Ubiquitination and proteasomal degradation of the BRCA1 tumor suppressor is regulated during cell cycle progression.

The BRCA1 tumor suppressor and the BARD1 protein form a stable heterodimeric complex that can catalyze the formation of polyubiquitin chains. Expression of BRCA1 fluctuates in a cell cycle-dependent manner, such that low steady-state levels of BRCA1 gene products are found in resting cells and early G1 cycling cells and high levels in S and G2 phase cells. Although transcriptional activation of the BRCA1 gene can account for induction of BRCA1 expression at the G1/S transition, the mechanisms by which BRCA1 is down-regulated during cell cycle progression have not been addressed. Here we show that the steady-state levels of BRCA1 protein remain elevated throughout mitosis but begin to decline at the M/G1 transition. This decline in BRCA1 levels coincides with the appearance of proteasome-sensitive ubiquitin conjugates of BRCA1 at the onset of G1. Formation of these conjugates occurs throughout G1 and S, but not in cells arrested in prometaphase by nocodazole. The proteasome-sensitive ubiquitin conjugates of BRCA1 appear to be distinct from BRCA1 autoubiquitination products and are probably catalyzed by the action of other cellular E3 ligases. Interestingly, co-expression of BARD1 inhibits the formation of these conjugates, suggesting that BARD1 serves to stabilize BRCA1 expression in part by reducing proteasome-sensitive ubiquitination of BRCA1 polypeptides. In summary, these data indicate that the cell cycle-dependent pattern of BRCA1 expression is determined in part by ubiquitin-dependent proteasomal degradation.

Germline mutations of the BRCA1 gene are responsible for a substantial proportion of hereditary breast and ovarian cancers (1,2). In this clinical setting, BRCA1 serves as a tumor suppressor that contributes to tumorigenesis through loss of function. The protein it encodes has been implicated in a number of biological processes, including the cellular response to DNA damage (3,4). In particular, BRCA1 is required for several checkpoints that control cell cycle progression (5,6) and inhibit mRNA processing (7,8) after genotoxic stress, as well as for certain modes of DNA repair such as nucleotide excision repair (9,10) and homology-directed repair of double-strand DNA breaks (11)(12)(13). As a key regulator of the DNA damage response, BRCA1 presumably promotes tumor suppression by preserving genomic stability. However, the molecular mechanisms by which it carries out these functions are not understood and, as a consequence, it is still unclear why inherited mutations of the BRCA1 gene predispose women to breast and ovarian cancer.
The BRCA1 polypeptide contains two recognizable amino acid motifs: a RING domain near the N terminus and two tandem copies of the BRCT domain at the C terminus (14). In vivo, BRCA1 exists as a heterodimer with BARD1, a distinct protein that harbors a similar array of RING and BRCT motifs (15). Since the phenotypes of mice null for either Brca1 or Bard1 are essentially indistinguishable, the functions of both proteins are likely to be mediated through the BRCA1/BARD1 heterodimer (16), and indeed BARD1 has already been implicated with BRCA1 in homology-directed repair of chromosomal breaks (17). BRCA1 and BARD1 associate by assembling a stable 4-helix bundle from the ␣ helices that flank their respective RING domains (18), and together they form an enzymatic complex that can catalyze ubiquitin polymerization in vitro (19 -24). This enzymatic activity implies that BRCA1/BARD1 functions as an E3 ligase that promotes ubiquitin modification of specific substrate proteins, and that these are likely to include important effectors of BRCA1-mediated tumor suppression (25,26). Although definitive substrates of BRCA1/ BARD1 have not yet been identified, autoubiquitination of the BRCA1 subunit is observed during in vitro reactions catalyzed by BRCA1/BARD1 (22). In vitro, BRCA1/BARD1 directs the formation of ubiquitin polymers through an unconventional isopeptide linkage involving lysine residue K6 of ubiquitin (27,28). These K6-linked polyubiquitin chains are distinct from the more common K48-linked chains that target substrate proteins for proteasomal degradation and the K63-linked chains implicated in various signaling pathways (29). K6-linked chains are also generated by BRCA1 autoubiquitination in vivo (28), and they appear to aggregate at sites of DNA damage (30). However, the functional consequences of BRCA1 autoubiquitination and the cellular role of K6-linked polyubiquitin chains are still unclear.
Early studies established that BRCA1 expression fluctuates in a cell cycle-dependent manner. While the steady-state levels of BRCA1 polypeptides are low in resting (G 0 ) cells and G 1 cycling cells, these levels increase considerably as cycling cells enter S phase (31,32). The induction of BRCA1 protein expression that occurs near the G 1 /S boundary is probably driven by transcriptional activation, since it is preceded by a sharp increase in the levels of BRCA1 mRNA that begins in late G 1 (33,34). Indeed, the promoter region of BRCA1 harbors binding sites for E2F transcription factors that may be responsible for the cell cycle-dependent induction of BRCA1 transcription (35). However, the mechanisms by which BRCA1 levels are down-regulated as cells progress from S phase to the next G 1 phase have not been examined. Here we show that, after peaking in late G 1 , the steady-state levels of BRCA1 mRNA steadily decline as cycling cells divide and enter the subsequent G 1 phase. In addition, we find that BRCA1 polypeptides are subject to ubiquitination and proteasome-mediated degradation at specific stages of cell cycle progression and that dimerization with BARD1 serves to protect BRCA1 polypeptides from ubiquitination in vivo. These data indicate that the cell cycle regulation of BRCA1 expression is determined in part by active degradation through the ubiquitin/proteasome pathway.

EXPERIMENTAL PROCEDURES
Cell Culture-The T24, HeLa, and 293 cell lines were obtained from the American Type Tissue Culture Collection and maintained in Mc-Coy's 5A, Dulbecco's modified Eagle's medium, and Iscove's media, respectively, supplemented with 100 g/ml penicillin/streptomycin, 2 mM L-glutamine, and 10% fetal bovine serum. The following drugs were obtained from Calbiochem and used at the noted concentrations unless otherwise specified: cycloheximide (100 g/ml), MG132 (20 M), ALLN (100 M), ALLM (100 M), and lactacystin (25 M). Cells were treated with drugs at ϳ50% confluence.
Cell Cycle Analysis-T24 cells were arrested in G 0 by contact inhibition as described (36). HeLa cells were synchronized in prometaphase by a thymidine/nocodazole block that entailed a 12 h treatment with 2 mM thymidine followed by a 14 -18 h treatment with 50 ng/ml nocodazole; the mitotic cells were then harvested by shake-off, and released into nocodazole-free medium after washing twice with phosphate-buffered saline and once with complete Dulbecco's modified Eagle's medium. Both the attached and unattached cells were harvested at each time point and used for FACS, Western, and Northern analyses. For FACS, the cells were stained with propidium iodide as described (36) and analyzed on a BD Biosciences FACSCalibur using CellQuest software. For Western analysis, the cells were lysed in low salt Nonidet P-40 buffer (10 mM HEPES pH 7.6, 250 mM NaCl, 0.1% Nonidet P-40, 5 mM EDTA, 10% glycerol) supplemented with 1 mM dithiothreitol, protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, BMB protease inhibitor pellet), and phosphatase inhibitors (10 mM ␤-glycerophosphate, 5 mM NaF, and 0.1 mM NaVO 4 ) and analyzed as described above. For Northern analysis, cytoplasmic RNA was isolated using the RNeasy Mini Kit (Qiagen) and 20-g aliquots were evaluated by Northern filter hybridization with radiolabeled cDNA probes for BRCA1, BARD1, and ␤-actin using ExpressHyb solution (Clontech).
Identification of Ubiquitinated Conjugates of Exogenous BRCA1 in Vivo-The FLAG-BRCA1 expression plasmid (Fl4-BRCA1-wt/pCMV) encodes full-length human BRCA1 with an N-terminal extension of 39 residues that contains four tandem FLAG epitopes; this plasmid was generated by inserting a cDNA sequence for single FLAG-tagged BRCA1 into the pCMV-Fl3-Not vector (derived from the 3ϫFLAG-CMV-7 plasmid; Sigma). The Fl4-BR771/pCMV expression vector encodes a truncated BRCA1 fragment (⌬BRCA1; amino acids 1-771) with the same N-terminal extension containing four tandem FLAG epitopes; it was prepared by inserting a cDNA sequence for single FLAG-tagged ⌬BRCA1 into the pCMV-Fl3-Not vector. The C61G and I26A mutants of BRCA1 were generated using the Stratagene Chameleon doublestranded site-directed mutagenesis kit. The mammalian expression plasmids pMT107 and pMT123 were a gift from Dr. Dirk Bohmann; these encode polyproteins comprised of 8 tandem copies of the His 6tagged or HA-tagged ubiquitin, respectively (37). To examine in vivo ubiquitination of exogenous BRCA1 polypeptides, 2 ϫ 10 6 293 cells were seeded onto a 100-mm plate. After 12 h the cells were transfected using 25-30 l of FuGENE 6 reagent (Roche Applied Science) with 10 -12 g of total plasmid DNA encoding FLAG-BRCA1, BARD1, and/or His 6 -Ub, all under the control of a CMV promoter. At 30 h after transfection, the cells were treated with MG132 or the specified drug for an additional 6 h. The cells (ϳ2 ϫ 10 7 cells at ϳ70% confluence) were then lysed by a brief sonication in 700 l of FLAG lysis buffer (50 mM Tris-HCl, pH 7.9, 137 mM NaCl, 1% Triton X-100, 0.2% Sarkosyl, 10% glycerol) supplemented with 1 mM dithiothreitol, protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, BMB protease inhibitor pellet), phosphatase inhibitors (10 mM NaF, 1 mM NaVO 4 ), and deubiquitinase inhibitors (5 mM N-ethylmaleimide, 10 nM ubiquitin aldehyde). Immunoprecipitations were performed by rotating 630 l of cell lysate with 50 l of anti-FLAG M2 beads (Sigma, 20% slurry) at 4°C for 10 -12 h. The beads were washed four times with FLAG lysis buffer and eluted for 12 h at 4°C with 500 g/ml FLAG peptide in 40 l of FLAG lysis buffer. Samples were separated on 3-8% Tris-acetate gels (Invitrogen), and Western analyses were performed with antibodies that recognize BRCA1 (Oncogene Ab-1), BARD1 (polyclonal 669D), and ubiquitin (Santa Cruz P4D1).
Purification of Endogenous Ubiquitinated Proteins-To generate a cell line that stably expresses His 6 -tagged ubiquitin, HeLa cells were transfected using LipofectAMINE (Invitrogen) with a mammalian expression plasmid (His-Ub 8 /pCIN4) that encodes a polyprotein comprised of 8 tandem copies of His 6 -Ub; this plasmid was generated by excising the polyprotein cDNA sequence from pMT107 (37) and inserting it into the pCIN4 vector (pIRES-neo; Clontech). HeLa cell transformants were selected with 0.5 mg/ml G418, and maintained in 0.25 mg/ml G418. The thymidine/nocodazole block and release were performed as described above, with only unattached cells collected at the mitotic time points and only attached cells at the interphase time points. Double thymidine blocks were performed by seeding 2 ϫ 10 6 cells onto each 100-mm plate, culturing the cells in 2 mM thymidine for 14 h, releasing the cells in thymidine-free medium for 10 h, treating for another 14 h with 2 mM thymidine, and releasing in thymidine-free medium containing 50 ng/ml nocodazole. MG132 addition and cell collection were performed at the time points described (six 100-mm plates at each time point for nocodazole release, four 100-mm plates for double thymidine release). Approximately 1 ϫ 10 6 cells from each time point were used for FACS analysis, and the remaining cells were lysed in 10 ml of lysis buffer (6 M guanidinium HCl, 100 mM NaH 2 PO 4 , 10 mM Tris-HCl, 10 mM imidazole, 10 mM ␤-mercaptoethanol, pH 8.0) and sonicated for 2 min on ice. To analyze the lysates directly by Western blotting, trichloroacetic acid precipitation was performed by diluting 10 l of lysate into 500 l of water containing 80 g/ml sodium deoxycholate as a carrier. After adding 71 l of 100% trichloroacetic acid, the sample was incubated on ice for 1 h and then microcentrifuged for 15 min. The pellet was washed twice with cold acetone, dried, resuspended in 1ϫ protein dye, and denatured for Western analysis. To isolate ubiquitin-conjugated proteins, the remainder of the cell lysate was incubated with 50 l of Ni-NTA agarose beads (Qiagen) for 2 h. The beads were then washed eight times for 5 min each with 4 ml of buffer containing 100 mM NaH 2 PO 4 , 10 mM Tris-HCl, 10 mM imidazole, 10 mM ␤-mercaptoethanol supplemented with the following: 1) 6 M guanidinium HCl, pH 8.0; 2) 8 M urea, pH 8.0; 3) 8 M urea, 0.2% Triton X-100, pH 6.3; 4) 8 M urea, pH 6.3; 5) 8 M urea, 0.1% Triton X-100, pH 6.3; 6) 8 M urea, pH 6.3; 7) 8 M urea, 0.5 M NaCl, pH 6.3; or 8) 8 M urea, pH 6.3. The ubiquitin conjugates were then eluted with 50 l NTA elution buffer (0.15 M Tris-HCl pH 6.7, 200 mM imidazole, 5% SDS, 0.72 M ␤-mercaptoethanol, 30% glycerol). After adding 5ϫ protein loading dye and denaturing at 70°C for 10 min, the eluates were fractionated by SDS-PAGE and analyzed by Western blotting.
RNAi Transfections-The siRNAs specific for human BACH1 (AGCUUACCCGUCACAGCUUdTdT) (38) and human BARD1 (AA-CAGUAACAUGUCCGAUGAAdTdT) were synthesized by Dharmacon RNA Technologies (Lafayette, CO). In addition, a SMARTpool of four siRNAs specific for BARD1 was also obtained from Dharmacon. HeLa cells were grown in 6-well plates in complete Dulbecco's modified Eagle's medium without antibiotics. At 30% confluence, the cells were transfected with 80 pmol of siRNA and 4 l of LipofectAMINE 2000 (Invitrogen) according to the manufacturer's protocol. After 6 h, the transfection mixtures were replaced with complete medium. After culturing the cells for an additional 18 h, the transfection procedure was repeated, and the cells were harvested for analysis 48 h after the initial transfection.

RESULTS
Cell Cycle Regulation of BRCA1 and BARD1-To confirm the cell cycle regulation of BRCA1 and BARD1 expression, T24 bladder carcinoma cells were arrested in G 0 by contact inhibition, released into fresh medium, and collected at various time points. In accord with previous studies (31,32), BRCA1 polypeptides are undetectable in arrested T24 cells but begin to appear by 16-h post-induction as cells approach the G 1 /S transition (Fig. 1A). The levels of BRCA1 increase further during S and G 2 /M, before decreasing as the cells enter the next G 1 phase by 36 h after induction. BRCA1 exists predominantly in a hypophosphorylated state in G 1 , but becomes hyperphosphorylated in S and G 2 /M, and reverts to a hypophosphorylated form again in the next G 1 phase (31,32,39,40). As reported previously (33,34), expression of BRCA1 mRNA is maximal in late G 1 and decreases as the cells traverse the cell cycle and enter the next G 1 phase (Fig. 1B).
Using a BARD1-specific polyclonal antibody, we find that BARD1 protein expression is also regulated with respect to the cell cycle (Fig. 1A). Protein levels follow a similar pattern as for BRCA1, appearing in late G 1 , increasing during S and G 2 /M, and decreasing as the cells enter the next G 1 phase. The cell cycle-dependent nature of BARD1 protein expression differs from that described previously using a monoclonal antibody raised against BARD1 (36); this discrepancy arose because the monoclonal antibody cross-reacts with a nonspecific band that co-migrates with BARD1 (see Supplementary Data and Discussion). The mRNA expression of BARD1 follows the same pattern as that of BRCA1, peaking in late G 1 and decreasing during the remainder of the cell cycle (Fig. 1B). Thus, BRCA1 and BARD1 expression are co-induced at both the mRNA and protein level as cells progress from G 1 to S phase.
To determine the timing and kinetics of the decrease in BRCA1 and BARD1 protein levels from the G 2 /M to G 1 phase, HeLa cells were arrested in prometaphase by a thymidine/ nocodazole block, released into nocodazole-free medium, and collected at hourly time points. Upon release from mitotic arrest, we observed a steady decrease in the levels of BARD1 and especially BRCA1 ( Fig. 2A). Although the steady-state levels of their mRNAs also decline after mitosis (Fig. 2B), this result suggests that BRCA1 and BARD1 polypeptides are actively degraded in vivo. Therefore, the kinetics of BRCA1 and BARD1 expression were compared with those of other proteins known to be proteolyzed during mitotic progression, including factors that begin to degrade in prometaphase (cyclin A), at the metaphase/anaphase transition (cyclin B1 and the securin PTTG), and at the M/G 1 transition (Plk1) (41). As seen in Fig. 2A, the decline in BRCA1 and BARD1 levels starts later (2-3 h after nocodazole release) than that of cyclin B1 or PTTG (1-2 h after release), and continues into the next G 1 phase. This pattern more closely resembles that of Plk1, a protein kinase that undergoes proteasome-mediated degradation beginning at the M/G 1 transition (42,43). BRCA1 Degradation Is Mediated by the Proteasome-To test whether endogenous BRCA1 and BARD1 polypeptides are actively degraded, their steady-state levels in asynchronous 293 cells were monitored after treatment with cycloheximide, an inhibitor of protein synthesis. As shown in Fig. 3A, BRCA1 protein levels declined within a few hours and displayed a much shorter half-life than ␣-tubulin. This implies that proliferating 293 cells possess an active mechanism for BRCA1 deg- radation. BARD1 protein levels decrease only slightly after 8 h of cycloheximide treatment, with a longer half-life and slower rate of turnover than BRCA1. Thus, BRCA1 may be the more tightly regulated subunit of the BRCA1/BARD1 heterodimer.
To determine whether BRCA1 degradation is mediated by the proteasome, we tested whether the cycloheximide-induced decrease in endogenous BRCA1 levels can be prevented by various proteasome inhibitors. MG132, ALLN, and ALLM are peptide aldehydes that inhibit the proteasome, as well as other cellular proteases, including cathepsins and calpains, with varying degrees of efficacy (44); while MG132 and ALLN appear to inhibit most of these proteases, ALLM is a relatively poor inhibitor of the proteasome. When 293 cells were treated for 8 h with cycloheximide and various concentrations of these inhibitors, the cycloheximide-induced decrease in BRCA1 levels was prevented by increasing concentrations of MG132 and ALLN, but not ALLM (Fig. 3B). The decrease was also prevented by lactacystin, a more specific inhibitor of the proteasome, implying that BRCA1 degradation after cycloheximide treatment is indeed mediated by the proteasome in 293 cells.
BRCA1 Is Degraded by Ubiquitin-mediated Proteolysis-Since the effect of proteasome inhibitors on cellular BRCA1 levels may be direct or indirect, we sought to determine whether the BRCA1 protein can be ubiquitinated in vivo. Therefore, 293 cells were co-transfected with expression plasmids encoding His 6 -tagged ubiquitin (to increase the cellular pool of free ubiquitin) and FLAG-tagged BRCA1. After treatment with various proteasome inhibitors, the cells were lysed in the presence of deubiquitinase inhibitors (N-ethylmaleimide and ubiquitin aldehyde). The FLAG-BRCA1 polypeptides were then immunoprecipitated with anti-FLAG M2 beads and eluted with FLAG peptide. In these experiments, the cell lysis and immunoprecipitations were conducted under stringent conditions (1% Triton X-100 and 0.2% Sarkosyl) to minimize coimmunoprecipitation of BRCA1-associated proteins. Western analysis of the eluates with a BRCA1-specific antibody indicates that similar amounts of FLAG-BRCA1 were recovered from each sample (Fig. 4A, lanes 2-6), while immunoblotting with a ubiquitin-specific antibody reveals the presence of ubiquitin-conjugated BRCA1 species that migrate more slowly than the unmodified FLAG-BRCA1 polypeptide (lanes 8 -12). Significantly, these species are most prominent in cells treated with an effective proteasome inhibitor, such as MG132 (lane 9), ALLN (lane 10), or lactacystin (lane 12). Formation of these ubiquitinated forms increases in a dose-dependent manner with the level of exogenous BRCA1 expression (Fig. 4B), and ubiquitinated BRCA1 is detectable even in the absence of exogenous His 6 -ubiquitin, though to a significantly lesser extent (Supplemental Figs. C and D). These results indicate that exogenous BRCA1 can indeed be ubiquitinated in vivo and that ubiquitinated BRCA1 polypeptides are subject to proteasomal degradation.
BARD1 Expression Stabilizes Exogenous BRCA1 by Specifically Reducing BRCA1 Ubiquitination-Previous studies have shown that BARD1 expression can stabilize BRCA1 in vivo. For example, experimental reduction of endogenous BARD1 protein levels by genetic targeting in mice (16) or by RNAi in frogs (45) results in a corresponding decrease in BRCA1 levels, while co-expression of exogenous BARD1 increases the steady-state levels of exogenous BRCA1 (21). Consistent with these data, we see a modest increase in exogenous BRCA1 levels with increasing BARD1 expression (Fig. 5, lanes 5-8). Significantly, this increase is accompanied by a corresponding decrease in ubiquitinated BRCA1 (lanes 9 -12), suggesting that BARD1 stabilizes BRCA1 by protecting it from ubiquitination. This effect is specific for BRCA1, since BARD1 expression does not influence the steady-state levels or ubiquitination of other proteins known to be degraded by the ubiquitin/proteasome pathway, such as Plk1 and cyclin A (Supplementary Figs. E and F).
Exogenous BRCA1 Polypeptides Are Subjected to Proteasomesensitive Polyubiquitination in Vivo -Since BRCA1 is autoubiquitinated during BRCA1/BARD1-dependent in vitro ubiquitination reactions (22), the in vivo ubiquitination of BRCA1 detected in our assay may be due to the catalytic activity of BRCA1 itself or to that of a distinct cellular E3 ligase(s). Nishikawa et al. (28) recently demonstrated that a truncated form of BRCA1 (amino acid residues 1-772) can catalyze autoubiquitination through lysine 6 of ubiquitin in vivo and showed that the levels of autoubiquitinated BRCA1 are not enhanced in the presence of a proteasome inhibitor. To address whether the ubiquitination detected in our assay is also due to autoubiquitination, we expressed a similar BRCA1 segment (⌬BRCA1, amino acids 1-771) in the presence of BARD1 and either His 6 -Ub (as in our previous experiments) or HA-Ub (as in those of Nishikawa et al.). We then compared the in vivo ubiquitination of FLAG-tagged ⌬BRCA1 segments that do or do not harbor the I26A mutation, an amino acid substitution that impairs the catalytic activity of BRCA1 by disrupting its interaction with cognate E2 enzymes (46). Lysates of transfected 293 cells were immunoprecipitated with FLAG-specific antibodies, and the presence of FLAG-tagged ⌬BRCA1 and its ubiquitin conjugates was measured by immunoblotting with BRCA1-specific (lanes 1-4) and Ub-specific (lanes 5-8) antibodies, respectively. As shown in Fig. 6, the steady-state levels of the wild-type (lanes 1 and 2) and mutant (lanes 3 and 4) ⌬BRCA1 segments increase modestly in the presence of the proteasome inhibitor MG132. A ladder of ubiquitinated ⌬BRCA1 forms is also apparent in these lanes, but it is more prominent with wild-type ⌬BRCA1 (lanes 1 and 2) than with the corresponding I26A mutant (lanes 3 and 4). Likewise, when detected by immunoblotting with Ub-specific antibodies, the levels of ubiquitinated conjugates are also clearly more abundant with wild type (lane 5) than mutant (lane 7) ⌬BRCA1. In this respect, our data are consistent with the previous results of  1-4). B, cell lysates were then immunoprecipitated with anti-FLAG M2 beads, followed by elution with the FLAG peptide, and the levels of FLAG-BRCA1 (lanes 5-8) and ubiquitin-conjugated FLAG-BRCA1 (lanes 9 -12) polypeptides were determined by Western analysis of the eluates with BRCA1-or ubiquitin-specific antibodies, respectively.
Nishikawa et al. (28), and support their contention that BRCA1 undergoes autoubiquitination in vivo. However, we also observe that low steady-state levels of ubiquitinated conjugates are formed from the catalytically impaired ⌬BRCA1 segment (lane 7). Moreover, because the levels of these conjugates are dramatically enhanced in the presence of a proteasome inhibitor (compare lanes 7 and 8), they are likely to serve as substrates for proteasomal degradation. Therefore, in addition to autoubiquitination, ⌬BRCA1 is also subjected to transubiquitination in vivo, presumably by cellular E3 ligases that promote BRCA1 turnover.
To address whether full-length BRCA1 polypeptides are also subject to ubiquitination, we transfected 293 cells with expression vectors that encode full-length BRCA1 bearing either the wild-type sequence, the catalytically impaired I26A mutation, or the C61G missense mutation-a cancer-predisposing lesion that also ablates the enzymatic activity of BRCA1. As shown in Fig. 7, in vivo ubiquitination of the mutants is only modestly reduced relative to that of wild-type BRCA1 (lanes 13, 15, and 17). Thus, under these conditions the polyubiquitination of full-length BRCA1 polypeptides is largely independent of its own enzymatic activity.
In addition to inactivating the catalytic activity of BRCA1, the C61G mutation also impairs its in vivo interaction with BARD1 (15). In contrast, the I26A mutation does not affect formation of the BRCA1/BARD1 heterodimer (46). As shown in Fig. 7, co-expression of BARD1 with either wild-type BRCA1, C61G-BRCA1, or I26A-BRCA1 increases the levels (lanes 8, 10, and 12) and decreases the ubiquitination (lanes 14, 16, and 18) of each of these polypeptides. However, the effect of BARD1 expression on the C61G mutant (lanes 10 and 16) is less pronounced than for either wild-type BRCA1 (lanes 8 and 14) or the I26A mutant (lanes 12 and 18), likely because of its weaker interaction with BARD1.
Endogenous BRCA1 Polypeptides Are Also Subject to Proteasome-sensitive Polyubiquitination-We have demonstrated that exogenously expressed BRCA1 polypeptides can be ubiquitinated in vivo (Figs. 4 -7). To determine whether endogenous BRCA1 can also be ubiquitinated, we created a stable line of HeLa cells that expresses His 6 -tagged ubiquitin (Fig. 8A); as a result, ubiquitin-conjugated cellular proteins can be purified from lysates of these cells by affinity chromatography on nickel agarose (NTA) beads. The His 6 -Ub expressing cells and the parental HeLa cells were cultured in the presence or absence of MG132 and lysed under denaturing conditions. Ubiquitinated BRCA1 polypeptides were then recovered on NTA beads, and detected by immunoblotting with BRCA1-specific antibodies. As seen in Fig. 8B, trace amounts of BRCA1 bind non-specifically to the NTA beads, even from parental HeLa cells that do not express His 6 -Ub (lanes 1 and 3). In addition, however, slower migrating forms of BRCA1 are readily detected in the His 6 -Ub expressing cells (lane 5), and the appearance of these forms is enhanced by the addition of a proteasome inhibitor (lane 7). Thus, endogenous BRCA1 can be ubiquitinated in vivo, and at least some of this ubiquitination is associated with proteasomal degradation.
In Vivo Ubiquitination of Endogenous BRCA1 Is Regulated With Respect to the Cell Cycle-BRCA1 protein expression is cell cycle regulated, with a steady decrease in levels as cells exit mitosis ( Fig. 2A). Whereas the concomitant reduction in mRNA levels suggests a transcriptional component to this decrease (Fig. 2B), we sought to determine whether ubiquitin-mediated proteasomal degradation also plays a role. Thus, HeLa cells that stably express His 6 -Ub were arrested in prometaphase by thymidine/nocodazole block and released into nocodazole-free medium for various lengths of time (0 -20 h), after which the cells were incubated with MG132 for an additional 6 h. The cells were then lysed and the ubiquitinated proteins were recovered by NTA chromatography as described above. As a control, we also monitored the ubiquitination of PTTG, a mammalian securin whose kinetics of degradation during mitosis have been well characterized (47). Proteasomal degradation of PTTG is required for the activation of separase, a protease that cleaves the cohesins that link sister chromatids. As a result, metaphase cells treated with a proteasome inhibitor do not progress into anaphase because the sister chromatids cannot separate. Therefore, cells released directly from nocodazole block into medium containing MG132 will arrest in metaphase and accumulate ubiquitinated PTTG that cannot be degraded. As shown in Fig. 8C, we do not detect ubiquitinated PTTG in cells arrested in prometaphase by nocodazole (lane 1). As expected, however, ubiquitinated forms of PTTG accumulate in cells arrested in metaphase (lane 3), but disappear if these cells are allowed to progress into the next G 1 phase prior to MG132 treatment (lanes 5 and 7).
BRCA1 ubiquitination follows a different pattern from that of PTTG (Fig. 8C). Although some unmodified BRCA1 from each time point binds non-specifically to the NTA beads, little, if any, of the slowly migrating ubiquitinated forms of BRCA1 are seen in cells arrested in prometaphase (Fig. 8C, lane 1) or   FIG. 7. Exogenous full-length BRCA1 is subjected to proteasome-sensitive ubiquitination by cellular E3 ligases. 293 cells were transfected with plasmids encoding His 6 -Ub and either the wild type (lanes 1 and 2, 7 and 8, 13 and 14), C61G mutant (lanes 3 and 4, 9 and 10, 15 and  16), or I26A mutant (lanes 5 and 6, 11 and  12, 17 and 18) form of the FLAG-tagged full-length BRCA1. Some cells were also co-transfected with a plasmid encoding BARD1 (even numbered lanes). After 30 h, the transfected cells were treated with 20 M MG132 for an additional 6 h. A, levels of BARD1 were measured by Western analysis of cell lysates with BARD1-specific antiserum (lanes 1-6). B, cell lysates were then immunoprecipitated with anti-FLAG M2 beads, followed by elution with the FLAG peptide, and the levels of FLAG-BRCA1 (lanes 7-12) and ubiquitin-conjugated FLAG-BRCA1 (lanes 13-18) polypeptides were determined by Western analysis of the eluates with BRCA1-or ubiquitin-specific antibodies, respectively. metaphase (lane 3). However, ubiquitinated BRCA1 becomes detectable as the cells exit mitosis and enter the next G 1 phase (lanes 5 and 7). Surprisingly, the levels of ubiquitinated BRCA1 continue to rise as cells progress into S phase (lane 9), even though the post-mitotic decrease in steady-state BRCA1 protein levels has ended and BRCA1 levels are increasing again with the onset of S phase.
High Turnover of Endogenous BRCA1 Polypeptides during S phase-To determine whether the S phase-specific ubiquitination of BRCA1 is also related to proteasomal degradation, His 6 -Ub expressing HeLa cells were arrested at the G 1 /S boundary by a double thymidine block and then released into fresh medium containing nocodazole (to prevent cells from progressing into the subsequent G 1 phase). At 4-hour intervals after release, parallel cultures were either harvested immediately or treated with MG132 for an additional 6 h. As the cells traverse S phase, ubiquitinated forms of BRCA1 are much more prominent in cells treated with a proteasome inhibitor (Fig. 9, lanes 10 and 12) than without (lanes 9 and 11), implying that the S phase ubiquitination of BRCA1 is indeed associated with proteasomal degradation. The levels of ubiquitinated BRCA1 steadily decrease as the cells progress from G 1 /S (lanes 10 and 12) to G 2 /M (lanes 14 and 16) while protein levels progressively increase (lanes 1, 3, 5, and 7), suggesting that BRCA1 is stabilized as cells enter mitosis. Interestingly, the ubiquitination and proteasomal degradation of BRCA1 seems most active around early S phase, a stage of the cell cycle when BRCA1 mRNA levels are elevated and BRCA1 protein begins to accumulate (Fig. 1). Thus, the results of Figs. 8 and 9 indicate that turnover of the BRCA1 protein is greatest, and its expression is most tightly regulated, during S phase, but that the formation of proteasome-sensitive ubiquitinated conjugates of BRCA1 is reduced or ceases in nocodazole-arrested cells.
Interestingly, after proliferating HeLa cells are subject to 6 h of treatment with MG132, most BRCA1 polypeptides exist in a hypophosphorylated state (Fig. 8B, lanes 4 and 8). We also observe this phenomenon after MG132 treatment of cells syn-chronized (Fig. 9, lane 2) or released (lanes 4 and 6) from a G 1 /S block, and some hypophosphorylated BRCA1 appears after MG132 treatment even in mitotic cells (lane 8). This may occur because 1) the proteasome is responsible for degradation of either BRCA1 phosphatases or BRCA1 kinase inhibitors, 2) MG132 has effects on cellular BRCA1 in HeLa cells that are independent of its inhibition of the proteasome, or 3) hypophosphorylated BRCA1 is more susceptible to proteasomal degradation than hyperphosphorylated BRCA1. The latter possibility is consistent with the data of Fig. 3A in which hypophosphorylated BRCA1 decays faster than hyperphosphorylated BRCA1 in 293 cells treated with cycloheximide.
In cells arrested at prometaphase by nocodazole, BRCA1 exists as a single band (Fig. 10, lane 2) that migrates with a FIG. 9. High levels of proteasome-sensitive ubiquitination of endogenous BRCA1 during S phase. His 6 -Ub expressing HeLa cells (His-Ub cells) were arrested at the G 1 /S boundary by double thymidine block, and released into fresh medium containing nocodazole (to prevent cells from progressing into the next G 1 phase). At 4-hour intervals after release, parallel cultures were either harvested immediately (lanes 1, 3, 5, and 7) or treated with MG132 for an additional 6 h ( lanes  2, 4, 6, and 8). The harvested cells were then analyzed by FACS, and the ubiquitin-conjugated polypeptides were purified from cell lysates by binding to Ni-NTA agarose beads and eluting in a buffer containing imidazole. Western analyses of the chromatography eluates and untreated cell lysates were performed with BRCA1-specific antibodies.  1 and 3). The BRCA1 polypeptides of mitotic cells presumably bear a distinctive pattern of post-translational modifications distinct from those of the hypo-and hyperphosphorylated BRCA1 (48). If the hypophosphorylated forms of BRCA1 are indeed more susceptible to proteasomal degradation, then the absence of these forms in prometaphase may explain why ubiquitination of BRCA1 is reduced at this stage of the cell cycle ( Fig. 9, lanes 15 and 16). DISCUSSION Early studies established that BRCA1 expression is low or undetectable in resting (G 0 ) cells, but that the steady-state levels of its mRNA and protein products increase markedly at the G 1 /S transition after resting cells are induced to proliferate (31)(32)(33)(34). These findings revealed the cell cycle-dependent nature of BRCA1 expression and indicated that transcriptional mechanisms are important for the increase in BRCA1 levels that occurs near the G 1 /S boundary. Here we show that BRCA1 levels are also regulated with respect to the cell cycle by ubiquitindependent proteolysis. Blagosklonny et al. (49) had previously examined the mechanism of BRCA1 turnover in cell lines expressing basal levels of BRCA1 protein that are low or undetectable in cycling cells, and found that BRCA1 degradation in these cells does not depend on the proteasome but is instead mediated primarily by a cathepsin-like protease(s). In contrast, it was observed that the steady-state levels of BRCA1 are increased upon treatment of 293 cells with the proteasomal inhibitor MG132 (45). While other proteases may also contribute to degradation of cellular BRCA1 polypeptides, our demonstration that ubiquitinated forms of BRCA1 exist in HeLa and 293 cells, and that their levels are elevated by proteasome inhibition, indicates that the ubiquitin/proteasome pathway also serves to regulate BRCA1 expression in vivo.
When unsynchronized proliferating cell populations are examined by immunostaining, the levels of BRCA1 protein appear to be significantly higher in S and G 2 phase cells than in G 1 cells (40). This implies that BRCA1 expression is downregulated as cells progress from G 2 to the subsequent G 1 stage. However, the timing and nature of this down-regulation had not been addressed. By monitoring synchronized cells released from mitotic arrest, we now show that the steady-state levels of BRCA1 polypeptides start to decline around the M/G 1 transition ( Fig. 2A). Parallel analysis of other proteins known to be proteolyzed at specific times during mitosis indicates that BRCA1 down-regulation begins subsequent to prometaphase (marked by the onset of cyclin A degradation) and the metaphase/anaphase transition (cyclin B1 and securin/PTTG degradation) and coincident with the proteolysis of Plk1, a protein kinase that undergoes ubiquitin-dependent degradation beginning at the M/G 1 transition (42,43). Our data also suggest that at least two processes can potentially contribute to cell cycleassociated BRCA1 down-regulation. First, the levels of BRCA1 mRNA continue to decline after peaking in S phase, perhaps reflecting decreased initiation of BRCA1 transcription and/or increased turnover of BRCA1 mRNA (Fig. 1B). Second, ubiquitinated conjugates of endogenous BRCA1 polypeptides begin to appear as cells proceed from mitosis to the subsequent G 1 phase (Fig. 8). Since these conjugates accumulate markedly in cells treated with specific proteasome inhibitors, they are likely to reflect increased proteolysis of BRCA1 at the M/G 1 transition and probably contribute to the down-regulation of BRCA1 expression that occurs in early G 1 .
Because the steady-state levels of BRCA1 protein increase at the G 1 /S transition (31,32), we had expected that the levels of proteasome-sensitive ubiquitinated BRCA1 conjugates would decrease as cycling cells progressed from G 1 to the subsequent S phase. Surprisingly, however, these levels remained high throughout S phase, and did not decrease until G 2 or M (Fig. 8). This indicates that high rates of BRCA1 ubiquitination and degradation are occurring during S phase, a stage of cell cycle progression when the steady-state levels of BRCA1 are high. BRCA1 has been implicated in a number of important S phase functions, such as homology-directed DNA break repair (11)(12)(13) and the DNA damage-induced S phase checkpoint (5). Thus, it is not obvious why BRCA1 synthesis and degradation should be elevated simultaneously in S phase. The high rate of BRCA1 turnover in S phase may allow for more stringent control of BRCA1 activity by promoting rapid changes in BRCA1 levels, localization, and/or post-translational modification. In any case, these data reveal that BRCA1 expression is controlled in part by the ubiquitin/proteasome pathway and that turnover of BRCA1 polypeptides is elevated during S phase.
In contrast to G 1 and S phase cells, ubiquitin conjugates of endogenous BRCA1 are barely detectable in cells arrested at prometaphase by nocodazole (e.g. see lane 1, Fig. 8C). This implies that BRCA1 ubiquitination is suspended during mitosis, at least while the spindle checkpoint is enforced. It appears that BRCA1 ubiquitination is also reduced during subsequent stages of mitosis since the level of ubiquitinated BRCA1 conjugates remains low after cells are released from nocodazole arrest in the presence of proteasome inhibitor (lane 3, Fig. 8C), a treatment that blocks cells in a subsequent stage of mitosis, but terminates the spindle checkpoint as manifested by the ubiquitination of the securin PTTG. Thus, although ubiquitination and degradation of BRCA1 occurs during G 1 and S, it is specifically suspended during mitosis. This mitotic stabilization of BRCA1 is intriguing given recent evidence that BRCA1 performs important functions during mitosis. In particular, Brca1-deficient mouse embryonic cells display mitotic lesions, including spindle abnormalities and centrosome amplification (6,50), and BRCA1 polypeptides are reported to bind ␥-tubulin and localize to the centromeres and spindle microtubules of mitotic cells (51)(52)(53)(54). Moreover, when Brca1-or Bard1-null blastocysts are cultured in vitro, cells of the inner cell mass fail to proliferate, while trophoblast giant cells, which endoreplicate their DNA without undergoing mitosis, appear to be unaffected (16,55). These observations indicate that BRCA1 has critical functions in mitosis, and suggest that disruption of these functions may be responsible for some aspects of the genomic instability that characterize Brca1-deficient cells (6,50).
In vitro ubiquitination reactions driven by the BRCA1/ BARD1 heterodimer generate conjugates of BRCA1 that harbor primarily K6-linked polyubiquitin chains (27, 28). Nishikawa et al. (28) have shown that K6-linked polyubiquitin conjugates of BRCA1 are also produced in vivo by autoubiquitination in FIG. 10. The phosphorylation state of BRCA1 in mitosis. Lysates of asynchronous HeLa cells (lanes 1, 3, 5, and 7) and HeLa cells released from a thymidine/nocodazole block for 0, 2, and 5 h (lanes 2, 4, and 6, respectively) were examined by immunoblotting with a BRCA1specific antibody. To obtain bands of comparable intensity, different quantities of lysate were fractionated from asynchronous cells (20 g) and from cells released for 0 h (2.0 g), 2 h (2.5 g), and 5 h (5.0 g). cells transfected with exogenous BRCA1 polypeptides. In addition, they found that the levels of these conjugates are not affected by proteasome inhibitors (28). Our present data indicate that exogenous BRCA1 can also be polyubiquitinated in vivo in a manner independent of its own enzymatic activity (Figs. 6 and 7), presumably through the action of other cellular E3 ligases. The resulting ubiquitin conjugates are likely to serve as intermediates for proteasomal degradation of BRCA1 since they can be stabilized by treatment with proteasome inhibitors. Indeed, by using cells that express epitope-tagged ubiquitin, we were able to detect ubiquitinated conjugates of endogenous BRCA1 and show that their levels are also enhanced by proteasome inhibition (Fig. 8B). Thus, the in vivo pool of ubiquitinated BRCA1 species appears to include both proteasome-insensitive and proteasome-sensitive conjugates. On one hand, the proteasome-insensitive BRCA1 conjugates are generated by autoubiquitination, a process that generates K6-linked chains and may impart critical, but as yet unknown, effects on BRCA1 activity (28). On the other hand, the proteasome-sensitive conjugates represent BRCA1 intermediates destined for degradation and as such are likely to harbor ubiquitin modifications that are recognized by the proteasome, such as K48-linked chains. Interestingly, the balance between proteasome-insensitive and proteasome-sensitive conjugates differed depending on the nature of the BRCA1 polypeptide under analysis. While proteasome-insensitive conjugates were prominent in cells transfected with truncated BRCA1 (28) (Fig. 6), proteasome-sensitive conjugates were more abundant in cells transfected with full-length BRCA1 (Fig. 7) and upon analysis of endogenous BRCA1 (Fig. 8). Although we do not know the basis for this phenomenon, the truncated BRCA1 polypeptides may represent enzymatically-deregulated forms with a heightened capacity for autoubiquitination.
Previous studies had shown that BRCA1 and BARD1 exert a reciprocal stabilizing effect on one another, such that expression of one subunit of the heterodimer serves to increase the steady-state levels of the other subunit (16,21,45). In light of this phenomenon, as well as the fact that the enzymatic activity of the heterodimer is dramatically higher than that of either BRCA1 or BARD1 alone (21), it seems unlikely that BRCA1 degradation would be triggered by autoubiquitination. Instead, the proteasome-sensitive ubiquitin conjugates of BRCA1 are probably generated by other cellular E3 ligases. At present, however, we do not know which E3 ligases are responsible for BRCA1 degradation. The anaphase promoting complex (APC) is an intriguing candidate since the kinetics of BRCA1 decay at the M/G 1 transition are reminiscent of late APC substrates, such as Plk1 and Cdc20 (41). However, in contrast to these proteins, proteasome-sensitive ubiquitination of BRCA1 continues as cells progress into S phase and does not cease until the following G 2 or M phase. Although other cell cycle-regulated E3 ligases, such as the skp1/cullin/F-box (SCF) complex, might promote BRCA1 ubiquitination in S phase, further studies will be required to identify the cellular enzymes responsible for BRCA1 ubiquitination. Indeed, these enzymes may be relevant in sporadic cases of breast cancer. While BRCA1 mutations are not observed in sporadic breast cancer, the tumor cells of these patients often exhibit reduced steady-state levels of BRCA1 polypeptides (56). In some cases this has been attributed to impaired synthesis due to hypermethylation of the BRCA1 promoter. However, an increased rate of BRCA1 degradation, conceivably mediated by a deregulated E3 ligase, might constitute another mechanism by which BRCA1 levels are reduced in sporadic breast cancer.
It is intriguing to note that the formation of proteasomesensitive ubiquitin conjugates of BRCA1 is reduced by co-ex-pression of BARD1 (Fig. 7). Although we do not know the molecular mechanism by which this is achieved, the ability of BARD1 to protect BRCA1 polypeptides from ubiquitination may account in part for the mutual stabilization that is observed upon co-expression of BRCA1 and BARD1 (16,21,45).
The results of Fig. 1, which were obtained by immunoblotting cell lysates with a BARD1-specific polyclonal antiserum, demonstrate that BARD1 protein expression is regulated with respect to the cell cycle. However, our previous data using a monoclonal antibody raised against BARD1 did not reveal cell cycle fluctuations in BARD1 protein levels by immunoblotting, despite the fact that immunofluorescence analyses with the same antibody, as well as with the BARD1-specific polyclonal antiserum, had shown perfect co-localization of BRCA1 and BARD1 polypeptides during S phase but little, if any, BARD1 staining during G 1 (36). We now demonstrate that this discrepancy arose because the monoclonal antibody cross-reacts strongly with an unrelated species in Western blots of human cell lysates; this species co-migrates with BARD1 during SDS-PAGE and its levels do not change during cell cycle progression (Supplemental Figs. A and B). However, when the same analysis is conducted with a polyclonal antiserum that specifically recognizes BARD1, the steady-state levels of BARD1 are found to fluctuate with cell cycle progression in parallel with those of BRCA1, at least with respect to the G 1 /S induction (Fig. 1). Interestingly, however, the half-life of BARD1 appears to be longer than that of BRCA1, indicating a higher turnover of BRCA1 than BARD1 in asynchronously cycling cells (Fig. 3). This implies that BRCA1 is the more tightly controlled subunit of the BRCA1/BARD1 heterodimer and raises questions as to how and why regulation of the two subunits is accomplished differently. While we have been able to demonstrate in vivo ubiquitination of exogenous BARD1 polypeptides (data not shown), we have not successfully detected ubiquitinated conjugates of endogenous BARD1, possibly due to the limitations of current BARD1 antibodies. Thus, it is unclear whether the same active mechanisms that degrade BRCA1 also apply to BARD1 or whether the more gradual decline in BARD1 levels during G 1 reflects a passive reaction to changes in BRCA1 levels.