Ternary complexes and cooperative interplay between NCoA-62/Ski-interacting protein and steroid receptor coactivators in vitamin D receptor-mediated transcription.

The vitamin D receptor (VDR) is a ligand-dependent transcriptional factor that binds to vitamin D-responsive elements as a heterodimer with retinoid X receptor (RXR) to regulate target gene transcription. The steroid receptor coactivator (SRC) proteins are coactivators that interact with the AF-2 domain of VDR to augment 1,25-dihydroxyvitamin D3-dependent transcription. In contrast, NCoA-62/Ski-interacting protein (SKIP) is a distinct, activation function-2-independent coactivator for VDR. The current study examined whether these two distinct classes of coactivators impact functionally on VDR-mediated transcription. Using a ternary complex binding assay, we observed a marked preference for the direct interaction of NCoA-62/SKIP with the VDR-RXR heterodimer as compared with the VDR-VDR homodimer or VDR monomer. The liganded VDR also formed a ternary complex with NCoA-62/SKIP and SRC proteins in vitro. Competition experiments using LXXLL peptides showed that NCoA-62/SKIP and SRC coactivators contact different domains of the VDR-RXR heterodimer. Synergistic interplays were observed between NCoA-62/SKIP and SRC coactivators in VDR-mediated transcriptional assays, and protein interference assays indicated a requirement for both NCoA-62/SKIP and SRCs in VDR- mediated transcription. These studies suggest that the ligand-dependent and simultaneous interaction of NCoA-62/SKIP and SRC coactivators with distinct interaction domains within the VDR-RXR heterodimer results in cooperative interplays between coactivators in VDR-mediated transcription.

The vitamin D receptor (VDR) 1 is a member of the nuclear hormone receptor (NR) superfamily that regulates target gene transcription in response to 1,25-(OH) 2 D 3 (1). Like most mem-bers of the NR superfamily, VDR consists of several functional domains including the highly conserved DNA-binding domain and a C-terminal ligand binding domain. The multifunctional ligand binding domain mediates dimerization of VDR with retinoid X receptor (RXR) and contains the ligand-dependent activation function-2 (AF-2) domain. A ligand-induced conformational change of AF-2 on the surface of the VDR ligand binding domain results in the formation of a binding interface that allows protein-protein contacts between the receptor and various coactivator proteins (2)(3)(4)(5)(6).
A number of putative NR coactivator proteins have been identified and characterized (7). A general property of these transcriptional cofactors is their ability to selectively interact with liganded receptors and modulate their transcriptional activity. The SRC family of coactivator protein includes steroid receptor coactivator-1 (SRC-1) (8), SRC-2 (GRIP, TIF2) (9, 10), and SRC-3 (ACTR, TRAM, AIB1) (11)(12)(13). Each member of this family has a centrally located receptor interaction domain containing three copies of a consensus leucine-rich motif, LXXLL, with conserved spacing between the motifs (14,15). Crystallographic and biochemical studies reveal that the surface of a single LXXLL motif directly contacts the ligand-activated AF-2 domain of NRs, thereby providing a molecular basis for NRcoactivator recruitment (2,5,14,16,17). Although the precise mechanism is unclear, these coactivators may function as bridging proteins that link the receptor to RNA polymerase II and the basal transcription machinery and potentially recruit limiting components into preinitiation complex assembly. Indeed, SRC-1 interacts with general transcription factors such as transcription factor IIB and TATA-binding protein (18). SRC-1 also interacts with the general transcriptional activator CBP/p300 through a distinct C-terminal domain, most likely as part of a large multi-protein complex assembled at the target gene promoter (19,20). Furthermore, coactivator proteins such as SRC-1 and CBP/p300 possess intrinsic histone acetyltransferase activity, suggesting that ligand-activated receptors may recruit coactivators that function to remodel chromatin structure, thus permitting greater accessibility of the transcriptional machinery to DNA (21)(22)(23).
A novel coactivator protein termed NCoA-62 was isolated in our laboratory as a VDR-interacting protein (24). NCoA-62 enhances VDR and other NR-mediated gene transcription. Although NCoA-62 is not related to the SRC coactivator family, it is highly related to BX42, a Drosophila melanogaster nuclear protein putatively involved in ecdysone-stimulated transcription (25). NCoA-62 was independently isolated as a protein that interacts with the v-Ski oncoprotein, and it was termed Skiinteracting protein, SKIP (26). Ski is a potent cellular differentiation factor that has been shown to function in part through NR-signaling pathways (27). Therefore, these studies also imply a role for NCoA-62/SKIP in cellular differentiation pathways including steroid hormone-mediated cellular differentiation. The mechanisms through which NCoA-62/SKIP function and the potential interplay between distinct coactivators such as NCoA-62/SKIP and the SRC coactivators in VDRmediated transcription are unknown. In the present study, we demonstrate a preferential, ligand-mediated interaction of NCoA-62/SKIP with the VDR-RXR heterodimer as compared with the VDR monomer or homodimer. Moreover, we show cooperative interactions between VDR, NCoA-62/SKIP, and SRC coactivators and a resulting synergistic effect of these two distinct coactivators in VDR-mediated transcription. These cooperative interplays are mediated through interactions of the SRCs and NCoA-62/SKIP coactivators with distinct domains of the VDR that permit ternary complex formation between the liganded VDR/RXR heterodimer and the NCoA-62/SKIP and SRC coactivator proteins.
NCoA-62 Deletion Constructs-Deletion mutants of NCoA-62 were generated by polymerase chain reaction. Primers with unique restriction site sequences were used to amplify selected regions of the NCoA-62 and were cloned into the SG5 expression plasmid. 5Ј primers contained a strong Kozak consensus sequence for efficient translation. All constructs were verified by DNA sequencing.
In Vitro Protein Interaction Assay-[ 35 S]Methionine-labeled protein was synthesized in a TNT T7 coupled in vitro transcription-translation system using rabbit reticulocyte lysate (Promega, Madison, WI). 10 g of GST or GST fusion protein was bound to 50% glutathione-agarose beads (Sigma) and equilibrated with 2ϫ GBB (20 mM Tris, pH 7.6, 50 mM NaCl, 1 mM dithiothreitol, 0.2% Nonidet P-40, and protease inhibitors (0.2 mM phenylmethylsulfonyl fluoride, 4.0 g/ml aprotinin, 2.0 g/ml leupeptin, and 1 g/ml pepstatin A)). Equivalent volumes of [ 35 S]methionine-labeled proteins were incubated with the immobilized GST fusion proteins in 100 l of 2ϫ GBB for 2 h at 4°C with gentle rocking in the presence or absence of 10 Ϫ8 M 1,25-dihydroxyvitamin D 3 . Histidine-tagged baculovirus-expressed human VDR was added in select experiments to examine ternary protein complex formation. Several experiments examined the effect of NR Box II and III peptides as competitors for protein complex formation. The beads were washed twice with 0.3 ml of 1ϫ GBB and once with 0.3 ml of 50 mM Tris-HCl, pH 8.0, buffer. Bound proteins were eluted with 35 l of 20 mM reduced glutathione in 50 mM Tris buffer. Eluted proteins were resolved by SDS-PAGE and visualized by autoradiography.
Baculovirus-mediated Expression of VDR-Human VDR cDNA was subcloned into the pVL-1392 polyhedrin transfer plasmid containing a polyhistidine tag and a protein kinase A consensus site at the N terminus. Recombinant baculovirus was isolated and plaque-purified by standard methods. A 1-liter culture of Sf-9 cells (1 ϫ 10 6 cells/ml) was infected for 48 h with the VDR-expressing recombinant baculovirus (multiplicity of infection ϭ 3). Whole cell extracts were prepared, and recombinant, His-tagged VDR was purified by nickel affinity chromatography.
Cell Culture and Transient Transfection-The vitamin D-responsive (VDRE) 4 -TK-GH growth hormone reporter plasmid contains four copies of the rat osteocalcin VDRE upstream of the viral thymidine kinase promoter. The (VDRE) 4 -TATA-GH construct utilizes the rat osteocalcin TATA region (Ϫ30 to ϩ10) as the basal promoter. The pSG5 human VDR expression plasmid was described previously (24). All cells were cultured in Dulbecco's modified Eagle's medium containing 5% charcoal-stripped, heat-inactivated calf bovine serum for 4 days before transfection as well as throughout the entire transfection protocol. COS-7 cells were transfected by standard calcium phosphate precipitation procedures as described previously (32). Carrier DNA (pBlueScript II-KS ϩ ) was added to bring the total DNA content to 15 g/60-mm plate. After a 16-h incubation, the precipitate was removed with two washes of phosphate-buffered saline, and the cells were replenished with fresh media containing 5% charcoal-stripped, heat-inactivated serum. The cells were treated with ligand or ethanol vehicle as indicated for 24 h, and the amount of secreted growth hormone was determined with an immunoassay kit (Nichols Institute, San Juan Capistrano, CA).

NCoA-62 Selectively Binds to the VDR-RXR Heterodimer in a
Ligand-dependent Manner-Baudino et al. (24) demonstrate that NCoA-62 interacts with nuclear receptor superfamily members and augments the transcriptional activities of liganded VDR, GR, retinoic acid receptor, and estrogen receptor (24). In that study, Far Western analysis and GST pull-down assays revealed a direct interaction of NCoA-62 with monomeric VDR. Here, we directly compared the abilities of NCoA-62 and SRC coactivators (GRIP-1/SRC-2) to interact with the VDR-VDR homodimer or the VDR-RXR heterodimer complexes using an in vitro ternary complex assay. As illustrated in Fig. 1, 35 Slabeled NCoA-62 interacted weakly with the GST-VDR fusion protein (lane 5). The addition of purified baculovirus-expressed VDR modestly enhanced NCoA-62 interaction with the GST-VDR protein (lane 6). This modest increase likely is the result of weak homodimerization of the immobilized GST-VDR with the baculovirus expressed VDR (BEVS-VDR) and subsequent stronger interaction of NCoA-62 with dimeric VDR. This is supported by the observations that additional BEVS-VDR had no effect on interaction with the GST control matrix (lane 4), and unrelated proteins such as bovine serum albumin or ovalbumin had no effect on NCoA-62 binding to GST-VDR (data not shown). NCoA-62 interaction with the homodimer was independent of the 1,25-(OH) 2 D 3 ligand (compare lanes 6 and 7). In contrast, 35 S-labeled GRIP interacted with the VDR complexes in a strict, ligand-dependent manner (compare lanes NCoA-62 interaction with the liganded heterodimer was ϳ4-fold higher than with the liganded VDR homodimer complex (compare lanes 7 and 10). In contrast, GRIP showed no significant difference between its ability to interact with the heterodimer or with the homodimer (middle panel, compare lanes 7 and 10). As expected, no interaction was observed for calcium-binding protein (CaBP) in any of these binding conditions (lower panel in Fig. 1). These data suggest that NCoA-62 selectively binds to the VDR-RXR heterodimer, and its interaction with the heterodimer is enhanced by the 1,25-(OH) 2 D 3 ligand.
The selective interaction between NCoA-62 and the VDR-RXR heterodimer was also indicated in transient expression studies in COS-7 cells transfected with a VDR expression vector and the 1,25-(OH) 2 D 3 -responsive VDRE 1 -TATA-GH reporter construct. As illustrated in Fig. 2, NCoA-62 and RXR expression, each, resulted in a modest 2-3-fold increase in 1,25-(OH) 2 D 3 -dependent growth hormone reporter expression under these conditions. Coexpression of NCoA-62 and RXR resulted in a 6-fold increase in VDR-mediated transcription, suggesting that both NCoA-62 and RXR expression were important for the maximal responsiveness in this system.
Residues 274 -342 of NCoA-62 Mediate Interactions with the VDR-RXR Heterodimer-A series of NCoA-62 deletion constructs was used to define the region of NCoA-62 that is involved in mediating interactions with the VDR-RXR heterodimer. This was assessed using the in vitro ternary complex assay ( Fig. 3A and summarized in Fig. 3B). Deletions of Nterminal residues 1-309 (N1, N2, N3, and N4) had no effect on the ability of the NCoA-62 proteins to bind to the VDR/RXR heterodimer. Moreover, the formation of the ternary complexes was enhanced by the addition of the 1,25-(OH) 2 D 3 ligand, as observed with the wild type protein. Further deletions of amino acids 309 -341 (mutant N5) resulted in the loss of formation of the ternary complex. C-terminal deletion mutants containing amino acids 87-309, 87-342, and 87-395 (C2, C3, and C4, respectively) each retained the ability to interact with VDR/ RXR. In contrast, mutant C1, containing amino acids 87-274, did not bind to VDR/RXR. These results suggest the presence of two binding sites on the NCoA-62 protein. One site is located between amino acids 274 and 309, and the other is located between residues 309 and 342. The presence of either binding site is sufficient to promote binding to the heterodimer in vitro.
The ability of the NCoA-62 deletion mutants to bind to a VDR monomer was examined in a yeast 2-hybrid system (Fig.  3B). The N1-N3 N-terminal deletion mutants were each active in this system. Further deletion of the amino acids 274 -309 in N4 resulted in a loss of interaction with VDR. Likewise, the C-terminal mutants C2-C4 were active, whereas the loss of amino acids 274 -309 in C1 abolished binding activity. These results support the in vitro data and demonstrate that amino acids 274 -309 are necessary for binding to the VDR monomer. Taken together, these data further suggest that amino acids 309 -342 are involved in binding to either the RXR monomer or to the VDR/RXR heterodimer but are not sufficient for binding to the VDR monomer.
NCoA-62 and SRC Coactivators Utilize Different Interaction Domains on the VDR-RXR Heterodimer-SRC coactivators such as GRIP-1 and SRC-1 interact with NRs in a ligand-dependent and AF-2 (helix H12)-dependent manner. Several signature LXXLL motifs (referred to as NR boxes) within the SRC coactivator primary sequence are necessary and sufficient for coactivator interaction with a hydrophobic cleft on the surface on the NRs consisting of helices H3, H4, H5, and H12, the AF-2 domain (5,14,16,33). However, NCoA-62 lacks any of the signature LXXLL or NR-box motifs, and the interaction between NCoA-62 and VDR is independent of the AF-2 domain of helix H12 (24). These observations suggest that NCoA-62 and SRC coactivators utilize different interaction domains on the VDR-RXR heterodimer to influence VDR-mediated transcription. To test this hypothesis, synthetic LXXLL-containing peptides were used as competitors for the binding of GRIP-1 or NCoA-62 to the VDR-RXR heterodimer complex in GST pulldown assays (Fig. 4). Peptides containing either NR Box II (KHKILHRLLQDSS) or NR Box III (ENALLRYLLDKDD) of mouse GRIP-1 were used.
As shown in Fig. 4, both NCoA-62 and GRIP-1 formed a ligand-enhanced ternary complex with the GST-RXR-VDR heterodimer (compare lanes 4 and 5). When increasing concentrations of NR Box III peptide were added, there was a dose-dependent decrease in the amount of 35 S-GRIP that complexed with the heterodimer (lanes 6 -10). Importantly, no competition of the NR Box III peptide for binding of NCoA-62/SKIP to the liganded GST-RXR-VDR heterodimer was observed (lanes 6 -10). Similar results were obtained using 35 S-SRC-1 in this assay (data not shown). In contrast to NR Box III, the NR Box II peptide did not compete with either GRIP, SRC-1, or

NCoA-62/SKIP and VDR-mediated Transcription
NCoA-62 in the formation of ternary complex (data not shown), thus confirming the results of Chen et al. (34) who showed that VDR selectively interacts with NR Box III peptides. The data in Fig. 4 clearly indicate that NCoA-62 and SRC coactivators dock with distinct regions of the VDR-RXR heterodimer.
Ternary Complex Formation among NCoA-62, VDR, and SRC Coactivators-The observation that NCoA-62 and SRC coactivators interact with the VDR-RXR heterodimer through distinct domains suggests that these proteins have the potential to form a complex. To test this possibility, purified GST-NCoA-62 was bound to glutathione-agarose incubated with 35 S-labeled, full-length GRIP-1, and protein-protein complexes were analyzed by SDS-PAGE and autoradiography. As shown in Fig. 5, lanes 1-5, GRIP did not interact directly with GST-NCoA-62 in this assay (lane 3), and unliganded VDR did not affect the ability of GRIP to coprecipitate with GST-NCoA-62 (lane 4). However, the addition of the 1,25-(OH) 2 D 3 -liganded VDR dramatically increased coprecipitation of GRIP and GST-NCoA-62 (lane 5). This ternary complex was also observed using GST-GRIP and 35 S-labeled, full-length NCoA-62 (lanes 6 -10). As expected, GST-GRIP did not interact directly with NCoA-62 in this assay (lane 8). However, GST-GRIP efficiently interacted with 35 S-labeled NCoA-62 in a liganded VDR-dependent manner (lane 10). Full-length SRC-1 also formed a ternary complex with VDR and NCoA-62 in a manner similar to GRIP-1 (data not shown). These data indicated that although NCoA-62 does not interact directly with SRC coactivators, it forms a 1,25-(OH) 2 D 3 -dependent ternary complex with VDR and SRC coactivators.

Synergistic Interplay between NCoA-62 and SRC Coactivators in VDR-mediated Transcription-
To address the functional consequences of this interaction, NCoA-62 and SRC coactivators were examined in a VDR-mediated reporter gene assay (Fig. 6). Expression of GRIP-1 or SRC-1 augmented VDRmediated transcription by ϳ2-fold, a level of enhancement similar to that observed with NCoA-62 expression under these conditions of reduced expression vectors (Fig. 6). Strikingly, coexpression of NCoA-62 and either GRIP or SRC-1 resulted in a synergistic enhancement in 1,25-(OH) 2 D 3 -activated expression of the reporter gene construct (8-or 7-fold, respectively). As expected, coexpression of GRIP and SRC-1 resulted in merely an additive effect on vitamin D-mediated transcription compared with each coactivator alone. This observation shows that there is synergistic interplay between NCoA-62 and SRC coactivators in VDR-mediated transcription.
To address the relative functional importance of the cooperative actions of NCoA-62 or SRC coactivators in vitamin Dmediated transactivation, protein inhibitors were developed based on the SRC receptor interaction domain (RID) and the NCoA-62 RID. A similar system was used previously to define NR-box peptides that selectively compete for nuclear receptor transactivation in transient expression studies (35,36). The SRC RID consisted of residues 595-780 of mouse GRIP-1 (SRC-2) containing the 3 NR-box motifs. The NCoA-62 inhibitor consisted of residues 87-342 of human NCoA-62 containing the NCoA-62 RID between residues 274 and 342. The NCoA-62 (87-342) derivative binds to both the VDR monomer and the VDR-RXR heterodimer (see Fig. 3). As illustrated in Fig. 7, the SRC RID selectively interfered with VDR-activated transcription, inhibiting 1,25-(OH) 2 D 3 -activated transcription by ϳ90%. This presumably occurs by interfering with native SRCs and blocking their interaction with the VDR in this cell system. As expected, coexpression of intact GRIP-1 with this inhibitor completely rescued the VDR response in this system. However, coexpressing intact NCoA-62 had little, if any, effect. These data indicate that NCoA-62 lacks significant coactivator activity in the absence of SRC interaction with VDR. In a complimentary experiment, the NCoA-62 RID was used to inhibit VDR-mediated transcription. This inhibition was effectively rescued by intact NCoA-62. However, coexpressing intact GRIP-1 in this system had only minimal effects of VDR-activated transcription, suggesting that SRC coactivators also show minimal effectiveness in the absence of intact NCoA-62  1, 5, 8 , 11, 14, 17, 20, 23, 26, 29, and 32) represent 10% of each 35  interaction. Thus, these data indicate that both coactivator classes may be required for each other's activity in this VDRmediated system and highlight the importance of both coactivators in the mechanism of VDR-activated transcription.

DISCUSSION
Vitamin D-mediated transactivation is a multistep event in which the interaction of VDR or VDR-RXR heterodimers with coactivators is an essential step. SRC coactivators such as GRIP-1 and SRC-1 interact with NRs in a ligand-dependent manner through the AF-2 domain, and this interaction is important in modulating the transcriptional response of the liganded receptor (8,9,37). However, a number of NR coactivators are distinct from the SRC family and may function through alternate pathways. Several examples include the vitamin D receptor-or thyroid hormone receptor-interaction protein complex (DRIP TRAP) (6,38) and CBP/p300 (19,21,22). NCoA-62/ SKIP is a distinct coactivator whose primary sequence is unrelated to the SRC family of proteins or to other nuclear receptor coactivators that have been described thus far. Moreover, NCoA-62/SKIP lacks the signature LXXLL motifs of many NR coactivators, and it does not require the AF-2 domain (helix H12) of VDR for interaction (24). These facts argue strongly for a functional role of NCoA-62/SKIP in VDR-activated transcriptional that is distinct from that of the SRC coactivator family, but evidence for this possibility has been lacking. Here, we report the original finding that selective, multiprotein complexes form between VDR, RXR, NCoA-62/ SKIP, and SRC coactivators and that these interactions result in a cooperative coactivator effect on VDR-activated transcription. The significant findings of this study are 1) that NCoA-62/SKIP interacts preferentially with the liganded VDR-RXR heterodimer compared with the liganded VDR monomer or homodimer complex, 2) that NCoA-62/SKIP forms a ternary complex with VDR and SRC coactivators that ultimately results in synergistic effects of NCoA-62/SKIP and SRC coactivators in VDR-mediated transcription, and 3) that NCoA-62/ SKIP and SRC coactivators are both required for optimal VDRmediated transcription in this cell culture system.
In agreement with our previous in vivo two-hybrid and in vitro protein interaction assays, NCoA-62/SKIP was shown here to interact with GST-VDR and GST-RXR fusion proteins. However, a direct comparison with the SRC coactivators (GRIP-1 or SRC-1) revealed the relatively weak nature of NCoA-62/SKIP interaction with the VDR or RXR monomer in this assay (Fig. 1). An important limitation of our previous interaction studies with NCoA-62/SKIP is that they examined only monomer interaction with coactivators. However, the VDR functions as a homodimer or as a heterodimer with RXR to regulate target gene transcription. To test the possibility that NCoA-62/SKIP might preferentially interact with the VDR-VDR homodimer or with the VDR-RXR heterodimer as a functional unit, the full-length VDR protein was expressed and purified as a partner for GST-VDR or GST-RXR in in vitro binding assays. As shown in Fig. 1, NCoA-62/SKIP interaction with the liganded VDR-RXR heterodimer was about 4-fold higher than NCoA-62/SKIP interaction with the liganded VDR homodimer complex and 12-fold higher than that interaction seen with the VDR monomer. This was in contrast to the SRC coactivators that interacted equally well with the liganded VDR monomer, VDR homodimer, and the VDR-RXR heterodimer under the conditions used in this assay.
There are several potential mechanisms that may explain the preferential interaction of NCoA-62/SKIP with the VDR-RXR heterodimer. One is that NCoA-62/SKIP may express two binding domains, one that contacts VDR (residues 274 -309) and the other that contacts RXR (residues 309 -342). The yeast 2-hybrid studies in Fig. 3B provide some initial support for this possibility, but more refined mapping studies are required. In this model, heterodimer formation may produce the proper configuration to permit high affinity binding of these two NCoA-62/SKIP domains to the VDR-RXR heterodimer. Such a mechanism is analogous to SRC coactivator interaction in which one NR-box contacts one monomer and another contacts the other partner. Recent structural analysis of a complex of liganded peroxisome proliferator-activated receptor-␥-ligand binding domain with a peptide encompassing two LXXLL motifs of SRC-1 generated a model in which NR boxes II and III each bind to one monomer of the peroxisome proliferator-activated receptor-␥-RXR heterodimer, yielding a stoichiometry of one SRC-1 molecule per receptor dimer (17). The observation that NCoA-62/SKIP has no obvious repeated sequences such as the three LXXLL motifs in the SRC coactivators may tend to argue against such a model for NCoA-62/SKIP interaction with heterodimers. Alternatively, residues 309 -342 of NCoA-62/ SKIP may contact an interface that is formed by the VDR-RXR heterodimerization event. Interestingly, the addition of ligand dramatically enhanced the stability of the VDR⅐RXR⅐NCoA-62 complex though a mechanism that has not been defined as yet. It is possible that the 1,25-(OH) 2 D 3 ligand causes a conformational change in the VDR-RXR heterodimer, thereby enhancing the ability of NCoA-62/SKIP to associate with the complex. The 1,25-(OH) 2 D 3 ligand is also known to promote VDR-RXR heterodimerization (39). Thus, it is equally possible that 1,25-(OH) 2 D 3 simply promotes VDR-RXR heterodimerization, leading to more stable heterodimers that can bind NCoA-62/SKIP. Additional studies will be required to determine the level at which the ligand exerts its effect on the interaction and to finely map the NCoA-62/SKIP interaction surface on the VDR-RXR heterodimer. Regardless, the current data clearly point to an important role of NCoA-62/SKIP in selective transcriptional events mediated by the VDR-RXR heterodimer.
Although we have not yet defined the precise NCoA-62/SKIP interaction surface, our recent data can exclude three important regions of the VDR; they are 1) the DNA binding domain (region C), 2) the heterodimerization interface (within and around helix H11), and 3) the SRC coactivator interaction surface (helices H3-5 and H12). The first can be excluded since the GST-VDR used in these in vitro studies lacks regions A-C and our previous studies showed that deletion of residues 1-116 (containing the DNA binding domain of the VDR) does not affect NCoA-62/SKIP interaction with VDR (24). The heterodimerization surface is also not involved since, as discussed above, VDR-RXR heterodimerization does not interfere but actually promotes NCoA-62/SKIP interaction with VDR. Finally, the SRC interaction surface can be excluded as well since deletion of helix H12 (the AF-2 domain) does not affect NCoA-62 interaction (24) and the NR-box peptide competition studies in Fig. 4 showed that NCoA-62 binds to a surface on the VDR-RXR heterodimer that is distinct from the LXXLL binding surface. The in vitro binding assays in Fig. 5 also demonstrated that both NCoA-62 and SRC coactivators may simultaneously interact with VDR, forming a ternary complex whose formation is mediated through distinct interaction surfaces on the VDR. The formation of this ternary complex is supported by the observations that NCoA-62 interacted weakly, if at all, with GRIP-1 or SRC-1. The addition of unliganded VDR to this assay had little effect, but the addition of liganded VDR dramatically enhanced the association of NCoA-62, SRCs, and VDR. Taken together, these data indicate that liganded VDR recruits SRC coactivators to the NCoA-62⅐VDR complex. In this scenario, VDR is the bridging protein that links NCoA-62/SKIP and SRCs to this complex. The SRCs dock with the ligand-dependent H3-5/H12 surface of VDR, whereas NCoA-62/SKIP simultaneously docks to a still undefined surface(s) on the VDR-RXR heterodimer.
The formation of this NCoA-62⅐VDR⅐SRC ternary complex is functionally relevant, and it provides a potential molecular basis for the functional interplays that are apparent between NCoA-62/SKIP and SRC coactivators in VDR-activated transcription. As evidenced in our transient reporter gene assays in Fig. 6, coexpression of NCoA-62 and SRC coactivators resulted in a synergistic enhancement of VDR-activated transcription, whereas coexpression of the two related SRCs (SRC-1 and GRIP-1) resulted in additive effects. One suggestion from these data is that NCoA-62/SKIP and SRCs function through distinct mechanisms. The SRCs are known to possess histone acetyltransferase activity and to recruit other proteins such as CBP/ P300 that possess histone acetyltransferase activity (22, 23). We have not been able to detect significant histone acetyltransferase activity in immunopurified NCoA-62/SKIP complexes, suggesting that NCoA-62/SKIP, like the purified DRIP⅐TRAP complex, lacks this important chromatin modifying activity. Thus, it is possible that NCoA-62/SKIP acts at another level of the transactivation process. The importance of both NCoA-62/ SKIP and SRCs in VDR-mediated transactivation is highlighted in the protein interference studies in Fig. 7. As reported previously for several NRs (35,36), expression of the hSRC-1 RID (residues 595-780) blocked VDR-activated transcription by preventing endogenous SRCs and perhaps other LXXLLcontaining proteins from interacting with the H3-5/H12 coactivator interaction surface. Importantly, NCoA-62/Skip is unable to rescue this inhibition (Fig. 7) despite the fact that it retains the ability to interact with the VDR (Fig. 4). This observation agrees well with VDR AF-2 mutations that abolish FIG. 6. NCoA-62 and SRC coactivators act synergistically to enhance VDR-mediated transcription. COS-7 cells were plated in 60-mm dishes and transfected with 2 g of (VDRE) 4 -TK-GH and 10 ng of SG5-VDR. Each group received pSG5 parent plasmid or 333 ng of pSG5-NCoA-62, pSG5-GRIP-1, pCR-SRC-1, or a combination of these as indicated. All transfections were equalized with the appropriate amounts of parent expression plasmids. After 24 h of treatment with 10 Ϫ8 M 1,25-(OH) 2 D 3 , GH values were determined. GH values are presented as the mean Ϯ S.D. The numbers above each bar represent fold induction as compared with ethanol-treated controls.

NCoA-62/SKIP and VDR-mediated Transcription
SRC coactivator interaction and transactivation while maintaining interaction with NCoA-62 (24). Thus, NCoA-62/Skip interaction with VDR is not sufficient to enhance the transactivation process; SRC contact with the receptor is needed for NCoA-62/Skip to function. In this sense, NCoA-62/SKIP might be classified as a comodulator of the SRC coactivator proteins. However, the data in Fig. 7 also highlight the requisite role of NCoA-62/SKIP in this process since expression of the NCoA-62/SKIP RID interferes with VDR transactivation, and this interference cannot be rescued with SRC coactivators. Although the precise roles of NCoA-62/SKIP and SRCs in this process have not been determined, the data here highlight the importance of both classes of coactivators and their potential assembly into a complex of proteins in the overall transactivation process mediated by VDR and perhaps other nuclear receptors. Understanding whether these observations extend into whole animal physiology or into native promoter contexts are important future research goals.
In summary, our data provide strong supportive evidence for the role of NCoA-62/SKIP as a novel coactivator protein in VDR-mediated transcription. NCoA-62/SKIP functions by selectively binding to the VDR-RXR heterodimer in a ligandenhanced manner. The subsequent synergistic interplay between NCoA-62/SKIP and SRC coactivators in VDR-mediated transcription may result from simultaneous interaction of NCoA-62/SKIP and SRC coactivators with distinct interaction domains within the liganded VDR-RXR heterodimer. More detailed functional studies are needed to obtain a better understanding of the precise nature of the ternary complex formations and the signaling pathways in which NCoA-62/SKIP may be involved. In addition, interactions of NCoA-62/SKIP with other coactivators, repressors, or with the components of preinitiation complex will be required to further understand the distinct role that NCoA-62/SKIP and SRCs play in nuclear receptor-mediated transcriptional pathways.