Histone Ubiquitination Associates with BRCA1-Dependent DNA Damage Response^{▿ †}

Plasmids, antibodies, and other materials.

S-Flag-biotin (SFB)-tagged full-length RAP80, a UIM domain deletion mutant of RAP80 (ΔUIM), a zinc finger domain deletion mutant of RAP80 (ΔZnF), and SFB-tagged full-length HSJ1A were described previously (25). The RAP80 UIM domain (amino acids [aa] 1 to 200), RAP80 ΔUIM (aa 73 to 126 deleted), RAP80 ΔUIM1 (aa 78 to 96 deleted), and RAP80 ΔUIM2 (aa 103 to 120 deleted) were cloned into the pGEX-4T-1 vector (Amersham) to generate glutathione S-transferase (GST) fusion proteins. H2A and H2B cDNAs were subcloned into a modified pCDNA3 vector to generate constructs encoding hemagglutinin (HA)-tagged H2A and H2B. The point mutants of H2A and H2B were generated by using the QuikChange site-directed mutagenesis kit (Stratagene). Rabbit anti-mouse RNF8, RAP80, and BRCA1 polyclonal antibodies were raised against GST-RNF8 (aa 1 to 324), GST-RAP80 (aa 1 to 354), and GST-BRCA1 (aa 1445 to 1812) fusion proteins, respectively. Rabbit anti-human RAP80, BRCA1, and phospho-H2AX antibodies were previously described (25). Antibodies to H2A, the ubiquitinated form of H2A (ub-H2A), H2B, and H4 antibodies were purchased from Upstate. Anti-HA and anti-β-actin antibodies were purchased from Covance and Sigma. Anti-RING1B and anti-RNF20 antibodies were purchased from MBL and Bethyl, respectively.

The small interfering RNA (siRNA) duplexes were purchased from Dharmacon Research (Lafayette, CO). The siRNA sequences targeting Ring1B and RNF20 were 5′-AAC UCA GUU UAU AUG AGU UAC-3′ and 5′-AAG AAG GCA GCU GUU GAA GAU-3′, respectively. siRNAs were transfected into the cells using Oligofectamine (Invitrogen) according to the manufacturer's instructions.

Cell culture and treatment with ionizing radiation.

293T and HeLa cells were cultured in RPMI 1640 medium with 10% fetal bovine serum. RNF8-deficient mouse embryo fibroblasts (MEFs) were cultured in Dulbecco modified Eagle medium with 10% fetal bovine serum. For ionizing radiation (IR) treatment, cells were irradiated at the indicated doses by using a JL Shepherd ¹³⁷Cs radiation source. Cells were then maintained in the culture conditions for the time points specified in the figure legends. H2AX^−/− MEFs were a gift from Andre Nussenzweig.

Cell lysis, immunoprecipitation, GST pull-down assay, and Western blotting.

Cells were lysed with NETN buffer (0.5% NP-40, 50 mM Tris-HCl [pH 8.0], 2 mM EDTA, and 100 mM NaCl). For immunoprecipitation and GST pull-down assay, insoluble lysates were collected and sonicated, followed by micrococcal nuclease (Sigma) treatment at room temperature for 10 min. Chromatin-associated proteins were eluted for further analyses. For acid extraction of histones, insoluble pellets were resuspended in 0.2 M HCl; the acid was neutralized with 1 M Tris-HCl (pH 8.5) for further Western blot analysis. Immunoprecipitation, GST pull-down assay, and Western blotting were performed following standard protocols as described previously (25).

Immunofluorescence staining.

Cells grown on coverslips were fixed in 3% paraformaldehyde for 20 min and permeabilized in 0.5% Triton X-100 in phosphate-buffered saline (PBS) for 5 min at room temperature. Samples were blocked with 5% goat serum and then incubated with primary antibody for 60 min. Samples were washed three times and incubated with secondary antibody for 30 min. The coverslips were mounted onto glass slides and visualized with a fluorescence microscope. For ubiquitinated H2A staining, cells were treated 0.5% Triton X-100 for 5 min before fixation. To visualize IR-induced foci, cells were cultured on coverslips and treated with 10 Gy IR (1 Gy = 100 rads), followed by recovery for 4 h.

Construction of the lysine-less H2AX mutant.

The lysine-less H2AX mutant (NOK) was constructed by overlap PCR to change all lysine resides to arginine residues using primer pairs KO1F (5′ ATG TCG GGC CGC GGC AGG ACT GGC GGC AGG GCC CGC GCC AGG GCC AGG TCG CGC TCG TCG CGC GCC GGC CTC CAG TTC CCA GTG GGC CGT GTA CAC CGG 3′) and KO1B (5′ AGC GGT GAG GTA CTC CAG CAC TGC CGC CAG GTA CAC TGG CGC GCC GGC GCC AAC GCG CTC GGC GTA GTG GCC CCT CCG CAG CAG CCG GTG TAC ACG G 3′), KO2F (5′ AGT ACC TCA CCG CTG AGA TCC TGG AGC TGG CGG GCA ATG CGG CCC GCG ACA ACA GGA GGA CGC GAA TCA TCC CCC GCC ACC TGC AGC TGG 3′) and KO2B (5′ ATG TTG GGC AGG ACG CCT CCC TGG GCG ATC GTC ACG CCG CCC AGC AGC CTG TTG AGC TCC TCG TCG TTG CGG ATG GCC AGC TGC AGT GG 3′), and KO3F (5′ CGT CCT GCC CAA CAT CCA GGC CGT GCT GCT GCC CAG GAG GAC CAG CGC CAC CGT GGG GCC GAG GGC GCC CTC G 3′) and KO3B (5′ TTA GTA CTC CTG GGA GGC CTG GGT GGC CCT CCT GCC GCC CGA GGG CGC CCT CGG 3′). Thereafter, wild-type H2AX, S139A mutant, or NOK mutant cDNA was PCR amplified using primers with flanking restriction sites and were subcloned into the pBabe-puro retroviral vector and packaged in BOSC23 cells for infecting H2AX-deficient cells.

Affinity purification of SFB-tagged RAP80 and mass spectrometry.

SFB-tagged RAP80 was stably expressed in K562 cells. Two-liter cultures of cells were harvested and treated with or without 20 Gy IR. At 4 h after IR treatment, cells were lysed with 40 ml NETN buffer on ice for 10 min. Cell lysates were centrifuged at 12,000 × g at 4°C for 20 min. The soluble fraction was collected. The insoluble fraction was washed three times with PBS and then treated with 50 units of micrococcal nuclease on ice for 1 h and centrifuged at 12,000 × g at 4°C for 20 min. The supernatant was the chromatin fraction combined with the NETN soluble faction. The cell lysates were incubated with 500 μl streptavidin-conjugated beads (Amersham) at 4°C for 2 h. The beads were washed three times with NETN buffer, and then bead-bound proteins were eluted with 1 ml PBS containing 2 mM biotin (Sigma). The eluted supernatant was incubated with 50 μl S beads (Novagen) at 4°C for 2 hours. The beads were washed three times with NETN buffer and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The gels were digested, and the peptides were analyzed by liquid chromatography-tandem mass spectrometry. (The common nonspecific associated proteins, including acetyl coenzyme A carboxylase 1, pyruvate carboxylase, methylcrotonoyl coenzyme A carboxylase subunit alpha, and heat shock protein 70, are not listed in the mass spectrometry results shown in Table 1, but these proteins were present in each purification.)

Histones H2A and H2B are ubiquitinated following DNA damage.

To examine the molecular mechanism by which BRCA1 is recruited to DNA damage sites, we focused our studies on RAP80, since the RAP80 UIM domain plays a key role in BRCA1 complex focus formation following DNA damage. Previously, we have found that RAP80 associated with chromatin following DNA damage (25), suggesting that there are RAP80 binding partners on the chromatin. By using protein affinity purification, we analyzed RAP80-associated proteins. Besides three known RAP80 partners, CCDC98, BRCC45, and BRCC36, we identified redundant peptides of histones H2A and H2B from the same purification (Table 1), suggesting that RAP80 may interact with histones. In fact, histones H2A and H2B are the most abundant ubiquitinated proteins on the chromatin; 10 to 15% of histone H2A and 1 to 5% of H2B are ubiquitinated under normal conditions. Thus, we hypothesize that the RAP80 UIM domain recognizes ubiquitinated histones H2A and H2B at the DNA damage sites, which loads the BRCA1 complex to DNA damage lesions on the chromatin. To examine our hypothesis, we first determined whether histones H2A and H2B are ubiquitinated at DNA damage sites. HeLa cells were treated with 10 Gy of IR. By using immunofluorescence staining with an antibody against ubiquitinated H2A, we found that the ubiquitinated H2A formed DNA damage-induced foci, which colocalized with phospho-H2AX (γH2AX) foci, a marker of DNA damage sites (Fig. 1A). This suggests that H2A is highly ubiquitinated at the DNA damage sites, which also has been recently reported by other groups (5, 33, 40). H2AX, a variant of H2A that controls DNA damage-induced focus formation by various of other proteins, was also recently shown to be ubiquitinated following DNA damage (19, 22, 68). To exclude the possibility that the observed ubiquitinated H2A foci could be solely ubiquitinated H2AX foci, due to a potential risk that ubiquitinated H2A antibody might also recognize ubiquitinated H2AX, we utilized an H2AX mutant (NOK) with all lysine residues mutated to arginine, which could not be ubiquitinated before and after DNA damage (see Fig. S1 in the supplemental material). H2AX^−/− cells were reconstituted with either wild-type H2AX or the NOK mutant. DNA damage-induced ubiquitinated H2A foci could still be observed in cells expressing the nonubiquitinatable NOK mutant, suggesting that ubiquitinated H2A contributes to at least part, if not all, of the foci detected by ubiquitinated H2A antibody (Fig. 1B). Moreover, these ubiquitinated H2A foci were controlled by γH2AX, since reintroducing the S139A mutant of H2AX into H2AX^−/− cells that had H2AX phosphorylation abolished following DNA damage also disrupted damage-induced ubiquitinated H2A foci (Fig. 1B).

Due to antibody limitation, we could not directly assess ubiquitinated H2B foci. Instead, we examined H2B ubiquitination following DNA damage by Western blotting. Both 293T and HeLa cells were treated with 0, 10, or 20 Gy of IR. Global H2B ubiquitination increased with the increase of IR dose (Fig. 1C), suggesting that DNA damage induces H2B ubiquitination. However, we did not observe any significant increase in H2A ubiquitination following DNA damage in the Western bolt analyses (Fig. 1C), which could probably be masked by a high endogenous H2A ubiquitination level without DNA damage. Nevertheless, the observed ubiquitinated H2A foci indicate that ubiquitinated H2A is highly concentrated at double-strand breaks after IR, making them distinct from background ubiquitinated H2A present before DNA damage. Since only ∼1% of H2B was ubiquitinated before IR treatment, we could observe the increase in the total ubiquitinated H2B level with a relatively high dose of IR. We also examined the time course of H2B ubiquitination following DNA damage and found that at 5 h after IR treatment, H2B ubiquitination increased by a maximum of twofold (Fig. 1D; see Fig. S2 in the supplemental material). Taken together, these results demonstrate that histone H2A and H2B are ubiquitinated following DNA damage. DNA double-strand breaks also induce most RAP80 association with chromatin (25). The increase of chromatin-bound RAP80 correlates with IR-induced H2B ubiquitination, indicating that RAP80 may recognize ubiquitinated histones following DNA damage (Fig. 1D; see Fig. S2 in the supplemental material).

RAP80 interacts with monoubiquitinated histones H2A and H2B.

Next we examined whether RAP80 recognizes ubiquitinated H2A and H2B. Even without DNA damage, a small portion of RAP80 already associates with chromatin (25). Thus, we first examined whether RAP80 interacts with ubiquitinated H2A and H2B on the chromatin under normal conditions. Coimmunoprecipitation results indicated that RAP80 associated only with the ubiquitinated form of H2A and not with unmodified H2A (Fig. 2A; see Fig. S3 in the supplemental material). Although a small amount of unmodified H2B associated with RAP80, RAP80-associated ubiquitinated H2B was dramatically enriched in this coimmunoprecipitation assay (Fig. 2A), suggesting that RAP80 associates with both ubiquitinated histones H2A and H2B on the chromatin. The reverse coimmunoprecipitation also confirmed these results (Fig. 2B). Since the UIM domain of RAP80 recognizes ubiquitinated proteins (1, 25, 51, 59, 64), we deleted the UIM domain (aa 73 to 126) and examined whether a UIM domain deletion mutation of RAP80 (ΔUIM) could abolish its interaction with ubiquitinated histones H2A and H2B. As shown in Fig. 2C, only wild-type RAP80, and not the ΔUIM mutant, interacted with ubiquitinated histones H2A and H2B, suggesting that it is the RAP80 UIM domain that recognizes ubiquitinated H2A and H2B. Ubiquitinated histones are the most abundant ubiquitinated proteins in the cell. To examine the specificity of the interaction between the RAP80 UIM domain and ubiquitinated histones, we also checked HJS1A, a UIM domain-containing protein in the nucleus. However, only RAP80, and not HJS1A, could associate with ubiquitinated histones, indicating that the UIM domain of RAP80 specifically recognized ubiquitinated histones (see Fig. S4 in the supplemental material).

Following IR treatment, most RAP80 associates with chromatin (25). Since the RAP80 UIM domain recognizes ubiquitinated histones H2A and H2B, we examined whether the interaction between RAP80 and ubiquitinated H2A and H2B was increased after DNA damage. As expected, DNA damage induced RAP80's association with ubiquitinated H2A and H2B in the chromatin fraction (Fig. 2D). This is consistent with the purification results for RAP80-interacting proteins (Table 1), suggesting that DNA damage-induced histone ubiquitination recruits RAP80 to the chromatin.

In this RAP80 UIM domain, there are two tandem UIMs. These two UIMs have conserved residues to interact with ubiquitin, but they differ in the flanking regions (Fig. 3A), indicating that these two UIMs may have different affinities for different ubiquitinated histones (18). We performed a GST pull-down assay to examine this possibility. The GST-UIM domain, but not its mutant with both UIMs deleted (GST-ΔUIM 1&2), could pull down ubiquitinated histones H2A and H2B, further confirming that RAP80 UIMs interact with ubiquitinated histones. However, deletion of either UIM abolished the interaction between the UIM domain and ubiquitinated H2A and H2B (Fig. 3B), suggesting that both UIMs have to cooperate together to interact with either ubiquitinated H2A or ubiquitinated H2B.

In the coimmunoprecipitation assay, we noticed that RAP80 associated with a small amount of nonubiquitinated H2B (Fig. 2A), suggesting that other regions of RAP80 could also interact with H2B regardless of whether it is ubiquitinated or not. Such interaction may determine the binding specificity of the RAP80 UIM domain with ubiquitinated H2A and H2B over other ubiquitinated proteins. Besides the N-terminal UIM domain, RAP80 also contains two other domains: the Abraxas/CCDC98-interacting region (AIR) in the middle and the C-terminal zinc finger domain (58). We generated GST fusion proteins of all three domains and performed a pull-down assay by using nonubiquitinated free histones. We found that only the zinc finger domain, and not the others, specifically interacted with H2B, but not H2A or H4 (Fig. 3C). Moreover, abolishing the Zn binding ability of this zinc finger domain by mutating a conserved cysteine to alanine (C508A) totally disrupted this interaction (Fig. 3D). We postulate that the interaction between the RAP80 zinc finger domain and H2B further increases the affinity between RAP80 and ubiquitinated H2B. To further analyze the function of the zinc finger domain of RAP80 in the DNA damage response, we examined IR-induced focus formation of mutant RAP80 in the absence of the zinc finger domain (ΔZnF). Compared with wild-type RAP80, which formed IR-induced foci within 15 minutes, the ΔZnF mutant had significantly delayed IR focus formation and started to relocate to DNA damage sites after 1 hour of DNA damage (Fig. 3E). Consistent with our and other previous reports (1, 25, 51, 64), mutant RAP80 without UIM failed to relocate to DNA damage sites (Fig. 3E). These results indicate that the RAP80 zinc finger domain facilities IR-induced RAP80 focus formation, possibly by additional interaction with H2B other than with the UIM domain.

Histone ubiquitination controls RAP80/BRCA1 complex association with chromatin following DNA damage.

To further examine the role of histone ubiquitination in targeting the RAP80/BRCA1 complex to DNA damage sites, we planned to abolish H2A and H2B ubiquitination in vivo. Recent studies have identified that an E3 ligase, RNF8, controls H2A ubiquitination, especially DNA damage-induced H2A ubiquitination (19, 33). We have generated RNF8-deficient MEFs (35). To our surprise, not only 80% of H2A ubiquitination but also 90% of H2B ubiquitination was lost in RNF8^−/− MEFs (Fig. 4A), suggesting that RNF8 not only controls H2A ubiquitination but also regulates H2B ubiquitination. The loss of ubiquitinated H2A and H2B correlated with the reduction of chromatin-associated RAP80 in RNF8^−/− cells (Fig. 4B), further confirming the interaction between RAP80 and ubiquitinated histones H2A and H2B. Consistently, DNA damage-induced H2A and H2B ubiquitination was also abrogated in the absence of RNF8 (Fig. 4C and D), and this in turn abolished the damage-induced association of RAP80 and BRCA1 with chromatin in RNF8-deficient cells (Fig. 4E). These data together suggest that ubiquitinated H2A and H2B could be docking sites for recruiting the RAP80/BRCA1 complex to the chromatin following DNA damage.