ADA3-containing complexes associate with estrogen receptor alpha

Abstract

Transcriptional repression and activation by nuclear receptors (NRs) are brought about by coregulator complexes. These complexes modify the chromatin environment of target genes and affect the activity of the basal transcription machinery. We have previously implicated the yeast ADA3 protein in transcriptional activation by estrogen and retinoid X receptors in yeast and mammalian cells. Here we report the cloning of the mouse homolog of ADA3 and its characterization with respect to the estrogen receptor alpha (ERα) function. Mouse mADA3 is 23% identical and 47% similar to yeast yADA3, and mADA3 in contrast to yADA3 does not interact with NRs directly even though it contains two LxxLL NR boxes. However, the ADA3-containing TBP-free-TAF-containing complex (TFTC) can interact with ERα in a ligand-independent manner, indicating that other subunits of the complex are sufficient to mediate interaction with NRs.

INTRODUCTION

Nuclear receptors (NRs) are ligand-dependent transcriptional regulators that have evolved from an ancestral orphan receptor into a highly diverse family present throughout the entire animal kingdom and encompassing receptors for steroid and non-steroid hormones, vitamins and metabolic intermediates (1,2). They have a wide variety of responsive genes to which they bind as mono-, homo- and heterodimers through response elements. NRs are composed of five to six independent domains that encode specific functions, including transcriptional activation and repression, DNA and ligand binding, cellular compartmentation and dimerization (1). NRs can activate transcription through two independent activation functions located in the N-terminal AB domains (AF-1) and the C-terminal ligand-binding domain (LBD, AF-2) (1–3). Binding of the ligand induces a major conformational change in the LBD, which modulates coregulator binding to NRs (3–5). Direct transcriptional repression by some NRs is mediated by co-repressor complexes that are associated with the unliganded receptor and condense the chromatin environment of the promoter through histone deacetylation (4–6). Upon ligand binding, co-repressors dissociate from the NR, and co-activators are recruited (4–6). Co-activators recognize the holo-LBD via conserved LxxLL motifs and in some cases the N-terminal activation function AF-1 (5–7). We have previously shown that the yeast yADA3 protein can act as a NR co-activator in yeast and transfected mammalian cells (8). ADA3 belongs to a group of proteins that were first characterized in yeast, and later identified in higher eukaryotes (9–14). ADA proteins have been found to be required for transcriptional activation by a number of yeast activators (15–17 and references therein). In yeast, several ADA protein complexes have been identified (18–21). ADA3 is found in vivo within multisubunit complexes of different size (0.2, 0.9 and 1.9 MD) and complexity that contain at least three to four additional proteins: ADA1, ADA2, ADA5 and GCN5 (20,21). In higher order complexes, different TAFs and Spt proteins were also found (16,22). In mammalian cells, the majority of ADA3 protein also seems complexed with Spt and TAF or TAF-like factors (11), making up the P/CAF, GCN5, STAGA and TBP-free-TAF-containing complexes (TFTCs) (10–14). These complexes are thought to be functional homologs of the yeast ADA complexes (11,12,14). Although at present not convincingly demonstrated, these complexes probably are recruited by different transcriptional activators, and have stimulatory activity on transcription in vitro (23–26). Interestingly, these complexes contain besides ADA2 and ADA3 additional subunits that have previously been implicated in NR signaling. TAF_II30 is present in P/CAF, GCN5, STAGA and TFTC complexes, and has been shown to act on estrogen receptor alpha (ERα) function (27). Furthermore, we have shown that ERα transactivation is impaired in yeast when yADA3 is deleted, and yADA2 and yGCN5 are required in addition to yADA3 for estrogen and retinoid X receptor function (8). GCN5 and the related protein P/CAF are also found in all four previously described mammalian complexes, and were reported to interact either directly or indirectly with NRs (26,28–30). Moreover, TAF_II135 and TAF_II55, both present in the TFTC, have been reported to have effects on NR transcriptional activation (31,32). Finally, it was possible to demonstrate that the major glucocorticoid receptor transactivation domain τ-1 can function by recruiting the STAGA complex (23,26). Intriguingly, ADA2 is involved in connecting both molecular entities, but does not seem to be the only factor capable of binding to and recruiting STAGA to the glucocorticoid receptor (26).

Here we present the cloning of the mouse homolog of ADA3. Surprisingly, although structurally related to yeast ADA3 and coding for two NR boxes, mADA3 unlike yADA3 does not bind NRs directly. We present evidence, that instead, mADA3 can be recruited to human estrogen receptor alpha (hERα) as part of the TFTC, and is found in intact cells to be associated with hERα. These results support our initial observation that NRs might function by recruiting ADA3 complexes to activate transcription.

MATERIALS AND METHODS

Cloning of mADA3

The 301 nt long mouse EST sequence we identified as being homologous to the C-terminus of yADA3 has GenBank accession no. AA289046 (33). The synthetic oligonucleotide designed according to this EST sequence for the nested RT–PCR reaction had the following sequences: (i) outer 5′, 5′-AGAGTCAGGGGCAGATGGAG-3′; (ii) inner 5′, 5′-TCCCCGCGCAATCAGAACAA-3′; (iii) inner 3′, 5′-TTCTTGGTGCGGTTGTGTGC-3′; (iv) outer 3′, 5′-GCCGGCTCACCTCCTCCTTT-3′. The RT–PCR reaction was performed with primers 2 and 3 on poly(A)⁺ RNA from 9.5–10.5-day-old mouse embryos. The reaction product was purified and a second PCR reaction using primers 1 and 4 was performed. The final reaction product was then purified on an acrylamide DNA gel, and used as a template for a random priming reaction using radiolabeled [α-³²P]dCTP and Klenow polymerase. The products from the random priming reaction were directly used to screen a random primed λZAPII-based mouse embryonic cDNA library. Positive clones were re-screened and then amplified and subjected to automated DNA sequencing.

Plasmid constructions

Plasmid constructions were done according to standard protocols. Details on plasmid constructions are available upon request.

Protein purification

Recombinant pGEX based bacterial expression vectors were propagated in Escherichia coli strain BL21. Expression of recombinant proteins was induced by treatment of exponential cultures with 0.5 mM IPTG for 2–3 h at ambient temperature. Cells were harvested and extracts prepared according to protocols provided by the manufacturer of the pGEX system (Pharmacia).

Antibodies

The rabbit polyclonal anti-hADA2 antibody was generated using a synthetic ovalbumine-coupled peptide (amino acids 1–17 of hADA2). Similarly, the mADA3 antibody was generated against a synthetic peptide of mADA3. After 4 weeks, rabbits were boosted with a second injection of antigen. Sera were collected starting 6 weeks after the initial immunization and tested for cross-reactivity with purified GST–hADA2 and GST–mADA3 proteins. The antibodies were then purified over antigen-affinity matrices and subsequently ammonium sulfate precipitated. Dialyzed fractions were stored as stock solutions. Antibodies against hERα (αF3), hTAF_II30 and the FLAG epitope have been described earlier. Antibodies against hPAF65β and hSpt3 were gifts from Y. Nakatani and R. G. Roeder, respectively. The FLAG-affinity matrix (αM2) was commercially obtained from IBI-Kodak.

Yeast two hybrid

Yeast two-hybrid experiments were done as previously described (8). We employed the L40 strain (34) harboring a LacZ reporter.

GST-based interaction assay

Gluthathion–Sepharose beads were incubated for 4 h with bacterial extracts containing GST alone or GST-fusion proteins, and subsequently washed four times with GST buffer (50 mM Tris–Cl pH 7.9, 150 mM NaCl, 5% glycerol, 0.1% NP40, 1 mM EDTA, 1 mM DTT). For interaction assays, loaded beads were incubated with either 200 ng of purified TFTC, or 2 µl of rabbit reticulocyte lysates containing translated protein radiolabeled with ³⁵S-methionine (Promega Coupled Transcription/Translation Kit, Promega Corp.). The beads were incubated together with the proteins in a total volume of 100 µl GST buffer for 30 min at ambient temperature. After three to five washes with GST buffer to remove unbound material, beads were resuspended in a suitable volume of 3× SDS loading buffer and subjected to denaturing SDS–polyacrylamide gel electrophoresis (SDS–PAGE). Samples were analyzed by Coomassie staining, western blotting or autoradiography of dried gels.

Transient transfections and immunoprecipitations

COS-1 cells were seeded onto 150 mm cell culture dishes at 1 × 10⁷ cells/plate in Dulbecco’s minimal essential medium lacking phenol red and supplemented with 5% of charcoal treated fetal calf serum and antibiotics. Calcium phosphate precipitates containing 15 µg of DNA were immediately added to the cells. After 12 h, cells were washed and supplemented with fresh media. Following an additional 24 h of incubation the cell layer was washed twice with cold PBS, and cells were collected and resuspended in IP buffer (50 mM Tris–Cl pH 7.9, 150 mM NaCl, 10% glycerol, 0.1% NP40, 1 mM EDTA, 1 mM DTT, 0.5 mM PMSF, PIC). Cell lysis was achieved by two cycles of freeze thawing, cellular debris was removed by centrifugation, and the resulting whole-cell extracts were pre-cleared using protein G Sepharose (150 µl per ml extract). Supernatants were incubated 4 h with 5 µl of anti-FLAG M2 matrix (IBI-Kodak), beads were washed four times with IP buffer and resuspended in 3× Laemmli SDS buffer. Following SDS–PAGE, immunoprecipitated proteins were blotted to nitrocellulose membranes and identified using rabbit polyclonal antibodies and mouse monoclonal antibodies.

MCF7 cells were cultured to confluency in 150 mm cell culture dishes in Dulbecco’s minimal essential medium (again lacking phenol red) supplemented with 10% charcoal-treated and delipidized fetal calf serum and antibiotics. After 12 h of 10^–6 M ligand treatment ∼5 × 10⁸ cells were collected and a nuclear extract was prepared as previously described (35). 300 µl (720 µg of protein) of MCF7 nuclear extracts were incubated with a monoclonal antibody raised against the F domain of ERα (αF3), or as control, with a monoclonal Gal4 antibody (αGal4). Following four washes with IP buffer, bound proteins were analyzed by SDS gel electrophoresis and western blotting.

RESULTS AND DISCUSSION

Cloning of a mouse cDNA homologous to yADA3

As ADA2 and GCN5 homologs had already been found to be present in man, we searched EST databases with sequences derived from the yeast ADA3 open reading frame. We identified a 301 nt mouse EST clone (GenBank accession no. AA289046) bearing homologies to the C-terminal domain of yADA3. Oligonucleotides were designed and used to specifically amplify a nested fragment of the EST sequence by RT–PCR from 9 to 12.5 days post-coitum mouse embryo poly(A)⁺ RNAs. The purified fragment was radioactively labeled in a random primed reaction and used to screen a λZAPII random primed cDNA library made from mouse heart tissue. Positive clones encompassing an entire open reading frame of 1296 nt with an in-frame termination codon just upstream of the initiating ATG were identified and fully sequenced (the sequence has been submitted to GenBank, accession no. AF383154). The sequence-predicted protein is 432 amino acids in length and shows 23% overall identity (47% similarity) with yeast ADA3, 35% identity (61% similarity) with its Drosophila counterpart and 99% identity with the recently cloned human ADA3-like protein (12). As shown in Figure 1A, the putative mouse ADA3 harbors a much shorter N-terminus than yADA3, and the two LxxLL motifs found in the sequence are not located in positions similar to those of yeast ADA3. A northern blot with different mouse tissues reveals the expression pattern of this newly cloned cDNA, which is highest in heart and testis tissues and barely detectable or absent in spleen (Fig. 1B).

Figure 1.

A novel cDNA clone closely related to yeast, Drosophila and human ADA3. (A) Sequence comparison between yADA3 and mADA3. Degrees of conservation (% identity and similarity) are indicated for the different regions of both proteins. Indicated are the LxxLL motifs. The C-terminal ADA2 interaction domain in yeast ADA3 was also found to encode the NR interface (8). (B) Northern blot analyses of mADA3 expression. A Clontech mouse multiple tissue northern blot was probed with a radiolabeled sonde for mADA3 revealing a broad but uneven expression pattern. Note the high levels of mADA3 RNA in heart and testis and the virtual absence of expression in the spleen. Size markers are on the left.

The novel cDNA clone encodes the mouse homolog of ADA3

To ascertain that this novel cDNA indeed encodes a functional homolog of the yeast ADA3 protein, we first unsuccessfully attempted to complement the growth defects of an ada3 deletion strain (8) with low or high-copy-number plasmid carrying the mouse cDNA clone (data not shown). In this respect, it is noteworthy that human ADA2 and GCN5 cDNAs were also unable to complement the ada2 and gcn5 deletion strains for growth (9). This reflects a high degree of evolutionary diversification between yeast and mammalian organisms, and has been observed for other transcription factors and adaptors as well.

Yeast ADA3 is known to be recruited to yeast ADA complexes through interaction with yeast ADA2 protein (17). Thus, we speculated that this interaction might be conserved between the different species and hence tested the putative mADA3 clone for interaction with yeast and human ADA2 in yeast two-hybrid assays. As shown in Figure 2A, a VP16 fusion of the putative mADA3 interacts strongly with lexA–yADA2 and lexA–hADA2. This interaction in both cases is stronger than the interaction of yADA3 with both proteins. To corroborate this result, and exclude the presence of a bridging factor in yeast, we tested the interaction of yADA3 and the putative mADA3 with GST-fused hADA2 in vitro. In vitro translated and ³⁵S-labeled yADA3 and mADA3 were incubated with glutathione matrix-bound GST–hADA2 or, as a control, unfused GST protein alone. An autoradiography of a representative experiment is shown in Figure 2B. Clearly, the putative mADA3 does interact with the hADA2 as does yADA3, but it does not bind the unfused GST, indicating that interaction is specific. This result, together with the above mentioned yeast two-hybrid assays, and the sequence homology to yeast ADA3 strongly support the notion that the novel cDNA indeed encodes the/a mouse homolog of the yeast ADA3 and is henceforth referred to as mADA3.

Figure 2.

mADA3 interacts with ADA2 in vivo and in vitro. (A) Yeast and mouse ADA3 interact with yeast and human ADA2 in yeast. cDNAs expressing the indicated VP16 and lexA fusion proteins were transformed into the yeast strain L40, and the activity of the lexA binding site-containing LacZ reporter was quantified by measuring relative β-galactosidase activity. Given are the mean β-galactosidase values (in relative units) for two independent transformations. (B) Yeast and mouse ADA3 interact with hADA2 in vitro. In vitro ³⁵S-labeled yADA3 (lane 1) or mADA3 (lane 4) was incubated with ‘control’ GST (lanes 2 and 5) or GST fused to hADA2 (lanes 3 and 6). An autoradiography of a dried SDS–polyacrylamide gel is shown. The positions of marker proteins relative to the samples are shown on the left. Arrows indicate the positions of mADA3 and yADA3.

mADA3 does not interact directly with ERα in yeast or in vitro

Since we had shown that yADA3 functions as a bona fide co-activator for retinoid X receptor and especially ERα in yeast (8), we investigated whether mADA3 would be capable of performing a similar role. When performing yeast two-hybrid interaction studies similar to those shown in Figure 2A using lexA fusions of ERα DEF domains, we detected only a marginal interaction between mADA3 and hERαDEF in the presence of estrogen, while yADA3 strongly interacted with liganded hERαDEF (Fig. 3A). Similar results were obtained with a lexA fusion of retinoid X receptor (data not shown).

Figure 3.

mADA3, unlike yADA3, does not interact directly with hERα. (A) mADA3 failed to interact with hERα in yeast. Plasmids expressing yeast and mouse ADA3 fused to VP16 were introduced into the L40 reporter strain together with lexA fused to hERα DEF domains (encompassing the C-terminal LBD). Transformants were grown in the presence or absence of estrogen (500 nM). β-Galactosidase activities are expressed as in Figure 2A. (B) mADA3 also failed to interact with hERα in vitro. A GST-based interaction assay similar to that described in Figure 2B was performed using in vitro ³⁵S-labeled yADA3 (lane 1) or mADA3 (lane 5) incubated with unfused GST (lanes 2 and 5) or GST–hERαCDEF (lanes 3 and 6).

Furthermore, when similar experiments were performed in vitro using GST fusions, we were also unable to detect an interaction between mADA3 and hERαCDEF (Fig. 3B, lanes 7 and 8), while a ligand-independent interaction between yADA3 and hERαCDEF was readily detectable (Fig. 3B, lanes 3 and 4). These results demonstrate that, in contrast to yADA3, mADA3 has no ERα-binding capability on its own.

mADA3-containing complexes interact with ERα in mammalian cells and in vitro

We next investigated whether complexes between ADA3 and ERα might be formed in vivo. COS-1 cells were transfected with expression vectors for Flag-tagged yADA3 or mADA3 and hERα, and immunoaffinity co-purification experiments were performed from estrogen-treated or untreated cells. Western blot analysis revealed that hERα efficiently associates with yADA3 in the immunoprecipitated complexes from co-transfected cells (Fig. 4A, lanes 9 and 10). However, this interaction is not ligand-dependent and therefore differs from that observed in the yeast two-hybrid system (Fig. 3A) (8), indicating that the mechanisms by which yADA3 associates with hERα in mammalian and yeast cells are slightly different. Note that in in vitro interaction assays (Fig. 3B), we also observe ligand-independent binding of yADA3 to the ERαCDEF domains (Fig. 3B, lanes 3 and 4).

Figure 4.

hERα associates with mADA3 in vivo and interacts with the mADA3-containing TFTC in vitro. (A) hERα coimmunoprecipitates with FLAG-tagged yADA3 from transfected COS-1 cells. COS-1 cells were transiently transfected with FLAG-yADA3 and hERα cDNAs. Transfected cells were treated with estradiol (E2, +) or vehicle alone (–) for 12 h. Extracts were prepared and subjected to immunoprecipitation with a monoclonal antibody against the FLAG epitope (αM2). Protein eluates were separated by SDS–PAGE and blotted to nitrocellulose membranes. Proteins were identified using the anti-FLAG antibody to detect the FLAG-tagged mADA3 and a monoclonal antibody (αF3) to detect the ERα. Lanes 1–5 show the input controls for the different cellular extracts, lanes 6–10 the immunopurified samples. The different co-transfected cDNA are indicated on top of the panel, size markers are shown to the left. (B) hERα coimmunoprecipitates with FLAG-tagged mADA3 from transfected COS-1 cells. The FLAG-tagged mADA3 cDNA was transfected alone or in combination with the hERα cDNA (as indicated above the panel). Extracts were obtained as in (A) and subjected to immunoprecipitation with the monoclonal anti-FLAG M2 (IBI-Kodak) agarose. Bound proteins (along with input controls) were identified by western blotting. (C) mADA3 can bind to hERα in vitro as part of the human TFTC. About 200 ng of purified human TFTC was incubated with unfused GST (lane 2) or GST–hERα (lanes 3 and 4) in the presence of either OHT or estrogen (E2). Lane 1 represents 40 ng of hTFTC loaded as input control. Shown is a western blot probed with monoclonal or polyclonal antibodies against the TFTC subunits hDDB1L, hGCN5, hPAF65β, hSpt3 and a polyclonal mADA3 antibody that cross-reacts with the human homolog.

In the case where we transfected an expression vector encoding FLAG-tagged mADA3 instead of yADA3 and performed similar co-immunoprecipitation experiments, we were also able to detect an interaction between mADA3 and hERα (Fig. 4B, lanes 6 and 7). This result indicates that similar to yADA3, mADA3 associates with ERα in mammalian cells. From the yeast two-hybrid data (Fig. 3A) and the in vitro interaction studies (Fig. 3B), it is evident that while yADA3 can bind directly to ERα the interaction of mADA3 with the receptor is mediated by a third, yet unidentified factor, present in the mammalian cell. Note that the in vivo interaction between yADA3 and NRs, albeit direct, might be stabilized by a third protein as well. Since both ADA3 and the related ADA2 have been reported to be present in several high-order complexes in mammalian cells (10–12), we speculated that an interaction between such complexes and ERα is responsible for the apparent indirect association between mADA3 and hERα.

To test this hypothesis, we first tested whether the TBP-free TAF-containing complex TFTC [(14); a gift from Dr Lazlo Tora, highly related to STAGA (11) and P/CAF (12) complexes] would be able to associate with hERα in vitro. Thus, we performed GST-based binding experiments similar to those described above (Fig. 3B) replacing free mADA3 with purified ADA3-containing TFTC. We opted to analyze TFTC interactions with the GST–hERαCDEF in both the presence (Fig. 4C, lane 4) and absence (data not shown) of estrogen and in the presence of anti-estrogen 2-hydroxytamoxifen (OHT; Fig. 4C, lane 3). As can be shown by western blotting of bound fractions, several major components of the TFTC can be easily detected to interact in a ligand-insensitive fashion with ERα (Fig. 4C, lanes 3 and 4). TFTC is complexed in a similar fashion with ERα also in the absence of any ligand (data not shown). The fact that OHT has no effect on the TFTC interaction (Fig. 4C, lane 3) highlights and confirms the observation that association does not depend on the ligand state of the receptor. In conclusion, these experiments demonstrate that TFTC, and most likely other ADA3-containing complexes related to TFTC, can mediate an interaction between mammalian ADA3 and ERα. Several subunits of TFTC and related complexes have already been shown to interact with estrogen receptors and/or with other NRs: hTAF_II30, the hallmark subunit of TFTC, interacts directly with the hERα D-domain (27). The GCN5-related P/CAF subunit has been shown to functionally interact with retinoid receptors, and is considered a CBP (28) and SRC1/TIF2 (28,33) co-activator, which itself acts to stimulate the transcriptional activities of a variety of NRs (28,29,33). These notions are further corroborated by the fact that GCN5 and P/CAF are the histone-acetyl transferases of the above mentioned ADA protein-containing complexes in mammalian cells (11,12,14). ADA2, which is found tightly associated with ADA3 and GCN5 in yeast, has been shown to interact and stimulate the τ-1 activation function of glucocorticoid receptor (26). Furthermore, we have recently been able to show a direct functional interaction between hADA2 and estrogen receptors (A.Benecke, R.Losson and P.Chambon, manuscript in preparation).

To put our observation that ADA3 protein-containing complexes are associated with ERα to a final test, we immunopurified endogenous hADA3 protein from MCF7 human breast cancer cells by virtue of a specific polyclonal antibody (α1328, directed against mADA3), and probed the immunoprecipitates for the presence of hERα. Nuclear extracts from cells treated with OHT, with estrogen (E2), and vehicle alone (–) were used (Fig. 5A). As a control, we also performed immuno-affinity co-purification experiments from the same extracts with an irrelevant monoclonal anti-Gal4 antibody (αGal4). It is evident from Figure 5A that endogenous hERα co-purifies with hADA3 from MCF7 extracts. Note that we could not reveal the presence of hADA3 since the specific band co-migrates with the immunoglobulin heavy chains (revealed in the lower panel to confirm similar loading). We also purified endogenous hERα from MCF7 nuclear extracts to investigate whether the above described interactions are also detectable in this direction. Unfortunately, we are unable to detect hADA3 in these immunoprecipitates since the amounts are beyond the detection limit of our antibody. However, as can be seen in Figure 5B (lanes 5 and 6), both the ADA3 partner ADA2 and the core TFTC component TAF_II30 can be readily co-immunopurified together with the estrogen receptor. This interaction is again insensitive to the ligand the cells were exposed to prior to harvest, confirming our initial observations in vitro (Fig. 4C) and in transfected COS cells (Fig. 4A and B). Thus, we have demonstrated that ADA3-containing complexes are associated with ERα both in vitro and in intact cells.

Figure 5.

Endogenous hADA3 associates with hERα in MCF7 cells. (A) Detection of endogenous hERα in hADA3 immunoprecipitates. MCF7 cells were grown in charcoal-treated and delipidized FCS-containing medium, and treated for 6 h either with OHT, with estrogen (E2) or with vehicle alone (–). Nuclear extracts were prepared and analyzed by western blotting either directly (input) or following immunoprecipitation with a purified polyclonal antibody raised against a mADA3 peptide (α1328) or with an irrelevant antibody raised against the DNA-binding domain of the yeast transcription factor Gal4 (αGal4). Western blots were probed with a monoclonal antibody against hERα (αB10) and subsequently with an anti-light chain peroxidase-coupled secondary antibody (top), or with a mixture of anti-mouse and anti-rabbit heavy chain peroxidase-coupled secondary antibodies (bottom). (B) Detection of TFTC subunits in hERα immunoprecipitates. Immunoprecipitations from MCF7 cell nuclear extracts with a monoclonal antibody against hERα (αF3) co-purify hADA2 (as revealed by polyclonal antibody α1275) and hTAF_II30 (as revealed by monoclonal antibody αTAF_II30) (lanes 5 and 6). Control experiments with the αGal4 antibody confirm the specificity of the interactions (lanes 3 and 4). FT, flow-through from the affinity column (lanes 7 and 8). Note that hERα is quantitatively removed from the extracts.

CONCLUSION

We have previously implicated the yeast ADA3 protein in transcriptional activation by estrogen and retinoid X receptors in yeast and mammalian cells (8). Here we report the cloning of the mouse homolog of ADA3 and its characterization with respect to ERα function. Mouse mADA3 is 23% identical and 47% similar to yeast yADA3, and mADA3 in contrast to yADA3 does not interact with NRs directly even though it contains two LxxLL NR boxes. However, the ADA3-containing TFTC can interact with ERα in a ligand-independent manner, indicating that other subunits of the complex are sufficient to mediate interaction with NRs. These results further implicate ADA3 proteins as critical NR co-factors and indicate that their transcriptional activity is most likely exerted by high molecular weight complexes of the TFTC/PCAF/STAGA group.

It will be of utmost interest to demonstrate the function of this interaction between ERα and the mADA3-containing complexes. Our attempts to demonstrate a co-activator role for mADA3 in transient transfection experiments did not show significant effects of an over-expression of mADA3 on hERα function (data not shown). In this respect, it is interesting to note that hADA3 has been shown to stimulate transactivation by the transcription factor p53 in transient transfection experiments (36), which demonstrates that hADA3 can act as an independent co-activator. The fundamental molecular difference between p53 transactivation by hADA3 and ERα transactivation by mADA3 is that hADA3 interacts directly with p53 while mADA3 interacts indirectly via other proteins with hERα. Over-expression of mADA3 in the ERα case can only indirectly affect transcription by stabilizing the ADA3-containing complex(es) that interact with ERα (11,12,14). Similarly, ADA3 over-expression could also lead to destabilizing such complexes by titrating components to which it binds directly. We have not observed any significant changes in ER activity in the presence of exogenous mADA3, indicating that neither effect is significant or dominant, and ADA3 is not the sole limiting component in the cell. We suggest that the ADA-complex independent activity of ADA3 on p53 is also at the basis of the co-activator function that yADA3 exerts on ERα since yADA3 in contrast to mADA3 can directly interact with ER (8).

DDBJ/EMBL/GenBank accession no. AF383154