Oncogene (2008) 27, 3641–3652 & 2008 Nature Publishing Group All rights reserved 0950-9232/08 $30.00 www.nature.com/onc ORIGINAL ARTICLE FANCG promotes formation of a newly identified complex containing BRCA2, FANCD2 and XRCC3

JB Wilson1, K Yamamoto2,9, AS Marriott1, S Hussain3, P Sung4, ME Hoatlin5, CG Mathew6, M Takata2,10, LH Thompson7, GM Kupfer8 and NJ Jones1

1Molecular Oncology and Stem Cell Research Group, School of Biological Sciences, University of Liverpool, Liverpool, UK; 2Department of Immunology and Medical Genetics, Kawasaki Medical School, Kurashiki, Okayama, Japan; 3Department of Biochemistry, University of Cambridge, Cambridge, UK; 4Department of Molecular Biophysics and Biochemistry, Yale University School of Medicine, New Haven, CT, USA; 5Division of Biochemistry and Molecular Biology, Oregon Health Sciences University, Portland, OR, USA; 6Department of Medical and Molecular Genetics, King’s College London School of Medicine, Guy’s Hospital, London, UK; 7Biosciences and Biotechnology Division, L441, Lawrence Livermore National Laboratory, Livermore, CA, USA and 8Department of Pediatrics, Division of Hematology-Oncology, Yale University School of Medicine, New Haven, CT, USA

Fanconi anemia (FA) is a human disorder characterized intricate interface between FANC and HRR in by cancer susceptibility and cellular sensitivity to DNA maintaining stability. crosslinks and other damages. Thirteen complementation Oncogene (2008) 27, 3641–3652; doi:10.1038/sj.onc.1211034; groups and are identified, including BRCA2, which published online 21 January 2008 is defective in the FA-D1 group. Eight of the FA proteins, including FANCG, participate in a nuclear core complex Keywords: ; ATR; interstrand cross- that is required for the monoubiquitylation of FANCD2 links; DNA repair; RAD51 paralog; replication restart; and FANCI. FANCD2, like FANCD1/BRCA2, is not epistasis part of the core complex, and we previously showed direct BRCA2–FANCD2 interaction using yeast two-hybrid analysis. We now show in human and hamster cells that expression of FANCG protein, but not the other core complex proteins, is required for co-precipitation of Introduction BRCA2and FANCD2.We also show that phosphoryla- tion of FANCG serine 7 is required for its co-precipitation Fanconi anemia (FA) is characterized clinically by with BRCA2, XRCC3 and FANCD2, as well as the direct progressive aplastic anemia, multiple congenital ab- interaction of BRCA2–FANCD2. These results argue normalities and a predisposition to malignancy includ- that FANCG has a role independent of the FA core ing acute myeloid leukaemia and squamous carcinomas complex, and we propose that phosphorylation of serine 7 of the head and neck (Alter, 1996). FA cells universally, is the signalling event required for forming a discrete but not exclusively, display hypersensitivity to DNA complex comprising FANCD1/BRCA2-FANCD2- interstrand crosslinking agents (Carreau et al., 1999; FANCG-XRCC3 (D1-D2-G-X3). Cells that fail to Kennedy and D’Andrea, 2005). As we discussed express either phospho-Ser7-FANCG, or full length previously (Thompson et al., 2005), this phenotype is BRCA2protein, lack the interactions amongst the four likely due to the absolute dual requirement for RAD51- component proteins. A role for D1-D2-G-X3 in homo- mediated repair (HRR) and logous recombination repair (HRR) is supported by our translesion synthesis (TLS) during crosslink repair at finding that FANCG and the RAD51-paralog XRCC3 are broken replication forks (Hinz et al., 2007; Patel and epistatic for sensitivity to DNA crosslinking compounds Joenje, 2007). As many fanc mutants show a suppression in DT40 chicken cells. Our findings further define the of base-substitution mutagenesis (reviewed by Hinz et al., 2006), it is clear that the ‘FA pathway’ has a much broader role than promoting just the repair of crosslink damage. Correspondence: Dr NJ Jones, Molecular Oncology and Stem Cell Thirteen FA genes are now identified (Levitus et al., ResearchGroup, Schoolof Biological Sciences, University of Liverpool, Biosciences Building, Crown Street, Liverpool L69 7ZB, UK. 2006; Smogorzewska et al., 2007), but the exact E-mail: [email protected] mechanistic function of many FA proteins remains 9Current address: Department of Hematology, Okayama Citizen’s unclear. FANCD1 is identified as the breast cancer Hospital, Okayama City, Okayama 700-8557, Japan. susceptibility protein BRCA2 (Howlett et al., 2002), 10Current address: Radiation Biology Center, Kyoto University, Yoshida-konoe, Sakyo-ku, Kyoto 606-8501, Japan. whose role lies in regulating RAD51 in HRR (Davies Received 20 September 2007; revised 20 November 2007; accepted 11 et al., 2001; Gudmundsdottir and Ashworth, 2006). December 2007; published online 21 January 2008 BRCA2 function requires a binding partner PALB2 FANCG required for BRCA2–FANCD2–XRCC3 interactions JB Wilson et al 3642 (Xia et al., 2006), recently identified as FANCN (Reid et al., 2007; Xia et al., 2007). The FANCA/B/C/E/F/G/ L/M proteins, together with FAAP24/100, are subunits of a nuclear core complex (Mathew, 2006; Ciccia et al., HeLa (WT)BD220 (A)PD20 (D2)BD215 (C)EUFA409EUFA121 (E) EUFA143 (F) CRL-1583 (G) (WT) 2007; Ling et al., 2007) required for the interdepen- BRCA2 dent monoubiquitylation of FANCD2 and FANCI IP: Anti-FANCD2 (Smogorzewska et al., 2007). IgG Increasing evidence implicates FANCD2 and other 1 2 3 4 5 6 7 8 FANC proteins, including FANCG, being coupled to HRR (Niedzwiedz et al., 2004; Yang et al., or ctor t e tor FANCG 2005; Taniguchi and D’Andrea, 2006). For example, FANCG we previously showed direct interactions between the vec FANCG following pairs of proteins: FANCG–BRCA2; FANCD2– 40BP6 KO40 NM3+ NM3+vec143+ 143+ 326SV+v326SV+ BRCA2; FANCG–XRCC3 (Hussain et al., 2003, 2004, BRCA2 2006). XRCC3 is one of five RAD51 paralogs, which have IP: Anti-FANCD2 nonredundant roles in HRR (Thacker, 2005). A reduced efficiency of HRR, measured after the induction of IgG I-SceI-mediated double-strand breaks, was reported in human and chicken mutants (Yamamoto et al., FANCG 2003; Nakanishi et al., 2005). FANCG protein has at least WB seven tetratricopeptide repeats (TPRs), which are critical ß-Actin for the function of FANCG and its interaction with FANCA, FANCF and the HRR proteins BRCA2 and 1234 5678 XRCC3 (Blom et al., 2004; Hussain et al., 2006). These Figure 1 FANCG is required for the interaction of FANCD2 and TPRs indicate that FANCG likely acts as a scaffold for BRCA2 in mammalian cells. Cell lines were treated with50 n M mitomycin C for 18 hprior to preparation of lysates (for thisfigure the assembly or stabilization of protein complexes (Lamb and subsequent figures unless stated). Cell lysates were immuno- et al., 1995; Groves and Barford, 1999). precipitated withantibody to FANCD2, samples loaded for FANCA/D1/D2/G/M (Yamashita et al., 1998; electrophoresis on a SDS–polyacrylamide gel electrophoresis Taniguchi et al., 2002; Collins and Kupfer, 2005; Esashi (PAGE) gel and blots probed withan anti-BRCA2 C-terminal et al., 2005; Meetei et al., 2005) are all phosphoproteins. antibody (amino acids 3245–3418). (a) FANCD2 and BRCA2 co- precipitate in wild-type (HeLa, CRL-1583) and FA-A (BD220), FANCG phosphorylation occurs at serines 7, 383 and FA-C (BD215), FA-E (EUFA409) and FA-F (EUFA121) human 387, and these modifications are functionally important cell lines, but not in the FA-G cell line EUFA143. (b) FANCD2 for cellular resistance to crosslinking damage (Mi et al., and BRCA2 interaction is restored in fancg mutant cell lines 2004; Qiao et al., 2004). The mutant FANCG-S7A transduced withwild-type human FANCG cDNA, but not withan empty pMMP vector (EUFA143, 326SV and NM3). It is also protein expressed in FA-G EUFA143 lymphoblasts was restored in a knockout hamster cell line (KO40) transfected with able to bind and stabilize FANCA and FANCC, the genomic hamster (40BP). Western blot with anti-FANCG although FANCD2 monoubiquitylation was slightly shows FANCG expression in the same cell lines. reduced (Qiao et al., 2004). While it is well established that FANCG is critical for the assembly of a functional FA nuclear core complex (Garcia-Higuera et al., 1999; FA-G lymphoblast cell lines (Hussain et al., 2004). The Waisfisz et al., 1999; Gordon and Buchwald, 2003), here interaction was further investigated by examining other we define an essential role for FANCG in mediating FA complementation groups. BRCA2 co-immunopre- the key interaction between BRCA2 and FANCD2 cipitated using FANCD2 antisera in bothwild-type cells (Hussain et al., 2004; Wang et al., 2004). Moreover, we (HeLa and CRL-1583) and cells from groups FA-A/C/ present evidence that phosphorylation of FANCG at E/F (Figure 1a), while it was absent in FA-D2 cells serine 7 is essential, not only for its own direct (PD20). Western blotting confirmed that each FA cell interactions withBRCA2 and XRCC3, but also the line utilized failed to express detectable levels of the pair-wise interactions among BRCA2, FANCD2 respective FA protein (data not shown). Co-precipita- and XRCC3. We propose that this new complex tion between FANCD2–BRCA2 failed to occur in CHO consisting of at least these four FANC and HRR FancG mutants NM3 and KO40, and in human FA-G proteins (D1-D2-G-X3) is responsible for promoting the fibroblast (326SV) and lymphoblast (EUFA143) cells HRR component of crosslink repair. (Figure 1b). Introduction of wild-type FANCG (either human cDNA or genomic hamster for 40BP6), and expression of FANCG protein in the four cell lines Results restored the interaction of FANCD2–BRCA2. The co- precipitation of FANCD2–BRCA2 was also present Interaction between BRCA2 and FANCD2 is dependent when cell extracts were incubated with DNase prior to on FANCG, but not other core complex proteins performing the immunoprecipitation (data not shown). We previously demonstrated that FANCD2 and As FANCD2–BRCA2 co-precipitated in human BRCA2 co-immunoprecipitated in wild-type mamma- FA-A/C/E/F cells, that fail to monoubiquitylate FANCD2 lian cells and that co-precipitation was absent in two (Gregory et al., 2003), we sought to determine which

Oncogene FANCG required for BRCA2–FANCD2–XRCC3 interactions JB Wilson et al 3643 isoform(s) of FANCD2 interact withBRCA2. We withco-precipitation of BRCA2 and FANCD2 in utilized a FA-D2 cell line PD20 that has been all cell lines tested, except EUFA143 þ vector, transfected withK561R- FANCD2 (Garcia-Higuera EUFA143 þ S7A and PD20 (Figure 3b). Thus, BRCA2 et al., 2001) and therefore expresses a FANCD2 protein and FANCD2 failed to co-precipitate in bothhamster that cannot be monoubiquitylated (Figure 2a). Co- and human cell lines that express FANCG-S7A, precipitation of BRCA2 withFANCD2 was observed indicating that phosphorylation of Ser7 is most likely in wild-type cell lines and PD20-3-15 (functionally complemented withhumanchromosome 3p) and also in PD20 þ K561R-FANCD2 and an additional FA-A cell line (HCS72). This result confirms that FANCD2-S S387A S383/387A interacts withBRCA2 in vivo (Figure 2b). Interaction +vector +S7A +S383A between BRCA2 and FANCD2-L (Wang et al., 2004) AA8 (WT)NM3 NM3 EUFA14340BP6 (G) KO40 (G)NM3 NM3+ NM3+ was corroborated in a reciprocal experiment using IP: Anti-FANCD2 BRCA2 BRCA2 antibody (Figure 2c). Neither FANCD2 isoform co-precipitated withBRCA2 in FancG mutants 12345 678 9 NM3, KO40 and UV40. The two isoforms of FANCD2 A can be distinguished in the wild-type cells, but only K561R FANCG FANCD2-S co-precipitated withBRCA2 in humancells +vector+S7A +S383A +S387 BD180 (WT)143 143 143+ 143 143 PD20-3-15PD20+PD20 (D2) of complementation groups FA-A/C/E/F as expected BRCA2 (Figure 2c). IP: Anti-FANCD2 123456789 The role of FANCG phosphorylation in mediating the Figure 3 Phosphorylation of FANCG at serine 7 is required for the interaction of FANCD2 and BRCA2. The generation of NM3 interaction between FANCD2 and BRCA2 and EUFA143 cell lines expressing human FANCG protein with Transduction of NM3 and EUFA143 withhuman mutations in the phosphorylation sites at serines 7, 383 and 387 was FANCG cDNA mutated at phosphorylation sites S7, described previously (Mi et al., 2004; Qiao et al., 2004). The level of S383 and S387 (substituted withalanine), results in only expression of the various forms of the FANCG protein in both sets of NM3 and EUFA143 are similar and this was confirmed here partial correction of their mitomycin C (MMC) (data not shown). These cell lines show intermediate sensitivity to hypersensitivity (Mi et al., 2004; Qiao et al., 2004), so MMC. In NM3, the D37 values of the various cell lines following we examined FANCD2–BRCA2 co-precipitation in treatment withMMC were: NM3 þ vector ¼ 28 nM,NM3þ S7A- these cell lines (Figure 3). BRCA2 co-precipitated using FANCG ¼ 40 nM, NM3 þ S383A-FANCG ¼ 50 nM, NM3 þ S387A- FANCD2 antisera in AA8, 40BP, NM3 þ FANCG and FANCG ¼ 46 nM and NM3 þ FANCG ¼ 120 nM, as compared with the wild-type parental cell line, AA8 ¼ 124 nM. For the human cells the two ‘mitotic’ mutants NM3 þ S383A and NM3 þ the dose of MMC required to produce 50% inhibition in a growth S387A (Figure 3a). Even mutation of FANCG at both assay was approximately 70 nM in EUFA143 þ FANCG, while S383 and S387 failed to disrupt the interaction between it was between 15 and 20 nM for the three cell lines expressing BRCA2–FANCD2 (Figure 3a, lane 9). In contrast, the mutant forms of FANCG.(a and b) FANCD2 and BRCA2 fail to interact in hamster NM3 cells and human EUFA143 BRCA2 failed to co-precipitate in NM3 expressing the cells transduced with S7A-FANCG cDNA (lanes a3 and b3), but FANCG-S7A protein or in NM3 withempty pMMP do co-precipitate in mitotic mutants NM3 þ S383A and vector. Identical results were observed in human cells, NM3 þ S387A (lanes 7 and 8).

tor

+K561R +FANCG

PD20 (D2)PD20 PD20-3-15HeLa (WT) +K561R FANCD2-L FANCD2-S BD180 PD20(WT) (D2)PD20 PD20-3-15EUFA143+vecEUFA143HSC72HeLa (A) (WT) WB IP: Anti-FANCD2 BRCA2 β-actin 1234 12345 678

)

UV40 (G)AA8 (WT)40BP6 KO40 (G)NM3 (G)V79 (WT) BD220 (A)HSC72 (ABD215 (C)EUFA409 PD20(E) (D2)EUFA121 (F) FANCD2-L IP: Anti-BRCA2 FANCD2-S

1234 56 7 8 9 101112 Figure 2 BRCA2 interacts withboththemonoubiquitylated (FANCD2-L) and nonubiquitylated (FANCD2-S) isoforms of FANCD2. (a) Western blot to confirm expression of FANCD2 isoforms in PD20 (FA-D2) cell lines. PD20-3-15, functionally complemented with human chromosome 3p, expresses both isoforms, while PD20 transfected with mutant K561R-FANCG cDNA expresses only the FANCD2-S isoform. (b) Co-precipitation of FANCD2 and BRCA2 in PD20 þ K561R-FANCG (lane 3) and an additional FA-A cell line HSC72 confirms co-precipitation of FANCD2-S withBRCA2. ( c) Reciprocal immunoprecipitation with antibody to BRCA2 and subsequent blotting withanti-FANCD2 indicates thatbothFANCD2-S and FANCD2-L interact with BRCA2 in wild-type (AA8 and V79) and FancG-corrected cells (40BP6). However, only the FANCD2-S isoform co-precipitates in FA- A (bothBD220 and HSC72), FA-C, FA-E and FA-F cells (lanes 7–10, 12) and neitherisoform in FA-G mutant cells (lanes 1, 4 and 5).

Oncogene FANCG required for BRCA2–FANCD2–XRCC3 interactions JB Wilson et al 3644 required for FANCG protein to mediate the interaction lanes 3 and 11). In addition, co-precipitation of between BRCA2–FANCD2. FANCG and XRCC3 was observed in wild-type cells (AA8, V79, BD180 and HeLa), HSC72 (FA-A) and Phosphorylation of Serine 7 is required for FANCG PD20 (FA-D2), but was absent in NM3 þ vector, interaction withBRCA2 and XRCC3, but not thecore EUFA143 þ vector and Xrcc3 mutant irs1SF. complex proteins FANCA and FANCF In contrast, co-precipitation of FANCG with As FANCG-S7A fails to mediate the interaction FANCA and FANCF was observed in all NM3 and between FANCD2 and BRCA2 in vivo, we next EUFA143 cell lines expressing a FANCG protein, determined whether this phosphomutant co-precipitated including FANCG-S7A (Figures 4c and 4d, lanes 3 withFANCA, FANCF, BRCA2 and XRCC3, all of and 11). Indeed, co-precipitation of FANCG-FANCA which directly interact with FANCG. We tested for co- and FANCG-FANCE was only absent in the respective immunoprecipitation of the various phosphomutant mutant cell lines HSC72 (FA-A) and EUFA121 (FA-F), forms of FANCG in whole cell extracts from cells NM3 þ vector, EUFA143 þ vector and KO40. Co-pre- treated with50 n M MMC (Figure 4). The S7A mutant cipitation of FANCG-S7A withFANCA and FANCC form of FANCG failed to co-precipitate withBRCA2 in was previously demonstrated in EUFA143 þ S7A cells NM3 þ S7A and EUFA143 þ S7A cells (Figure 4a, (Qiao et al., 2004). Therefore, phosphorylation of Ser7 lanes 3 and 11). Cells expressing the S383A and S387A appears required for FANCG’s interaction withthe forms of FANCG did show co-precipitation with HRR proteins BRCA2 and XRCC3 in vivo, but not for BRCA2 (Figure 4a, lanes 5, 6, 13 and 14). Co- interaction withtheFA core complex proteins FANCA precipitation was present in positive control cell lines and FANCF. (including NM3 þ FANCG and EUFA143 þ FANCG), but absent in negative controls. BRCA2 is required for complex formation amongst The interactions of XRCC3 with the mutant forms of FANCD2, FANCG and XRCC3 FANCG were identical to that shown by BRCA2 Having established that phosphorylation of FANCG at (Figure 4b). XRCC3 was found to co-precipitate in Ser7 was required for its own interaction withBRCA2 NM3 and EUFA143 cell lines expressing the wild-type, and XRCC3, and for the co-precipitation of BRCA2– S383A and S387A forms of the FANCG protein FANCD2, we further investigated the interactions (Figure 4b, lanes 4–6, 12–14), but was absent in between these four proteins. We previously described NM3 þ S7A and EUFA143 þ S7A cells (Figure 4b, the co-precipitation of BRCA2–XRCC3 (Hussain et al.,

(G)

(WT) +vector+S7A +FANCG+S383A+S387A +vector+S7A +FANCG+S383A+S387A BP6 (WT) 3 AA8 (WT)NM3 NM3 NM3 NM3 NM3 40 KO40 (G)BD180 143(WT) 14 143 143 143 EUFA673HSC72 HeLa (A) IP: Anti-BRCA2 FANCG

1234 5 67891110 12 13 14 15 16 17

) ctor +vector+S7A +FANCG+S383A+S387A +ve +S7A +FANCG+S383A+S387A 3 AA8 (WT)NM3 NM3 NM3 NM3 NM3 irs1SFV79 (X3 (WT) BD180 143(WT) 14 143 143 143 HSC72HeLa (A) (WT)PD20 (D2)

IP: Anti-FANCG XRCC3 1234 567891110 12 13 14 15 16 17

) CG 3A ) 2) 7A 38 (WT) 383A S387A (WT) +vector+S +FAN +S +S387A +vector+S7A +FANCG+S + AA8 (WT)NM3 NM3 NM3 NM3 NM3 KO40 (G)40BP6 (WTBD180143 143 143 143 143 HSC72HeLa (A PD20 (D IP: Anti-FANCA FANCG 1234 567891110 12 13 14 15 16 17

) or 387A (G) ct 7A 387A WT) +vector+S7A +FANCG+S383A+S S BP6 (WT +ve +S +FANCG+S383A+ AA8 (WT)NM3 NM3 NM3 NM3 NM3 KO40 40 BD180143 (WT) 143 14z 143 143 PD20 (D2)EUFA121HeLa (F) ( IP: Anti-FANCG FANCF

1234 567891110 12 13 14 15 16 17 Figure 4 Phosphorylation of FANCG at serine 7 is required for its interaction with BRCA2 and XRCC3, but not with the core complex proteins FANCA and FANCF. (a and b) FANCG failed to co-precipitate witheitherBRCA2 or XRCC3 in NM3 þ S7A and EUFA143 þ S7A cells (lanes 3 and 11). These interactions are present in all other cell lines (including the mitotic mutants S383A and S387A) other than the mutants for FANCG (NM3 þ vector, KO40, EUFA143 þ vector) or XRCC3 (irs1SF). (c and d) FANCG co-precipitates withFANCA and FANCF in all cell lines (including NM3 þ S7A and EUFA143 þ S7A; lanes 3 and 11) apart from the mutants for the respective protein (NM3 þ vector, KO40, EUFA143 þ vector, HSC72 for FANCA and EUFA121 for FANCF).

Oncogene FANCG required for BRCA2–FANCD2–XRCC3 interactions JB Wilson et al 3645 ) ) ) XRCC3 A + S7 brca2

AA8 (WT)irs1SF (xrcc3irs1SF HeLa (WT) AA8 (WT)NM3 NM3+ V-C8 ( CAPAN-1 ( FANCG FANCG IP: Anti-BRCA2 IP: Anti-XRCC3 IgG IgG

FANCG FANCG IP: Anti-FANCD2 IP: Anti-FANCD2 IgG IgG

BRCA2 XRCC3 IP: Anti-FANCD2 IP: Anti-FANCD2 IgG IgG

XRCC3 BRCA2 IP: Anti-BRCA2 IP: Anti-FANCD2 IgG IgG 1234 5 6 789 Figure 5 BRCA2 and phospho-Ser7-FANCG, but not XRCC3, are required for the formation of the D1-D2-G-X3 complex. (a) Immunoprecipitation was performed withantibodies to BRCA2 or FANCD2 and aliquots subsequently immunoblotted with either FANCG, BRCA2 or XRCC3 antibodies. Co-precipitation was observed between FANCG-BRCA2, FANCG-FANCD2 and BRCA2-FANCD2 in all four cell lines tested, including irs1SF which fails to express the XRCC3 protein. (b) Immunoprecipitation was performed withantibodies to XRCC3 or FANCD2 and aliquots subsequently immunoblotted witheitherFANCG, BRCA2 or XRCC3 antibodies. Co-precipitation of FANCG-XRCC3, FANCG-FANCD2 and FANCD2-XRCC3 was absent in both brca2 mutant cell lines (V-C8 and CAPAN-1, lanes 8 and 9) and also in NM3 and NM3 þ S7A (lanes 6 and 7).

2006), and we can also now report the co-precipitation Interactions of phospho-Ser7-FANCG of FANCD2 withbothFANCG and XRCC3 in Having established that FANCG-S7A fails to hamster and human cells (Figure 5, lanes 1, 4 and 5). co-precipitate withBRCA2, FANCD2 and XRCC3, While a direct interaction between FANCG and we next examined the interactions of the endogenous FANCD2 has not been reported, both interact directly phospho-S7-FANCG (FANCGS7-P) protein using a withBRCA2 (de Winter et al., 1998; Hussain et al., phospho-specific antibody raised against phosphoserine 2003, 2004). Each of these four proteins can therefore be 7 of FANCG (Qiao et al., 2004). This antibody does not co-precipitated with the other three, suggesting that they detect FANCG-S7A and treatment of extracts with exist together in a single protein complex (D1-D2- phosphatase abolishes its ability to detect wild-type G-X3). Besides being required for co-precipitation of FANCG. Immunoprecipitation was performed with BRCA2-FANCD2 (Figures 1–3), FANCG is also antibodies to BRCA2, FANCD2, FANCA, XRCC3 required for the co-precipitation of BRCA2-XRCC3 and FANCG itself and subsequently immunoblotted (Hussain et al., 2006) and FANCD2-XRCC3 (Figure 5, with either the phospho-specific antibody (Figure 6a, lanes 6 and 7). lanes 5–8) or the standard/nonspecific anti-FANCG as a We next determined whether BRCA2 or XRCC3 were control (Figure 6a, lanes 1–4). Blotting withthe also required for the formation of the D1-D2-G-X3 phospho-specific antisera confirmed that FANCGS7-P complex. To achieve this we examined co-precipitation does co-precipitate withBRCA2, FANCD2 and of pairs of proteins in cell lines mutant for BRCA2 XRCC3 in AA8 and NM3 þ FANCG cells (Figure 6a, (hamster V-C8 and human CAPAN-1) and XRCC3 lanes 5 and 8). The specificity of the antibody was (irs1SF). XRCC3 was not required for the co-precipita- confirmed by immunoprecipitating withnonspecific tion of BRCA2–FANCD2, FANCG–BRCA2 or anti-FANCG. The phospho-specific antibody failed to FANCG–FANCD2 (Figure 5a). Eachof thesepairs of detect FANCG-S7A (Figure 6a, lane 7), while it was proteins co-precipitated in irs1SF (Figure 5, lane 2) and detected using the nonspecific FANCG antibody the positive controls (Figure 5, lanes 1, 3 and 4). We (Figure 6a, lane 3). similarly determined that co-precipitation between Interestingly, the phospho-specific antibody failed to FANCG–BRCA2, FANCG–XRCC3 and BRCA2– detect FANCG following immunoprecipitation with XRCC3 was present in PD20 cells (Figure 4b and data anti-FANCA (Figure 6a, lanes 5 and 8), while interac- not shown), indicating that FANCD2 was not required tion between FANCA and FANCG was clearly for these interactions. However, as co-precipitation of observed using the nonspecific antibody (Figure 6a, FANCG–XRCC3, FANCG–FANCD2 and FANCD2– lanes 1, 3 and 4). To confirm the lack of interaction of XRCC3 was absent in bothV-C8 and CAPAN-1 cells FANCGS7-P and FANCA we used increased doses of (Figure 5, lanes 8 and 9), and that both these cell lines MMC (Figure 6b). Following immunoprecipitation with are predicted to express truncated BRCA2 (Goggins anti-FANCA, FANCG was detected at all the doses et al., 1996; Wiegant et al., 2006), full lengthBRCA2 utilized (0–250 nM) in AA8 and HeLa. However, upon protein would seem to be required for eachof blotting with the phospho-specific FANCG antibody, these interactions, and thus also for the formation an interaction was present only in cells treated withthe of the D1-D2-G-X3 protein complex. highest MMC dose (250 nM). Therefore, FANCGS7-P

Oncogene FANCG required for BRCA2–FANCD2–XRCC3 interactions JB Wilson et al 3646 Western blot: Anti-FANCG Anti-Phospho-Ser7 FANCG AA8 HeLa Cell line:

S7A FANCG S7A FANCG IPs: A A/A A/A/G A A/A A/A/G AA8 (WT)NM3+vectorNM3+ NM3+ AA8 (WT)NM3+vectorNM3+ NM3+ FANCG IP: Anti-BRCA2 FANCG pFANCG IP: Anti-FANCD2 FANCG BRCA2 IP: Anti-FANCG FANCG XRCC3

IP: Anti-FANCA FANCG 123 456 Figure 7 The D1-D2-G-X3 complex is independent of FANCG’s FANCG interaction withFANCA and thecore complex. Cell lysates of IP: Anti-XRCC3 AA8 and HeLa were successively immunoprecipitated withanti- IgG body to FANCA ( Â 2) and finally withthenonspecific FANCG 1234 5678 antibody. The resulting precipitates: A (from first Ip with anti- FANCA; lanes 1 and 4), A/A (from second IP withanti-FANCA; Cell line: AA8 HeLa lanes 2 and 5) and A/A/G (from final IP withanti-FANCG; lanes 3

M M and 6) were subsequently immunoblotted withnonspecific FANCG M M antibody, phospho-specific FANCG antibody, anti-BRCA2 or 0 nM 20 nM 50 nM 100 n 250 nM 0 nM 20 n 50 n 100 nM250 n anti-XRCC3. Depletion of the lysates of FANCA and FANCA- IP: Anti-FANCA FANCG bound FANCG does not remove phospho-Ser7-FANCG protein IP: Anti-FANCA pFANCG or FANCG-bound BRCA2 and XRCC3 (lanes 3 and 6).

IP: Anti-BRCA2 pFANCG anti-FANCA; (A/A) after a further IP of the resulting 1234 5 6789 10 first supernatant withanti-FANCA; (A/A/G) after IP of Figure 6 Interactions of phospho-Ser7-FANCG. (a) Immunopre- the second supernatant with anti-FANCG (A/A/G). cipitation was performed withantibodies to BRCA2, FANCD2, FANCA, XRCC3 and FANCG and aliquots subsequently Following the initial immunoprecipitation with anti- immunoblotted with either the nonspecific FANCG antibody FANCA (Figure 7, lanes 1 and 4), FANCG protein was (H:83–622, R no. 2942-44) as a control (lanes 1–4), or a detected, but there was no co-precipitation of phospho-specific antibody (lanes 5–8). Western blotting with the FANCGS7-P, BRCA2 and XRCC3. The second immu- phospho-specific antisera shows co-precipitation of phospho-Ser7- noprecipitation (A/A) withanti-FANCA (Figure 7, FANCG withBRCA2, XRCC3 and FANCD2 in AA8 and NM3 þ FANCG cells, while no co-precipitation was observed with lanes 2 and 5) confirms that very little FANCA- FANCA with the conditions utilized here (cells treated with 50 nM). associated FANCG protein remains. The final immu- (b) FANCA only interacts with phospho-Ser7-FANCG in AA8 noprecipitation withanti-FANCG (A/A/G) should and HeLa cells treated with250 n M MMC, withno interaction therefore only detect FANCG protein that was not detected at doses between 0–100 nM. Conversely phospho-Ser7- bound to FANCA in the cellular extracts. Western FANCG interacts withBRCA2 at all doses tested (0–250 n M). blotting of this third precipitate (Figure 7, lanes 3 and 6) reveals the presence of FANCG protein, which is detected using both the nonspecific and phospho-specific only appears to interact with FANCA at higher levels of antibodies. Also present were BRCA2 and XRCC3, MMC-induced DNA damage, and at a dose where less indicating that these were in complex with the FANCG than 10% of cells remain viable. In contrast, immuno- protein that remained (presumably FANCGS7-P) precipitation between BRCA2 and FANCGS7-P was following the depletion of FANCA. observed at all doses, including untreated cells.

FANCG Ser7-phosphorylation is ATR-dependent The D1-D2-G-X3 complex is independent of FANCG’s The kinase that phosphorylates FANCG at Ser7 interaction withFANCA remains to be determined (Qiao et al., 2004). Given The lack of interaction between FANCGS7-P and that phosphatidylinositol-3 kinase related kinases FANCA suggests that the D1-D2-G-X3 complex forms (PIKK) ATR and ATM have been shown to phosphor- independently of FANCA and is discrete from the FA ylate FANCD2 (Taniguchi et al., 2002; Ho et al., 2006), nuclear core complex. We confirmed this by examining we investigated their possible role in the phosphoryla- HeLa and AA8 whole cell extracts for the presence of tion of FANCG. Expression of FANCGS7-P was FANCGS7-P, BRCA2 and XRCC3 by western blotting, determined by western blotting in wild-type after they had been immunodepleted of FANCA. We (GM02188), ataxia telangiectasia (AT) and ATR-Seckel performed two successive immunoprecipitations with cells (O’Driscoll et al., 2003; Alderton et al., 2004) antisera to FANCA to deplete extracts of FANCG following treatment with50 n M MMC, withand without protein bound to FANCA, and then performed a third the addition of the PIKK inhibitor wortmannin. IP withthenonspecific/standard FANCG antibody Wortmannin has been shown to inhibit ATM at a dose (Figure 7). Therefore, three immunoprecipitates of 20 mM, while ATR is inhibited at doses from 100 mM were produced for eachextract: (A) after IP with (Sarkaria et al., 1998). We failed to detect FANCGS7-P in

Oncogene FANCG required for BRCA2–FANCD2–XRCC3 interactions JB Wilson et al 3647 the cisplatin and MMC sensitivity of knockout mutants

0 µM 20 µM 0 µM 20 µM 100 µM created in chicken DT40 cells, whose properties make 0 µM 20 µM 100 µM – – 100 µM– – – – – – – them amenable to determining genetic interactions WT WT WT ATR ATR ATR ATM ATM ATM pFANCG (Takata et al., 2001; Yamashita et al., 2002; Niedzwiedz Western blot Exp. 1 et al., 2004). Double mutants of fancg and were FANCG made by disrupting FANCG in a conditional xrcc3 Western blot pFANCG background, followed by excision of the Xrcc3-EGFP Exp. 2 FANCG expression cassette by activating MerCreMer recombi- 123456 789 nase (Zhang et al., 1998). The removal was ensured by subcloning and was further verified by the loss of GFP M M fluorescence and by Southern blotting using human µ µM - 0 - 200 µ XRCC3 probe (data not shown). Mutant xrcc3 cells 0 - G G NC NC show a greater sensitivity to killing by cisplatin 0 µM – FA FA compared to fancg cells in a clonal assay (Figure 9a). WT WT - 200EUFA143 µM 143+ 143+ However, two independent clones of xrcc3/fancg cells IP: Anti-FANCD2 BRCA2 display the same cisplatin sensitivity as the xrcc3 single 12345 mutant. To confirm epistasis, we examined growth Figure 8 Phosphorylation of FANCG at Ser7 is ATR dependent. inhibition by MMC (Figure 9b). The xrcc3 cells showed (a) Cell lines were treated with50 n M mitomycin C and the greater sensitivity to MMC than fancg cells (50% indicated dose of wortmannin for 18 hprior to preparation of growthinhibition at 4.5 and 12 n M, respectively, lysates. Western blotting was performed with the phospho-specific compared to 65 nM for wild-type cells). The responses or nonspecific antibodies to FANCG in repeat experiments. ATR- Seckel (DK0064) cells failed to express a detectable level of of xrcc3 and xrcc3/fancg cells were the same, demon- phospho-Ser7-FANCG (lanes 4–6), while it was clearly evident in strating epistasis and a functional overlap between bothwild-type (GM02188) and ATM-deficient (GM01389D) cells FANCG and XRCC3 in this model system. (lanes 1–3 and 7–9). Use of a wortmannin dose that inhibits ATM (20 mM) failed to reduce the expression of phosphorylated FANCG in wild-type cells (lane 2), although the higher dose of 100 mM, that inhibits ATR, does result in reduced expression (lanes Discussion 3 and 9). (b) Incubation with200 mM abolishes the co-precipitation of FANCD2–BRCA2 in wild-type (BD180) and EUFA143 þ We showed that the key FANCD2–BRCA2 interaction FANCG cells. requires FANCG. FANCD2–BRCA2 co-precipitation in cells from complementation groups FA-A/C/E/F ATR-Seckel cells in two independent experiments (and also in PD20 þ K561R cells) indicates that (Figure 8a, lanes 4–6). Expression of FANCG was FANCD2-S, in addition to the monoubiquitylated detected using the nonspecific antibody, indicating that isoform (Wang et al., 2004), interacts withBRCA2, Ser7 is not phosphorylated in ATR-Seckel cells. Con- which is consistent with our earlier finding that trastingly, expression of FANCGS7-P was detected in nonmonoubiquitylated FANCD2 and BRCA2 interact wild-type and ATM-deficient cells, including those directly in the yeast two-hybrid system (Hussain et al., treated with20 mM wortmannin (Figure 8a, lanes 1, 2, 2004). Importantly, these results also show that 7, and 8). Expression did appear reduced in cells treated FANCD2 interacts withBRCA2 independently of the with100 mM wortmannin (Figure 8a, lanes 3 and 9), a FA nuclear core complex, which may seem counter- dose reported to inhibit ATR. These data suggest that intuitive because FANCD2-L has been perceived to be phosphorylation at Ser7 occurs in an ATR-dependent the active isoform for ‘FA pathway’ function (Wang manner. Either ATR phosphorylates FANCG directly, et al., 2004). We address this apparent paradox below, or it may be phosphorylated by a kinase downstream of but first we propose that this interaction regulates the ATR, such as CHK1. We also found that when wild- activity and/or subcellular localization of BRCA2, type human and hamster cells were incubated with which becomes active in chromatin upon the critical 200 mM wortmannin, co-precipitation of FANCG with FANCD2–FANCI interdependent monoubiquitylation BRCA2 or XRCC3 was not observed (data not shown). step(s) (Wang et al., 2004; Smogorzewska et al., 2007). Consistent withthisobservation, treatment of wild-type Phosphorylation of FANCG-S7 is essential for the and EUFA143 þ FANCG cells abolished the co-precipi- direct FANCD2–BRCA2 interaction, as well as for the tation of FANCD2 and BRCA2 (Figure 8b). interactions of FANCG withBRCA2, XRCC3 and FANCD2 (see Table 1 summary). Commensurate with FANCG’s proposed role in assembling multi-protein Xrcc3 and FancG are epistatic for interstrand crosslink complexes, FANCGS7-P is also required for the interac- sensitivity in DT40 cells tions of BRCA2–XRCC3 and FANCD2–XRCC3 If the putative D1-D2-G-X3 complex is responsible for (Table 1). However, FANCGS7-P is not required promoting the HR component of crosslink repair, for protein interactions in the core complex, since mutations abolishing FANCG and XRCC3 are expected FANCG-S7A co-precipitates withcore complex to behave in an epistatic manner in response to proteins FANCA/C/F (Table 1), and we only detect crosslinking agents. The genetic relationship between an interaction of FANCGS7-P withFANCA in cells FANCG and XRCC3 was investigated by determining exposed to excess MMC (Figure 6b).

Oncogene FANCG required for BRCA2–FANCD2–XRCC3 interactions JB Wilson et al 3648 100 100

10

10

1 GROWTH (%) SURVIVAL (%)

0.1 1 0120 50 100 150 200 250 CISPLATIN (µM) MMC (nM)

WT fancg #1 fancg #1/xrcc3 WT fancg xrcc3 fancg #2 fancg #2/xrcc3 xrcc3 fancg/xrcc3

Figure 9 FANCG and XRCC3 are epistatic in DT40 chicken cells. (a) In a clonal assay, mutant xrcc3 cells exhibit a greater sensitivity to cisplatin compared to two independently generated fancg mutant cell lines (1 and 2). Knockout of XRCC3 in the fancg cell lines create double mutants (fancg no. 1/xrcc3 and fancg no. 2/xrcc3) that display the same degree of cisplatin sensitivity as the xrcc3 single mutant, indicating epistasis. (b)Thexrcc3 cells show a higher degree of growth inhibition by MMC than fancg cells. The response of the double mutant fancg/xrcc3 was almost identical to the xrcc3 cell line, again clearly demonstrating epistasis for sensitivity to interstrand crossing agents.

Table 1 Summary of the in vivo interactions of FANCG, BRCA2, the D1-D2-G-X3 complex. We found that ATR-Seckel FANCD2 and XRCC3 as determined by co-immunoprecipitation of cells lack FANCGS7-P and that wortmannin inhibits this endogenous proteins in cells treated with50 n M mitomycin C phosphorylation. Because S7 phosphorylation occurs in Proteins tested for Phosphorylation status of FANCG S-phase, and is DNA-damage inducible (Qiao et al., co-precipitation 2004), the ATR kinase is the likely candidate for Non-phospho Phospho-Ser7 Figure performing this step. (FANCG-S7A) (FANCGS7-P) There has been much conjecture and conflicting FANCG–BRCA2 À + 4a; 6a results in the literature on the involvement of the FA FANCG–FANCD2 À + 5a; 6a proteins in HRR (and single-strand annealing based on FANCG–XRCC3 À + 4b; 6a HRR) of I-SceI induced DNA double strand breaks FANCG–FANCA + À 4c; 6; 7 FANCG–FANCF + À 4d (DSBs) in artificial direct-repeat substrates (Niedzwiedz FANCG–FANCC + À NAa et al., 2004; Bridge et al., 2005; Nakanishi et al., BRCA2–FANCD2 À + 3a and b; 6a; 7 2005; Ohashi et al., 2005; Taniguchi and D’Andrea, BRCA2–XRCC3 À + 5a; 7 2006). Sucha defect in FA cells is typically small or FANCD2–XRCC3 À + 5b absent compared to that in mutants of BRCA2 and RAD51 paralogs as discussed in detail (Hinz et al., Abbreviation: NA, not applicable. Plus (+) or minus (À) indicates whether co-precipitation of proteins was observed, dependent on the 2006). Moreover, given that most FA cell lines show a phosphorylation status of FANCG at Serine 7. Interactions of the modest or no sensitivity to agents that induce direct mutant FANCG-S7A protein determined in hamster NM3+S7A- DSBs, the relevance of I-SceI-based HRR assays to FANCG or human EUFA143+S7A-FANCG cells. Interactions of crosslink repair is not clear. Models for interstrand phospho-Ser7-FANCG (FANCGS7-P) determined in wild-type hamster (AA8) or human cells (HeLa and/or BD180) using a phospho-specific crosslink repair at replication forks propose that HRR antibody. aIt was previously shown that FANCG-S7A co-precipitates acts in conjunction withcrosslink unhooking by withFANCC (Qiao et al., 2004). ERCC1-XPF and translesion synthesis in order to restart replication, withFA proteins mediating these processes (Dronkert and Kanaar, 2001; Thompson, Our studies point to the existence of a protein 2005; Thompson et al., 2005; Mirchandani and complex (D1-D2-G-X3) comprising FANCG, the D’Andrea, 2006; Niedernhofer, 2007; Patel and Joenje, RAD51 paralog XRCC3 and two noncore complex 2007; Zhang et al., 2007). Data from the CHO hamster FA proteins, FANCD1/BRCA2 and FANCD2. Evi- model system argue strongly that the FA ‘pathway’ is dence for a functional link between XRCC3 and some important for diverse damages that cause blocked or components of the Fanconi pathway is further suggested broken replication forks (Wilson et al., 2001; Tebbs by our finding that Xrcc3 and FancG are epistatic for et al., 2005; Hinz et al., 2007). XRCC3 is required for crosslinker sensitivity in DT40 chicken cells. Moreover, replication fork slowing on cisplatin (a crosslinker)- D1-D2-G-X3 complex formation occurs independently damaged (Henry-Mowatt et al., 2003). In of FANCD2 monoubiquitylation and other core com- the light of our findings of Xrcc3/FancG epistasis and plex proteins, but depends on FANCGS7-P and full the direct XRCC3–FANCG interaction, the D1-D2- lengthBRCA2. We propose a model in whichbinding of G-X3 complex may control this modulation of fork FANCGS7-P to BRCA2 is the critical event in forming progression. Other proteins, such as FANCE, RAD51C

Oncogene FANCG required for BRCA2–FANCD2–XRCC3 interactions JB Wilson et al 3649 and PALB2/FANCN, may associate withtheD1-D2- extends beyond the interaction of monoubiquitylated G-X3 complex. FANCE directly interacts with FANCD2-FANCI and BRCA2. A ‘replication restart FANCD2 and co-immunoprecipitates withBRCA2 complex’, comprising at least D1-D2-G-X3 likely pro- (Pace et al., 2002; Wang et al., 2004), RAD51C is the motes HRR and cell survival in response to replication direct binding partner of XRCC3 in one of two distinct forks that break upon encountering DNA damage. RAD51 paralog complexes (Masson et al., 2001; Liu et al., 2002; Yamada et al., 2004) and PALB2 enables key biochemical functions of BRCA2 (Xia et al., 2006). Materials and methods The dual requirement for HRR and TLS during Cell lines crosslink repair (Thompson et al., 2005) may be Chinese hamster ovary (CHO) cell line AA8 and derived envisioned as follows withrespect to thetwo major mutants UV40, NM3 and irs1SF (Xrcc3-mutated) have been FANCG-containing protein complexes such that the FA described previously (Busch et al., 1996; Liu et al., 1998; ‘pathway’ facilitates crosslink repair in a branched Lamerdin et al., 2004). KO40 is a FancG knockout (Tebbs manner. The newly identified (minimal) D1-D2-G-X3 et al., 2005) and 40BP6 is its complemented counterpart complex likely serves to initiate or modulate HRR at (corrected withgenomic CHO FancG). Human cell lines HeLa broken replication forks in a FANCD2-FANCI mono- and CRL-1583, the FA cell lines 326SV (FA-G), PD20 ubiquitylation-dependent manner (Smogorzewska et al., (FA-D2) and complemented counterparts 326SV þ FANCG, 2007), whereas the FANCA/B/C/E/F/G/L/M/FAAP24/ PD20-3-15, PD20 þ K561R-FANCD2 have been described FAAP100 core complex (which is required for this (Whitney et al., 1995; Garcia-Higuera et al., 2001; Hussain et al., 2004; Qiao et al., 2004). Hamster V-C8 and human monoubiquitylation) promotes TLS. Mutant fancg CAPAN-1 cells, mutant for BRCA2, have been described human and hamster cells expressing FANCG-S7A (Goggins et al., 1996; Wiegant et al., 2006). Fibroblast cells protein retain core complex formation and display were maintained as previously described (Johnson and Jones, sensitivity to MMC that is intermediate between that 1999; Johnson et al., 2000). Lymphoblastoid BD180, of wild-type and fancg null cells (Qiao et al., 2004). This GM02188 (wild-type), GM01389D (ATM-deficient cells) and genetic separation of FANCG function is consistent DK0064 (Seckel syndrome cells withimpaired ATR function), with our hypothesis that there is an intact TLS pathway and FA cell lines HSC72, BD220 (FA-A), BD215 (FA-C), in FANCG-S7A expressing cells, which presumably lack EUFA409 (FA-E), EUFA121 (FA-F), EUFA673 and the functionality of D1-D2-G-X3 in HRR. EUFA143 (FA-G) were grown in RPMI1640 medium with The concept of a FA ‘pathway’, which branches into 15% fetal calf serum. NM3 and EUFA143 cell lines expressing FANCG with phosphorylation site mutations have been HR and TLS components that can operate indepen- described and characterized previously (Mi et al., 2004; Qiao dently, downstream of an intact nuclear core complex, et al., 2004). Cell lines were generously provided by Alan in the repair of interstrand crosslinks (Wang, 2007) is D’Andrea (PD20’s), Margaret Zdzienicka (V-C8), Penny consistent withdata for otherDNA lesions. For Jeggo (GM02188/ATM-deficient/ATR-Seckel) and Hans example, hamster fancg cells have significant hypersen- Joenje (some EUFA lines). sitivity to killing by simple alkylating agents (ENU, MMS), UV-C and bleomycin (Wilson et al., 2001; Tebbs Immunoblotting and co-immunoprecipitation et al., 2005), while several human complementation Western blotting was performed as described previously groups, including FA-G cells, also show appreciable (Wilson et al., 2001). Antibodies to FANCD2 (ab2187), b- bleomycin and MMS sensitivity (Carreau et al., 1999). actin (ab6276) and IgG (ab6802) were obtained from Abcam We propose that this enhanced killing may be attributed Limited (Cambridge, UK) and BRCA-2 antibody (Ab-2) was to compromised HR repair of broken replication forks obtained from Oncogene ResearchProducts (Merck, West in the FANCG-defective cells. Conversely, reduced hprt Drayton, UK). FANCA, FANCG (H:83-622, R no. 2942-44) fancg and FANCF (H:1-374, R no. 6305-07) antibodies were as mutagenesis observed in hamster cells after ENU described previously (Kupfer et al., 1997; Waisfisz et al., 1999; and UV-C, results from loss of FANCG-mediated TLS de Winter et al., 2000). Preparation of a phospho-specific (Hinz et al., 2006, 2007). This branched role of the FA antibody raised against phosphoserine 7 of FANCG (Qiao pathway in HR-mediated repair of broken replication et al., 2004) and the XRCC3 antibody (Hussain et al., 2006) forks was also recently discussed in the context of a were described previously. Co-immunoprecipitation was per- requirement for gH2AX (Lyakhovich and Surralles, formed using Sigma’s (Poole, UK) EZview red protein affinity 2007), which interacts with FANCD2 in a UV-C and gel system (Hussain et al., 2004, 2006). Total cell extracts were BRCA1 dependent manner (Bogliolo et al., 2007). prepared from 1–2 Â 107 exponentially growing cells either In conclusion, our studies indicate a function for untreated or treated withMMC for 18 h.Antibodies described FANCG that is distinct from its role in the nuclear core above and FANCA-antisera (Abcam Limited, ab5063) were used for immunoprecipitation. For some experiments DNAse complex. It is apparent that protein–protein interactions at 1 mgmlÀ1 was added to total cell extracts and incubated at mediated by FANCG depend on its TPR motifs (Hussain 37 1C overnight prior to the addition of antibody. When used, et al., 2006), which act as scaffolds to facilitate two wortmannin was added to cells 18 hbefore thepreparation of functionally discrete complexes, the core complex and D1- extracts. D2-G-X3. The assembly of this latter complex depends on ATR-mediated FANCG Ser7 phosphorylation, suggesting Chicken DT40 cell lines and gene targeting that discrete phosphorylation sites determine the separate The xrcc3 cell line and the subsequent generation of a functions of FANCG. Our studies indicate that the conditional xrcc3-deficient DT40 cell line were previously functional link between certain FA proteins and HRR described (Takata et al., 1998; Hirano et al., 2005). This cell

Oncogene FANCG required for BRCA2–FANCD2–XRCC3 interactions JB Wilson et al 3650 line harbours human Xrcc3-IRES-EGFP expression cassette, (MT) for expert technical assistance, the Fanconi Anemia as well as Cre recombinase fused withtheligand-binding ResearchFund for nurturing researchcollaborations and the domain of the oestrogen receptor and designated MerCreMer NorthWest Cancer ResearchFund (UK) for financial (Zhang et al., 1998). The FANCG gene was disrupted in this support. This work was funded by grants NWCRF-CR624 cell line using a FANCG targeting vector (Yamamoto et al., and NWCRF-CR751 (NJJ). MT laboratory was supported in 2003), then the hXrcc3 expression cassette was deleted by part by grants from the Ministry of Education, Culture, addition of 4-hydroxy tamoxifen, leading to removal of Xrcc3 Sports, Science and Technology of Japan with financial expression. Cell culture and determination of sensitivity to support also provided by the Naito Foundation, and the cisplatin by colony formation was as described (Takata et al., Sagawa Foundation for Promotion of Cancer Research. SH 2001; Yamamoto et al., 2003). For MMC-induced growth and CGM acknowledge support from the Medical Research inhibition, 5 Â 105 cells were seeded into six-well dishes and Council and the Daniel Ayling Fanconi Anemia Trust. The surviving cells were counted after 72 h. contribution by LHT was performed under the auspices of the US Department of Energy by Lawrence Livermore National Acknowledgements Laboratory under Contract DE-AC52-07NA27344 and funded by the DOE Low-Dose Program and NCI/NIH grant We thank all individuals who generously provided reagents, CA112566. GMK acknowledges grant funding from NHLBI Yuxuan Xiao (NJJ), Keiko Namikoshi and Masayo Kimura R01 HL063776.

References

Alderton GK, Joenje H, Varon R, Borglum AD, Jeggo PA, O’Driscoll interact in a functional nuclear complex. Mol Cell Biol 19: M. (2004). Seckel syndrome exhibits cellular features demonstrating 4866–4873. defects in the ATR-signalling pathway. Hum Mol Genet 13: Garcia-Higuera I, Taniguchi T, Ganesan S, Meyn MS, Timmers C, 3127–3138. Hejna J et al. (2001). Interaction of the Fanconi anemia proteins and Alter BP. (1996). Fanconi’s anemia and malignancies. Am J Hematol BRCA1 in a common pathway. Mol Cell 7: 249–262. 53: 99–110. Goggins M, Schutte M, Lu J, Moskaluk CA, Weinstein CL, Petersen Blom E, van de Vrugt HJ, de Vries Y, de Winter JP, Arwert F, Joenje GM et al. (1996). Germline BRCA2 gene mutations in patients H. (2004). Multiple TPR motifs characterize the Fanconi anemia withapparently sporadic pancreatic carcinomas. Cancer Res 56: FANCG protein. DNA Repair 3: 77–84. 5360–5364. Bogliolo M, Lyakhovich A, Callen E, Castella M, Cappelli E, Ramirez Gordon SM, Buchwald M. (2003). Fanconi anemia protein complex: MJ et al. (2007). Histone H2AX and Fanconi anemia FANCD2 mapping protein interactions in the yeast 2- and 3-hybrid systems. function in the same pathway to maintain chromosome stability. Blood 102: 136–141. EMBO J 26: 1340–1351. Gregory RC, Taniguchi T, D’Andrea AD. (2003). Regulation of the Bridge WL, Vandenberg CJ, Franklin RJ, Hiom K. (2005). The BRIP1 Fanconi anemia pathway by monoubiquitination. Semin Cancer functions independently of BRCA1 in the Fanconi anemia Biol 13: 77–82. pathway for DNA crosslink repair. Nat Genet 37: 953–957. Groves MR, Barford D. (1999). Topological characteristics of helical BuschDB, Zdzienicka MZ, Natarajan AT, Jones NJ, Overkamp WJI, repeat proteins. Curr Opin Struct Biol 9: 383–389. Collins A et al. (1996). A CHO mutant, UV40, that is sensitive to Gudmundsdottir K, Ashworth A. (2006). The roles of BRCA1 and diverse mutagens and represents a new complementation group of BRCA2 and associated proteins in the maintenance of genomic mitomycin C sensitivity. Mutat Res 363: 209–221. stability. Oncogene 25: 5864–5874. Carreau M, Alon N, Bosnoyan-Collins L, Joenje H, Buchwald M. Henry-Mowatt J, Jackson D, Masson JY, Johnson PA, Clements PM, (1999). Drug sensitivity spectra in Fanconi anemia lymphoblastoid Benson FE et al. (2003). XRCC3 and Rad51 modulate replication cell lines of defined complementation groups. Mutat Res 435: fork progression on damaged vertebrate chromosomes. Mol Cell 11: 103–109. 1109–1117. Ciccia A, Ling C, Coulthard R, Yan ZJ, Xue YT, Meetei AR et al. Hinz JM, Nham PB, Salazar EP, Thompson LH. (2006). The Fanconi (2007). Identification of FAAP24, a Fanconi anemia core complex anemia pathway limits the severity of mutagenesis. DNA Repair 5: protein that interacts with FANCM. Mol Cell 25: 331–343. 875–884. Collins N, Kupfer GM. (2005). Molecular pathogenesis of Fanconi Hinz JM, Nham PB, Urbin SS, Jones IM, Thompson LH. (2007). anemia. Int J Hematol 82: 176–183. Disparate contributions of the Fanconi anemia pathway and Davies AA, Masson JY, McLlwraithMJ, Stasiak AZ, Stasiak A, homologous recombination in preventing spontaneous mutagenesis. Venkitaraman AR et al. (2001). Role of BRCA2 in control of Nucleic Acids Res 35: 3733–3740. the RAD51 recombination and DNA repair protein. Mol Cell 7: Hirano S, Yamamoto K, Ishiai M, Yamazoe M, Seki M, Matsushita N 273–282. et al. (2005). Functional relationships of FANCC to homologous de Winter JP, van der Weel L, de Groot J, Stone S, Waisfisz Q, Arwert recombination, translesion synthesis, and BLM. EMBO J 24: F et al. (2000). The Fanconi anemia protein FANCF forms a 418–427. nuclear complex withFANCA, FANCC and FANCG. Hum Mol Ho GPH, Margossian S, Taniguchi T, D’Andrea AD. (2006). Genet 9: 2665–2674. Phosphorylation of FANCD2 on two novel sites is required for de Winter JP, Waisfisz Q, Rooimans MA, van Berkel CGM, mitomycin C resistance. Mol Cell Biol 26: 7005–7015. Bosnoyan-Collins L, Alon N et al. (1998). The Fanconi anaemia Howlett NG, Taniguchi T, Olson S, Cox B, Waisfisz Q, de Die- group G gene FANCG is identical withXRCC9. Nat Genet 20: Smulders C et al. (2002). Biallelic inactivation of BRCA2 in Fanconi 281–283. anemia. Science 297: 606–609. Dronkert MLG, Kanaar R. (2001). Repair of DNA interstrand cross- Hussain S, Wilson JB, Blom E, Thompson LH, Sung P, Gordon SM links. Mutat Res 486: 217–247. et al. (2006). Tetratricopeptide-motif-mediated interaction of Esashi F, Christ N, Gannon J, Liu YL, Hunt T, Jasin M et al. FANCG withrecombination proteins XRCC3 and BRCA2. DNA (2005). CDK-dependent phosphorylation of BRCA2 as a Repair 5: 629–640. regulatory mechanism for recombinational repair. Nature 434: Hussain S, Wilson JB, Medhurst AL, Hejna J, Witt E, Ananth S 598–604. et al. (2004). Direct interaction of FANCD2 withBRCA2 in Garcia-Higuera I, Kuang Y, Naf D, Wasik J, D’Andrea AD. (1999). DNA damage response pathways. Hum Mol Genet 13: Fanconi anemia proteins FANCA, FANCC, and FANCG/XRCC9 1241–1248.

Oncogene FANCG required for BRCA2–FANCD2–XRCC3 interactions JB Wilson et al 3651 Hussain S, Witt E, Huber PAJ, Medhurst AL, Ashworth A, Mathew Pace P, Johnson M, Tan WM, Mosedale G, Sng C, Hoatlin M et al. CG. (2003). Direct interaction of the Fanconi anaemia protein (2002). FANCE: the link between Fanconi anaemia complex FANCG withBRCA2/FANCD1. Hum Mol Genet 12: 2503–2510. assembly and activity. EMBO J 21: 3414–3423. Johnson MA, Bryant PE, Jones NJ. (2000). Isolation of camptothecin- Patel KJ, Joenje H. (2007). Fanconi anemia and DNA replication sensitive Chinese hamster cell mutants: phenotypic heterogeneity repair. DNA Repair 6: 885–890. within the ataxia telangiectasia-like XRCC8 (irs2) complementation Qiao FY, Mi J, Wilson JB, Zhi G, Bucheimer NR, Jones NJ et al. group. Mutagenesis 15: 367–374. (2004). Phosphorylation of Fanconi anemia (FA) complementation Johnson MA, Jones NJ. (1999). The isolation and genetic analysis of group G protein, FANCG, at serine 7 is important for function of V79-derived etoposide sensitive Chinese hamster cell mutants: two the FA pathway. J Biol Chem 279: 46035–46045. new complementation groups of etoposide sensitive mutants. Mutat Reid S, Schindler D, Hanenberg H, Barker K, Hanks S, Kalb R et al. Res 435: 271–282. (2007). Biallelic mutations in PALB2 cause Fanconi anemia subtype Kennedy RD, D’Andrea AD. (2005). The Fanconi anemia/BRCA FA-N and predispose to childhood cancer. Nat Genet 39: 162–164. pathway: new faces in the crowd. Genes Dev 19: 2925–2940. Sarkaria JN, Tibbetts RS, Busby EC, Kennedy AP, Hill DE, Abraham Kupfer GM, Naf D, Suliman A, Pulsipher M, D’Andrea AD. (1997). RT. (1998). Inhibition of phosphoinositide 3-kinase related The Fanconi anaemia proteins, FAA and FAG, interact to form a kinases by the radiosensitizing agent wortmannin. Cancer Res 58: nuclear complex. Nat Genet 17: 487–490. 4375–4382. Lamb JR, TugendreichS, Hieter P. (1995). Tetratrico peptide repeat Smogorzewska A, Matsuoka S, Vinciguerra P, McDonald ER, Hurov interactions—to Tpr or not to Tpr. Trends Biochem Sci 20: 257–259. KE, Luo J et al. (2007). Identification of the FANCI protein, a Lamerdin JE, Yamada NA, George JW, Souza B, Christian AT, Jones monoubiquitinated FANCD2 paralog required for DNA repair. NJ et al. (2004). Characterization of the hamster FancG/Xrcc9 gene Cell 129: 289–301. and mutations in CHOUV40 and NM3. Mutagenesis 19: 237–244. Takata M, Sasaki MS, Sonoda E, Morrison C, Hashimoto M, Utsumi Levitus M, Joenje H, de Winter JP. (2006). The Fanconi anemia H et al. (1998). Homologous recombination and non-homologous pathway of genomic maintenance. Cell Oncol 28: 3–29. end-joining pathways of DNA double-strand break repair have Ling C, Ishiai M, Ali AM, Medhurst AL, Neveling K, Kalb R et al. overlapping roles in the maintenance of chromosomal integrity in (2007). FAAP100 is essential for activation of the Fanconi vertebrate cells. EMBO J 17: 5497–5508. anemia-associated DNA damage response pathway. EMBO J 26: Takata M, Sasaki MS, Tachiiri S, Fukushima T, Sonoda E, Schild D 2104–2114. et al. (2001). Chromosome instability and defective recombinational Liu N, Lamerdin JE, Tebbs RS, Schild D, Tucker JD, Shen MR et al. repair in knockout mutants of the five Rad51 paralogs. Mol Cell (1998). XRCC2 and XRCC3, new human Rad51-family members, Biol 21: 2858–2866. promote chromosome stability and protect against DNA cross-links Taniguchi T, D’Andrea AD. (2006). Molecular pathogenesis of and other damages. Mol Cell 1: 783–793. Fanconi anemia: recent progress. Blood 107: 4223–4233. Liu N, Schild D, Thelen MP, Thompson LH. (2002). Involvement of Taniguchi T, Garcia-Higuera I, Xu B, Andreassen PR, Gregory RC, Rad51C in two distinct protein complexes of Rad51 paralogs in Kim ST et al. (2002). Convergence of the Fanconi anemia and ataxia human cells. Nucleic Acids Res 30: 1009–1015. telangiectasia signaling pathways. Cell 109: 459–472. Lyakhovich A, Surralles J. (2007). New roads to FA/BRCA path- Tebbs RS, Hinz JM, Yamada NA, Wilson JB, Salazar EP, way—H2AX. Cell Cycle 6: 1019–1023. Thomas CB et al. (2005). New insights into the Fanconi anemia Masson JY, Tarsounas MC, Stasiak AZ, Stasiak A, Shah R, pathway from an isogenic FancG hamster CHO mutant. DNA McIlwraithMJ et al. (2001). Identification and purification of two Repair 4: 11–22. distinct complexes containing the five RAD51 paralogs. Genes Dev Thacker J. (2005). The RAD51 gene family, genetic instability and 15: 3296–3307. cancer. Cancer Lett 219: 125–135. Mathew CG. (2006). Fanconi anaemia genes and susceptibility to Thompson LH. (2005). Unraveling the Fanconi anemia-DNA repair cancer. Oncogene 25: 5875–5884. connection. Nat Genet 37: 921–922. Meetei AR, Medhurst AL, Ling C, Xue YT, Singh TR, Bier P et al. Thompson LH, Hinz JM, Yamada NA, Jones NJ. (2005). How (2005). A human ortholog of archaeal DNA repair protein Hef is Fanconi anemia proteins promote the four Rs: replication, defective in Fanconi anemia complementation group M. Nat Genet recombination, repair, and recovery. Environ Mol Mutagen 45: 37: 958–963. 128–142. Mi J, Qiao F, Wilson JB, HighAA, SchroederMJ, Stukenberg PT Waisfisz Q, de Winter JP, Kruyt FAE, de Groot J, van der Weel L, et al. (2004). FANCG is phosphorylated at serines 383 and 387 Dijkmans LM et al. (1999). A physical complex of the Fanconi during mitosis. Mol Cell Biol 24: 8576–8585. anemia proteins FANCG/XRCC9 and FANCA. Proc Natl Acad Sci Mirchandani KD, D’Andrea AD. (2006). The Fanconi anemia/BRCA USA 96: 10320–10325. pathway: a coordinator of cross-link repair. Exp Cell Res 312: Wang WD. (2007). Emergence of a DNA-damage response network 2647–2653. consisting of Fanconi anaemia and BRCA proteins. Nat Rev Genet Nakanishi K, Yang YG, Pierce AJ, Taniguchi T, Digweed M, 8: 735–748. D’Andrea AD et al. (2005). Human Fanconi anemia monoubiqui- Wang XZ, Andreassen PR, D’Andrea AD. (2004). Functional tination pathway promotes homologous DNA repair. Proc Natl interaction of monoubiquitinated FANCD2 and BRCA2/FANCD1 Acad Sci USA 102: 1110–1115. in chromatin. Mol Cell Biol 24: 5850–5862. Niedernhofer LJ. (2007). The Fanconi anemia signalosome anchor. Whitney M, Thayer M, Reifsteck C, Olson S, Smith L, Jakobs PM Mol Cell 25: 487–490. et al. (1995). Microcell mediated chromosome transfer maps the Niedzwiedz W, Mosedale G, Johnson M, Ong CY, Pace P, Patel KJ. Fanconi-anemia-group-D gene to chromosome-3p. Nat Genet 11: (2004). The Fanconi anaemia gene FANCC promotes homologous 341–343. recombination and error-prone DNA repair. Mol Cell 15: 607–620. Wiegant WW, Overmeer RM, Godthelp BC, van Buul PPW, O’Driscoll M, Ruiz-Perez VL, Woods CG, Jeggo PA, Goodship JA. Zdzienicka MZ. (2006). Chinese hamster cell mutant, (2003). A splicing mutation affecting expression of ataxia-telangiec- V-C8, a model for analysis of BRCA2 function. Mutat Res 600: tasia and Rad3-related protein (ATR) results in Seckel syndrome. 79–88. Nat Genet 33: 497–501. Wilson JB, Johnson MA, Stuckert AP, Trueman KL, May S, Bryant Ohashi A, Zdzienicka MZ, Chen JJ, Couch FJ. (2005). PE et al. (2001). The Chinese hamster FANCG/XRCC9 mutant Fanconi anemia complementation group D2 (FANCD2) functions NM3 fails to express the monoubiquitinated form of the FANCD2 independently of BRCA2- and RAD51-associated homologous protein, is hypersensitive to a range of DNA damaging agents and recombination in response to DNA damage. J Biol Chem 280: exhibits a normal level of spontaneous sister chromatid exchange. 14877–14883. Carcinogenesis 22: 1939–1946.

Oncogene FANCG required for BRCA2–FANCD2–XRCC3 interactions JB Wilson et al 3652 Xia B, Dorsman JC, Ameziane N, de Vries Y, Rooimans MA, Sheng phosphorylation and FAA/FAC nuclear accumulation. Proc Natl Q et al. (2007). Fanconi anemia is associated witha defect in the Acad Sci USA 95: 13085–13090. BRCA2 partner PALB2. Nat Genet 39: 159–161. Yamashita YM, Okada T, Matsusaka T, Sonoda E, Zhao GY, Araki Xia B, Sheng Q, Nakanishi K, Ohashi A, Wu JM, Christ N et al. K et al. (2002). RAD18 and RAD54 cooperatively maintenance of (2006). Control of BRCA2 cellular and clinical functions by a genomic stability cells. EMBO J 21: 5558–5566. nuclear partner, PALB2. Mol Cell 22: 719–729. Yang YG, Herceg Z, Nakanishi K, Demuth I, Piccoli C, Michelon J Yamada NA, Hinz JM, Kopf VL, Segalle KD, Thompson LH. (2004). et al. (2005). The Fanconi anemia group A protein modulates XRCC3 ATPase activity is required for normal XRCC3-Rad51C homologous repair of DNA double-strand breaks in mammalian complex dynamics and homologous recombination. J Biol Chem cells. Carcinogenesis 26: 1731–1740. 279: 23250–23254. Zhang N, Liu X, Li L, Legerski R. (2007). Double-strand breaks Yamamoto K, Ishiai M, Matsushita N, Arakawa H, Lamerdin JE, induce homologous recombinational repair of interstrand cross- Buerstedde JM et al. (2003). Fanconi anemia FANCG protein in links via cooperation of MSH2, ERCC1-XPF, REV3, and the mitigating radiation- and enzyme-induced DNA double-strand Fanconi anemia pathway. DNA Repair 6: 1670–1678. breaks by homologous recombination in vertebrate cells. Mol Cell Zhang Y, Wienands J, Zurn C, Reth M. (1998). Induction of the Biol 23: 5421–5430. antigen receptor expression on B lymphocytes results in rapid Yamashita T, Kupfer GM, Naf D, Suliman A, Joenje H, Asano S competence for signaling of SLP-65 and Syk. EMBO J 17: et al. (1998). The Fanconi anemia pathway requires FAA 7304–7310.

Oncogene