Redistribution of BRCA1 among Four Different Protein Complexes following Replication Blockage*

The BRCA1 protein is known to participate in multiple cellular processes. In these experiments, we resolved four distinct BRCA1-containing complexes. We found BRCA1 associated with the RNA polymerase II holoenzyme (holo-pol), a large mass complex called the fraction 5 complex, the Rad50-Mre11-Nbs1 complex, and a complex that has not been described previously. We observed this new complex after treating cells with hydroxyurea, suggesting that the hydroxyurea-induced complex (HUIC) is involved with the response to DNA replication blockage. After hydroxyurea treatment of cells, BRCA1 content decreased in the holo-pol and the fraction 5 complex, and BRCA1 was redistributed to the HUIC. The HUIC was shown not to contain a number of holo-pol components or the Rad50-Mre11-Nbs1 complex but was associated with the BRCA1-associated RING domain protein BARD1. These data suggest that BRCA1 participates in multiple cellular processes by multiple protein complexes and that the BRCA1 content of these complexes is dynamically altered after DNA replication blockage.

Mutations in the BRCA1 tumor suppressor gene are associated with about 4% of all breast cancers and about 50% of all familial cases (1,2). Emerging data indicate that BRCA1 is likely to serve as an important central component in multiple biological pathways that regulate transcription, repair of DNA damage, the cell cycle, polyadenylation of mRNAs, and chromatin remodeling (3)(4)(5)(6)(7). It is not clear whether all of these processes are due to one biochemical mechanism or to multiple mechanisms with BRCA1 functioning in multiple protein complexes.
Several BRCA1-containing complexes have been purified using different methods. BRCA1 involvement in transcription is indicated by its association with the RNA polymerase II holoenzyme (holo-pol) and by activation of transcription by BRCA1 in cell free reactions (12,(22)(23)(24). BRCA1 is associated with the chromatin-remodeling SWI-SNF (mating type switch/ sucrose non-fermenters) complex, either in association with holo-pol (23) or independent of Pol II (25). BRCA1 association with Rad50-Mre11-Nbs1 may contribute to repair of DNA damage. The BRCA1-associated genome surveillance complex contains various proteins for DNA repair, including the Rad50 complex, cell cycle check point, and DNA replication factors (26). Because BRCA1-associated genome surveillance complex is derived from a single-step immunoprecipitation (IP) from unpurified nuclear extracts, it is unclear whether it represents multiple complexes or a single complex.
BRCA1 protein dynamically changes its subcellular localization, depending on the cell cycle or whether the genome has been damaged. In S phase, BRCA1 localizes to discrete nuclear foci (27), but treatment with hydroxyurea (HU), UV irradiation, or ␥-irradiation leads to dispersal of these BRCA1 foci (28). After HU and UV treatment, BRCA1 colocalizes with BARD1 and RAD51 in proliferating cell nuclear antigen-containing replication structures (29). After HU treatment or irradiation, BRCA1 forms a complex with Rad50, Mre11, and Nbs1 in discrete nuclear foci (irradiation-induced foci) (19,20). It is unknown whether these changes in subcellular position reflect changes in BRCA1 protein complexes.
In this study, we observed BRCA1 associated with three protein complexes in asynchronously cycling cells, and BRCA1 shifted to a fourth protein complex after cells were treated with HU. These data support a concept that the multiple processes with which BRCA1 is involved reflect multiple protein complexes with which it associates.

MATERIALS AND METHODS
Cell Culture and Biochemical Purification-HeLa-S3 cells and 293S cells were passaged in suspension culture using standard procedures. About 5 ϫ 10 9 cells were infected with recombinant adenovirus at a multiplicity of infection of about 1-2 plaque-forming units/cell, and cells were harvested 44 h after infection. The purification from whole-cell extracts by chromatography on a Biorex70 ion exchange matrix and sucrose gradient sedimentation have been described previously (22,23).
Adenovirus Construction-HA epitope-tagged full-length and deleted mutant BRCA1 were inserted into AdEasy (Quantum Biotechnology, Inc.) shuttle vectors such that the BRCA1 gene would be under the control of the cytomegalovirus promoter. Full-length HA epitope-tagged BRCA1 (HA epitope tagged to the amino terminus, HA-BRCA1; HA epitope tagged to the carboxyl terminus, BRCA1-HA) was subcloned from constructs in the pcDNA3 vector (27). The 775-1292 deletion was constructed by digestion of KpnI and NheI from HA-BRCA1 followed by ligated insertion of the following forward and reverse linkers: forward, 5Ј-CTGGTGGACCAAAGAAGAAGCGTAAGACCGGTG-3Ј; and reverse, 5Ј-CTAGCACCGGTCTTACGCTTCTTCTTTGGTCCACCAGGT-AC-3Ј. This fragment was inserted into the AdEasy shuttle vector. With each shuttle vector construct, recombination occurred in bacteria to recover adenoviral genomic DNA with the BRCA1 gene, and virus was recovered after transfection into 293A cells.
Immunoprecipitation-400 -500 l of protein from sucrose gradient fraction or 150 -180 l of eluted protein from the Biorex70 column was immunoprecipitated with the specific antibody for the HA epitope (12CA5), Med17, or Mre11. 450 -750 l of binding reactions was incubated with rotation for 4 h at 4°C in buffer H (20 mM Tris-OAc, pH 7.9, 1 mM EDTA, and 5% glycerol), 0.12 M KOAc, 0.1% Nonidet P-40, 0.1 mM dithiothreitol, 0.2 mg/ml bovine serum albumin, and 0.5 mM phenylmethylsulfonyl fluoride in the presence of protein extract, 3-5 l of antibody, and 20 l of protein A beads. When affinity-purified anti-Med17 antibody (30,31) was used, these steps were performed with or without antigenic peptide (0.1 mg/ml). With all IPs, supernatant was removed, and protein beads were then washed three times using 450 -800 ml of wash buffer (120 or 300 mM KOAc, 20 mM Tris-OAc, pH 7.9, 0.1% Nonidet P-40, 0.1 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride). For Western blot analysis, samples were subjected to electrophoresis in 5% or 6% SDS-polyacrylamide gels and immunoblotted using the indicated antibodies.

Identification of the Hydroxyurea-induced Complex (HUIC)-We
have previously identified two separate protein complexes containing BRCA1 in HeLa cell extracts (22,23). One complex was identified as the holo-pol, and the second complex was a more massive complex, which may have contained a nucleic acid component (23). A third BRCA1-containing complex has been described that contains Rad50-Mre11-Nbs1 along with BRCA1 (20) or contains other repair factors (26). Data have suggested that BRCA1 shifts its subnuclear localization and perhaps associates with new protein complexes after HU-mediated replication block or exposure to DNA-damaging agents in the S phase of the cell cycle (26,28). To induce a replication block in cells in large-scale culture, we treated HeLa-S3 cells with 10 mM HU for 16 h. We reasoned that, regardless of the position of a given cell in the cell cycle, this incubation time was sufficient to drive all cells into S phase, whereupon DNA replication is blocked due to depletion of deoxynucleotide triphosphates. Extracts were then prepared by standard procedures (23). When comparing asynchronous cell extracts with HU-treated extracts, we observed an increased abundance of cyclin A relative to cyclin E, suggesting that the HU-treated extract was indeed enriched for S-phase proteins (Fig. 1A). Extracts were subjected to ion exchange chromatography using Biorex70 matrix, and comparison to our previously published observations from untreated extracts (23) revealed that BRCA1 and Pol II fractionation on this matrix were unchanged by HU treatment. Most of the BRCA1 eluted in the 0.6 M KOAc step elution, and a smaller amount of BRCA1 was observed in the 0.3 M KOAc step elution. Most of Pol II was observed in the 0.6 M KOAc elution (Fig. 1B).
We next fractionated BRCA1 using sucrose gradient sedimentation. Whereas Pol II was observed to sediment at about 30 S in the fractions consistent with the holo-pol in both extracts (ϩ and ϪHU), BRCA1 was shifted in its sedimentation pattern. In untreated cell extracts, BRCA1 is observed in two pools of protein, one in fractions 1-3 at the bottom of the gradient (Ͼ60 S), and one pool in fractions 11-17 consistent with the holo-pol (Fig. 1C, top). The HU-treated extract resulted in an observed shift in BRCA1 from these two complexes to a third complex with lower mass that we call the HUIC (Fig.  1C). The BRCA1 content in fractions 1-3 was markedly decreased, and the BRCA1 content in fractions 11-15 was also diminished, but a new peak of BRCA1 protein was observed in fractions 17-21. The holoenzyme-specific marker cycC (32) eluted primarily in fractions 9 -15, regardless of the exposure of the cells to HU, consistent with those fractions containing the holo-pol (Fig. 1C). It is thus inferred that this third peak seen only after the addition of HU is not associated with the holopol. Whereas the observations with the untreated extracts are consistent with previously published observations (22,23), the HUIC has not yet been described.
To further characterize the HUIC as well as other BRCA1containing protein complexes, we established a system for overexpressing epitope-tagged BRCA1 in preparative extracts.
Overexpressed BRCA1 Copurifies with holo-pol-Recombinant adenoviruses, which express full-length BRCA1 with an amino-terminal HA tag, were used to infect ϳ5 ϫ 10 9 293S cells in suspension culture. Whole-cell extracts were prepared by standard procedures (23)  C, identification of HUIC by sucrose gradient sedimentation. The protein peak eluted from Biorex70 at 0.6 M KOAc was subjected to sucrose gradient sedimentation. Immunoblots of these fractions were stained for BRCA1, Pol II large subunit, and cycC. In fractions 17-21 of the HU-treated sample, a new peak was observed. Fractions on the left (low numbers) represent rapidly sedimenting complexes, and fractions on the right (high numbers) represent low molecular mass complexes. Approximate sedimentation coefficients were determined by analyzing eukaryotic and bacterial ribosomes on similar sucrose gradients.
washes of increasing concentrations of potassium acetate ( Fig.  2A). Immunoblotting of eluted proteins for HA-BRCA1 and Pol II revealed that overexpressed BRCA1 cofractionated in the 0.6 M KOAc elution in a manner similar to that observed with the holo-pol from HeLa cells ( Fig. 2B; Ref. 22). This 0.6 M KOAc peak fraction was then subjected to sucrose gradient sedimentation (Fig. 2C). Immunoblotting for HA reveals that HA-BRCA1 has two peaks of protein concentration, in fraction 5 and fractions 13-17. By contrast, Pol II has a broad peak with highest concentrations that correspond to the peak of HA-BRCA1 in fractions 13-17 (Fig. 2C), suggesting that the peak of BRCA1 in fraction 5 is a distinct pool of the protein and is not the holo-pol. Overexpression of BRCA1 did not result in significant aberrant pools of BRCA1 because this pattern of fractionation of BRCA1 is almost identical to that of the endogenous BRCA1 in HeLa cells ( Fig. 1; Ref. 23). The pattern of the overexpressed BRCA1 does result in sharp peaks on the sucrose gradient, and this will be shown to be due in part to the absence of the HUIC when the full-length protein is overexpressed (see below). In addition, overexpression of a similar BRCA1 protein that is HA-tagged on the carboxyl terminus resulted in an identical purification over these two steps (data not shown).
To confirm that overexpressed HA-BRCA1 interacts with Pol II, sucrose gradient-purified holo-pol (fraction 14) was epitopepurified using the HA epitope-specific antibody and compared with the equivalent fraction from uninfected 293S cells (Fig.  2D). Pol II was purified in the sample with the overexpressed HA-BRCA1, but only a background level of Pol II was detected in the IP from the uninfected cells. These results showed that overexpressed epitope-tagged BRCA1 is associated with holopol, as is the endogenous BRCA1.
Characterization of the HUIC and Association with BARD1-We applied the overexpression of tagged BRCA1 to the analysis of the HUIC. After infecting 293S cells with fulllength BRCA1, we treated the cells with 10 mM HU for 16 h before extraction. Chromatography on Biorex70 was un-changed (data not shown), and analysis of the 0.6 M KOAc peak on sucrose gradient revealed that there was no shift of the overexpressed HA-BRCA1 into the HUIC, in contrast to the shift observed with endogenous BRCA1 in HeLa cells (compare Fig. 3B with Fig. 2C). When we analyzed uninfected 293S cells for endogenous BRCA1 and the effect of HU treatment on these cells, it was apparent that without HU treatment, BRCA1 sedimented in three peaks. The BRCA1-containing fractions were consistent with the fraction 5 complex and the holo-pol having the highest BRCA1 content, but BRCA1 was also detected at low levels in fractions consistent with the HUIC (Fig.  3B). This analysis was complicated by a nonspecific band migrating slightly faster than the BRCA1 (indicated by an asterisk) in fractions 17-23. Treatment with HU decreased the BRCA1 content of the first peak and the second peak and increased the BRCA1 content in the HUIC. This result was similar to the shift of BRCA1 protein into the HUIC observed in HU-treated HeLa cells. It is unclear why overexpression of full-length BRCA1 should interfere with the formation of the HUIC when the less expressed endogenous protein apparently does associate with the HUIC. Overexpressing BRCA1 causes G 1 arrest and growth suppression (data not shown; Refs. [33][34][35]. Because the HUIC was observed in a replication blockade that occurs in S phase, it is presumed that overexpression of full-length BRCA1 might be incompatible with BRCA1 association with the HUIC. Because it is reported that several deletion mutants do not induce G 1 arrest and growth suppression (34,35), we constructed several deletion mutants of HA-BRCA1. One of the deletion mutants, HA-BRCA1(⌬775-1292) (Fig. 3A), was inserted into recombinant adenovirus and used to infect 293S cells. HA-BRCA1(⌬775-1292) did not suppress the growth of cells (data not shown). Infected whole-cell extracts were chromatographed on Biorex70 matrix and analyzed by immunoblotting using an antibody specific for the HA tag. HA-BRCA1(⌬775-1292) eluted primarily in the 0.6 M KOAc peak, consistent with the wild type BRCA1 (data not shown). When we subjected the 0.6 M KOAc fraction of HA-BRCA1(⌬775-1292) to sucrose gradient sedimentation, three peaks were observed in the same positions on the gradient as observed with endogenous BRCA1 (Fig. 3B). HU treatment decreased the BRCA1 content present in the fraction 5 and holo-pol peaks and increased the BRCA1 content in fractions consistent with the HUIC (Fig. 3B). This pattern of redistribution of BRCA1 content after HU treatment from complexes of higher mass to the HUIC, which was originally observed in HeLa cells, was repeated in 293S cells.
We analyzed these fractions for BARD1 and for Rad50 because these proteins are known to interact with BRCA1. We found that the sedimentation of BARD1 did not change after HU treatment. BARD1 was present throughout the gradient, but with peaks consistent with the fraction 5 and the HUIC (Fig. 3B). Rad50 sedimented in fractions at the top of the gradient regardless of HU treatment (Fig. 3B), suggesting that either BARD1 or Rad50 may be associated with BRCA1 in the HUIC. By comparison, Pol II did not cosediment with the HUIC (Fig. 1; data not shown).
We tested whether the HUIC was a derivative of the holo-pol by immunoprecipitation using the holoenzyme-specific affinitypurified antibody directed against Med17 (30) (Fig. 4A). Fractions 5, 17, and 25 containing HA-BRCA1(⌬775-1292) (from Fig.  3B) were immunoprecipitated with the Med17-specific antibody, and the blot was stained for the HA epitope (Fig. 4A). These fractions were chosen because they represent the peaks of the fraction 5, holo-pol, and HUIC complexes. When analyzing fraction 17, which had sedimentation consistent with the holo-pol, significant purification of HA-BRCA1 by the Med17-specific antibody was observed (lanes 4 -6), indicating association of the HA-BRCA1(⌬775-1292) with the holo-pol. By contrast, the HA-BRCA1(⌬775-1292) in fraction 25, the HUIC-containing fraction, was not at all associated with Med17 (lanes 7-9). The very faint band in lane 9 probably resulted from contamination of the HUIC by the holo-pol, which sedimented in adjacent fractions. This result suggests that the HUIC is a distinct complex from the holo-pol. The results of the IP from fraction 5 were weakly positive (lanes 1-3), suggesting that the protein complex in this sample may be derived from the holo-pol. BRCA1 has been reported to interact with the Rad50-Mre11-Nbs1 repair complex and also with BARD1 (8,20), and these proteins all copurify with the HUIC (Fig. 3B). We tested whether the HUIC is the same as the BRCA1-Rad50-Mre11-Nbs1 complex by immunopurification using Mre11-specific antiserum. Whereas the Mre11 antibody could purify Mre11 and Rad50, HA-BRCA1(⌬775-1292) and BARD1 were not detected (Fig. 4B). Thus, the HUIC is not the same as the Rad50containing complex. By contrast, we have identified a BRCA1associated protein in the HUIC. Anti-HA epitope immunopurification from the HUIC-containing samples revealed that BARD1 was present in this complex (Fig. 4C).
BRCA1-Rad50-Mre11-Nbs1 Complex-It has been reported that the Rad50-Mre11-Nbs1 complex associates with BRCA1 (20), but we did not observe this complex in HUIC fractions even though the proteins cosediment. A small pool of BRCA1 was observed in the 0.3 M KOAc elution of the Biorex70 column (Fig. 2B), and most of the total Rad50 in the extract was in the 0.3 M KOAc elution (data not shown). The 0.3 M KOAc Biorex70 eluate was analyzed by sucrose gradient sedimentation (Fig.  5A). Samples expressing full-length HA-BRCA1 were subjected to Western blot analysis and stained for the HA epitope, Rad50, Mre11, and Nbs1. These patterns are similar to Rad50, Mre11, and Nbs1 sedimentation in fractions 15-23. HA-BRCA1 was present in these samples but also shifted to a higher molecular mass and fractionated in fractions 13-23. This shift in BRCA1 content was likely due to contamination by holo-pol (data not shown). Next, we tested whether BRCA1 is associated with Rad50 in this 0.3 M KOAc chromatographic fraction using anti-Mre11 antibody. In this fraction, full-length HA-BRCA1 was associated with Mre11, as was Rad50 (Fig. 5B). Finally, we tested whether HU, which causes damaged DNA, would stimulate association of endogenous BRCA1 with the Rad50-Mre11-  4 -6). Immunoblots were stained using antibody specific for BRCA1, Rad50, Mre11, and Nbs1. Nbs1 complex. Although it has been reported that HU treatment causes a colocalization of BRCA1 and Rad50 complex (26), HU treatment did not stimulate BRCA1 association with the Rad50 complex (Fig. 5C).

DISCUSSION
Here we report the chromatographic separation of four BRCA1-containing complexes. Upon Biorex70 chromatography, the BRCA1-Rad50 complex elutes in the 0.3 M KOAc fraction and is thus readily resolved from the other three complexes. By sucrose gradient sedimentation analysis of the Bi-orex70 0.6 M KOAc fraction, we purified the fraction 5 complex, holo-pol complex, and the HUIC (summarized in Fig. 6).
We identified the HUIC, and we demonstrated that it contains BARD1. The subnuclear localization of BRCA1 and BARD1 changes in the S phase after treatment with HU or UV to proliferating cell nuclear antigen-containing replication structures (28,29). Kleiman and Manley (5) found that BRCA1/ BARD1/CstF represses nuclear mRNA polyadenylation in vitro and that mRNA 3Ј processing is transiently inhibited in cells treated with HU or UV. CstF50 and proliferating cell nuclear antigen interact strongly in a two-hybrid screen. We suggest that the HUIC protein complex identified in this study likely functions as the protein complex responsible for the activity that represses polyadenylation of mRNAs (4).
The fraction 5 complex is massive, with a sedimentation coefficient of 60 S or greater. Previous experiments showed that this complex is associated with RNA (23), and in this study, we show that it does contain the Med17 holo-pol component. It is possible that the fraction 5 complex results from artifactual association with ribosomes or hnRNP (heterogeneous nuclear ribonucleoprotein) particles, both of which sediment in this portion of the gradient. Further characterization of the fraction 5 complex will clarify this. It is shown in this study that the BRCA1 content in the fraction 5 is regulated because HU treatment results in a profound decrease in BRCA1 protein in this fraction. For this reason, we suggest that BRCA1 association with the fraction 5 complex is likely to be biologically meaningful.
The association of BRCA1 with the Rad50-Mre11-Nbs1 complex was not stimulated by HU. We anticipated that this association would be stimulated by HU because HU causes DNA replication arrest with DNA gaps that would likely induce a damage response. Our result with no increase in the association between BRCA1 and the Rad50 repair complex after HU treatment is consistent with similar results from another laboratory after ␥-irradiation or methyl methanesulfonate treatment (20). If anything, HU treatment resulted in a decrease in BRCA1 content in the Rad50 complex, suggesting that BRCA1 is also dynamically shifting from this complex to the HUIC. Current data have not yet determined whether the BRCA1-Rad50-Mre11-Nbs1 complex we have identified is the same as the BRCA1-associated genome surveillance complex.
BRCA1-BARD1 interactions are demonstrated in the most abundant BRCA1-containing complexes. As mentioned above, the redistribution of BRCA1 among protein complexes is consistent with changes noted in subnuclear localization (28,29). We observed clearly that BRCA1 content shifted from the holopol complex and fraction 5 complex to the HUIC after HU treatment of cells. This shift was less apparent for the BRCA1-Rad50 complex. Is the HUIC a product derived from the holopol complex or from the fraction 5 complex? We found that the HUIC does not contain the holo-pol components Med17, Pol II, and BRG1 (data not shown), and in sucrose gradient sedimentation of the 0.6 M KOAc elution, BARD1 fractionated broadly throughout the gradient, including two peaks consistent with the fraction 5 complex and the HUIC. It is possible that HU treatment leads to degradation of holo-pol and the fraction 5 complexes, leaving a residual BRCA1-BARD1-containing complex in the HUIC. This hypothesis is consistent with a model in which transcription functions in surveillance for DNA damage by transcription-coupled repair and recombination (6). Upon encountering DNA damage, the BRCA1-BARD1 ubiquitin ligase activity destroys the holo-pol, leaving the HUIC at the site of damage, and the HUIC then recruits repair factors.
BRCA1 is a dynamic protein because it is present in untreated cycling cells in three complexes, but after replication blockage by HU treatment, the BRCA1 content shifts to a new protein complex, the HUIC. It is likely that BRCA1 protein participates in multiple cellular pathways by different functions in different complexes.