Oncogene (2000) 19, 4441 ± 4445 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc SHORT REPORT The BRCA2 activation domain associates with and is phosphorylated by a cellular protein kinase

Jonathan Milner1,3, FrancËois Fuks1,3, Luke Hughes-Davies1,2 and Tony Kouzarides*,1

1Wellcome/CRC Institute and Department of Pathology, Cambridge University, Tennis Court Road, Cambridge, CB2 1QR, UK; 2CRC Human Cancer Genetics Research Group, Addenbrooke's Hospital, Cambridge, UK

A substantial proportion of familial breast cancers have In addition to a role in DNA repair, several lines of mutations within the BRCA2 gene. The product of this evidence suggest that BRCA2 may function to regulate gene has been implicated in DNA repair and in the transcription. We have previously shown that the regulation of transcription. We have previously identi®ed amino-terminal exon 3 of BRCA2 has the ability to at the amino-terminus of BRCA2 a transcriptional stimulate transcription when fused to GAL4-DNA activation domain whose importance is highlighted by binding domain (Milner et al., 1997). The activation the presence of predisposing mutations and in-frame domain within exon 3 was found to be bipartite, with a deletions in breast cancer families. This activation primary activating region (residues 18 ± 60) and an domain shows sequence similarity to a region of c-Jun auxiliary activating region (residues 60 ± 105). Impor- which has been de®ned as a binding site for the c-Jun N- tantly, the introduction of a missense mutation in exon terminal kinase. Here, we show that the analogous 3 (mutation Y42C) which is detected in familial breast region in BRCA2 is also a binding site for a cellular cancer disrupts the activation potential of BRCA2 kinase, although this kinase is distinct from JNK. The (Milner et al., 1997). This ®nding suggest that BRCA2 associated enzyme is able to phosphorylate elimination of BRCA2 transactivation capacity might residues within the BRCA2 activation domain. Consis- be an important event in the generation of a subset of tent with this observation, we ®nd that the activation inherited breast cancers. This concept is reinforced by domain of BRCA2 is phosphorylated in vivo. Our results the recent identi®cation of a deletion disrupting the indicate that the BRCA2 activation domain possesses a whole BRCA2 exon 3 as the disease-causing mutation binding site for a kinase that may regulate BRCA2 in a breast cancer family (Nordling et al., 1998). It is activity by phosphorylation. Oncogene (2000) 19, worth mentioning that not only fragments of BRCA2 4441 ± 4445. are able to regulate transcription but that also the intact protein is able to do so. Indeed, exogenous Keywords: BRCA2; protein kinase; phosphorylation expression of full-length BRCA2 was found to regulate p53's transcriptional activity (Marmorstein et al., 1998). Further evidence that BRCA2 may have a role in transcriptional regulation comes from our recent An estimated 5 ± 10% of all breast cancer cases result observation that BRCA2 N-terminus interacts with a from a hereditary predisposition to the disease. Germ- transcriptional co-activator protein, P/CAF, and dis- line mutations in the human BRCA2 gene account for plays associated acetyltransferase activity when bound a substantial proportion of familial, early-onset breast to it (Fuks et al., 1998). cancers (Wooster et al., 1995; Tatvtigian et al., 1996). We have previously recognized that the auxiliary BRCA2 has exons and encodes a very large protein of activating region of exon 3 (residues 60 ± 105) shows 3418 amino-acids (Tatvigian et al., 1996). The func- sequence similarity to the transcription factor c-Jun tions of BRCA2 remain unclear but accumulating (Milner et al., 1997). In particular, the homology spans evidence suggest that it may have a role in DNA the d-region of c-Jun activation domain which contains repair. BRCA2 nullizygous mouse embryos are highly the binding site for the c-Jun kinase JNK (Hibi et al., sensitive to ionizing radiation (Sharan et al., 1997) and 1993; De rijard et al., 1994) (see Figure 1a). In the are defective in DNA repair (Suzuki et al., 1997; present study, we have further investigated this Connor et al., 1997; Patel et al., 1998). Moreover sequence similarity of exon 3 activation domain to several groups have reported an interaction between the c-Jun d-region. We ®nd that BRCA2 exon 3 does BRCA2 and Rad51, a protein involved in homologous not bind to JNK but instead binds a distinct kinase. recombination and DNA double-strand break repair We also show that this protein kinase phosphorylates (Sharan et al., 1997; Mizuta et al., 1997; Wong et al., BRCA2 and that the phosphorylation event requires 1998). The sites of Rad51 interaction in BRCA2 have the activation domain of BRCA2. These results suggest been ®ne-mapped to the eight evolutionarily conserved that the mechanism by which BRCA2 activity is BRC motifs (Wong et al., 1998; Chen et al., 1998) and regulated may be, at least partly, through its to a C-terminal region (Sharan et al., 1997; Mizuta et association with, and phosphorylation by protein al., 1997). kinase. The sequence homology we have previously noted between the activation domain of BRCA2 and the *Correspondence: T Kouzarides JNK-binding site in c-Jun prompted us to examine 3These authors contributed equally to this work whether JNK is binding to BRCA2. Given that JNK Received 25 May 2000; revised 10 July 2000; accepted 10 July 2000 can bind to c-Jun and phosphorylate it, we asked if Phosphorylation of BRCA2 activation domain by a cellular kinase J Milner et al 4442

Figure 1 BRCA2 c-Jun homology region does not bind to JNK1. (a) BRCA2 exon 3 activation domain shows sequence similarity with the JNK-binding site in c-Jun (d-region). The sequence alignment is shown between BRCA2 activation domain from amino- acids 48 ± 105 and c-Jun activation domain from amino-acids 1 ± 55, which encompass the JNK-binding site (d-region). Residues with identity or similarity are shaded. Below the alignment is shown a schematic representation of BRCA2 with its amino-terminal exon 3 underlined. Exon 3 contains, as previously described (Milner et al., 1997), a bipartite activation domain with a primary activating region (PAR, residues 18 ± 60) and an auxiliary activating region (AAR, residues 60 ± 105). (b) A solid-phase kinase assay was used to test for binding of BRCA2 to JNK1. Recombinant JNK1 was incubated with equivalent amounts of bacterially expressed GST (lane 1) or GST BRCA2 60 ± 105 (containing the c-Jun homology region; lane 2) or, as a positive control, GST c-Jun 1 ± 79 (lane 3). After 1 h at 4 8C, the beads were washed in kinase bu€er (25 mM HEPES pH 7.4, 5 mM b-mercaptoethanol, 10 mM 32 MgCl2,3mM Na3VO4). The beads were then incubated with 1 mCi for [g- P]ATP in kinase bu€er for 15 min at 308C. The phosphorylated proteins were separated by SDS ± PAGE and visualized by autoradiography. Expression and puri®cation of GST fusions were done as previously reported (Fuks et al., 2000)

BRCA2 can bind JNK and become phosphorylated by GST BRCA2 60 ± 105 no phosphorylation by recombi- it. To this end, we performed solide-phase kinase nant JNK1 was detected (Figure 1b, lane 2). Thus, assays (Hibi et al., 1993; De rijard et al., 1994), in which these data indicate that JNK1 does not bind to JNK was incubated with immobilized-bacterially ex- BRCA2 activation domain and are in agreement with pressed GST BRCA2, extensively washed and radi- May et al. (1999). olabelled ATP added to test for phosphorylation of JNK1 belongs to the JNK family of protein kinases, BRCA2. The N-terminal activation domain of c-Jun which includes at least 10 isoforms, and which are all (residues 1 ± 79) linked to GST was used as a positive able to bind to and phosphorylate the c-Jun d-domain control. As expected, GST c-Jun 1 ± 79 was eciently (Gupta et al., 1996). Thus, although BRCA2 was not phosphorylated by recombinant JNK1, indicating that binding to JNK1, it was still possible that another JNK there was an association between c-Jun and JNK1 family member, or even a distinct kinase with a similar (Figure 1b, lane 3). No phosphorylation of the control recognition site to JNK, was binding to and GST alone was observed (lane 1). In contrast to GST phosphorylated the BRCA2 c-Jun homology domain. c-Jun, residues 60 ± 105 of BRCA2 (encompassing the To address this possibility, we carried out solide-phase sequence homology with the c-Jun d-domain) failed to kinase assays as described above, this time with Hela associate with JNK1 kinase activity. Indeed, when we nuclear extracts. Figure 2a (lane 1) shows that used the solide-phase kinase assay with immobilized immobilized GST BRCA2 18 ± 141, which contains

Oncogene Phosphorylation of BRCA2 activation domain by a cellular kinase J Milner et al 4443 the d-homology region, is eciently phosphorylated in examined whether the kinase eluted from GST vitro when incubated with Hela nuclear extracts. In BRCA2-loaded beads can phosphorylate exogenous contrast, no kinase activity was detected when GST BRCA2 18 ± 141 in solution. To this end, we performed alone was used (lane 3). These results indicate that a a two step experiment in which we ®rst puri®ed the protein kinase present in Hela cells is speci®cally able kinase from Hela nuclear extracts using GST BRCA2, to bind and phosphorylate in vitro BRCA2 exon 3 then we attempted to elute the kinase from BRCA2 by activation domain. To examine whether the BRCA2- the addition of cold ATP. Elution of the kinase was associated kinase activity was mediated through the d- examined by phosphorylation in solution of GST homology domain (i.e. residues 60 ± 105), we used a BRCA2 18 ± 141 in the presence of radiolabelled mutated version of BRCA2 18 ± 141 in which the d- ATP. Figure 2b indicates that the kinase binds homology region (60 ± 105) was deleted (GST BRCA2 speci®cally to the d-homology region of BRCA2. 18 ± 141 Dd). As depicted in Figure 2 (lane 2), removal Indeed, the kinase is eciently eluted from GST of this region markedly reduced the kinase activity BRCA2 18 ± 141 with cold ATP, as indicated by the associated to BRCA2 exon 3. ability of the kinase to subsequently phosphorylate We next wished to de®ne the kinase-binding-site in BRCA2 in solution (Figure 2b, lane 1). In contrast, no BRCA2. The above-described experiment using the elution of the kinase was found when we used the mutant GST BRCA2 18 ± 141 Dd (Figure 2a) did not mutant GST 18 ± 141 (dD), which is deleted o€ the c- distinguish whether this mutant a€ected the phos- Jun homology region (Figure 2b, lane 2). Taken phoacceptor sites or the binding site for the kinase. together, our data indicate that, in vitro, BRCA2 is Thus, to directly determine whether the d-homology binding to and phosphorylated by a kinase present in region of BRCA2 was binding to the kinase, we Hela nuclear extracts, and that the c-Jun homology region located within exon 3 activation domain constitutes the kinase binding site. Since all JNK family members bind and phosphor- ylate the d-domain of c-Jun (Gupta et al., 1996), the above-mentioned data still left open the possibility that one member of the JNK family, distinct of JNK1, was the kinase associated with BRCA2 c-Jun homology domain. Activation of JNK protein kinases by exposure to environmental stress, such as UV irradia- tion, is a characteristic feature of these kinases (Hibi et al., 1993; Gupta et al., 1996). We therefore tested whether the BRCA2-associated kinase is inducible by

Figure 2 A kinase from Hela cells binds and phosphorylates in vitro the BRCA2 activation domain. (a) Hela nuclear extracts were incubated with GST fusions of BRCA2 (lanes 1 and 2) or GST alone (lane 3). After 12 h of incubation followed by extensive washing in kinase bu€er, the solid-phase kinase assay was performed as described in Figure 1b. The phosphorylated proteins were resolved by SDS ± PAGE and visualized by autoradiography. (b) Elution of the cellular kinase from BRCA2 activation domain. Hela nuclear extracts were incubated with Figure 3 The BRCA2-associated kinase is not induced by UV. GST BRCA2 fusions or GST alone for 12 h. After extensive U2OS cells were irradiated or not with UV-C (70 J/m2 for 5 s). washing in kinase bu€er, GST fusion proteins bound to Unstimulated or UV-stimulated cells were then harvested and gluthatione-sepharose beads were subjected to elution in kinase lysed in IPH bu€er (Bannister and Kouzarides, 1996). Whole cell bu€er containing 1 mM of cold ATP. After centrifugation, the extracts were incubated for 12 h at 48C with equivalent amounts eluted fraction was examined by phosphorylation of GST BRCA2 of either GST BRCA2 18 ± 141 (lanes 1 and 2) or, as a positive 18 ± 141 in kinase bu€er containing 1 mCi of [g-32P]ATP for control, GST c-Jun 30 ± 78 (lanes 3 and 4). Following the solid- 15 min at 308C. Phosphorylation of BRCA2 18 ± 141 was phase kinase assay, performed as described in Figure 1b, the analysed by SDS ± PAGE and autoradiography proteins were separated by SDS ± PAGE

Oncogene Phosphorylation of BRCA2 activation domain by a cellular kinase J Milner et al 4444 binding site for JNK does suggest however that this kinase may have structural features analogous to JNK. Having shown in vitro that BRCA2 activation domain is binding to and phosphorylated by a cellular kinase, we next sought to con®rm these results in vivo. To this end, we generated mammalian expression vectors containing either the entire BRCA2 activation domain (residues 18 ± 105) or only residues 18 ± 60 (i.e. lacking its d-homology region), tagged with an Ha epitope. U2OS cells were transfected with these Ha- tagged BRCA2 constructs and metabolically labelled with 32P-orthophosphate. Following the labelling period, labelled fusion proteins were immunoprecipi- tated from extracts using anti-Ha antibody. As shown in Figure 4, a 32P-labelled protein of the appropriate molecular-weight was precipitated with anti-Ha anti- body in cells transfected with Ha-BRCA2 18 ± 105 (lane 2). No precipitate was detected after substitution of the Figure 4 BRCA2 activation domain is phosphorylated in vivo. Ha antibody with the control anti-Flag antibody (lane U2OS cells were transiently transfected with 10 mgofan 1). In contrast to Ha-BRCA2 18 ± 105, no labelled expression vector containing an Ha-epitope fused to either protein was precipitate from cells transfected with Ha- BRCA2 18 ± 105 (lanes 1 and 2; Ha-BRCA2 18 ± 105) or BRCA2 BRCA2 18 ± 60 which lacks the c-Jun homology region 18 ± 60 (lanes 3 and 4; Ha-BRCA2 18 ± 60). Forty-eight hours after transfection, cells were labelled with 10 mCi of [32P]ortho- (Figure 4, lanes 3 and 4). Thus, these data indicate that phosphate for 24 h. The cells were then lysed in 500 ml of IPH BRCA2 is phosphorylated in vivo and, consistent with bu€er (Bannister and Kouzarides, 1996). Whole cell extracts were the in vitro results presented in Figure 2, the exon 3 precipitated with either anti-Ha antibody (3F10, Boehringer; lanes activation domain is required for BRCA2 phosphor- 2 and 4) or, as a negative control, anti-Flag antibody (M2, ylation. Kodak; lanes 1 and 3). Precipitated proteins were separated by SDS ± PAGE and visualized by autoradiography In summary, we show that the BRCA2 activation domain associates with a cellular kinase which can phosphorylate the BRCA2 activation domain. The binding site for the BRCA2-associated kinase corre- UV-light. The results shown in Figure 3 indicate that sponds to residues which have homology to the JNK this is not the case. For these experiments, extracts of binding site in c-Jun. While the nature of the BRCA2- UV-treated or untreated U2OS cells were incubated associated kinase is currently unknown, the present with GST fusion proteins, followed by a solid-phase data provide insights into the regulation of the BRCA2 kinase assay. As expected, UV irradiation led to a large protein and suggest that phosphorylation may be one increase in the phosphorylation of GST c-Jun (30 ± 78) mechanism by which BRCA2's transcriptional function (Figure 3, compare lanes 3 and 4). In contrast, when is regulated. we used GST BRCA2 18 ± 141, no signi®cant enhance- ment of in vitro phosphorylation could be detected from UV-irradiated cells as compared to untreated cells (Figure 3, compare lanes 1 and 2). Thus these results, Acknowledgments together with the failure of JNK1 to bind and J Milner was supported by a grant from the Cancer Research Campaign; F Fuks was supported by fellowships phosphorylate BRCA2 (see Figure 1b), indicate that from the European Community Training and Mobility of BRCA2 is phosphorylated independently of the JNK Researchers Fellowship and the Wiener-Anspach Funda- signalling pathway. The fact that the BRCA2-asso- tion; L Hughes-Davies was funded by a CRC Clinical ciated kinase recognizes sequences homologous to the Fellowship.

References

Bannister AJ and Kouzarides T. (1996). Nature, 384, 641 ± Hibi M, Lin A, Smeal T, Minden A and Karin M. (1993). 643. Genes Dev., 7, 2135 ± 2148. Chen PL, Chen CF, Chen Y, Xiao J, Sharp D and Lee WE. Marmorstein LY, Ouchi T and Aaronson SA. (1998). Proc. (1998). Proc. Natl. Acad. Sci. USA, 95, 5287 ± 5292. Natl. Acad. Sci. USA, 95, 13869 ± 13874. Connor F, Smith A, Wooster R, Stratton M, Dixon A, May GH, Harris F, Gillespie D and Black DM. (1999). Genes Campbell E, Tait TM, Freeman T and Ashworth A. Chromosomes Cancer, 25, 407 ± 409. (1997). Hum. Mol. Genet., 6, 291 ± 300. Milner J, Ponder B, Hughes Davies L, Seltmann M and Derijard B, Hibi M, Wu IH, Barrett T, Su B, Deng T, Karin Kouzarides T. (1997). Nature, 386, 772 ± 773. M and Davis RJ. (1994). Cell, 76, 1025 ± 1037. Mizuta R, LaSalle JM, Cheng HL, Shinohara A, Ogawa H, Fuks F, Milner J and Kouzarides T. (1998). Oncogene, 17, Copeland N, Jenkins NA, Lalande M and Alt FW. (1997). 2531 ± 2534. Proc. Natl. Acad. Sci. USA, 94, 6927 ± 6932. Fuks F, Burgers WA, Brehm A, Hughes-Davies L and Nordling M, Karlsson P, Wahlstrom J, Engwall Y, Wallgren Kouzarides T. (2000). Nat Genet., 24, 88 ± 91. A and Martinsson T. (1998). Cancer Res., 58, 1372 ± 1375. Gupta S, Barrett T, Whitmarsh AJ, Cavanagh J, Sluss HK, Patel KJ, Yu VPCC, Lee H, Corcoran A, Thistlewaite FC, Derijard B and Davis RJ. (1996). EMBO J., 15, 2760 ± Evans MJ, Colledge WH, Friedman LS, Ponder BAJ and 2770. Venkitaraman AR. (1998). Mol. Cell, 1, 347 ± 357.

Oncogene Phosphorylation of BRCA2 activation domain by a cellular kinase J Milner et al 4445 Sharan SK, Morimatsu M, Albrecht U, Lim DS, Regel E, Wong AKC, Pero R, Ormonde PA, Tavtigian SV and Bartel Dinh C, Sands A, Eichele G, Hasty P and Bradley A. PL. (1998). J. Biol. Chem., 272, 31941 ± 31944. (1997). Nature, 386, 804 ± 810. Wooster R, Bignell G, Lancaster J, Swift S, Seal S, Mangion Suzuki A, de la Pompa JL, Hakem R, Elia A, Yoshida R, Mo J, Collins N, Gregory S, Gumbs C and Micklem G. (1995). R, Nishina H, Chuang T, Wakeham A, Itie A, Koo W, Nature, 378, 789 ± 792. Billia P, Ho A, Fukumoto M, Hui CC and Mak TW. (1997). Genes Dev., 11, 1242 ± 1252. Tavtigian SV, Simard J, Rommens J, Couch F, Shattuck Eidens D, Neuhausen S, Merajver S, Thorlacius S, Ot K, StoppaLyonnetD,BelangerC,BellR,BerryS,Bogden R,ChenQ,DavisT,DumontM,FryeC,HattierT, Jammulapati S, Janecki T, Jiang P, Kehrer R, Leblanc JF, Mitchell JT, McArthur-Morrison J, Nguyen K, Peng Y, Samson C, Schroeder M, Snyder SC, Steele L, String- fellow M, Stroup C, Swedlund B, Swensen J, Teng D, Thomas A, Tran T, Tranchant M, Weaver- Feldhaus JE, Wong AKC, Shizuya H, Eyfjord JE, Cannon-Albright L, Labrie F, Skolnick MH, Weber B, Kamb A and Goldgar DE. (1996). Nat Genet., 12, 333 ± 337.

Oncogene