Oligomerization of transcriptional intermediary factor 1 regulators and interaction with ZNF74 nuclear matrix protein revealed by bioluminescence resonance energy transfer in living cells.

Transcriptional intermediary factor 1 (TIF1) alpha and KAP-1/TIF1beta, two members of the TIF1 family of nuclear cofactors, are ubiquitous co-regulators of nuclear receptors and KRAB motif-containing zinc finger transcription factors, respectively. Despite the functional evidence suggesting a role for TIF1 proteins as modulators of transcription, the study of their interactions with transcriptional machineries in physiologically relevant systems has been difficult. Here, we have developed a bioluminescence resonance energy transfer (BRET) biophysical approach to study protein-protein interactions in the nuclear compartment of living mammalian cells. We report that TIF1alpha and KAP-1 form homo- and hetero-oligomers in intact mammalian cells. BRET titration experiments indicate that both homo- and hetero-oligomers occur with relatively high affinity suggesting that they could co-exist in cells. Furthermore, we demonstrate that KAP-1 but not TIF1alpha interacts with the KRAB multifinger ZNF74 in the nuclear matrix. Splice variants and point mutants of ZNF74 that lack transcriptional activity were found not to interact with KAP-1 confirming the physiological importance of this interaction in living cells. The interaction of ZNF74 with KAP-1 did not prevent KAP-1 homomerization indicating that the oligomers most likely represent the transcriptionally active species. Furthermore, the detection of ternary ZNF74.KAP-1.TIF1alpha complexes suggests the existence of cross-talk between KAP-1-interacting KRAB proteins and TIF1alpha-interacting nuclear receptors. In addition to providing new insights into the molecular interactions involved in the transcriptional activities of these proteins, this study shows that BRET can be advantageously used as a non-transcription-based oligomerization detection system to study the interaction of transcriptionally active proteins, including nuclear matrix proteins, in living cells.

Whereas a few studies have suggested that TIF1 family members can act as homo-or hetero-oligomers (12), the nature of the complexes in which these proteins are engaged remain largely unknown. Indeed, despite the evidence suggesting an important role for TIF1 proteins as modulator of transcription, the study of their interactions with the transcriptional machinery in physiologically relevant systems has been difficult. Studies addressing their physical interactions included in vitro assays carried out mainly with soluble fragments (to minimize aggregation) of recombinant proteins (1,3,4,(12)(13)(14)(15)(16), transcription-based yeast two-hybrid assays (1,3,5,8,9,17) and coimmunoprecipitation in overexpression systems (4,14,18). However, results obtained with these various approaches led to conflicting conclusions. For instance, in vitro and co-immunoprecipitation studies suggested that TIF1␣ and KAP-1 can homo-oligomerize (14) although no such homophilic interaction was detectable in yeast two-hybrid interaction assays (8). Conversely, whereas heteromerization of TIF1␣ with KAP-1 was evidenced by yeast two-hybrid methods (5), it was not detected using in vitro and co-immunoprecipitation approaches (14).
The analysis of the interactions involving TIF1 family members is further complicated when considering that the transcriptional activities of TIF1 and KRAB-containing proteins drastically differ between yeast, where these proteins are not normally expressed, and mammalian cells. In contrast to what is found in mammalian cells, TIF1␣, KAP-1, nor KRAB domain-containing proteins have transcriptional repressor activity in yeast (8) with the consequence that many functionally relevant interactions may not be found in this system (2).
In one case, a physiologically more appropriate transcrip-tion-based mammalian two-hybrid system has been successfully used to demonstrate the interaction of a non-repressive KAP-1 fragment with the KRAB domain of KRAZ1 and KRAZ2 (10). However, such assay cannot be used with full-length proteins that have intrinsic transcriptional activity, significantly limiting its usefulness. Although co-immunoprecipitation could alleviate this problem, this approach has the inherent drawback of possibly revealing complexes established during or following cell lysis. Furthermore, co-immunoprecipitation could not be readily used with proteins tightly bound to the nuclear matrix such as the KRAB-containing proteins that are resistant to solubilization in non-denaturating detergents. Taken together, the above considerations stress the need for the development of new approaches to study protein-protein interactions involving transcriptional regulators and/or nuclear matrix proteins. Fluorescence and bioluminescence resonance energy transfer (FRET or BRET) have recently been used to assess proteinprotein interactions in living cells (20,21). In particular, BRET was successfully used to study homomerization of the Kai transcription factors, in the nuclear-less bacterial cell Escherichia coli (22), and the oligomerization of the membrane-bound G protein-coupled membrane receptors in mammalian cells (23,24). Here, using nuclear and nuclear-matrix targeting sequences as controls, BRET was used for the first time to quantitatively study protein interactions in the nuclear compartments of living mammalian cells.
We report that two TIF1 family members, TIF1␣ and KAP-1, can form homo-and hetero-oligomers in intact mammalian cells and that KAP-1, unlike TIF1␣, interacts with the KRAB multifinger protein ZNF74 in the nuclear matrix. In addition to providing new insights into the molecular interactions involved in the transcriptional activities of the TIF1 and KRAB-containing proteins, our study demonstrates that BRET can advantageously be used as a non-transcription-based detection system to study the interactions of full-length transcriptional regulators, including nuclear matrix proteins, in living cells.
Rluc-GFP10 and GFP-NLS Constructs-Rluc-GFP10 construct corresponding to a fusion of the luciferase to GFP was a gift from PerkinElmer Life Sciences. To target GFP protein into the nucleus, a nuclear localization signal (NLS) (31) was inserted after the GFP in pGFP10 C2 vector digested with ApaI and BamHI (pGFP10-NLS). PCR-derived regions as well as blunted cloning regions were sequenced in all constructs.

Cell Culture and Transfections
Human embryonic kidney 293T cells (32) maintained in Dulbecco's modified Eagle's medium supplemented by 10% fetal bovine serum (Invitrogen), 100 g/ml penicillin and streptomycin, 1 mM L-glutamine were seeded at a density of 1 ϫ 10 6 cells per 100-mm dish. Transient transfections of plasmids were performed the following day by using the calcium phosphate precipitation method (30).

BRET Assay
Transient transfections with Rluc and/or GFP10 fusion or other constructs were performed in 293T as described above. The total amount of DNA used for transfection was adjusted to 10 g by adding pGEM4 vector (Promega). Forty-two hours post-transfection (or 48 h for transfections including TIF1␣ constructs), cells were detached with 2 mM phosphate-buffered saline/EDTA and washed twice with 1 mM phosphate-buffered saline/glucose. Cells were resuspended in 1 mM phosphate-buffered saline/glucose at ϳ2 ϫ 10 6 cells/ml and 90 l (ϳ20 g) were distributed in 96-well microplates (white Optiplate from Packard). Upon the addition of 5 M of the cell permeant luciferase substrate, coelenterazine deep blue (10 l) (PerkinElmer Life Sciences), the bioluminescence resulting from its degradation (emission peak 400 nm) was detected using a 370 -450-nm band pass filter (donor emission). The energy transferred to GFP10 (emission peak 510 nm) was detected using a 500 -530-nm band pass filter. Readings were collected with a modified top count apparatus (BRETCount, PerkinElmer Life Sciences) that allows sequential integration of the signals detected at 370 -450 nm (bioluminescence signal) and at 500 -530 nm (fluorescence signal). The BRET signal (BRET ratio) was quantified by calculating the fluorescence/luminescence ratio as previously reported (33). The BRET ratio was found to be stable over several readings performed at different times after addition of the substrate (here evaluated in a 5-20-min range) (23). Expression level of each construct was determined by direct measurements of total fluorescence and luminescence on aliquots of transfected cell samples. The GFP10 total fluorescence was measured using a FluoroCount (PerkinElmer Life Sciences) with an excitation filter at 400 nm, an emission filter at 510 nm, and the following parameters: gain 1; PMT 1100 V; time 1.0 s. After the fluorescence measurement, the same cells were incubated for 10 min with coelenterazine H (Molecular Probes) at a final concentration of 5 M and the total luminescence of cells was measured using a LumiCount (PerkinElmer Life Sciences) with the following parameters: gain 1; PMT 700 V; time 0.5 s. In contrast to deep blue coelenterazine, coelenterazine H does not lead to energy transfer to GFP10 and thus allows assess to Rluc activity without GFP10 emission quenching. The BRET ratios were plotted as a function of the GFP/LUC fusion protein expression ratio, both fusion proteins expression being assessed with the same cells as described above, to take into account the potential variations in the expression of individual constructs from transfection to transfection. BRET titration curves were fitted using a non-linear regression equation (GraphPad Prism).

Co-immunoprecipitation
Cells were lysed with RIPA buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 20 mM MgCl 2 , 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml aprotinin, pepstatin, and leupeptin) at 4°C for 20 min and centrifuged at 15,000 ϫ g for 15 min. The extracts were precleared over protein A-Sepharose for 1 h at 4°C. Co-immunoprecipitations of the precleared extracts were carried out with protein A-Sepharose (50 l) and mouse anti-HA monoclonal antibody (12CA5). Washed immunoprecipitates were resuspended in Laemmli buffer and the recovered proteins were analyzed by SDS-PAGE and electrotransferred for Western blotting.

MBP Pull-down Assays
Pull-down assay with MBP and MBP fusions were performed as described by Gebelein and Urrutia (34). Briefly, 293T cells transiently transfected with pCGN-HA-KAP1 (5 g) were solubilized for 20 min in RIPA buffer (1 ml/ϳ10 7 cells). After centrifugation at 12,000 ϫ g for 10 min, the equivalent of 3 ϫ 10 6 solubilized cells was first incubated with an equimolar quantity of purified MBP or MBP fusion protein (0.5-1 g) for 2 h, re-centrifuged to eliminate any potential insoluble material, and then incubated with an amylose resin for 1 h. Complexes retained by the washed affinity resin were analyzed by SDS-polyacrylamide gel electrophoresis and electrotransferred for Western blotting. After the anti-HA immunodetection, the blot was dehybridized and rehybridized with an anti-MBP antibody to confirm the loading of equivalent amounts of MBP fusion proteins (data not shown).

Western Blotting
The expression of epitope-tagged and fusion proteins immobilized on nitrocellulose membranes was verified by Western blotting using the 12CA5 mouse anti-HA antibody (35), rabbit polyclonal anti-GFP (Clontech), rabbit anti-MBP (New England Biolabs), rabbit anti-ZNF74 antibody raised against aa 139 to 261 and aa 584 to 643 of ZNF74-II, and mouse polyclonal anti-KAP1 raised against aa 381 to 564 (36). Following addition of the appropriate secondary antibody (either a sheep anti-mouse or goat anti-rabbit horseradish peroxidase), a chemiluminescence reagent was used (Renaissance kit, Amersham Biosciences).

Oligomerization of TIF1 Family Members Detected in Living
Cells-Although the presence of interaction motifs such as a ring finger, B boxes, and a coiled-coil domain (RBCC) suggests that members of the TIF1 family could oligomerize, studies assessing such interactions led to conflicting results (5,8,(12)(13)(14). Here, using a newly developed BRET assay (22,23,33), homo-and heteromerization of KAP-1 and TIF1␣ were investigated in living mammalian cells. Because the transfer of energy between a bioluminescence donor luciferase and a fluorophore acceptor GFP occurs with a R 0 of ϳ50 Å and that no transfer would be detected for distances above 100 Å (22), BRET was used to monitor intermolecular interactions between these potential partners. To this end, full-length KAP-1 and TIF1␣ were tagged at their amino terminus with either Renilla reniformis luciferase (Rluc) or Aequorea GFP green variant (GFP10) (Fig. 1A). Homomeric and heteromeric pairs were then co-expressed in 293T cells and the occurrence of BRET was determined by measuring the ratio of the light emitted by the GFP (410 Ϯ 40 nm) and the luciferase (515 Ϯ 15 nm) upon addition of the membrane-permeable luciferase substrate coelenterazine. The BRET ratio was plotted as a function of the GFP/LUC fusion protein expression ratio (both fusion protein expression being individually assessed in each sample as described under "Experimental Procedures") to take into account the potential variations in the expression of individual constructs from transfection to transfection. As shown in Fig.  1A, significant BRET ratios, indicative of protein proximity, were detected for the Rluc-KAP1/GFP10-KAP1, Rluc-TIF1␣/ GFP10-TIF1␣, and Rluc-TIF1␣/GFP10-KAP1 pairs. In each case, when a fixed amount of Rluc fusion was transfected, the BRET ratio increased as a function of the amount GFP fusion transfected and reached a maximum when the amount of expressed GFP fusion (acceptor) was no longer limiting compared with Rluc fusion (donor). Such saturation is indicative of a specific interaction. When considering the heteromeric TIF1␣/ KAP1 pair, identical results were obtained for the two possible BRET orientations (i.e. Rluc-TIF1␣/GFP10-KAP1 and Rluc-KAP1/GFP10-TIF1; data not shown). Because as most techniques, BRET cannot distinguish between dimers and higher order oligomeric species, the term oligomers is used to describe the detected interactions with the understanding that some of the complexes may be simple dimers.
The saturation curves obtained for the various pairs can be used to estimate the relative affinity of the partners for each other (33). For this purpose, BRET 50 values, defined as the relative amount of GFP fusion over Luc fusion, GFP/LUC in arbitrary units, required to obtain half the maximum BRET ratio (BRET max ) were determined. Considering these values, it appears that the homomeric pairs (Rluc-TIF1␣/GFP10-TIF1␣, BRET 50 ϭ 0.32 Ϯ 0.07; Rluc-KAP1/GFP10-KAP1, BRET 50 ϭ 0.40 Ϯ 0.15) tend to have a slightly higher binding affinity than the heteromeric pair (Rluc-TIF1␣/GFP10-KAP1, BRET 50 ϭ 0.74 Ϯ 0.31). In an attempt to determine the molar ratio of GFP10/RLuc constructs needed to reach the BRET 50 , a construct covalently fusing the Rluc and the GFP10 (Rluc-GFP) was used. The GFP/LUC expression ratio obtained for this fusion, which by definition corresponds to an equimolar concentration of GFP10 and Rluc, was 0.125 Ϯ 0.020 arbitrary units (n ϭ 5). Assuming that this value can be used to determine the ratio of the various Rluc and GFP10 fusions tested, 2.5, 3, and 6 times more GFP fusion than Rluc fusion was required to reached half the maximum BRET ratio for TIF1␣ homomeric, KAP-1 homomeric, and TIF1␣/KAP-1 heteromeric pairs, respectively. These data indicate that the affinities between the partners are relatively high and that large excess is not needed for the interactions to occur. Furthermore, because the affinity of the homo-and heteromers are in the same range of magnitude, these complexes can potentially co-exist in living cells.
Co-immunoprecipitation experiments carried in 293T cells co-expressing the differentially tagged full-length KAP-1 proteins, GFP10-KAP1 and HA-KAP1, confirmed that KAP-1 oligomers represent stable complexes that resist cell lysis and sample preparation. Indeed, as shown in Fig. 1C, at least 15% of the total GFP-KAP1 could be co-immunoprecipitated with the HA-KAP-1. The low level of TIF1␣ expression did not allow co-immunoprecipitation experiments to be carried out using up to 1 mg of cell extract. Starting with larger amounts of cell extracts (ϳ4 mg of protein), Peng et al. (14) were able to detect TIF1␣ homomerization by co-immunoprecipitation but failed to detect interaction between TIF1␣ and KAP-1 (14). The detection by BRET of both TIF1␣ homo-and heteromerization with KAP-1, using as little as 100,000 cells (ϳ20 g of protein) illustrates the high sensitivity of the energy transfer assay. Considering the conflicting results obtained relative to the homo-and heteromerization of TIF1 family members using yeast two-hybrids, in vitro studies, and co-immunoprecipitation (as pointed out in the introduction), BRET provides here a unique means to assess these interactions and unambiguously demonstrates that TIF1␣ and KAP-1 form homo-and heteromers in intact mammalian cells.
The functional significance of TIF1 family member oligomerization still remains to be determined. However, it was found that purified fragments of the KAP-1 RBCC domain can oligomerize and that such oligomerization is required for interaction with the purified KRAB repressor domain of Kox-1 in vitro (13). Assuming that native proteins also depend on KAP-1 oligomerization for interaction, this suggests that the KAP-1 co-repressor oligomerization may be a sine qua non condition for KRAB proteins to exert their repressive function. From a study showing that TIF1␥ coimmunoprecipitates with TIF1␣ and prevents TIF1␣ repression of RXR nuclear receptor-mediated transcription, it was also suggested that the heteromerization of some TIF1 family members may constitute a transcription regulatory mechanism (14). Also, it was recently suggested that proper nuclear targeting of KAP-1 to transcriptionally silent centromeric regions may require its oligomerization (18). Indeed, such targeting was compromised by altering the RBCC domain of KAP-1 but preserved by replacing the RBCC domain with GAL4 DNA binding domain, which has the ability to dimerize. Whether the TIF1␣/KAP1 heteromerization shown in this study affects intranuclear localization of these individual TIF1 family members, or their respective transcriptional activity, remains to be determined.
Interactions between ZNF74, KAP-1, and TIF1␣ in Living Cells-ZNF74 is a nuclear matrix protein (27) that belongs to the large KRAB domain-containing multifinger family. Two isoforms, a long one (ZNF74-II) containing a full KRAB box and a shorter one (ZNF74-I) with an incomplete KRAB domain, are generated by alternative promoter usage and splicing (28). Whereas ZNF74-I is preferentially located to nuclear speckles enriched in splicing factors and is transcriptionally inactive, ZNF74-II has a more diffuse nuclear localization and its KRAB box has been shown to repress transcription (28). Given that the repressor activity of several KRAB multifinger proteins has been shown to require an interaction with the co-repressor KAP-1 (7), one could propose that ZNF74-II also mediates repression through a direct interaction with KAP-1. However, because of its tight attachment to the nuclear matrix and its intrinsic repressive activity, the in vivo interactions of ZNF74 with KAP-1 (or other candidate protein partners) could not be readily assessed in mammalian cells either by co-immunoprecipitation or by a transcription-based interaction assay. Here thus, BRET was used to assess the interaction between ZNF74-II and KAP-1 in living mammalian cells. As seen in Fig.  2A, significant BRET ratios revealing heteromerization of ZNF74-II and KAP-1 were obtained for the Rluc-ZNF74-II/ GFP10-KAP1 pair. Similar BRET ratios were obtained when the partners were tested in the reverse orientation (Rluc-KAP1/GFP10-ZNF74-II; see Fig. 3A). In contrast, no heteromerization between ZNF74-II and the related TIF1␣ was detected. Indeed, the marginal BRET signals obtained between Rluc-ZNF74-II and GFP10-TIF1␣ (or the reverse pair; not shown) was not different from the background signal observed with the negative control Rluc-ZNF74-II/GFP10-NLS pair ( Fig.  2A).
The specificity of Rluc-ZNF74-II/GFP10-KAP1 heteromerization was confirmed by the reduction in the BRET ratio observed upon increasing concentrations of competing HA-KAP1 (Fig. 2B). Interestingly, no competition occurred when HA-ZNF74-II was added to the Rluc-KAP1/GFP10-KAP1 pair and thus, the formation of BRET productive homomers of KAP-1 is not impaired by the interaction of ZNF74-II with KAP-1. This indicates that ZNF74-II can bind to oligomers of KAP-1 in intact mammalian cells and suggests that the oligomers most likely represent the transcriptionally active species. Such interaction between ZNF74 and oligomers of KAP-1 is consistent with the previous in vitro finding that purified RBCC fragments from KAP-1 interacted as multimers with a KRAB fragment from KOX1 (12,13).
The interaction of one KRAB zinc finger protein, KOX1, with TIF1␣ (5,8,17) previously led to the suggestion that cross-talk could exist between KRAB multifinger proteins and nuclear receptors known to interact with TIF1␣ (11,37). However, our data using full-length proteins in living mammalian cells suggest that ZNF74 does not interact directly with TIF1␣. Similarly, another study failed to demonstrate TIF1␣ interaction with the isolated KRAB domain of five different KRAB zinc finger proteins using yeast two-hybrid systems (17). Thus, for these KRAB zinc finger proteins as well as for ZNF74, a possible cross-talk with nuclear receptors is not likely to occur through a direct interaction with TIF1␣. To test the hypothesis that a cross-talk regulation between TIF1␣ and ZNF74 could result from the formation of a ternary complex between TIF1␣, ZNF74, and KAP-1, the ability of KAP-1 to promote a BRET productive interaction between TIF1␣ and ZNF74-II was assessed. As shown in Fig. 2C, the background BRET signal detected between TIF1␣ and ZNF74-II (see Fig. 2A for titration curve) was significantly potentiated by the addition of the unfused HA-KAP1. Such increase in the BRET signal was specific for the TIF1␣/ZNF74 pair because it was not observed when GFP10-NLS was used instead of GFP10-ZNF74 (data not shown). This indicated that KAP-1, TIF1␣, and ZNF74 are part of a protein complex placing TIF1␣ and ZNF74 in close enough proximity for BRET to occur. Thus, we suggest that formation of ternary complexes between the co-regulators KAP-1, TIF1␣, and KRAB multifinger proteins such as ZNF74 may allow cross-talk of KRAB multifinger proteins with nuclear receptors.
Differential Interaction of KAP1 with ZNF74 Isoforms: Requirement of an Intact KRAB Domain-As indicated above, the BRET signal obtained between KAP1 and ZNF74 was independent of the donor/acceptor orientation considered. Indeed, as seen in Fig. 3A, the BRET saturation curve obtained with the Rluc-KAP1/GFP10-ZNF74-II pair (BRET 50 ϭ 0.39 Ϯ 0.16; BRET max ϭ 0.31 Ϯ 0.04) was indistinguishable from that obtained with the reverse pair presented in Fig. 2A (BRET 50 ϭ 0.48 Ϯ 0.12; BRET max ϭ 0.35 Ϯ 0.03). As previously shown for ZNF74 (27), the GFP10-ZNF74-II fusion was exclusively targeted to the nuclear matrix and was not detected in the soluble fraction of the nucleus including the DNase-released chromatin (Ns) or in the cytoplasmic-enriched fraction (Cyt) (Fig. 3A). Thus, as a negative control, a minimal nuclear matrix targeting sequence (ZNF74-II aa246 -580 ) (38) fused to GFP was used. As expected, the fusion was exclusively recovered in the nuclear matrix and only a background signal was obtained for the Rluc-KAP1/GFP10-ZNF74-II aa246 -580 pair (Fig. 3A) ruling out that the BRET signal observed between Rluc-KAP1 and GFP10-ZNF74-II could represent bystander BRET resulting from crowding in the nuclear matrix.
Previous in vitro and yeast two-hybrid studies indicated that the KRAB domain of a few KRAB-containing proteins is necessary and sufficient for interaction with KAP-1 co-repressor (7,9,17). Because the ZNF74-I isoform encodes an incomplete KRAB domain deleted from its first 31 amino acids, we tested its interaction with KAP-1 in living mammalian cells. In con-trast with the results obtained for ZNF74-II, the marginal BRET signal obtained between Rluc-KAP1 and GFP10-ZNF74-I (Fig. 3A) (or the reverse pair, not shown) was not different from that observed with the nuclear matrix negative control. The importance of the KRAB box integrity for occurrence of the interaction was further assessed by mutating two amino acids highly conserved and proposed to be important for the interaction of KRAB motif-containing proteins with KAP-1 (29). Mutation of aspartate 47 and valine 48 to alanine residues, within the ZNF74-II KRAB domain, abrogated the interaction with KAP-1 as indicated by the absence of significant BRET between Rluc-KAP1 and GFP10-ZNF74-II DV3 AA (Fig.  3A). The absence of significant BRET was not because of inappropriate expression of GFP-ZNF74 constructs because similar GFP activity were observed (as accounted by the GFP/LUC ratio). The integrity of the fusion protein produced was also confirmed by immunoblot analysis that revealed the appropriate molecular weight for all constructs (Fig. 3B). Pull-down assays using bacterially expressed ZNF74 isoforms and the DV 3 AA mutant were in agreement with the BRET results obtained in intact cells (Fig. 3C).
The observation that the transcriptionally active ZNF74-II interacts with the corepressor KAP-1, whereas the transcriptionally silent ZNF74-I (28) does not, suggests that the repressive activity of ZNF74-II is mediated by its interaction with KAP-1. This confirms the proposed role of KAP-1 as a universal co-repressor for KRAB zinc finger proteins (19). CONCLUSION In this study, BRET is shown for the first time to represent a reliable, quantitative and highly sensitive method to assess the interaction of transcriptionally active nuclear proteins in living cells. We took advantage of this non-transcription-based method to clearly show that two members of the TIF1 family, TIF1␣ and KAP-1 co-regulators of transcription can homo-and heteromerize in living mammalian cells. Furthermore, we show that the ZNF74 nuclear matrix protein from the large KRAB multifinger family interacts with homomers of the corepressor KAP-1 but also with heteromers of KAP-1/TIF1␣. This suggests that such heteromers may mediate cross-talk between KRAB multifinger proteins and nuclear receptors known to interact with TIF1␣ coregulator. As BRET allows real time kinetic studies to be performed in living cells, it will now be possible to assess such interactions under various conditions affecting cell cycle and transcriptional activity.