Estrogen receptor α is a putative substrate for the BRCA1 ubiquitin ligase

Abstract

The breast cancer suppressor protein, BRCA1, is a ubiquitin ligase expressed in a wide range of tissues. However, inheritance of a single BRCA1 mutation significantly increases a woman's lifetime chance of developing tissue-specific cancers in the breast and ovaries. Recently, studies have suggested this tissue specificity may be linked to inhibition of estrogen receptor α (ERα) transcriptional activation by BRCA1. Here, we show that ERα is a putative substrate for the BRCA1/BARD1 ubiquitin ligase, suggesting a possible mechanism for regulation of ERα activity by BRCA1. Our results show ERα is predominantly monoubiquitinated in a reaction that involves interactions with both BRCA1 and BARD1. The regions of BRCA1/BARD1 necessary for ERα ubiquitination include the RING domains and at least 241 and 170 residues of BRCA1 and BARD1, respectively. Cancer-predisposing mutations in BRCA1 are observed to abrogate ERα ubiquitination. The identification of ERα as a putative BRCA1/BARD1 ubiquitination substrate reveals a potential link between the loss of BRCA1/BARD1 ligase activity and tissue-specific carcinoma.

The breast cancer suppressor protein, BRCA1, is essential for early development and associated with a number of cellular processes including cell proliferation, cell cycle progression, apoptosis, and DNA repair/recombination. Inheritance of a single mutation in BRCA1 increases a woman's lifetime risk of developing breast cancer from 1 in 9 to greater than 1 in 2 (1). Additionally, in sporadic nonfamilial cases of breast cancer, BRCA1 expression is significantly reduced, indicating loss of BRCA1 at the protein or mRNA level (2, 3). Therefore, BRCA1 inactivation either through mutation or repression of gene expression is implicated in all forms of breast cancer development.

The only known enzymatic activity associated with BRCA1 is its activity as a ubiquitin protein ligase (4). Attachment of ubiquitin to substrates is a versatile method of regulation involved in practically all aspects of cell biology. Substrate ubiquitination involves several steps and a well-known trio of enzymes called ubiquitin activating (E1), ubiquitin conjugating (E2), and ubiquitin ligase (E3). The E3 activity of BRCA1 has been localized to the N-terminal RING domain and depends on heterodimerization with the RING domain of BARD1 (5). Importantly, all BRCA1 cancer-predisposing mutations within the RING domain are observed to eliminate the BRCA1/BARD1 E3 activity both in vitro and in vivo (5–7). This strongly suggests a link between the tumor suppressor function of BRCA1 and its E3 activity. The ultimate fate of a ubiquitinated substrate depends on both the number of ubiquitins linked in a chain and the ubiquitin chain topology (8). BRCA1/BARD1 has been observed to monoubiquitinate proteins as well as build Lys-6-linked polyubiquitin chains. Lys-6-linked chains are recognized by the 26S proteasome for deubiquitination rather than degradation (9), suggesting ubiquitination by BRCA1/BARD1 may signal for processes other than degradation. To understand the biological function of the BRCA1/BARD1 E3 ligase and the relationship to cancer development, specific BRCA1/BARD1 ubiquitination substrates must be identified.

The BRCA1/BARD1 putative ubiquitination substrates identified to date (histones, γ-tubulin, RNA polymerase II, and CtIP) cannot explain the association of ubiquitously expressed BRCA1 with the development of tissue-specific breast and ovarian tumors (10–14). Recently, however, BRCA1 has been shown to interact with and regulate estrogen receptor α (ERα) and progesterone receptor (PR) transcriptional activation (15–17). Of the >50 proteins known to interact with BRCA1, only ERα and PR have expression profiles similar to BRCA1 mutation-associated tumors. ERα and PR are transcription factors whose activity is modulated by binding the hormones estrogen and progesterone, respectively. Hormone binding elicits a conformational change that enhances receptor dimerization and binding to hormone-responsive DNA elements located in the promoter region of target genes. Transcriptional activation is then promoted through recruitment of coregulators and transcriptional machinery. This results in estrogen and progesterone displaying tissue-selective actions through ERα and PR that are central in the growth, differentiation, and function of the breast and ovaries. Both clinical and experimental evidence indicates that endogenous exposure to female reproductive hormones is a central factor in the development of breast cancer (18, 19). This suggests that regulation of ERα and PR transcriptional activation by BRCA1 could have significant implications in controlling breast tissue proliferation.

Modulation of ERα and PR functions by BRCA1 is a compelling explanation for the linkage between BRCA1 and tissue-specific cancers. The mechanism of this regulation, however, is largely unknown. Both mono- and polyubiquitinated forms of ERα have been observed in vivo (20, 21), with ubiquitination occurring after transcription initiation and functioning to clear ERα from the promoter. The E3 responsible for ERα ubiquitination is unknown. In a recent study, proteasome-mediated degradation was found to regulate PR but not ER in mouse mammary gland tissue, with BRCA1/BARD1 implicated in the formation of polyubiquitin chains on PR (22). Here, we show the ligand-binding domain (LBD) of ERα is predominantly monoubiquitinated by BRCA1/BARD1, suggesting a possible mechanism for regulation of ERα activity by BRCA1. The regions of BRCA1, BARD1, and ERα-LBD necessary for ubiquitination are determined. In addition, the effects of BRCA1 cancer-predisposing mutations and the specificity of BRCA1/BARD1-mediated ERα ubiquitination are evaluated.

Results

ERα Is Ubiquitinated by BRCA1/BARD1.

The interacting regions of BRCA1 and ERα were previously localized to ERα-LBD and BRCA1₃₀₂ (BRCA1 residues 1–302) (16). Similar notation is used throughout the text to indicate the length of BRCA1 and BARD1 constructs, with all BARD1 constructs beginning at Met 26. Consistent with BRCA1₃₀₂ and ERα-LBD interacting, we observe that BRCA1₃₀₄/BARD1₃₂₇ ubiquitinates estrogen-bound ERα-LBD, when reconstituted in vitro with other ubiquitination enzymes. The BRCA1/BARD1 E3 ligase predominantly monoubiquitinates ERα-LBD as determined by Western blotting against both a V5-epitope on ERα-LBD and a T7-epitope on ubiquitin [Fig. 1A and supporting information (SI) Fig. 6]. Monoubiquitination regulates diverse processes ranging from membrane transport to transcriptional regulation; whereas polyubiquitination is known to signal proteins for events such as degradation and DNA repair (23). Therefore, monoubiquitination of ERα-LBD by BRCA1/BARD1 may suggest BRCA1/BARD1 regulates ERα activity by using processes other than degradation, consistent with a recent report of constant ER levels in mouse mammary gland tissue (22). Alternatively, other factors not present in the reconstituted system may be required for the formation of polyubiquitin chains on ERα. Importantly, removal of the V5-epitope on ERα-LBD before and after ubiquitination shows the V5-epitope is neither causal for ERα ubiquitination nor is the epitope ubiquitinated (SI Fig. 6).

Fig. 1.

Regions of BRCA1/BARD1 required for ERα-LBD ubiquitination. (A) Western blot analysis of ERα-LBD ubiquitination by BRCA1₃₀₄/BARD1₃₂₇ (lanes 1–4); BRCA1₃₀₄/BARD1₁₄₀ (lanes 5–8); BRCA1₁₁₂/BARD1₁₄₀ (lanes 9–12); and BRCA1₁₁₂/BARD1₃₂₇ (lanes 13–16). (B) BRCA1₃₀₄/BARD1₁₄₀ (lanes 1–3); BRCA1₂₅₈/BARD1₁₄₀ (lanes 4–6); BRCA1₂₄₁/BARD1₁₄₀ (lanes 7–9); and BRCA1₁₇₇/BARD1₁₄₀ (lanes 10–12). (C) BRCA1₁₁₂/BARD1₃₂₇ (lanes 1–3) and BRCA1₁₁₂/BARD1₁₇₀ (lanes 4–6). All blots were analyzed with anti-V5 to visualize V5-ERα-LBD.

Regions of BRCA1/BARD1 and ERα Necessary for Ubiquitination.

ERα-LBD ubiquitination requires regions beyond the RING domains of BRCA1/BARD1. BRCA1₁₁₂/BARD1₁₄₀, which contains the minimal RING domains and is an active E3 for building free ubiquitin chains (7), is not sufficient for ERα-LBD ubiquitination (Fig. 1A). This suggests specific interactions between BRCA1/BARD1 beyond the RING domains and ERα-LBD are required for ubiquitin transfer. The effects of both BRCA1 and BARD1 on ERα-LBD ubiquitination were determined by using varying lengths of BRCA1 and BARD1. BRCA1₃₀₄/BARD1₁₄₀ and BRCA1₁₁₂/BARD1₃₂₇ both ubiquitinate ERα-LBD, with only BRCA1₁₁₂/BARD1₃₂₇ showing slightly reduced ERα ubiquitination relative to BRCA1₃₀₄/BARD1₃₂₇ (Fig. 1A). This suggests ERα-LBD interacts with regions of both BRCA1 and BARD1 beyond the RING domains for ubiquitination. Because the RING domains of BRCA1/BARD1 contain the E2 binding site (7), these residues are necessary, although not sufficient, for ERα ubiquitination. BRCA1/BARD1 E3 activity depends on heterodimerization (5), therefore the ability of BRCA1 or BARD1 alone to ubiquitinate ERα cannot be determined.

To determine the region of BRCA1 necessary for ERα-LBD ubiquitination, BRCA1 length was varied while keeping BARD1 confined to the RING domain. Shortening BRCA1 to 241 residues results in slightly decreased ERα-LBD ubiquitination, however, decreasing BRCA1 to 177 residues or less greatly reduces or eliminates ERα ubiquitination (Fig. 1 A and B). This suggests BRCA1 residues between 177 and 241 along with the N-terminal RING domain are required for ERα ubiquitination. In addition, BRCA1 residues between 241 and 258 affect the level of ERα ubiquitination (Fig. 1B). Notably, BRCA1 constructs ending between amino acids 177 and 241 degraded rapidly during purification and were not assayed for ERα ubiquitination.

The region of BARD1 necessary for ERα ubiquitination was determined by decreasing the length of BARD1 while keeping BRCA1 confined to the RING domain. ERα ubiquitination mediated through BARD1 interactions require that BARD1 is longer than 170 residues, because truncating BARD1 to residues 26–170 significantly decreases ERα-LBD ubiquitination (Fig. 1C). Combined, these results indicate BRCA1/BARD1-mediated ERα ubiquitination requires regions of BRCA1 between residues 1–241 and regions of BARD1 extending beyond residue 170.

BRCA1/BARD1 ubiquitinates both agonist- and antagonist-bound ERα-LBD (Fig. 2A). Estrogen binds to ERα-LBD and acts as an agonist to promote ERα transcriptional activity, whereas tamoxifen acts as an antagonist in breast tissue (24). Estrogen and tamoxifen bind the same site in ERα, however, agonists and antagonists induce a different conformation of helix 12 at the LBD C terminus (25, 26). This conformational change either allows or inhibits binding of coregulators to the ERα coactivator-binding region (Fig. 2B), which is formed by a shallow hydrophobic groove containing residues from helices 3, 4, and 5 (26). Ubiquitination assays with either estrogen- or tamoxifen-bound ERα-LBD show both liganded states are ubiquitinated by BRCA1/BARD1 (Fig. 2A). Apo ERα-LBD was not amenable to purification and therefore not investigated for in vitro ubiquitination. There are two possible explanations for the ubiquitination of both estrogen- and tamoxifen-bound ERα-LBD. First, BRCA1/BARD1 interacts with a different surface of ERα than the coactivator-binding region, which is affected by the movement of helix 12 in estrogen- and tamoxifen-bound states. Alternatively, crystal structures of ligand-bound ERα indicate that the position of helix 12 is highly flexible, with one structure even showing helix 12 extended away from the main protein core (27). Therefore, the flexible nature of helix 12 may allow BRCA1/BARD1 to interact with the ERα coactivator-binding region and ubiquitinate ERα in both liganded states.

Fig. 2.

ERα-LBD region required for ubiquitination by BRCA1/BARD1. (A) Western blot of BRCA1/BARD1 ubiquitination of estrogen-bound (lanes 1–3; LBD-est) and tamoxifen-bound (lanes 4–6; LBD-tam) ERα-LBD. Lanes 7–9 show ubiquitination of estrogen-bound ERα-LBD I358A/Q375A by BRCA1/BARD1. Blot was analyzed by using anti-V5 antibody against V5-ERα-LBD. (B) Ribbon representation of ERα-LBD (PDB ID code 1ERE) showing the coactivator-binding region (red), I358 and Q375 (blue), helix 12 (labeled), and estrogen (green).

The region of ERα-LBD that interacts with BRCA1/BARD1 for ubiquitination was investigated by scanning mutagenesis, focusing on the ERα coactivator-binding region. Five single mutations in the ERα coactivator-binding region (I358A, Q375A, V376A, L378A, and L379A) were assessed for ubiquitination by BRCA1/BARD1. Of these mutations, both I358A and Q375A slightly weaken BRCA1/BARD1-mediated ERα ubiquitination (data not shown). Double mutation of I358A/Q375A, however, significantly decreases ERα-LBD ubiquitination by BRCA1/BARD1 (Fig. 2A), suggesting mutation of I358 and Q375 in the ERα coactivator-binding region decreases interactions with BRCA1/BARD1 and therefore decreases ubiquitination. An interaction between BRCA1/BARD1 and the ERα coactivator-binding region is consistent with previous studies showing peptides containing the ERα-LBD coactivator-binding region can interact with BRCA1₃₀₂ (28). ERα-I358A/Q375A is native-like, as determined by near- and far-UV CD and a thermal melt similar to WT (data not shown).

Effect of BRCA1 Cancer-Predisposing Mutations on ERα Ubiquitination.

Inheritance of a single BRCA1 mutant allele significantly increases a woman's lifetime risk of developing breast cancer (1). Therefore, effects of BRCA1 RING domain cancer-predisposing mutations on ERα ubiquitination were determined. Both BRCA1-C61G and -C64G are mutations observed in breast cancer patients and have previously been shown to eliminate substrate-independent BRCA1/BARD1 ligase activity (6, 7). Consistent with this, ERα-LBD is not ubiquitinated by either mutant protein (Fig. 3), indicating that these BRCA1 cancer-predisposing mutations disrupt BRCA1/BARD1-mediated ERα ubiquitination.

Fig. 3.

ERα-LBD ubiquitination by WT and mutant BRCA1/BARD1. Western blot analysis against the V5-epitope on ERα-LBD. WT BRCA1₃₀₄/BARD1₁₄₀ (lanes 1–3) ubiquitinates ERα-LBD; BRCA1 cancer-predisposing mutations, C61G (lanes 4–6) and C64G (lanes 7–9), do not. ERα-LBD is not ubiquitinated by the BRCA1 mutant, I26A (lanes 10–12), which maintains E3 structural integrity but disrupts E2 binding (7).

Mutation of Ile 26 to Ala (I26A) in BRCA1 also abrogates ERα-LBD ubiquitination. BRCA1-I26A is a designed mutant protein that is structurally identical to WT BRCA1 and heterodimerizes with BARD1. However, BRCA1-I26A cannot interact with E2 enzymes and therefore is dominant negative for BRCA1 ligase activity (7). BRCA1-I26A/BARD1 fails to ubiquitinate ERα-LBD (Fig. 3), indicating that ERα ubiquitination depends on the E3 ligase activity of BRCA1/BARD1. This is further supported by a lack of ERα ubiquitination in the absence of BRCA1/BARD1 (data not shown).

Substrate Specificity in Vitro.

BRCA1/BARD1 displays substrate specificity for ERα-LBD ubiquitination. Substrate specificity was assessed by using the ERα DNA-binding domain (ERα-DBD) as a BRCA1/BARD1 substrate. Eight of the ≈80 residues in ERα-DBD are Lys, however, in vitro, after 60 min, ERα-DBD is only weakly ubiquitinated by BRCA1₃₀₄/BARD1₃₂₇, whereas significant levels of ERα-LBD are ubiquitinated after 20 min (Fig. 4B). This suggests that, in vitro, the BRCA1/BARD1 ligase shows specificity for the LBD over the DBD. BRCA1/BARD1 substrate specificity was investigated further by competing ERα-LBD ubiquitination with DBD over a 1:1 to 50:1 molar ratio of DBD:LBD (Fig. 4A). ERα-LBD ubiquitination is unchanged at all DBD concentrations investigated, indicating BRCA1/BARD1 specifically ubiquitinates ERα-LBD even with a 50-fold molar excess of DBD present. Therefore, in vitro, BRCA1/BARD1 displays substrate specificity and specific interactions are required for ERα-LBD ubiquitination, rather than a simple diffusion-based mechanism.

Fig. 4.

Specificity of ERα-LBD ubiquitination. (A) BRCA1₃₀₄/BARD1₃₂₇-mediated ubiquitination of 1 μM ERα-LBD competed with 1, 10, or 50 μM ERα-DBD. ERα-LBD is ubiquitinated to a similar extent in the presence of all ERα-DBD concentrations. Ubiquitinated ERα-LBD and -DBD are marked with ■ and ▴, respectively. Unmodified ERα-LBD and -DBD are marked with □ and ▵, respectively. (B) BRCA1₃₀₄/BARD1₃₂₇-mediated ubiquitination of ERα-LBD (lanes 1–3) and ERα-DBD (lanes 4–6). The molecular weight of ubiquitinated ERα-DBD is marked (▴). ERα-DBD is not bound to DNA. (C) E2 specificity of BRCA1/BARD1-mediated ERα-LBD ubiquitination. ERα-LBD is ubiquitinated by BRCA₃₀₄/BARD1₁₄₀ in the presence of UbcH5c (lanes 1–3) but not UbcH7 (lanes 4–6). All Western blots were analyzed by using anti-V5 against V5-ERα-LBD and V5-ERα-DBD.

Substrate specificity was further assessed by the ability of BRCA1/BARD1 to ubiquitinate hen egg white lysozyme, which has previously been used as a generic ubiquitination substrate for E3s (11). Lysozyme is not ubiquitinated by BRCA1/BARD1 in our reconstituted system, confirming that specific interactions between the substrate and BRCA1/BARD1 are required for ubiquitination. Additionally, competition of ERα-LBD ubiquitination with a 50-fold molar excess of lysozyme does not affect ERα-LBD ubiquitination (SI Fig. 7). Furthermore, although the RING domains of BRCA1/BARD1 are an active E3 ligase, they do not ubiquitinate ERα (Fig. 1A), implying E3 specificity for ERα ubiquitination. Simply having an active E3 ligase is not sufficient for ERα ubiquitination; rather ubiquitination requires specific interactions between BRCA1/BARD1 sequences past the RINGs and ERα.

The specificity of BRCA1/BARD1-dependent ERα ubiquitination was also investigated by using two E2 conjugating enzymes, UbcH5c and UbcH7. UbcH5c and UbcH7 are ≈60% similar in protein sequences and nearly identical in structure. Both UbcH5c and UbcH7 can be charged with ubiquitin and interact with the same region of BRCA1/BARD1 (7). However, only UbcH5c builds polyubiquitin chains with BRCA1/BARD1 (5, 7). Similarly, only UbcH5c transfers ubiquitin to ERα (Fig. 4C), indicating that ERα ubiquitination by BRCA1/BARD1 requires interaction between BRCA1 and a specific E2.

Lys 302 in ERα Is Ubiquitinated by BRCA1/BARD1.

The site on ERα ubiquitinated by BRCA1/BARD1 was identified by trypsin digestion and mass spectrometry. Trypsin proteolysis of a ubiquitinated protein creates a distinct signature peptide at the ubiquitination site. This peptide contains a two-residue Gly-Gly remnant from the C terminus of ubiquitin that is covalently attached to the substrate Lys, resulting in a mass shift of 114.1 Da (29). The precise site of ubiquitin modification is determined by HPLC and mass spectrometry. To identify the ERα residue ubiquitinated by BRCA1/BARD1, reactions were analyzed ± ATP. No ubiquitination will occur in the absence of ATP, so peptides observed only with ATP arise from ubiquitination. A ubiquitination site in ERα was mapped to K302 in both + 2 and + 3 charge state mass spectra (Fig. 5A and SI Fig. 8). The peptide sequence and localization of the diglycine ubiquitin remnant on K302 are validated by a SEQUEST (30) XCorr of 3.5 and y- and b-ion series fragments observed for the majority of the sequence (Fig. 5A). In addition, the extracted ion chromatogram for this peptide shows a significant increase in the presence of ATP, further confirming that the peptide is ubiquitinated (Fig. 5B). Trypsin does not cleave at a modified Lys; therefore, ubiquitinated Lys residues will have a missed tryptic cleavage site. As determined from Western blots, ≈20% of the ERα in solution is ubiquitinated, therefore, the corresponding unmodified peptide should have a decrease in intensity between reactions in the absence and presence of ATP. Indeed, the intensity of the extracted ion chromatogram for the peptide containing unmodified K302 decreases ≈20% in the presence of ATP, further confirming K302 as a site of ubiquitin attachment on ERα (Fig. 5C).

Fig. 5.

Lys 302 in ERα is ubiquitinated by BRCA1/BARD1. (A) Tandem mass spectrum of the 2+ charged peptide (m/z = 1,002) STENLYFQGIDPFTK*K. The * indicates that the preceding Lys contains a diglycine ubiquitin remnant left after digestion. The sequence adjacent to ERα Lys 302 is from the N-terminal tag. Observed y- and b-ions are indicated in black. (B and C) Ion chromatograms extracted from μLC-MS runs for (B) modified, m/z = 668–669, and (C) unmodified, m/z = 630–631, +3 charged STENLYFQGIDPFTKK peptide. Solid lines represent mean signal for nine separate μLC-MS runs of in vitro ERα ubiquitination with (black) and without (gray) ATP. Dashed lines indicate ± 1 standard deviation. The gray shaded areas indicate regions that are statistically different (P ≤ 0.001) ± ATP, and the height indicates the −log₁₀ of the most significant P value from a two-tailed t test found over that region.

Mutagenesis was attempted to confirm ubiquitination of ERα K302 by BRCA1/BARD1. Mutation of K302 to Ala still results in ERα-LBD ubiquitination by BRCA1/BARD1. Therefore, each of the 13 ERα-LBD Lys residues was individually mutated to either Ala or Arg; in addition a K302A/K303A double mutation was investigated. BRCA1/BARD1 ubiquitinates each of the ERα-LBD Lys mutants (data not shown), indicating that if the primary ubiquitinated Lys is mutated, another nearby Lys can be ubiquitinated. This phenomenon is consistent with other E3s and ubiquitination substrates in vitro (31). Importantly, identification of ubiquitin on K302 does not preclude the possibility that a different Lys on a separate ERα is ubiquitinated but not detected by mass spectrometry. ERα-LBD-K302A/K303A is, however, still specifically ubiquitinated over the DBD (SI Fig. 9), suggesting that the observed specificity for ERα-LBD is not a result of having a Lys adjacent to the N-terminal tag. Finally, ERα is predominantly monoubiquitinated by BRCA1/BARD1, indicating that ubiquitination of a single Lys prevents other Lys residues within the same molecule from being ubiquitinated.

Discussion

Identifying the ubiquitination substrates of BRCA1/BARD1 is central to understanding the biological function of the BRCA1/BARD1 E3 ligase and its relationship to cancer development. Here, we show ERα-LBD is a putative ubiquitination substrate of BRCA1/BARD1. Our findings can be summarized as follows: (i) ERα-LBD is predominantly monoubiquitinated by BRCA1/BARD1 in vitro. (ii) BRCA1/BARD1-mediated ERα ubiquitination requires interactions between the ERα-LBD coactivator-binding region and BRCA1/BARD1 beyond the RING domains. (iii) ERα-LBD ubiquitination by BRCA1/BARD1 displays substrate specificity and E2 specificity in vitro. (iv) Both estrogen- and tamoxifen-bound ERα are ubiquitinated by BRCA1/BARD1. (v) BRCA1 cancer-predisposing mutations C61G and C64G and a designed mutant I26A do not ubiquitinate ERα-LBD. (vi) BRCA1/BARD1 ubiquitinates K302 of ERα.

The biological consequence of ubiquitination, the most famous of which is targeting substrates with polyubiquitin chains to the proteasome for degradation, depends on both the number of ubiquitins added to a substrate and ubiquitin chain topology (8). In contrast, substrate monoubiquitination regulates diverse proteins including histones, endocytic machinery, and transcription factors (23). In our study, we observed predominant monoubiquitination of ERα-LBD by BRCA1/BARD1 (Fig. 1A), consistent with previous observations of in vitro monoubiquitination of histones and γ-tubulin by BRCA1/BARD1 (10–12). Monoubiquitination of transcriptional activators has been suggested to be principal in regulating activator activity (32), suggesting ERα monoubiquitination may be functionally relevant. Alternatively, additional factors not present in the reconstituted system may be necessary for chain elongation on BRCA1/BARD1 substrates.

The putative region of ERα ubiquitinated by BRCA1/BARD1 is implicated in regulating ERα transcriptional activity. ERα-LBD is ubiquitinated by BRCA1/BARD1 at K302 as determined by mass spectrometry (Fig. 5). It is likely that BRCA1/BARD1 also ubiquitinates the adjacent residue K303, as ubiquitination enzymes do not display sufficient fidelity to ubiquitinate a particular Lys preferentially over an adjacent Lys. Ubiquitinated K303 was likely not detected by mass spectrometry because a missed trypsin cleavage at K303 would generate a large peptide with ≥4+ charge state within the measured m/z range. We did not consider peptides ≥4+ in our analysis because peptides of this length are underrepresented in analyses using low-resolution tandem mass spectrometry and generally are ignored in proteomics experiments (33, 34). K302 and K303 are located in the hinge that connects the ERα-DBD and -LBD. The hinge region has 43% identity from Homo sapiens to Xenopus laevis, whereas ERα-DBD and -LBD have 99% and 82% identity over the same species, respectively. Intriguingly, ERα hinge residues between K302 and S305, immediately before the start of the LBD, are ≈85% conserved between Homo sapiens and Xenopus laevis. The higher conservation of K302-S305 relative to the rest of the hinge suggests this region may be functionally important. Indeed, several modifications have been observed at these residues: sumoylation of K302 and K303, acetylation of K303, and phosphorylation of S305 (35, 36). In addition, a K303R mutation has been observed in as many as 34% of premalignant breast tumors, whereas WT ERα is expressed in normal tissue. The K303R mutation renders breast cells hypersensitive for proliferation in response to low doses of estrogen (37). The hypersensitivity has been suggested to result from a lack of acetylation at K303 and increased phosphorylation at S305 (36). However, inability of K303R to be ubiquitinated could also contribute to proliferation hypersensitivity.

BRCA1/BARD1-mediated ERα ubiquitination may represent the regulatory mechanism for repression of ERα transcriptional activation by BRCA1. This inhibition has been suggested to occur from an interaction between BRCA1₃₀₂ and ERα-LBD (16, 28). These studies found that the BRCA1 RING domain (residues 1–67) is not required for the observed interaction with ERα, and that the cancer-predisposing mutation, C61G, in full-length BRCA1, does not disrupt the BRCA1–ERα interaction. However, in apparent contradiction to these observations, BRCA1-C61G fails to repress ERα transcriptional activity (16). ERα-LBD ubiquitination by BRCA1/BARD1 requires the BRCA1/BARD1 RING domains and at least 241 residues of BRCA1 (Fig. 1). However, BRCA1-C61G disrupts ERα-LBD ubiquitination (Fig. 3). The observation that BRCA1-C61G interacts with ERα-LBD but does not repress ERα activity, together with the loss of ERα ubiquitination by BRCA1-C61G suggests ubiquitination of ERα by BRCA1/BARD1 could regulate ERα transcriptional activity.

Central to understanding the biological function of the BRCA1/BARD1 ligase and its relationship to cancer is the identification of BRCA1/BARD1 ubiquitination substrates. Several putative BRCA1/BARD1 ubiquitination substrates have been identified including histones, γ-tubulin, RNA polymerase II, and CtIP (10–14). Although each has advanced our understanding of the BRCA1/BARD1 E3 ligase, none provide a clear correlation between ubiquitously expressed BRCA1 and the development of tissue-specific cancers. Here, we show ERα is a putative substrate for the BRCA1/BARD1 E3 ligase, providing a compelling link between the loss of BRCA1 ligase activity and BRCA1-associated tissue-specific tumorigenesis. Exposure to estrogen is known to be a central factor in the development of breast cancer (19); therefore, regulation of ERα activity by BRCA1, possibly through ERα ubiquitination, could have significant implications in controlling breast tissue proliferation. Further studies investigating the biological role of BRCA1/BARD1-mediated ERα ubiquitination will surely reveal additional insights regarding the impact of these events on breast and ovarian cell proliferation.

Materials and Methods

Plasmids.

Plasmids for expression of BRCA1, BARD1, UbcH5c, UbcH7, ubiquitin, and Uba1 were derived from those described previously (7, 38). ERα-LBD (residues 302–552) and ERα-DBD (residues 180–262) were cloned into pET151D (Invitrogen, Carlsbad, CA). Point mutations were introduced by QuikChange (Stratagene, La Jolla, CA) and confirmed by DNA sequencing (University of Washington Biochemistry Sequencing).

Protein Expression and Purification.

All proteins were expressed in Escherichia coli as described (38). Briefly, pCOT7-His₆-FLAG-BRCA1 and pET28N-BARD1 of specified lengths were cotransformed and grown in BL21star(DE3). BRCA1/BARD1 was purified by Ni²⁺-affinity and size exclusion chromatography (Superdex 75 or Superdex 200) in 25 mM sodium phosphate, 150 mM NaCl, pH 7.0. BRCA1/BARD1 was used within 12 h of purification. pET28N-UbcH5c was transformed and grown in BL21(DE3) cells and purified by cation exchange (SP Sepharose) eluted with a 0–0.5 M NaCl gradient in 30 mM Mes and 1 mM EDTA (pH 6.0). UbcH5c fractions were purified further by size exclusion chromatography. pET28N-UbcH7 was transformed and grown in Rosetta cells (Novagen, Madison, WI) and purified as UbcH5c. pET28N-His₆-Uba1 was transformed and grown in BL21(DE3) cells and purified by Ni²⁺-affinity and anion exchange (Poros HQ; Applied Biosystems, Foster City, CA) eluted with a 0–3 M NaCl gradient. pET28N-His₆-T7-ubiquitin was transformed and grown in BL21(DE3) cells and purified as BRCA1/BARD1.

ERα-LBD was grown in the presence of 20 μM estrogen or tamoxifen in BL21(DE3) cells transformed with pET151D-ERα-LBD. Inclusion bodies were solubilized with 5 M urea, 25 mM Tris, 200 mM NaCl, 10 mM NH₄Cl, 20 μM estradiol or tamoxifen, pH 7.6, 4°C for 12 h. Soluble material was purified by Ni²⁺-affinity in 5 M urea and the eluant was refolded by serial dialysis at 4°C. ERα-LBD was purified by size exclusion chromatography (Superdex 75; Amersham Biosciences, Piscataway, NJ) in 25 mM sodium phosphate, 150 mM NaCl, pH 7.4. ERα-LBD is native-like as assessed by ¹H NMR, a two state unfolding transition, and near- and far-UV CD. ERα-LBD is a homodimer as determined by analytical ultracentrifugation, size exclusion chromatography, and sucrose gradient (data not shown). ERα-DBD was grown in BL21(DE3) cells transformed with pET151D-ERα-DBD and purified as BRCA1/BARD1.

Ubiquitination Assays.

Unless otherwise stated, ubiquitination reactions contain 1 μM estrogen-bound His₆-V5-ERα-LBD, 1 μM His₆-FLAG-BRCA1/BARD1 of lengths specified in the text, 1 μM UbcH5c, 0.5 μM Uba1, and 20 μM T7-ubiquitin at 37°C, pH 7.0. Reactions were initiated with 2.5 mM ATP, 5 mM MgCl₂, pH 7.0, and quenched with SDS/PAGE buffer. Products were resolved on SDS/12% PAGE gels and transferred to PVDF membranes (Bio-Rad, Hercules, CA). Membranes were blotted with anti-V5 (Invitrogen) or anti-T7 (Novagen) antibodies from mouse, followed by goat anti-mouse conjugated to Alexa Fluor680 (Molecular Probes, Eugene, OR). Results were detected on an Odyssey infrared imaging system (Licor, Lincoln, NE). All assays were repeated at least three times with at least two separate protein purifications.

Mass Spectrometry.

After 2 h at 37°C, ubiquitination reactions were denatured with RapiGest (Waters, Milford, MA) and digested with trypsin (Promega, Madison, WI). Peptides were separated by using an Agilent 1100 binary HPLC on a 75 μm ID microcapillary column packed in-house with 40 cm of C₁₂ Proteo Jupiter 4u, 90A (Phenomenex, Torrance, CA). Peptides were eluted by acetonitrile gradient and electrosprayed into an LTQ linear ion trap mass spectrometer (ThermoFinnigan, San Jose, CA). MS/MS spectra were acquired by using data-dependent acquisition. Regions of m/z and time that had differences in intensity between ubiquitination reactions ± ATP were identified by using in-house developed software. Details of the software will be reported elsewhere. Peptide identities were determined by database searching with SEQUEST (30).

Abbreviations

ERα

estrogen receptor α

E1

ubiquitin-activating enzyme

E2

ubiquitin-conjugating enzyme

E3

ubiquitin ligase

PR

progesterone receptor

LBD

ligand-binding domain

DBD

DNA-binding domain.