Direct Stimulation of Transcription Initiation by BRCA1 Requires Both Its Amino and Carboxyl Termini*

Published experiments suggest that BRCA1 interaction with RNAPII and regulation of a number of target genes may be central to its role as a tumor suppressor. Previous in vivo and in vitro work has implicated the carboxyl terminus of BRCA1 in transcriptional stimulation, but the mechanism of action remains unknown, and whether the full-length protein stimulates transcription is controversial. BRCA1 interacts with a number of enhancer-binding transcriptional activators, suggesting that these factors recruit BRCA1 to promoters, where it stimulates RNA synthesis. To investigate whether BRCA1 has intrinsic transcriptional activity, we established a fully purified transcription assay. We demonstrate here that BRCA1 stimulates transcription initiation across a range of promoters. Both the amino and carboxyl termini of BRCA1 are required for this activity, but the BRCA1-binding partner, BARD1, is not. Our data support a model whereby BRCA1 stabilizes productive preinitiation complexes and thus stimulates transcription.

Of the many functions attributed to BRCA1, 2 one of the first identified was transcriptional stimulation (1,2). BRCA1 copurifies with the RNA polymerase II (RNAPII) holoenzyme (3,4), and reporter assays and microarray studies show that it regulates the expression of a range of p53-dependent and -independent targets (5,6). Thus, one way in which BRCA1 may serve as a tumor suppressor is through up-regulation of growth-suppressive targets (7,8). While the mechanism of stimulation is unknown, the transcriptional activity of BRCA1 most likely depends in part on its reported interactions with a wide range of transcriptional activators. However, in a defined system assayed in vitro, a Gal4 fusion to the carboxyl terminus of BRCA1 activates transcription, independent of other activators (9), suggesting an intrinsic transcriptional activity for BRCA1. A subsequent study found that Gal4 fusions to full-length BRCA1 could not activate transcription in transfected cells and that the degree of transcriptional activation conferred by Gal4 fusions to the carboxyl terminus of bovine BRCA1 was much lower than human BRCA1 (10). Since the human carboxyl terminus is more acidic than the bovine version, the transcriptional activity may simply be a function of its acidity. Regardless, in vivo reporter assays using BRCA1 without a Gal4 fusion indicate that transcriptional stimulation by BRCA1 is dependent on its carboxyl terminus (6,11). To better understand whether BRCA1 might directly regulate transcription, we developed an assay to test the function of full-length human BRCA1 in transcription, independent of an artificial DNA-binding domain protein fusion. We demonstrate here that BRCA1 stimulates basal transcription by promoting initiation of RNA synthesis. This is the first demonstration of direct transcriptional activity by full-length BRCA1.

MATERIALS AND METHODS
Transcription Factors-The transcription factors used in these assays were purified using established techniques (9,12,13). BRCA1/BARD1, BRCA1, and the truncation mutants were purified from baculovirus infected insect cells as described previously (14,15). p53 was also purified from baculovirus infected insect cells (16).
Plasmid Templates-G-less cassette templates were based upon the p(C2AT) 19 vector (17) and have been described previously (18).

RESULTS AND DISCUSSION
Based on the prior evidence that BRCA1 is a coactivator of p53 transcriptional targets (5, 6), we first attempted to reconstitute coactivation by purified full-length BRCA1/BARD1 and p53 in vitro. We reasoned that in the absence of a Gal4 fusion, sequence specific p53 binding might serve to localize BRCA1/ BARD1 to the promoter region. Transcription reactions were performed with purified TFIID, TFIIB, TFIIA, RNAPII, TFIIE, TFIIF, TFIIH, and PC4. To detect transcriptional activation, a modified adenoviral E4 promoter with upstream p53 response elements (p53 G5E4) linked to a 384-base pair G-less cassette was used (16). As an internal control template for basal transcription, the adenoviral major late promoter (⌬ML) linked to a 210-base pair G-less cassette was used. Transcription from both templates was low in the absence of BRCA1/BARD1 and p53 (Fig. 1A, lane 1). To our surprise, addition of BRCA1/ BARD1 alone stimulated transcription from both templates (lane 2). Addition of p53 specifically activated transcription of the p53 G5E4 template (lane 3). Addition of both p53 and BRCA1/BARD1 resulted in the highest ratio of activated/basal transcription, demonstrating that a modest amount of coactivation can occur with these purified factors (lane 4). We were intrigued that BRCA1/BARD1 could stimulate transcription in the absence of p53 or a Gal4 fusion. In the following experiments we characterized the mechanism by which BRCA1 directly stimulates basal transcription.
In addition to leaving out p53, we found that by omitting PC4, a factor required for activated transcription (20), the level of RNA synthesis was significantly higher and the stimulatory effect on transcription by BRCA1 was apparent (Fig. 1B). We tested several promoters for effects by BRCA1/BARD1 on RNA synthesis. All of these templates were identical with the exception of the 50 base pairs of sequence in the core promoter immediately upstream of the G-less cassette sequence. The magnitude of the stimulation of RNA synthesis by BRCA1/BARD1 differed among templates, indicating that the effect of BRCA1/BARD1 varied dependent on core promoter sequences (Fig. 1B). Stimulation was highest (ϳ10-fold) for the p53 G5E4 promoter template (lanes 5 and 6), and we chose that template for subsequent experiments. The fact that BRCA1/BARD1 stimulated transcription from the IgG template, which does not require TFIIE/TFIIH, indicated that these factors were not required for transcriptional stimulation. Indeed, removal of TFIIE and TFIIH from the reaction and substitution of TBP for TFIID did not affect the stimulation of RNA synthesis by BRCA1/BARD1 (Fig. 1C).
One trivial explanation for these results would be if the BRCA1/BARD1 preparation used in our assay contained a contaminating general transcription factor that was limiting in the assay. The BRCA1/BARD1 protein was purified from insect cells and judged free of major contaminants by silver stained protein gels (15). However, to rule out this possibility, we tested whether the BRCA1/BARD1 preparation could complement transcription reactions lacking a single factor (Fig. 1D). Transcriptions were conducted using a linearized ⌬ML template that requires TBP, TFIIB, RNAPII, TFIIF, TFIIE, and TFIIH. BRCA1/BARD1 was present in all reactions at a 9 nM concentration. Transcription was observed only when all factors were present, and thus we exclude the possibility that the BRCA1/BARD1 preparation contained a general transcription factor.
Having established that BRCA1/BARD1 stimulated basal transcription in a minimal RNAPII transcription system, we next asked what stage of transcription BRCA1/BARD1 enhanced. We used a pulse/chase strategy to separate transcriptional initiation from elongation ( Fig. 2A). In the pulse phase, only ATP and [␣-32 P]CTP were added to the reaction mixture. The lack of UTP prevented elongation from occurring beyond four nucleotides, resulting in a stalled RNAPII complex. In the chase phase, a complete, unlabeled nucleotide mixture was added with excess CTP, allowing elongation of the labeled nascent transcripts. Any new initiations that occurred during the chase phase were unlabeled and thus not detected. Regardless of whether TFIID or TBP was used for TATA binding activity, inclusion of BRCA1/BARD1 during the pulse stimulated transcription, while addition during the chase had no effect (Fig. 2B). These results indicated that BRCA1/BARD1 stimulate basal transcription by promoting initiation. However, it was also possible that BRCA1/BARD1 load during the initiation phase but then promote transcriptional elongation. To determine whether this might be true, we examined transcription from very short templates (40 -50 nucleotides), reasoning that the importance of an elongation factor over such a short template would be greatly reduced. A similar level of stimulation of RNA synthesis was observed for these minitemplates (ϳ10-fold) as was seen for the ϳ400-base pair templates, thus supporting the idea that BRCA1/BARD1 promote the initiation of transcription (Fig. 2C).
Both BRCA1 and BARD1 copurify with the RNAPII holoenzyme (21), and thus we used the heterodimer in experiments to this point. The major functional outcome of the BRCA1/BARD1 interaction is to potentiate the E3 ubiq-uitin ligase activity of BRCA1 (22). We had no reason to believe this enzymatic function had a role in transcriptional stimulation because E1 and E2 enzymes and ubiquitin were omitted from the reactions. Therefore, we tested whether BARD1 was required for transcriptional stimulation by BRCA1. When comparing BRCA1/BARD1 to BRCA1 alone, we observed similar levels of stimulation of RNA synthesis, evident in each case at concentrations as low as 1 nM (Fig. 3A). We conclude that BARD1 is not required for transcriptional stimulation by BRCA1. Next we examined truncations of BRCA1 to determine what portion of the protein contains the stimulatory activity. Deletion of either the 300 amino-terminal residues or the 336 carboxyl-terminal residues of BRCA1 abolished stimulation of transcription (Fig. 3B). Both the amino and carboxyl termini of BRCA1 are known to interact with RNAPII (21), and these truncations may reduce association with RNAPII in our assay. In addition, previous reports localize transcriptional activity to the carboxyl terminus of BRCA1 (2,9). Since truncation of either terminus did not support transcriptional stimulation, we tested an additional two internal deletions spanning most of the intervening sequence (Fig. 3C). Both BRCA1-(⌬303-770)/BARD1 and BRCA1-(⌬770 -1290)/BARD1 stimulated transcription as well as or better than BRCA1/BARD1. At the highest concentration tested (9 nM), the BRCA1-(⌬770 -1290)/BARD1 actually repressed transcription, possibly reflecting a transcriptional squelching effect. In summary, the amino and carboxyl termini, but not internal domains of BRCA1, are required for transcriptional stimulation (Fig. 3D).
Our data to this point suggested that BRCA1 might be promoting formation of the initiation complex through contacts mediated by its amino and carboxyl termini. To determine which transcription factors might be affected by these contacts, we attempted to titrate factors downward in concentration, reasoning that the stimulatory activity should be enhanced by limiting conditions for the relevant factors. To our surprise, downward titration of TFIIA resulted in higher levels of basal transcription and a reduction in the stimulatory effect of BRCA1 (Fig. 4A). Without TFIIA (lanes 1 and 2), we observed a negligible stimulatory effect of BRCA1, but the inhibitory activity of TFIIA on basal transcription was relieved by addition of BRCA1. TFIIA is known to act as an anti-repressor for TBP-binding inhibitors and is a required factor in activated transcription systems utilizing TFIID (23, 24), so repression was unexpected.  3 and 4) and BRCA1/BARD1 (lanes 2 and 4) were assayed. The ratio of stimulated/basal transcription was determined by PhosphorImager analysis of the accumulated RNA from the p53 G5E4 template relative to the ⌬ML200 template. B, stimulation of transcription by BRCA1/BARD1 was tested in reactions containing TFIID, TFIIA, TFIIB, RNAPII, TFIIF, TFIIE, and TFIIH using a variety of single ϳ400-base pair G-less cassette templates. Full-length transcript is noted by an arrow. The following promoters linked to G-less cassette templates were used: ⌬ML, the core adenoviral major late promoter (lanes 1 and 2); IgG, the immunoglobulin heavy chain promoter (lanes 3 and 4); p53 G5E4, the adenoviral E4 promoter with a 20-bp p53 response element upstream of the TATA box (lanes 5 and 6); G5E4, the adenoviral E4 promoter without p53 response elements (lanes 7 and 8); HIV, the human immunodeficiency virus promoter (lanes 9 and 10); DH, the Drosophila heat shock promoter (lanes 11 and 12). However, this was not the first observation of basal repressive action by TFIIA. Prior to the cloning and recombinant expression of TFIIA, researchers reported on a repressive activity that purified closely with TFIIA (25). This activity repressed basal transcription from consensus TATA box promoters but stimulated transcription from non-consensus promoters. The authors (25) suggested a model whereby TFIIA interacts with TBP, altering its conformation and association with the promoter. If this conformational change altered the preference of TBP for the TATA box, then it could interfere with formation of the preinitiation complex on the correct DNA site and repress transcription (25).
Based on these previous findings, we speculated that BRCA1 might prevent improper TBP localization, either by disrupting non-TATA bound TBP or by stabilizing complex formation on bona fide TATA boxes. Precedent for regulation of TBP binding exists in the ATPase Mot1, which can dissociate TBP from DNA. Initial in vitro work cast Mot1 as a transcriptional inhibitor (26,27), but examination in vivo also demonstrated activation of several targets (28 -30). Subsequent in vitro work using lower concentrations of Mot1 recapitulated transcriptional stimulation, especially under conditions where excess non-promoter DNA was present (31). The authors (31) concluded that Mot1 acts by promoting dissociation of TBP from non-TATA DNA sequences and thereby raising the effective TBP concentration.
The plasmid templates used in our experiments have ϳ3000 base pairs of sequence, of which about 50 base pairs serve as promoter. Many suboptimal TATA boxes exist in the extraneous DNA, and we infer that TFIIA stabilizes TBP on these non-promoter sites, thus reducing the effective concentration of TBP. Our results show that BRCA1 counters TFIIA repression, and our results are consistent with this rescue occurring during preinitiation or initiation. To test whether BRCA1 could stimulate basal transcription in the absence of TFIIA, but under conditions that were unfavorable for initiation, we limited the general transcription factors involved in nucleation of the preinitiation complex, TBP and TFIIB. By limiting TBP 10-fold (from 16 to 1.6 nM), a modest stimulation of transcription by BRCA1 was revealed (Fig. 4B, lanes 1 and 2, compare with Fig. 4A, lanes 1 and 2). Under conditions where both TBP and TFIIB are limiting, the effect of BRCA1 was further enhanced (Fig. 4B,  lanes 3 and 4). This result demonstrates that the stimulatory activity of BRCA1  1-3) or TBP (lanes 4 -6) were assembled as described above. BRCA1/BARD1 (10 nM) was omitted (lanes 1 and 4), added during the pulse (lanes 2 and 5), or added during the chase (lanes 3 and 6). C, transcription reactions containing TFIID, TFIIA, TFIIB, RNAPII, TFIIF, TFIIE, and TFIIH and mini-templates (ϳ50 base pairs) were performed with (even lanes) or without (odd lanes) BRCA1/BARD1 (10 nM). The promoters correspond to those used with the ϳ400-base pair templates described in the legend to Fig. 1B.  4 and 7). B, transcriptional stimulation by BRCA1 and the truncations BRCA1-(300 -1863) and BRCA1-(1-1527) was tested as in A. C, transcriptional stimulation by BRCA1/BARD1 and internal deletions BRCA1-(⌬303-770)/BARD1 and BRCA1-(⌬770 -1290)/BARD1 was tested as in A. D, summary of transcriptional stimulation by BRCA1 variants. Amino-terminal domain (gray) denotes RING finger domains in BRCA1 and BARD1, and the carboxyl-terminal domain (black) corresponds to BRCT repeats in BRCA1.
is not limited to reversal of TFIIA basal repression but applies more generally to situations under which preinitiation complex assembly is a limiting step.
The challenges to proper initiation in our transcription assay likely underestimate the difficulties in vivo, where correct promoters must be discriminated from total genomic DNA. To test whether the presence of excess plasmid DNA could inhibit transcription, we titrated a competitor plasmid lacking promoter sequences into transcription reactions that were conducted in the presence or absence of BRCA1/BARD1 (Fig. 4C). With addition of 300 ng or more of competitor DNA, transcription levels were reduced, confirming that excess DNA can inhibit transcription (compare lanes 1 and 2 with lanes 5-8).
The most likely explanation for this effect was that the competitor DNA titrated initiation factors away from the bona fide TATA box. Although transcription levels were lower overall, we observed an increasing degree of transcriptional stimulation by BRCA1/BARD1 with increasing competitor plasmid. Without competitor DNA, the addition of BRCA1/BARD1 stimulated transcription only 1.2-fold (lanes 1 and 2). At the highest level of competitor plasmid tested (900 ng), RNA synthesis was stimulated by BRCA1/BARD1 over 4-fold (lanes 7 and 8). Therefore, the presence of excess competitor DNA inhibits transcription but increases the potential for stimulation by BRCA1.
We find that limiting the initiation factors TFIIB and TBP, either directly or by addition of excess competitor DNA, increases the stimulatory effect of BRCA1. This outcome could be explained by BRCA1 stabilization of productive initiation complexes or conversely by destabilization of non-productive complexes. Based on the known interaction between BRCA1 and RNAPII, the former possibility is, in our opinion, more likely. Taken together, our data support a model where BRCA1 stabilizes productive transcription initiation complexes, and this may be one mechanism by which it coactivates the transcription of gene targets. Stimulation by BRCA1 was observed in our assays with purified components and a range of promoters at concentrations as low as 1 nM BRCA1. However, in the cell, where BRCA1 concentration is likely even lower, it could be recruited to specific promoters by enhancer-binding factors. Once bound to a specific promoter, BRCA1 could stimulate assembly of the preinitiation complex through its interactions with RNAPII and perhaps other general transcription factors.