An hGCN5/TRRAP Histone Acetyltransferase Complex Co-activates BRCA1 Transactivation Function through Histone Modification*

It is well established that genetic mutations that impair BRCA1 function predispose women to early onset of breast and ovarian cancer. However, the co-regulatory factors that support normal BRCA1 functions remain to be identified. Using a biochemical approach to search for such co-regulatory factors, we identified hGCN5, TRRAP, and hMSH2/6 as BRCA1-interacting proteins. Genetic mutations in the C-terminal transactivation domain of BRCA1, as found in breast cancer patients (Chapman, M. S., and Verma, I. M. (1996) Nature 382, 678–679), caused the loss of physical interaction between BRCA1 and TRRAP and significantly reduced the co-activation of BRCA1 transactivation function by hGCN5/TRRAP. The reported transcriptional squelching between BRCA1 and estrogen receptor α (Fan, S., Wang, J., Yuan, R., Ma, Y., Meng, Q., Erdos, M. R., Pestell, R. G., Yuan, F., Auborn, K. J., Goldberg, I. D., and Rosen, E. M. (1999) Science 284, 1354–1356) was rescued by the overexpression of TRRAP or hGCN5. Histone acetyltransferase hGCN5 activity appeared to be indispensable for coregulator complex function in both BRCA1-mediated gene regulation and DNA repair. Biochemical purification of the hGCN5/TRRAP-containing complex suggested that hGCN5/TRRAP formed a complex with hMSH2/hMSH6, presumably as a novel subclass of hGCN5/TRRAP-containing known TFTC (TBP-free TAF-containing)-type histone acetyltransferase complex (hTFTC, hPCAF, and hSTAGA) (Yanagisawa, J., Kitagawa, H., Yanagida, M., Wada, O., Ogawa, S., Nakagomi, M., Oishi, H., Yamamoto, Y., Nagasawa, H., McMahon, S. B., Cole, M. D., Tora, L., Takahashi, N., and Kato, S. (2002) Mol. Cell 9, 553–562). Unlike other subclasses, the isolated complex harbored a previously unknown combination of components including hMSH2 and hMSH6, major components of the BRCA1 genome surveillance repair complex (BASC). Thus, our results suggested that the multiple BRCA1 functions require a novel hGCN5/TRRAP histone acetyltransferase complex subclass.

and two tandem copies of BRCT 2 at its C-terminal end (8). The major function of BRCA1 is thought to be as a tumor suppressor via DNA repair and transcriptional control (9) presumably involving chromatin remodeling and histone modification (10). The transcription factor function of BRCA1, as a co-regulator of other classes of sequence-specific regulators, is dependent on the BRCT autonomous transactivation domain (11,12). The physiological significance of the BRCT domain is further supported by the finding that a number of BRCA1 mutations found in breast cancer patients involve the BRCT domain with resultant loss of transactivation function (3,13). However, despite the pivotal role of BRCT function in BRCA1-mediated tumor suppression, little is known about the co-regulators and co-regulator complexes that support BRCT function (14).
Recent progress in cell biology has revealed that chromatin remodeling and modification are indispensable for events involving chromosomal DNA. A large number of chromosomal DNA-interacting factors and complexes have been identified, and most of them appear to exhibit specific enzyme activities and chromatin remodeling functions (10,15). For gene regulation by sequence-specific regulators and co-regulators, chromatin remodeling and modification are thought to be tightly coupled such that histone acetylation appears to often initiate gene regulation and is followed by further histone modification, including methylation and phosphorylation (10,16,17). However, the molecular mechanisms by which these processes of histone modification are controlled remain largely unknown with many histone modification complexes still to be identified. Although it is known that several histone acetyltransferases (HATs) and HAT-containing complexes co-regulate sequence-specific regulators (5, 18 -21), it is unclear whether each sequence-specific regulator requires a cognate HAT complex (or complexes) or can share common HAT complexes with other sequencespecific regulator classes.
It was reported recently that the transactivation function of BRCA1 was squelched by the transactivation of ER␣ (4,22), considered to be a critical regulator of estrogen-dependent breast cancer. ER␣ is a member of the nuclear receptor gene superfamily, acting as a hormone-dependent transcription factor, and is known to require a number of chromatin remodeling and histone modification complexes. As two HAT co-acti-vator complexes have already been reported to support the hormoneinduced transactivation function of hER␣ (19,21), it is possible that BRCA1 and ER␣ share a common co-regulator complex. To address this issue, we searched for a co-activator complex common to both hER␣ and BRCT and biochemically identified a TRRAP/hGCN complex (23,24). Both TRRAP and hGCN5 potently co-activated hER␣ transactivation function, and the addition of hGCN5 rescued the transcriptional squelching observed between BRCA1 and hER␣. Biochemical purification of the TRRAP/hGCN5-containing complexes associated with BRCA1 identified a novel class of TFTC-type HAT complex. Thus, our study suggested that a novel hGCN5/TRRAP HAT complex subclass is required for the multiple functions of BRCA1.
Cell Culture-Human MCF7 breast cancer cells were transformed to stably express hGCN5 as described previously (21) and maintained in Dulbecco's modified Eagle's medium with 10% fetal bovine serum con-taining 500 g/ml G418. The HCC1937 cell line was obtained from the ATCC and maintained in RPMI 1640 medium with 10% fetal bovine serum.
GST Pull-down Assay-GST fusion proteins were expressed in Escherichia coli and bound to glutathione-Sepharose 4B beads (Amersham Biosciences). In vitro translated proteins were prepared by in vitro translation using the T7 promoter of the pcDNA3 vector. Proteins were labeled using [ 35 S]methionine (Amersham Biosciences), and in vitro translation was carried out using the TNT-coupled rabbit reticulocyte lysate system (Promega). The in vitro translated proteins were then incubated with beads for 1 h at 4°C, and free proteins were removed by washing the beads with NET-N ϩ buffer (150 mM NaCl, 1 mM EDTA, 20 mM Tris-HCl (pH 7.5), 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, and 1 mM dithiothreitol). Bound proteins were eluted in boiling sample buffer, separated by SDS-PAGE, and visualized by autoradiography. For all GST pull-down assays, the input lanes show 10% of the protein amount used in the assays.
Chromatin Immunoprecipitation Assay-Preparation of soluble MCF7 chromatin for PCR amplification was performed as described previously (21). Specific primer pairs for the p21 promoter region used were 5Ј-TCCAGCGCACCAACGC-3Ј and 5Ј-AGCTGCTCACACCT-CAG-3Ј. Subconfluent MCF7 cells were infected with adenoviral vectors (Adeno-LacZ, V; Adeno-BRCA1, BRCA1; and Adeno-BRCA1 5382insC, BRCTmt) for 24 h at a multiplicity of infection of 20 to ensure expression in more than 90% of cells. Cells were then treated with 0.01% methylmethane sulfonate (MMS) for 60 min, washed with phosphatebuffered saline, and incubated for 60 min with fresh medium.
Northern Blot Analysis-Total cellular RNA was isolated from the indicated cells using ISOGEN reagent (Wako Co.), and 20 mg RNA were used for Northern blot analysis with digoxigenin-dUTP-labeled cDNA probes. Probes were synthesized using a PCR digoxigenin probe  2-4) was detected using an anti-GST-specific antibody specific for GST-fused BRCT (16). At the right side of the panel, an arrow shows TRRAP as a direct interactant; this was confirmed by Western blotting (data not shown). B, physical interaction between the complex components and BRCT was confirmed by GST pull-down assay. TRRAP binding was determined by SDS-PAGE autoradiography. C, mapping of the BRCT-or ER␣-interacting region of TRRAP by GST pull-down assay using TRRAP fragments and GST-BRCT or GST-ER␣-LBD. BRCT-interacting domains were designated as ID1 and ID2 as shown at the bottom of the panel. TRRAP ID1 interacts with both BRCT and ER␣-LBD. a.a., amino acids.
Cell Survival Assay-HCC1937 cells plated on 60-mm dishes at 40% confluency were transfected with pcDNA expression vectors (0.8 g) and pSh-RNAi vectors (0.2 g) using Effectene TM (Qiagen). Total DNA amounts were adjusted by supplementing up to 1.0 g with empty pcDNA vector or scrambled sequence-inserted RNAi vector. After 24 h, transfected cells were treated with medium containing 0.1% MMS for 50 min, washed with phosphate-buffered saline, and maintained for 8 days in fresh medium. Surviving cells were then counted (12).

Biochemical Purification of hGCN5 and TRRAP as BRCA1
Interactants-To identify the co-regulators responsible for BRCT function, we biochemically purified protein complexes that interacted with BRCT. Recombinant BRCT domains (residues 1563-1863) fused to GST were immobilized to glutathione-Sepharose beads and used as a purification probe (Fig. 1, A and B). A number of BRCT-associating proteins were purified from HeLa S3 nuclear extracts with a 400-kDa protein identified as TRRAP and a 95-kDa protein as hGCN5 by MALDI-TOF MS (Fig. 1C). The HAT hGCN5 and the Myc-interacting protein TRRAP (5) are known to be present in several subtypes of TFTC/STAGA complex in combination with other components (6,7).
Notably hMSH6 and hMSH2 were also identified as BRCT interactants. As hMSH2 and hMSH6 are thought to be major components of the BRCA1 genome surveillance repair complex (BASC) (29,30) along with other DNA repair response proteins, it was possible that BASC was trapped to the BRCT domain along with the hGCN5/TRRAP complex. Although a TFTC-like complex has been purified previously from HeLa cell nuclear extracts and shown to serve as a ligand-dependent co-activator complex for ER␣ in vitro (21), ER␣ was undetectable in the BRCT interactants by our purification procedure (Fig. 1C). This suggested that we had isolated a novel hGCN5/TRRAP complex subclass that interacted with BRCT but not with ER␣. Also as reported previously (31, 32), although TRRAP and hGCN5 were clearly detected by Western blot  2-4). Whole cell lysates were incubated with either free histones (closed bars) or bovine serum albumin (open bars) together with 3 H-labeled acetyl-CoA and assayed for HAT activity by filter binding assay (21,28). Lane 1 shows no cell lysate added. C, co-activation of BRCT by hGCN5 and TRRAP. Transient transfection assays of GAL-BRCT with hGCN5, TRRAP, and MSH2 using a luciferase reporter construct containing the GAL4 DNA binding site (17M8) showed specific enhancement of transcription. Subconfluent proliferating MCF7 cells in 12-well dishes were transfected overnight with the indicated expression vectors and reporter plasmid (17M8-luc) (0.2 g for each lane) using Lipofectamine 2000 TM (Invitrogen). Total amount of DNA in each transfection was adjusted to 1.0 g by supplementing with empty vectors. Luciferase activity is expressed relative to the pM and pcDNA vectors (lane 1). Results represent the average of at least three independent experiments; error bars indicate standard deviation. D, hGCN5/TRRAP is recruited to a p21 WAF1/CIP1 promoter along with BRCA1. MCF7 cells were infected with Adeno-LacZ (V), Adeno-BRCA1 (BRCA1), or Adeno-BRCA1 5382insC (BRCA1mt) and treated with 0.01% MMS for 60 min. Soluble chromatin fractions were applied to chromatin immunoprecipitation analysis. Induction of the p21 gene by DNA damage was confirmed by Northern blot analysis. a.a., amino acids; ChIP, chromatin immunoprecipitation. analysis, TIP60, a component of another TRRAP-containing HAT complex subtype involved in DNA repair, interacted with ER␣ but not with BRCT (Fig. 1C). Association of BRCA1 with TRRAP, hGCN5, and MSH2 in vivo was confirmed by co-immunoprecipitation (Fig. 1D), and RAD50, a BASC component, also co-precipitated with BRCA1 as expected.
TRRAP Was a Direct Interactant of BRCT-We then attempted to identify BRCT interactants in the hGCN5/TRRAP-containing complex by far Western blot analysis. Of the purified interactants bound to GST-BRCT probes (Fig. 1C), we observed significant binding between TRRAP and BRCT ( Fig. 2A). Direct and clear interaction of BRCT with TRRAP but not with hGCN5 was confirmed in vitro using a GST pulldown assay (Fig. 2B). This assay was repeated using TRRAP deletion mutants to map the interacting regions (Fig. 2C) and identified two domains in the N-and C-terminal regions, designated as ID1 and ID2, that appeared to serve as the direct interface for BRCT (Fig. 2C).
A BRCA1 Mutation Found in Breast Cancer Patients Abrogated the Physical Association with TRRAP-The LXXLL motif is well known to serve as a direct interface for liganded nuclear receptors via the C-terminal helix 12, and indeed three LXXLL motifs were found to be responsible for stable association with liganded hER␣ (see Fig. 2C) (21). Interestingly ID1, but not ID2, overlapped with the hER␣-interacting region (see Fig. 2C). Thus, the mode of the BRCT interaction with TRRAP appears to be distinct from that of hER␣. The physiological relevance of the observed interaction between BRCT and the hGCN5/TRRAP complex was then verified using a BRCA1 mutant (5382insC) from the breast cancer cell line HCC1937 that exhibits reduced transactivation function (33). When this mutant BRCA1 was used as the purification probe, hGCN5/TRRAP complex components could not be isolated (Fig.  3A). Indeed a GST pull-down assay showed that the TRRAP ID2 domain exhibited only poor interaction with the 5382insC mutant (Fig.  3B, lower panel). Moreover BRCT mutations found in patients resulted in the complete loss of TRRAP ID1 association (Fig. 3B, upper panel). Far Western blotting also showed no interaction between 5382insC and TRRAP ( Fig. 2A). Therefore, BRCT mutations often found in breast cancer patients may exhibit decreased transactivation function because of reduced association with TRRAP co-activator complexes.
Co-activation of BRCA1 Transactivation Function by hGCN5 through Histone Acetylation-To determine whether the TRRAP/ hGCN5 HAT complex served as a co-activator complex for BRCT, the co-activator functions of the purified interactants were tested using a transient expression assay in MCF7 cells using a chimeric BRCT domain linked to the yeast GAL4 DNA-binding domain. Clear potentiation of BRCT function by hGCN5 overexpression was observed (see Fig. 4C, lane 10), whereas the RNAi-mediated reduction of endogenous hGCN5 levels (see Fig. 4A, lane 3) (25,26) resulted in reduced BRCT transactivation function (Fig. 4C, lane 11). Likewise both TRRAP and MSH2 alone appeared to co-activate BRCT at least to some extent (Fig.  4C). Based on the findings that HAT activity is essential for HAT complex co-activator-mediated transactivation of sequence-specific regulators, we tested an hGCN5 point mutant known to abolish HAT activity through the replacement of a glutamic acid (Glu) with a glutamine (Gln) at position 575 (E575Q) (34). Results of the HAT assay confirmed the loss of HAT activity of this mutant even though the mutant clearly overexpressed (Fig. 4C, compare lanes 3 and 4 with wild-type hGCN5). The E575Q mutation clearly abrogated the hGCN5-mediated co-activator activity on BRCT (Fig. 4C, compare lane 10 with lane 13), which suggested that the GCN5 HAT complex was a critical component of the co-activator complex involved in BRCA1 transactivation function.
To test whether hGCN5 and TRRAP were indeed recruited to BRCA1, which could then act as a sequence-specific activator at target gene promoters, we performed a chromatin immunoprecipitation assay using the p21 WAF1/CIP1 gene promoter, a known BRCA1 target (35). As expected from previous reports, clear recruitment of exogenous BRCA1 to the target sequence (Ϫ266 to Ϫ7 bp) in the p21 promoter was observed in MCF7 cells after BRCA1 was activated by MMS-induced DNA damage (Fig. 4D). Reflecting this BRCT recruitment, hGCN5 and TRRAP were also detected in the promoter along with hyperacetylation of histone H3 presumably because of activity of hGCN5 HAT recruited to the complex (Fig. 4D).
hGCN5 and TRRAP Overexpression Abrogated Transcriptional Squelching of hER␣ by BRCA1-We then examined whether hGCN5 and TRRAP were able to rescue the reported transcriptional squelching between ER␣ and BRCA1. Estrogen-induced transactivation of hER␣ bound to consensus estrogen-responsive elements (EREs) was potentiated by hGCN5 (Fig. 5, lane 2). In contrast, BRCA1 overexpression squelched the ligand-induced transactivation of hER␣ (Fig. 5, compare lane 1 with lane 3) as previously reported. Interestingly the BRCA1 5382insC mutant failed to squelch hER␣ transactivation (Fig. 5, lane 7), suggesting the possibility of common co-activator(s) between hER␣ and BRCA1. This hypothesis was supported by the finding that hGCN5 and TRRAP were both able to abrogate the transcriptional squelching of hER␣ by BRCA1 (Fig, 5, lanes 4 and 5) and that an additive effect was observed when both factors were used together (Fig. 5, lane 7). These findings suggested that hGCN5 HAT and TRRAP are common coactivators for both hER␣ and BRCA1.
Biochemical Identification of a Novel hGCN5/TRRAP HAT Complex-To identify the components of the BRCT-interacting hGCN5/ TRRAP complex and to test whether the hGCN5/TRRAP complex was trapped to BRCA1, we established an MCF7 stable transformant cell line expressing FLAG-tagged hGCN5. Complexes from this cell line were purified by two-step column chromatography followed by glycerol density gradient sedimentation (Fig. 6A) (16). Unlike TFTC-type coactivator complexes, the proteins SAP130, TAF5, and TAF6 appeared to be absent (18,36), whereas TAF10 was present (37,38). Notably hMSH2 and hMSH6 were also identified by MALDI-TOF MS and were further confirmed by Western blot analysis (Fig. 6, B and C). However, other BASC components were not detected, suggesting that the purified hGCN5/TRRAP complex(es) did not associate with BASC such that hMSH2 and hMSH6 formed part of the hGCN5/TRRAP complex.
hGCN5 HAT Activity Was Required for hGCN5-mediated Co-activation of BRCA1 DNA Repair Function-Finally to test whether the BRCA1-associated hGCN5/TRRAP complex played a role in BRCA1mediated DNA repair, we measured cell survival following DNA damage in HCC1937 cells with modulated expression levels of hGCN5/ TRRAP components. Although ectopic BRCA1 expression clearly enhanced cell survival, the BRCA1 mutant 5382insC appeared to lower cell survival (Fig. 7, compare lane 6 with lane 11). Interestingly hGCN5 potentiated the DNA repair activity of BRCA1 (Fig. 7, lane 7), whereas the hGCN5 HAT mutant (hGCN5 E575Q) appeared to act as a dominant negative hGCN5 (see lane 10). Thus, our results suggested that the identified hGCN5/TRRAP complex acted as a co-regulator of BRCA1mediated DNA repair via its HAT activity.

Core Components of the hER␣-interacting TFTC-like Complex Were Identified Along with hMSH2 and hMSH6 as BRCT Domain
Interactants-BRCA1 has been well established as a tumor suppressor of breast and ovarian cancers through DNA repair control and gene regulation. Although several factors and complexes, including BASC, have been identified as being involved in the BRCA1-mediated DNA repair, it remains unclear how these factors/complexes support BRCA1 functions at specific stages of the DNA repair process. Similarly gene regulation by BRCA1 also appears to require several factors/complexes, but the functional significance of these factors/complexes in BRCA1 function is largely unknown. To address these issues, we biochemically purified and identified protein interactants for BRCT, the C-terminal domain thought to support the transactivation function of BRCA1. Our study identified TRRAP, hGCN5, hMSH2, and hMSH6 as BRCT interactants, and their association with BRCA1 was confirmed using both cellular and in vitro approaches. Importantly hGCN5 was found to coactivate BRCA1 function; the HAT activity of hGCN5 apparently was indispensable for BRCT co-activation. In support of these findings, the histone hyperacetylation detected using BRCA1 could be potentiated by hGCN5 overexpression, whereas an hGCN5 mutant lacking HAT activity was unable to co-activate BRCT transactivation function. As hGCN5 is well known to associate with TRRAP to form a transcriptional coactivator complex (6,7), it was likely that hGCN5 co-activated BRCT function with TRRAP as part of a complex. However, none of the subclasses of hGCN5/TRRAP-containing TFTC-like complexes reported so far also contain hMSH2 and hMSH6.
hGCN5 HAT Co-activated BRCA1 Function through Histone Modification-Biochemical purification of BRCA1-interacting complexes from MCF7 cells stably expressing hGCN5 identified a TFTC/  STAGA complex subclass that appeared to support the transactivation function of BRCA1 via the BRCT domain. The identified hGCN5/ TRRAP complex was distinct from other TFTC/STAGA complex subclasses (6,39) as it contained hMSH2 and hMSH6, major functional components of BASC (30). The presence of these DNA repair-related factors in the hGCN5/TRRAP complex suggested that BRCA1-mediated DNA repair required histone modification via the HAT activity of hGCN5 presumably as part of the TRRAP-containing complex. Given the finding that the physical and functional association between BRCA1 and TRRAP could be disrupted by using BRCT mutants observed in breast cancer patients, it is likely that the hGCN5/TRRAP complex identified in our study plays a pivotal role, via its histone acetylation activity, in the diverse functions of BRCA1, including DNA repair and gene regulation.
Although hMSH2 and hMSH6 were present in the biochemically purified hGCN5/TRRAP complex, other BASC components were not. Given that RAD50, a major BASC component (29,30), was co-immunoprecipitated with BRCA1 in agreement with previous reports (12), the hGCN5/TRRAP complex may form part of a larger complex with BASC to fully support BRCA1 functions. As BRCT mutations found in breast cancer patients prevented interaction with the hGCN5/TRRAP complex, it is possible that the hGCN5/TRRAP complex is recruited first to allow histone modification as part of the transcriptional process involving hGCN5-mediated histone acetylation and perhaps allowing the recognition of damaged DNA. BASC may then associate with BRCA1 bound to the modified chromosome. To better understand the molecular basis of BRCA1 function, it would be interesting to investigate the cooperative functions between the two complexes in terms of both transcriptional control and DNA repair involving histone modification.
Multiple Nuclear Complexes Support BRCA1 Functions?-It has become clear that DNA-binding transcriptional activators require a number of nuclear complexes to control transcription and that these complexes associate with activators in a sequential and "hit-and-run" manner. As BRCA1 harbors a transactivation function in its C-terminal domain, it is likely that BRCA1 recruits several transcriptional co-regulator complexes in a sequential but highly regulated manner. However, unlike other DNA-binding factors, BRCA1 is presumably able to recruit a number of factors/complexes involved in the DNA repair process.