SI Text).

Green plant tissue contains pigments that absorb more strongly in the blue than in the green–yellow regions of the spectrum, as can be seen by the luminescence spectra normalized to 530 nm (Fig. 3B) and by absorption spectra of extracted plant tissue (Fig. 3C). This information can be used to normalize spectra for use in quantitative BRET assays. For example, we calculate that the absorption at 480 nm is 1.27× that at 530 nm (see SI Text), so we can correct our BRET images for this differential absorption, as illustrated in Fig. 3. Fig. 3D is the uncorrected image of an RLUC-expressing seedling (same BRET ratio image shown in Fig. 1D), and Fig. 3E has been corrected over the area of the image that is the green cotyledon tissue (upper region encased in red box; the root region of seedlings is not pigmented). Within this corrected region, the BRET ratio for RLUC emission now is 0.57, which is close to the value obtained for RLUC emission from etiolated seedlings (Y:B = 0.54 ± 0.14; see preceding paragraph). For an RLUC·EYFP-expressing seedling, Fig. 3F is an uncorrected image (same BRET ratio image shown in Fig. 1H), whereas Fig. 3G has been corrected over the area of the image that is the pigmented cotyledon tissue (red box). Within this corrected region, the BRET ratio for RLUC emission now is 0.97, which is more similar to the values obtained for RLUC·EYFP emission from etiolated seedlings (Y:B = 0.98 ± 0.04; see preceding paragraph), RLUC·EYFP emission from bacterial cells (4), and “humanized” RLUC (i.e., sequence optimized for human codon bias) (hRLUC)·Venus (optimized YFP) emission from mammalian cells (see Fig. 4I). Note that after correction, the seedlings in Fig. 3 E and G show equivalent BRET ratios over the green cotyledon (red box areas) and nonpigmented root regions.

Fig. 4.

Subcellular imaging of BRET in single mammalian cells. Cells were imaged with an Olympus IX71 microscope by using optics arrangement 3 with a ×60 oil immersion objective (NA 1.45). The luminescence images are integrations of 20 sequential 100-msec exposures. Cells were treated with 10 μM ViviRen in DMEM with 10% FBS (36°C). (A–D) HEK293 cells expressing hRLUC; Y÷B = 0.32 ± 0.05 SD (n = 10) over the luminescent portion of the cell. (E) Spectra of hRLUC versus hRLUC·Venus emission from HEK293 cells. (F–J) HEK293 cells expressing hRLUC·Venus; Y÷B = 0.82 ± 0.07 SD (n = 8) over the luminescent portion of the cell. (K–O) HEK293 cells expressing unfused hRLUC and Venus; Y÷B = 0.36 over the luminescent portion of this cell. (P–S) Mouse GHFT1 cells expressing hRLUC·C/EBPα; Y÷B = 0.24 ± 0.06 SD (n = 11) over the luminescent portion of the cell. (T) Spectra of hRLUC·C/EBPα versus hRLUC·C/EBPα + Venus·C/EBPα emission from mouse GHFT1 cells. (U–Y) Mouse GHFT1 cells expressing hRLUC·C/EBPα + Venus·C/EBPα; Y÷B = 0.39 ± 0.04 SD (n = 9) over the luminescent portion of the cell. (Z–D′) Mouse GHFT1 cells expressing hRLUC + Venus·C/EBPα; Y÷B = 0.30 over the luminescent portion of this cell. A, F, K, P, U, and Z are bright-field images; B, G, L, Q, V, and A′ are blue luminescence images; C, H, M, R, W, and B′ are yellow luminescence images; D, I, N, S, X, and C′ are BRET ratios (Y÷B) over the entire image (pseudocolor scale shown above D); and J, O, Y, and D′ are fluorescence images from Venus in fusion proteins.

Subcellular Imaging of BRET Interactions in Single Mammalian Cells.

BRET imaging also is possible from single mammalian cells in serum-containing medium by using the coelenterazine analog ViviRen as the substrate for RLUC (see SI Text and SI Fig. 6 for a discussion of the relative merits of native coelenterazine versus its analogs as substrates for BRET imaging). Fig. 4 shows images from both human HEK293 and mouse GHFT1 cells transfected with various BRET constructs. A comparison of the bright-field images (Fig. 4 A, F, K, P, U, and Z) with the luminescence and fluorescence images shows that the transfection efficiency was rather low, with usually a single cell transfected within the field of view (Fig. 4 A–D is an exception, where two cells are transfected). Fig. 4 A–D shows HEK293 cells transfected with hRLUC; these cells exhibit a low BRET ratio (Y:B = 0.32 ± 0.05 SD) over the image (Fig. 4D). On the other hand, HEK293 cells transfected with the fusion protein hRLUC·Venus (Fig. 4 F–J) show a strong BRET ratio of 0.82 (Y:B = 0.82 ± 0.07 SD; Fig. 4I). To determine whether the BRET ratio is perturbed by expression of noninteracting proteins, we coexpressed unfused hRLUC and unfused Venus and found a low BRET ratio that indicates no spurious interaction (Y:B = 0.36; Fig. 4 K–O).

Furthermore, we have been able to image the subcellular distributions of interactions of the C/EBPα in single isolated mammalian cells. C/EBPα is a transcriptional factor that localizes to the heterochromatin in the nucleus and dimerizes (12, 13). A C/EBPα·hRLUC fusion protein localizes to the nucleus of mouse GHFT1 cells and demonstrates a low BRET ratio (Y:B = 0.24 ± 0.05 SD; Fig. 4 P–S). When C/EBPα·hRLUC is coexpressed with C/EBPα·Venus, however, the nuclear expression consistently reveals a higher BRET ratio within the nucleus (Y:B = 0.39 ± 0.04 SD; Fig. 4 U–Y). These observations confirm FRET studies of C/EBPα dimerization in GHFT1 cells that display nuclear regions of increased C/EBPα interaction (12). It is interesting to note that the fusion of C/EBPα to RLUC appears to influence the spectrum of RLUC emission slightly, as demonstrated by the reduction of the BRET ratio as compared with the RLUC control (Y:B = 0.24 versus 0.32). This observation illustrates the importance of using the appropriate control for BRET studies. Coexpression of unfused hRLUC with Venus·C/EBPα is another control to demonstrate the absence of spurious interaction; as shown in Fig. 4C′, the BRET ratio of hRLUC essentially is the same (Y:B = 0.30) in the cytoplasm as the hRLUC control in Fig. 4D (Y:B = 0.32) despite the expression of Venus·C/EBPα that is localized prominently to the nucleus but is also found in the cytoplasm (Fig. 4D′). Another interesting observation is that hRLUC or hRLUC·Venus sometimes appears to be excluded from the nucleus in mammalian cells (e.g., Fig. 4 G–J and A′–C′), whereas it entered the Arabidopsis nucleus (the bulge at the 10 o'clock position of the ring; Fig. 2 J–N). However, when hRLUC or Venus is fused to C/EBPα, it is able to translocate to the mammalian nucleus (Fig. 4 Q–S, V–Y, and D′). Finally, the BRET ratio is independent of the brightness of the luminescence signal; thus, the BRET ratio is not influenced by the level of expression of donor (RLUC) or acceptor (e.g., Venus or YFP; see SI Text and SI Fig. 7).

Discussion

Fluorescence techniques have become the predominant methodology for imaging cells, subcellular organelles, and small pieces of tissue. This technology has blossomed, leading to fluorescence tools such as FRET, fluorescence photobleaching, fluorescence lifetime imaging, confocal microscopy, etc. (3). On the other hand, because of its exquisite sensitivity and independence from excitation, luminescence technology often is preferable for high-throughput screening (7). An additional advantage of BRET is that it is superior for deep penetration of animal tissues (6). Luminescence is also favored over fluorescence for assays of promoter activity, especially for those involving temporal changes of promoter activity, because native luciferases tend to have a shorter half-life than native fluorescent proteins do and therefore are better reporters of time-dependent changes (14) [there are destabilized fluorescent proteins that have shorter half-lives, but they still are not as useful for time-course experiments in which cells are irradiated frequently, leading to phototoxicity (15)].

Therefore, BRET outshines FRET in applications where fluorescence excitation is undesirable. The major limitation that has caused luminescence imaging to lag behind fluorescence imaging has been the fact that luminescence signals usually are much dimmer than fluorescence signals are. However, our application of a modified EB-CCD camera coupled with a microimager demonstrates that BRET now can be imaged at tissue, cellular, and even subcellular levels. At the subcellular level, we show that nuclear and cytoplasmic BRET signals can be visualized (Figs. 2 and 4). With this technology, it should be possible to visualize mitochondrial and chloroplastidic BRET signals as well. We have applied this technology to image protein interaction of (i) COP1 in plant seedlings, and (ii) C/EBPα in the mammalian nucleus. We also show that tissues that absorb light differentially can be used for quantitative measurements of BRET by a correction using RLUC·EYFP emission profiles. Together with the demonstration that BRET imaging is possible in tissues of intact animals (6), our results enable a host of applications for imaging without fluorescent excitation, thereby avoiding the problems of photobleaching, autofluorescence, phototoxicity, and undesirable stimulation of photobiological processes.

Materials and Methods

Plant and Mammalian Cell Culture.

Arabidopsis and tobacco plant seedlings were transformed with expression cassettes of P_35S::RLuc, P_35S::RLuc·EYFP, and P_35S::RLuc·COP1 + P_35S::EYFP·COP1 (10). Plant seedlings were grown on 1/2 MS medium for 5–7 days (after germination) separately before measuring luminescence spectra and imaging. Single cells of Arabidopsis were generated from cell suspensions initiated from calli. Human HEK293 and mouse GHFT1 cells were grown in DMEM with 10% FBS at 37°C and transfected with plasmids using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). After 24 h, cells were washed and resuspended in DMEM with 10% FBS for imaging. Plasmids were insertions into pcDNA3.1 with the transgenes expressed under the control of the CMV promoter P_CMV. See SI Text for details.

BRET Methods.

Methods for measuring protein interactions in populations of cells with BRET have been described for bacteria (4, 16), plants (10, 17), and mammalian cells (6, 18–20). In this study, substrates for RLUC were native coelenterazine (NanoLight, Pinetop, AZ), ViviRen/EnduRen (Promega, Madison, WI), or Deep Blue C (BioSignal; PerkinElmer, Waltham, MA). For spectral measurements of BRET emission, a QuantaMaster QM-7/SE (Photon Technology International, Birmingham, NJ) spectrophotometer was used; for luminescence spectral measurements, the excitation beam was blocked, and the slit width was set to 16 nm.

Imaging of BRET.

BRET imaging was accomplished by using (i) a Dual-View microimager, and (ii) a modified EB-CCD camera. The Dual-View microimager (Optical Insights, Tucson, AZ) allows the simultaneous acquisition of luminescence images at two wavelengths. It consists of a dichroic mirror (in our case, to split at 505 nm with Q505LPxr) and interference filters to select for wavelengths <505 nm (HQ505SP; blue) and for wavelengths >505 nm (HQ505LP; yellow). Our EB-CCD camera had a GaAsP photocathode with low-ion feedback and cooling to −25°C (Hamamatsu Photonic Systems, Bridgewater, NJ) (see SI Text for details). The acquisition software was Photonics-WASABI (Hamamatsu). BRET imaging of plant seedlings was performed both in a light-tight box and through an inverted microscope. In the case of the light-tight box setup (used for Figs. 1 and 3), the EB-CCD camera was coupled to the Dual-View to which a ×4 noninfinity-corrected microscopic objective had been attached, and the entire apparatus was enclosed in the light-tight box. This apparatus did not allow excitation of fluorescence. For plant seedlings, cells, and mammalian cells, the Dual-View and EB-CCD were attached to the bottom port of an IX-71 inverted microscope (Olympus America Inc., Melville, NY). This setup allows the measurement of fluorescence with an epifluorescence attachment (excitation 500/20 nm, emission 520 long pass). The entire IX-71 microscope was enclosed in a temperature-controlled (22–37°C) light-tight box. For low-power imaging (e.g., Fig. 2 B–I), a Macro XLFLuor ×2 objective, numerical aperture (N.A.) 0.14 (Olympus) was used (optics arrangement 1), whereas for higher magnifications, a UPlanFl ×40 objective, N.A. 1.30 (oil immersion, Olympus) was used for Fig. 2 J–N (optics arrangement 2), or a Plan Apo ×60 objective, N.A. 1.45 (oil immersion, Olympus) was used for Fig. 4 (optics arrangement 3). For luminescence imaging of plant seedlings and cells, a single exposure (5–7.5 min) was taken; for single mammalian cells, 20 sequential 100-msec exposures were integrated by choosing the median value for each pixel over the sequence of 20 exposures. These integrations and the BRET ratio (Y:B) was calculated by using ImageJ. See SI Text for details.

Abbreviations

BRET

bioluminescence resonance energy transfer

CCD

charge-coupled device

COP1

constitutive photomorphogenesis 1

EB-CCD

electron bombardment-CCD

RLUC

Renilla luciferase

EYFP

enhanced yellow fluorescent protein

hRLUC

“humanized” RLUC

NA

numerical aperture.