BRCA1/BARD1 ubiquitinate phosphorylated RNA polymerase II.

The breast- and ovarian-specific tumor suppressor BRCA1, when associated with BARD1, is an ubiquitin ligase. We have shown here that this heterodimer ubiquitinates a hyperphosphorylated form of Rpb1, the largest subunit of RNA polymerase II. Two major phosphorylation sites have been identified in the Rpb1 carboxyl terminal domain, serine 2 (Ser-2) or serine 5 (Ser-5) of the YSPTSPS heptapeptide repeat. Only the Ser-5 hyperphosphorylated form is ubiquitinated by BRCA1/BARD1. Overexpression of BRCA1 in cells stimulated the DNA damage-induced ubiquitination of Rpb1. Similar to the in vitro reaction, the stimulation of Rpb1 ubiquitination by BRCA1 in cells occurred only on those molecules hyperphosphorylated on Ser-5 of the heptapeptide repeat. In vitro, the carboxyl terminus of BRCA1 (amino acids 501-1863) was dispensable for the ubiquitination of hyperphosphorylated Rpb1. In cells, however, efficient Rpb1 ubiquitination required the carboxyl terminus of BRCA1, suggesting that interactions mediated by this region were essential in the complex milieu of the nucleus. These results link the BRCA1-dependent ubiquitination of the polymerase with DNA damage.

BRCA1, the breast-and ovarian-specific tumor suppressor protein, has been found to regulate a number of processes central to the normal function of the cell, including transcription, chromatin dynamics, homologous recombination, and other forms of DNA damage repair (1,2). Because BRCA1 has been found associated with a wide range of proteins involved in these processes, it may function as a scaffold, organizing effector proteins in a context-dependent manner. However, when BRCA1 is associated with the BARD1 protein, it is also an enzyme, an E3 ubiquitin ligase (3,4). The realization that BRCA1 is an enzyme establishes the necessity of identifying its substrates in order to understand how the ubiquitination activity impacts these processes in the cell. BRCA1 and BARD1 are associated with the messenger RNAsynthesizing polymerase in a complex known as the RNA polymerase II holoenzyme (holo-pol) 1 (5)(6)(7). One function for BRCA1 in this holo-pol complex appears to be as a coactivator of transcription, because it has been shown that BRCA1 stimulates the activation signal of p53, NF-B, and others (8 -13). Previously, we modeled that the BRCA1 and BARD1 in the holo-pol complex may ubiquitinate the transcribing RNA polymerase II (RNAPII) when it encounters DNA damage, and we also suggested that this ubiquitination event would stimulate the repair process (14,15).
Rpb1 is the largest subunit of RNAPII, and its carboxylterminal domain (CTD) is highly conserved, consisting of multiple repeats (27 in budding yeast, 52 in humans) of the heptapeptide YSPTSPS. Serines 2 (Ser-2) and 5 (Ser-5) of multiple repeats are phosphorylated co-transcriptionally, Ser5*p predominating at the promoter and Ser2*p in the coding sequence (16,17). In response to DNA damage Rpb1 is also ubiquitinated, an event associated with changes in concentration of both the hypophosphorylated and the hyperphosphorylated Rpb1 (18). In budding yeast, the Rsp5 E3 ligase ubiquitinates Rpb1 independent of its phosphorylation state (19,20). In higher eukaryotes the ubiquitin ligase(s) that mediate this modification of RNAPII are unknown, and it is possible that multiple factors mediate the reaction. Because BRCA1 and BARD1 are associated with RNAPII in the holo-pol complex (6), BRCA1 is a reasonable candidate for the RNAPII ubiquitin ligase. In addition, after DNA damage BRCA1 and BARD1 also associate with the polyadenylation cleavage factor CstF (21), known to interact with RNAPII via Rpb1 hyperphosphorylated on Ser-2 (Ser2*p) of the YSPTSPS heptapeptide repeats (22,23). These results led us to speculate that a substrate for BRCA1-dependent ubiquitination could be the Ser2*p form of Rpb1.
In these experiments we tested whether BRCA1 in association with BARD1 could ubiquitinate RNAPII. We found that hyperphosphorylated RNAPII serves as a substrate for the BRCA1-dependent ubiquitination activity, and we found that overexpression of BRCA1 in cells stimulates the DNA damageinduced ubiquitination of hyperphosphorylated RNAPII. Strikingly, the ubiquitination reaction, when tested both in vitro and in vivo, was enhanced not by Ser2*p of the heptapeptide repeat but rather by Ser5*p. These results thus identify a substrate for ubiquitination by BRCA1/BARD1 that is correlated with the cellular response to DNA damage.

MATERIALS AND METHODS
Protein Purification-The expression and purification of BRCA1 and BARD1 from baculovirus-infected insect cells has been described, along with a description of the purification of the ubiquitination factors E1 and UbcH5c E2 (24). The core RNAPII was purified from calf thymus using an established protocol (25). The budding yeast Rpb1 CTD was expressed as a hexahistidine and GST fusion (26) and purified by nickel-nitrilotriacetic acid chromatography using standard techniques. Ubiquitin was obtained from a commercial vendor (Sigma).
The yeast Kin28, Ctk1, and Srb10 kinases were each expressed in Saccharomyces cerevisiae as HA-tagged fusion proteins. Active kinases were purified by immunoprecipitation using the 12CA5 monoclonal antibody specific for the HA tag (27,28).
Human TFIIH was purified from HEK-293 cells as described (29). In brief, ϳ10 12 cells were collected over a period of several months, and a whole cell extract was prepared for each. The whole cell extracts were bound to a Biorex70 matrix at 0.15 M KOAc in buffer A (20 mM Hepes, pH 7.9, 1 mM EDTA, 5% glycerol, 3 mM dithiothreitol), washed at 0.3 M KOAc, 0.6 M KOAc, and the peak was collected at 1.5 M KOAc. At each column step, TFIIH-containing fractions were identified by Western blotting using antibodies specific to the 89-kDa ERCC-3 subunit of TFIIH. The 1.5 M KOAc peak fraction was dialyzed to 0.1 M KCl in buffer A, bound to a DEAE fast flow matrix, and the protein peak at 0.3 M KCl was collected and dialyzed to 0.1 M KCl. The protein was bound to a 2-ml BioScale-Q column (Bio-Rad Laboratories), and protein was eluted in a gradient from 0.1 to 1.0 M KCl. TFIIH-containing fractions were subjected to gel filtration using a Superdex-200 (HR16/60; Amersham Biosciences) column in 0.3 M KCl in buffer A. The TFIIH migrated at a volume consistent with a 700-kDa complex, and samples were dialyzed in 0.1 M KCl in buffer A.
pCMV-Myc-ubiquitin was constructed as follows. Ubiquitin was amplified from cDNA of HeLa cells as a template using the primers 5Ј-GCCGAATTCGGATGCAGATCTTCGTGAAAAC-3Ј and 5Ј-CCGC-TCGAGCTAACCACCTCTCAGACGCAGG-3Ј that contain 5Ј-EcoRI site and 3Ј-XhoI site. The PCR product was then subcloned into the pCMV-Myc vector (Clontech). All constructs were verified by DNA sequence.
To test this hypothesis, we utilized purified RNAPII core enzyme that had been phosphorylated in vitro by TFIIH as a substrate in ubiquitination reactions. Purified RNAPII exists in two forms, the IIA form, in which the Rpb1 CTD has a low level of phosphorylation, and the IIO form, in which this domain is hyperphosphorylated and has significantly shifted migration on SDS-PAGE. Phosphorylation of this RNAPII preparation by TFIIH results in the labeling of both of these forms of Rpb1 (Fig. 1A, lanes 1 and 2). This labeled RNAPII was tested in ubiquitination reactions that contained purified E1, E2 UbcH5c, E3 BRCA1/BARD1, and ubiquitin. In the complete reaction, the RNAPIIO band disappeared and a slower migrating diffuse band was observed. Under these conditions, the hypophosphorylated RNAPIIA was not modified (Fig. 1A, lane 3). These results suggest that the hyperphosphorylated RNA-PII is a substrate for the BRCA1/BARD1 ubiquitin ligase.
The appearance of the slowly migrating RNAPIIO band was dependent upon the inclusion of each ubiquitination factor. Single omission of the substrate, E1, E2, E3, or ubiquitin failed to produce the slowly migrating RNAPIIO band (Fig. 1B). The appearance of the slowly migrating RNAPIIO band was thus consistent with modification by ubiquitination because only when all ubiquitination factors were included in reactions did this species appear (lane 1).
We tested whether the full 12-subunit RNAPII complex was required for ubiquitination by BRCA1/BARD1 or whether the phosphorylated CTD would suffice. The experiment of Fig. 1B was repeated using only the Rpb1 CTD fused to GST. This substrate was phosphorylated by purified TFIIH and [␥-32 P]ATP. When labeled GST⅐CTD was incubated with the complete reaction containing E1, E2 UbcH5c, ubiquitin and BRCA1/BARD1, the GST⅐CTD protein had markedly slowed migration. In this portion of the gel (Ͼ85 kDa), the resolution was imperfect, and we interpret the diffuse band with slowed migration to be consistent with the multiple additions of 8-kDa ubiquitin moieties (Fig. 1C, lane 1). The CTD of this substrate protein had no lysines to be modified by ubiquitination. We suggest that the CTD recruits the BRCA1/BARD1 E3 ligase for the ubiquitination of a separate domain of the polypeptide. These results indicate that both the 12-subunit RNAPII com-plex and the GST⅐CTD were substrates for the BRCA1/BARD1 E3 ubiquitin ligase.
The CTD used in these experiments was from the budding yeast S. cerevisiae, and contained 26 copies of the YSPTSPS heptapeptide. The CTD is co-transcriptionally phosphorylated in vivo on both Ser-2 and Ser-5. RNAPII containing unphosphorylated Rpb1 is preferentially recruited to preinitiation complexes but is phosphorylated during the transition from initiation to elongation. A Ser5*p form of the Rpb1 CTD predominates at the promoter, with Ser2*p CTD more prevalent in the coding sequence. TFIIH kinase activity is directed primarily at Ser-5 (23), with human Cdk7 and its homolog Kin28 in S. cerevisiae acting as the kinase in each case. The S. cerevisiae kinases Ctk1 and Srb10 have highest phosphorylation activity directed at Ser-2 (28). When the CTD is expressed and purified from bacteria, it is unphosphorylated, whereas RNAPII purified from eukaryotic cells is phosphorylated to different degrees on both serine positions. To test which phosphorylation event is required for ubiquitination, it was necessary to use the CTD purified from bacteria. Incubation of the CTD with each specific kinase results in differently phosphorylated products: predominantly Ser5*p when Kin28 is the kinase or Ser2*p when Ctk1 or Srb10 is used (28). We tested whether the ubiquitination activity of BRCA1/BARD1 was directed specifically at the Rpb1 CTD containing either Ser5*p or Ser2*p. In Fig. 2A, the GST⅐CTD was labeled by phosphorylation with Kin28, Ctk1, or Srb10 prior to incubation in the ubiquitination reaction. Ser5*p GST⅐CTD was multiply ubiquitinated in the presence of BRCA1/BARD1 (Fig. 2A, lane 2), but Ser2*p GST⅐CTD ubiquitination could not be detected (lanes 4 and 6). This result suggested that the ubiquitination of the CTD by BRCA1/ BARD1 was specific for substrates containing Ser5*p.
The specificity of the BRCA1/BARD1 E3 ligase in this reaction was striking. If the heterodimer was simply binding to and ubiquitinating a long polypeptide with multiple negative charges, as in the hyperphosphorylated CTD, then we would expect little or no preference for either the Ser2*p or Ser5*p forms. Instead, the ubiquitination by BRCA1/BARD1 was specific for the Ser5*p CTD. In binding experiments using the purified BRCA1/BARD1 and purified RNAPII, we found that the BRCA1 bound to RNAPII independent of phosphorylation ( Fig. 2B, right panel). This result was not surprising because it is known that BRCA1 binds to Rpb2 and Rpb12 of RNAPII (32). However, when comparing the effectiveness of the purification of RNAPII on a BRCA1/BARD1 affinity matrix, the recovery of the Ser5*p-RNAPII was more complete than was observed for the hypophosphorylated form (Fig. 2B, left panel). Thus, binding alone did not specify the ubiquitination substrate, but Ser5-specific phosphorylation enhanced both the level of binding and of ubiquitination by BRCA1/BARD1. Note that the Ser5*p form of the CTD is observed at the promoter, whereas the Ser2*p is associated with transcription elongation (17). Thus, the Ser5*p-specific modification of RNAPIIO by BRCA1/ BARD1 is not consistent with targeting the elongating polymerase for ubiquitination.
In assays using RNAPII as the substrate, ubiquitination was specific for the hyperphosphorylated Rpb1. Similar specificity was observed for all constructs tested with the exception of BRCA1 alone and the ⌬N construct, which had no detectable activity. Thus, BARD1 and the BRCA1 RING domain were each required for ubiquitination of RNAPII (Fig. 3B). The absence of activity seen with BRCA1 lacking BARD1 is consistent with previously published results. BARD1 is required for a high level of ubiquitination activity of BRCA1, and the isolated RING domains of each protein have been shown to have low levels of ubiquitination activity in vitro (3,33,34). However, the ubiquitination activity of BRCA1 is significantly potentiated by its interaction with BARD1 (3,4), and structural studies of the amino terminus of BRCA1 and BARD1 reveal extensive interaction between these domains (35). The ⌬N construct lacks a RING domain and was thus expected to lack ubiquitination activity.
All of the active truncations of BRCA1 specifically ubiquitinated the hyperphosphorylated form of RNAPII, whereas the hypophosphorylated form was relatively unmodified (Fig. 3B). We had previously hypothesized that the carboxyl terminus of BRCA1 mediates the specificity of its association with RNAPII because this domain of BRCA1 activates transcription (36 -38) and because it binds to two RNAPII subunits (32). Efficient ubiquitination of RNAPII, however, was observed even when the ubiquitin ligase was a BRCA1 truncation that lacked the carboxyl terminus, suggesting that the function of the BRCA1 carboxyl-terminal transcription activation domain is unrelated to its ubiquitination of phosphorylated RNAPII by BRCA1.
The RNAPII ubiquitination assay yields a qualitative result, indicating that hyperphosphorylated Rpb1 is a substrate for the ubiquitination activity of BRCA1/BARD1. We repeated the experiment using TFIIH-phosphorylated CTD (Ser5*p) as a substrate, and we found that there were no differences in the degree of ubiquitination obtained with the BRCA1 carboxylterminal truncations (Fig. 3C). Under these more sensitive conditions, weak ubiquitination was evident when BRCA1 lacking BARD1 was included in reactions (Fig. 3C, lane 3), whereas the ⌬N construct had no ubiquitination activity (Fig.  3C, lane 8). Therefore, in vitro, the carboxyl terminus of BRCA1 is not required for ubiquitination of hyperphosphorylated RNA-PII or Ser5*p-phosphorylated CTD.
BRCA1 Ubiquitinated Phosphorylated RNAPII in Vivo-We next asked whether BRCA1 could ubiquitinate hyperphosphorylated RNAPII in vivo. We transfected HEK-293T cells with plasmids encoding HA epitope-tagged BRCA1 and Myc epitopetagged ubiquitin. Transfected cell lysates were immunoprecipitated using antibody specific to the Myc epitope, thus purifying ubiquitinated proteins, and then immunoblots were probed using antibodies specific to RNAPII. The immunoblot was stained with the monoclonal antibody H14, which specifically binds to RNAPII phosphorylated on Ser-5 of the heptapeptide repeat in the CTD (18). The lysate (input) contained a phosphorylated RNAPII large subunit that migrated at a position consistent with 240 kDa (Fig. 4B, lane 1). Background levels of ubiquitinated phospho-RNAPII were detected in cells transfected with vector alone (lane 2). It is established that hyperphosphorylated RNAPII becomes ubiquitinated following ultraviolet (UV) irradiation of cells (18, 39 -41), and we detected the UV-dependent ubiquitination of RNAPII (Fig. 4B, lane 5). Most of the ubiquitinated species migrated on protein gels with a very small shift relative to the unmodified species (compare lanes 5 and 1), and this would be expected for a low number of ubiquitin moieties (about 8 kDa each) bound to a 240-kDa polypeptide. The resolution of these species was poor by SDS-PAGE, but we consistently observed stimulated recovery of the hyperphosphorylated Rpb1 band due to ubiquitination after UV irradiation. In addition, a diffuse band of ubiquitinated species was observed shifted at slower migration that we interpret to be multiply ubiquitinated RNAPIIO.
Transfection of full-length BRCA1 had minimal effect on RNAPII ubiquitination status (Fig. 4B, lanes 3 and 6). We had previously observed that overexpression of full-length BRCA1 dysregulated normal BRCA1 complex formation, presumably by altering the cell cycle (30). In those experiments, expression of a BRCA1 with an internal deletion, HA-BRCA1(⌬775-1292), allowed us to overexpress BRCA1 and observe all of the protein complexes seen with the endogenous protein (30). This internal deletion, here called HA-BRCA1(⌬M), strongly stimulated the ubiquitination of Ser5*p-hyperphosphorylated RNAPII independent of DNA damage (Fig. 4B, lane 4, top panel).
UV irradiation of the cells stimulated ubiquitination of phospho-RNAPII (Fig. 4B, lanes 5 and 6), and in UV-irradiated HA-BRCA1(⌬M)-expressing cells a significant increase in the intensity of the slowly migrating band was observed (lane 7) that we interpret to be multiply ubiquitinated RNAPIIO. These results indicate that overexpression of BRCA1(⌬M) stimulated ubiquitination of Ser5*p-Rpb1 independent of, but qualitatively modified by, DNA damage. When we tested the H5 monoclonal antibody that specifically binds to Ser2*p RNAPII or the 8WG16 monoclonal antibody that specifically recognizes hypophosphorylated RNAPII on immunoblots, ubiquitinated RNAPII was not detected (data not shown). These results were consistent with the in vitro experiments (Fig. 2) in which Ser-5 phosphorylation of the RNAPII CTD specifically stimulated its ubiquitination by BRCA1/BARD1. These results were also con-sistent with the previously established ubiquitination of Ser-5phosphorylated RNAPIIO after UV-induced DNA damage (18,40).
The consequences of BRCA1-dependent ubiquitination are unclear. BRCA1/BARD1 have been shown to direct the linkage of ubiquitin chains via either lysine 6, lysine 48, or lysine 63 isopeptide bonds (4,42). Appending ubiquitin chains via lysine  1-7). Cells were treated with 20 J/m 2 UV irradiation (lanes 5-7) and 50 M MG132 (lanes 1-7). Lysates were immunoprecipitated by anti-Myc antibody (lanes 2-7). Immunoblots were stained with H14 monoclonal antibody to recognize the Ser5*p version of RNA-PIIO (top panel). Input samples were immunoblotted and stained with H14, anti-HA antibody, and 8WG16 (8WG), the last to detect unphosphorylated RNAPII. The input sample (lane 1) was only included in the top panel. C, HEK-293T cells were transfected with vectors to express HA-BRCA1(⌬M) (lanes 1-4) and Myc ubiquitin (lanes 1-4) and treated with 20 J/m 2 UV (lanes 3, 4) in the presence of 50 M MG132 (lanes 2, 4) as described under "Materials and Methods." Lysates were immunoprecipitated using an anti-Myc antibody, and immunoblots were stained for Ser5*p-RNAPIIO using antibody H14 (top panel). Input samples were stained with antibodies 8WG16 and H14 to detect hypophosphorylated RNAPIIA and Ser5*p-RNAPIIO in the samples (bottom panel). D, HEK-293T cells were transfected with vectors to express Myc ubiquitin (lanes 1-4) and also HA-BRCA1(⌬M) (lanes 2 and 4) and subjected to UV irradiation (lanes 3 and 4). Lysates were immunoprecipitated using the Ser5*p-specific H14 antibody and immunoblots probed with the Myc-specific antibody to detect ubiquitin (top panel). Input samples were immunoblotted using H14 and 8WG16 antibodies (bottom panel). . Cells were irradiated in the presence of MG132 as above, and lysates were immunoprecipitated using the H14 monoclonal antibody (top panel). Input lysates were analyzed in the bottom two panels, and immunoblots were probed as indicated. 48 target the substrate for proteasome-mediated degradation; thus BRCA1/BARD1 ubiquitination may in some cases not lead to protein degradation. We tested whether inhibition of the proteasome, using MG132, could stabilize the ubiquitinated phospho-RNAPII. Proteasome inhibition resulted in longer chains of ubiquitin appended on the Rpb1 subunit of RNAPIIO (Fig. 4C, lane 4, top panel), suggesting that BRCA1-dependent ubiquitination may cause degradation of RNAPIIO. Interestingly, UV irradiation of cells resulted in a shift in the polymerase from RNAPIIA to RNAPIIO (Fig. 4C, bottom panel), a phenomenon that has been observed previously (18). Although quantitation using two different antibodies in immunoblots is imprecise, this result suggests that phosphorylation of Rpb1 to Ser5*p is a generalized response after DNA damage. Although proteasome inhibition stabilized the recovery of ubiquitinated RNAPIIO (lanes 3 and 4), the amount of RNAPIIO in the lysate was not markedly increased (Fig. 4C, lanes 3-4, bottom panel). We infer from this result that only a fraction of the total RNAPII is targeted for degradation following BRCA1-dependent ubiquitination.
Repeating the experiment, but using the H14 antibody to immunoprecipitate the RNAPIIO and the anti-Myc antibody on immunoblots to detect the ubiquitin, revealed that HA-BRCA1(⌬M) expression stimulated the appearance of ubiquitinated RNAPIIO (Fig. 4D, lane 2). As in panel B, expression of HA-BRCA1(⌬M) in UV-irradiated cells resulted in the recovery of higher levels of ubiquitinated RNAPIIO (Fig. 4D, lane 4). Compared with anti-Myc ubiquitin immunoprecipitation, use of the H14 antibody reproducibly yielded lower amounts of ubiquitinated RNAPIIO, even after UV irradiation. We interpret this lower level to be due to less effective immunoprecipitation reactions with the latter antibody.
We have previously shown that BRCA1 is a component of RNAPII holo-pol, and the carboxyl terminus of BRCA1 is important for this association (5,6). In the in vitro assays in this study (Fig. 3), the carboxyl terminus of BRCA1 was not required for ubiquitination of the polymerase. However, in the complicated environment of a cell, the carboxyl-terminal-mutated BRCA1 might not associate with the polymerase and thus not ubiquitinate it. We examined whether the carboxyl terminus of BRCA1 affected ubiquitination of phospho-RNAPII in tissue culture cells. We found that overexpression of BRCA1 lacking its carboxyl terminus resulted in only background levels of ubiquitinated RNAPIIO (Fig. 5B, compare lanes 1-4). We thus conclude that in cells the carboxyl terminus of BRCA1 is important for the UV damage-induced ubiquitination of RNAPIIO.
We also tested whether a specific missense mutation associated with breast cancer affects the ubiquitination of RNAPIIO. The disease-associated missense mutation M1775R in the BRCT domain of the carboxyl terminus of BRCA1 ablates the double strand break repair and transcription activation function of BRCA1 (43). BRCA1 proteins containing the M1775R mutation do not bind to histone deacetylases (44), BACH1 (45), and the transcriptional co-repressor CtIP (46,47). As shown in Fig. 5B, expression of BRCA1 with M1775R did not stimulate the ubiquitination of phosphorylated RNAPII (Fig. 5B, lane 5,  top panel). Although the mutation of BRCA1 at residue M1775R decreases the stability of the protein (48), the expression level of the HA-BRCA1(⌬M-M1775R) was equal to that of HA-BRCA1(⌬M) (Fig. 5B, middle panel). Furthermore, the M1775R mutation disrupted BRCA1 binding to RNAPIIO (Fig. 6). In transfected cells, immunopurification of HA-BRCA1(⌬M) also purified Ser5*p Rpb1 (Fig. 6, lane 2). Deletion of the carboxyl terminus of BRCA1 or the BRCA1 protein containing a missense mutation resulted in significantly de-creased binding to RNAPIIO (Fig. 6, lanes 3 and 4). Thus, an intact carboxyl terminus was required for BRCA1 to bind to RNAPIIO. These data suggest that ubiquitination of phosphorylated RNAPII by BRCA1 in response to DNA damage requires an intact BRCT domain.
The active site of BRCA1 for ubiquitin ligase activity is encoded in the RING domain of the amino terminus of the protein. Missense mutation of one of the zinc-coordinating residues, C61G, results in an inactive E3 ubiquitin ligase even in the presence of wild-type BARD1 (3,34,35). In patients, inheritance of this missense mutation is associated with breast cancer (49,50). Expression of HA-BRCA1(⌬M) containing the C61G missense mutation did not stimulate the ubiquitination of phosphorylated RNAPII (Fig. 5C, top panel).
The experiment in Fig. 5C was repeated, but the immunoprecipitating antibody was the Ser5*p-specific H14, and ubiquitinated species were detected using the Myc-specific antibody on immunoblots. As before, we observed that HA-BRCA1(⌬M) expression stimulated the recovery of ubiquitinated RNAPIIO (Fig. 5D, lane 3). Further, expression of BRCA1 variants containing the missense mutation C61G (lane 4) or a carboxylterminal truncation (lane 5) failed to stimulate the ubiquitination of RNAPIIO. As in Fig. 4D, this immunoprecipitation reaction was weaker than when the Myc antibody was used, and we only detected the ubiquitinated species when HA-BRCA1(⌬M) was expressed. Taken together, the data in Figs. 4 and 5 indicate that BRCA1 stimulates the ubiquitination of Ser5*p RNAPII after UV irradiation.

DISCUSSION
Identification of the substrates for BRCA1-dependent ubiquitination activity is important for understanding how mutation of BRCA1 is associated with loss of tumor suppression activity. The currently identified substrates include histone proteins, p53, Fanconi anemia protein D2, and centrosomal proteins including NPM1 and ␥-tubulin (24,(51)(52)(53)(54). Among these, only the modification of ␥-tubulin by BRCA1/BARD1 has been shown to affect the biology of breast cells. It has been shown that failure to ubiquitinate ␥-tubulin results in centrosome amplification (24). The BRCA1/BARD1 proteins are known to regulate multiple processes in the cell, including transcription, DNA repair, and centrosome dynamics (5,(55)(56)(57)(58)(59). Although the ubiquitination of ␥-tubulin may in part ex- plain the BRCA1-dependent regulation of centrosome dynamics, it was unclear whether the BRCA1-dependent ubiquitination activity also regulates the transcription and DNA repair function of BRCA1.
We had proposed that the BRCA1-dependent ubiquitination activity may function in DNA repair by modification of RNAPII that transcribes DNA near a lesion (14,15). This proposed role for BRCA1 in transcription-coupled repair could be important following UV damage or double strand breaks. One prediction of this model was that BRCA1/BARD1 ubiquitination activity would be targeted to the elongating, hyperphosphorylated form of RNAPII. Actively transcribing RNAPII is phosphorylated on Ser-5 proximal to the promoter and on Ser-2 further downstream (23). Thus, the principal form of RNAPII that elongates through a gene is the Ser2*p form, which we now show is not a substrate for BRCA1/BARD1. The model that BRCA1-dependent ubiquitination directly links transcription elongation to repair is thus not supported. Instead, we found that Ser-5 phosphorylation of RNAPII is a generalized response to UV irradiation, and BRCA1-dependent ubiquitination modifies the RNAPIIO. It has been observed that transcriptionally engaged RNAPII does become phosphorylated on Ser-5 by the action of extracellular signal-regulated kinases 1 and 2 (60). The data are most consistent with a model whereby DNA damage causes phosphorylation of a subpopulation of RNAPII, followed by ubiquitination by BRCA1/BARD1 and subsequent degradation at the proteasome.
In these experiments we found that overexpression of BRCA1 in cells could stimulate the damage-induced ubiquitination of RNAPII. When we inhibited BRCA1 expression by transfection of short interfering RNA specific for BRCA1, we did not observe a decrease in ubiquitination of RNAPII. 2 We interpret these results to indicate that one or more other ubiquitin ligases can execute this function. Several other factors have been implicated in the ubiquitination of RNAPII, including Cockayne syndrome proteins CSA and CSB (60,61). Even though other factors can also ubiquitinate RNAPII, our results overexpressing BRCA1 clearly indicate that it participates in this process.
In summary, we found in this study that BRCA1/BARD1 ubiquitinate RNAPII hyperphosphorylated via Ser-5 of the heptapeptide repeat. Rpb1 was multiply ubiquitinated. In experiments using highly purified factors in vitro, only the amino terminus of BRCA1, containing the catalytic RING domain, was required for ubiquitination of phospho-RNAPII. The BARD1 protein was not essential, but it was highly stimulatory. In cells, overexpression of BRCA1 could stimulate the ubiquitination of hyperphosphorylated RNAPII. In contrast to the in vitro reactions using purified factors, in the cell the carboxyl-terminal domain was important for the DNA damagestimulated ubiquitination of phosphorylated RNAPII by BRCA1. These results are consistent with our observations that both the amino-and carboxyl-terminal domains of BRCA1 are required for BRCA1 association with the polymerase complex.