BRCA1-Interacting Protein OLA1 Requires Interaction with BARD1 To

Published OnlineFirst June 1, 2018; DOI: 10.1158/1541-7786.MCR-18-0269

Oncogenes and Tumor Suppressors Molecular Cancer Research BRCA1-Interacting Protein OLA1 Requires Interaction with BARD1 to Regulate Centrosome Number Yuki Yoshino1, Huicheng Qi1, Hiroki Fujita1, Matsuyuki Shirota2, Shun Abe1, Yuhei Komiyama1, Kazuha Shindo1, Masahiro Nakayama3, Ayako Matsuzawa3, Akihiro Kobayashi1, Honami Ogoh4, Toshio Watanabe4, Chikashi Ishioka5, and Natsuko Chiba1

Abstract

BRCA1 functions as a tumor suppressor in DNA repair and lation, and two ATP-binding residues caused centrosome centrosome regulation. Previously, Obg-like ATPase 1 (OLA1) amplification and identified five missense mutants that are was shown to interact with BARD1, a heterodimer partner of deficient in the regulation of centrosome number. Three of BRCA1. OLA1 binds to BRCA1, BARD1, and g-tubulin and them did not bind to BARD1. Two phosphomimetic mutations functions in centrosome regulation. This study determined that restored the binding to BARD1 and the efficient centrosome overexpression of wild-type OLA1 (OLA1-WT) caused centro- amplification by their overexpression. Knockdown and over- some amplification due to centriole overduplication in mam- expression of BARD1 also caused centrosome amplification. mary tissue–derived cells. Centrosome amplification induced BARD1 mutant reported in cancer failed to bind to OLA1 and by overexpression of the cancer-derived OLA1 mutant, which is rescue the BARD1 knockdown-induced centrosome amplifica- deficient at regulating centrosome number, occurred in signif- tion and reduced its centrosomal localization. Combined, these icantly fewer cells than in that induced by overexpression of data reveal that the OLA1–BARD1 interaction is important for OLA1-WT. Thus, it was hypothesized that overexpression of the regulation of centrosome number. OLA1 with normal function efficiently induces centrosome amplification, but not that of OLA1 mutants, which are defi- Implications: Regulation of centrosome number by BRCA1/ cient at regulating centrosome number. We analyzed whether BARD1 together with OLA1 is important for the genome overexpression of OLA1 missense mutants of nine candidate integrity to prevent tumor development. Mol Cancer Res; 1–13. phosphorylation residues, three residues modified with acety- 2018 AACR.

Introduction functions as a ubiquitin ligase (3). BRCT domains are present in many proteins that function in DNA repair and at cell-cycle BRCA1 is a breast and ovarian cancer suppressor (1) and forms checkpoints and are responsible for the binding to phosphory- a heterodimer with BARD1 (2). Both BRCA1 and BARD1 have a lated proteins (4). RING domain in their amino (N)-terminal region and two BRCT The BRCA1/BARD1 heterodimer functions in DNA repair domains in their carboxy (C)-terminal region. The dimerization and centrosome regulation (5). Centrosomes are the major of BRCA1 and BARD1 is mediated by interaction of their microtubule nucleation centers and function in the formation N-terminal regions involving the RING domains and the dimer ofthemitoticspindle.ThesinglecentrosomeinG1 phase duplicates only once in S phase (6). The centrosome contains a pair of centrioles, called the mother and daughter centrioles, 1Department of Cancer Biology, Institute of Development, Aging and Cancer 2 and centrosome duplication is initiated by their physical sep- (IDAC), Tohoku University, Sendai, Japan. Division of Interdisciplinary Medical aration (centriole disengagement) in late mitosis-early G Science, Graduate School of Medicine, Tohoku University, Sendai, Japan. 1 3Department of Molecular Immunology, Institute of Development, Aging and phase. In S phase, a daughter centriole forms perpendicular Cancer (IDAC), Tohoku University, Sendai, Japan. 4Department of Biological to each mother centriole. Overduplication of centrioles leads to Science, Graduate School of Humanities and Sciences, Nara Women's University, centrosome amplification, which is the best-characterized cen- Nara, Japan. 5Department of Clinical Oncology, Institute of Development, Aging trosomal abnormality in cancer and causes aneuploidy and and Cancer (IDAC), Tohoku University, Sendai, Japan. genome instability (7). Note: Supplementary data for this article are available at Molecular Cancer BRCA1 and BARD1 localize to the centrosome throughout the Research Online (http://mcr.aacrjournals.org/). cell cycle (8, 9). BRCA1 directly binds to g-tubulin, a component Corresponding Author: Natsuko Chiba, Institute of Development, Aging and of centrosomes (10), and the BRCA1/BARD1 heterodimer ubi- Cancer, Tohoku University, 4-1 Seiryomachi Aoba-ku, Sendai 980-8575, Japan. quitinates g-tubulin and other centrosomal proteins (11). Fibro- Phone: 81-22-717-8477; Fax: 81-22-717-8482; E-mail: blasts of Brca1 knockout mouse embryos show centrosome [email protected] amplification (12). Suppression of BRCA1 results in centrosome doi: 10.1158/1541-7786.MCR-18-0269 amplification in cells derived from mammary tissue (11). BRCA1 2018 American Association for Cancer Research. with mutations found in familial breast cancers are deficient in

www.aacrjournals.org OF1

Yoshino et al.

the regulation of centrosome number (13). Centrosome ampli- 10 mg/mL of insulin, 20 ng/mL of epidermal growth factor, and fication is significantly associated with lack of BRCA1 expression 0.5 mg/mL of hydrocortisone. Hs578T, MCF7, PC3, Saos-2, HEK- in breast cancer tissue (14). 293T, HCC1937, T47D, and MCF10A were obtained from ATCC. We identified Obg-like ATPase 1 (OLA1) as a protein that Cell line identities were verified using the GenomeLab Human interacts with BARD1 (15). OLA1 is highly conserved from STR Primer set (Beckman Coulter). Cells were transfected with bacteria to humans and is a member of the Obg family and vectors using Lipofectamine LTX or with siRNAs using Lipofecta- YchF subfamily of P-loop GTPases (16–18), while OLA1 has mine RNAiMAX (Invitrogen) or TransIT-X2 Dynamic Delivery both ATPase and GTPase activities (19–23). OLA1 is com- System (Mirus Bio). posed of a central guanine nucleotide-binding domain (G domain), flanked by a coiled-coil domain and a ThrRS-GTP- Plasmid construction SpoT (TGS) domain (19). The G domain is the basic functional To generate the HA-OLA1 expression vector, pCY4B-FLAG- unit of GTP-binding proteins (G proteins) and contains five OLA1 was digested and the OLA1 fragment was subcloned into characteristic sequences, G1–G5, which are involved in nucle- pCY4B-HA (28). pCY4B-HA-OLA1 mutants were generated by otide binding and hydrolysis (24). The function of the TGS site-directed mutagenesis of pCY4B-HA-OLA1. For the pCY4B- domain is so far unknown. OLA1 is involved in multiple FLAG-OLA1 mutants and pGEX-OLA1 mutant, pCY4B-HA-OLA1 cellular processes including protein translation, the antioxi- mutants were digested and the OLA1 fragments with mutations dant response, signal transduction, and cell proliferation (16, were subcloned into pCY4B-FLAG and pGEX (GE Healthcare). For 25, 26), and its overexpression is observed in many malignant pEGFP-centrin, centrin2 fragments were amplified from cDNA tumors (22). from HeLa cells, and the PCR products were subcloned into a OLA1 directly binds to BRCA1, BARD1, and g-tubulin and pEGFP vector (Clontech). For pCY4B-HA-BARD1-FL-WT, the localizes to centrosomes and the spindle pole. Knockdown of BARD1 fragments were amplified from p3XFLAG-BARD1, and OLA1 causes centrosome amplification in cell lines derived from the PCR products were subcloned into pCY4B-HA. pCY4B-HA- mammary tissue. OLA1 with a substitution mutation (E168Q) BARD1 mutants were generated by site-directed mutagenesis. For observed in a breast cancer cell line fails to bind to the amino (N)- pCY4B-HA-BARD1-546–777-WT and mutants, BARD1 fragments terminal region of BRCA1 and does not rescue OLA1 knockdown- were amplified from pCY4B-HA-BARD1. PCR products were induced centrosome amplification. Moreover, a BRCA1 variant, subcloned into pCY4B-HA. For pET-BARD1-546–777-V695L, I42V, which is deficient at regulating centrosome number, pCY4B-HA-BARD1-546–777-V695L was digested and the BARD1 decreases the binding activity to OLA1. These suggest that OLA1 fragment was subcloned into pET vector (Novagen). For Myc- plays an important role in centrosome regulation together with BARD1-V695L, the BARD1 fragment was amplified from pCY4B- BRCA1 (15). In general, supernumerary centrosomes arise from HA-BARD1-V695L, and PCR products were subcloned into centriole overduplication, centrosome fragmentation, and/or pCMV-Myc (Clontech). pCY4B-FLAG-OLA1, pCY4B-FLAG- centrosome accumulation caused by aborted cell division, for OLA1-E168Q, pEGFP-OLA1, pCY4B-HA-g-tubulin, pET-g-tubu- example, through failure of cytokinesis. Inhibition of BRCA1 and lin, pET-BRCA1-1-304, pET-BARD1-546–777 pGEX-OLA1, knockdown of OLA1 causes centrosome fragmentation and cen- pCMV-Myc-BARD1, pcDNA3-HA-BRCA1-WT, pcDNA3-HA- triole overduplication (15, 27). BRCA1-I42V, pCY4B-HA-BRCA1-1–304, and p3XFLAG-BARD1 In this study, we found that overexpression of wild-type OLA1 have been described previously (15, 29, 30). All constructs were (OLA1-WT) causes centrosome amplification. The extent of cen- verified by DNA sequencing. trosome amplification induced by overexpression of OLA1– E168Q was significantly lower than that by OLA1-WT. We there- Immunocytochemistry fore hypothesized that overexpression of functional OLA1- Cells were fixed with methanol:acetone (1:1) for 10 minutes at induced centrosome amplification and analyzed the effect on the 20C. After blocking with phosphate-buffered saline (PBS) centrosome of overexpression of OLA1 mutants. We identified a containing 1% bovine serum albumin (PBS-1% BSA), cells were further five OLA1 mutants, which decreased the number of cells incubated with anti–g-tubulin (Sigma), anti-HA (Y-11, Santa with centrosome amplification. Three of these mutants abolished Cruz Biotechnology), and/or anti-Cep170 antibodies (Invitro- the direct binding to BARD1. Furthermore, we identified BARD1 gen) in PBS-1% BSA for 4 hours. After washing with PBS contain- mutations that abolished the binding activity to OLA1 and the ing 0.05% Tween 20 (PBS-T), cells were incubated with Alexa function in the regulation of centrosome number. Taken together, Fluor 488 and 568-conjugated antibodies (Molecular Probes) in these suggest that the OLA1–BARD1 interaction is important for PBS/1% BSA for 30 minutes. Cells were washed with PBS-T and the regulation of centrosome number in addition to the OLA1– mounted in mounting medium containing DAPI (Vector BRCA1 interaction. Laboratories).

siRNA Materials and Methods siRNAs were synthesized using a Silencer siRNA Construction Cell lines and transfection Kit (Ambion) or synthesized in Japan Bio-service. The siRNA Hs578T cells were grown in Dulbecco's modiﬁed Eagle's medi- sequence of BARD1-1 was 50-AAGUGUAUGCUUGGGAUUCUC- um (DMEM) with 10% fetal bovine serum (FBS) and 10 mg/mL of 30 (31), that of BARD1-2 was 50-GAGUAAAGCUUCAGUGCAAA- insulin. MCF7, PC3, HeLa, U2OS, Saos-2, and HEK-293T cells 30 (32), and that of BARD1-3 was 50-AAGUCUAAUGCCA- were grown in DMEM with 10% FBS. HCC1937 cells were grown GAUUAGGA-30. The siRNA sequence of OLA1-1 was 50-AAGU- in RPMI-1640 medium with 10% FBS. T47D cells were grown in UGGGGAGAUGGGAAUCU-30 and that of OLA1-2 was 50-AAGC- RPMI-1640 medium with 10% FBS and 10 mg/mL of insulin. GAACAUGACACAAAGUG-30 (15). The Silencer-negative control MCF10A cells were grown in DMEM/Ham's F-12 1:1 with 5% FBS, siRNA template set was used as the control.

OF2 Mol Cancer Res; 2018 Molecular Cancer Research

OLA1 Requires BARD1 to Regulate Centrosome Number

Immunoprecipitation and immunoblotting The effect of overexpression of OLA1 harboring the E168Q Cell lysates were prepared in lysis buffer (10 mmol/L Hepes, pH mutation (OLA1-E168Q) was also analyzed. Western blot anal- 7.6, 250 mmol/L NaCl, 0.1% Nonidet P-40 (NP-40), 5 mmol/L ysis confirmed that the expression level of HA-OLA1-E168Q was EDTA) containing a protease inhibitor cocktail (Sigma). Whole similar to that of HA-OLA1-WT (Fig. 1C). Overexpression of cell lysates were prepared in 1 sample buffer (50 mmol/L Tris– OLA1-E168Q caused centrosome amplification, but the percent- HCl pH 6.8, 2% sodium dodecyl sulfate (SDS), 0.67 mol/L 2- age of cells with extra centrosomes was markedly lower than that mercaptoethanol, 12% glycerol, 1% Bromphenol blue), sonicat- of the cells transfected with OLA1-WT (Fig. 1B). ed, and incubated at 95C for 10 minutes. For immunoprecipi- Because OLA1-E168Q does not bind to the N-terminal region tation, anti-FLAG (M2, Sigma) or anti-Myc (9E10, Covance) of BRCA1 (15), we presumed that BRCA1 is involved in the antibodies and Protein G-Sepharose beads (GE Healthcare) were centrosome amplification induced by the overexpression of added to each cell lysate. Mixtures were incubated at 4C for 4 to OLA1. The effect of overexpression of OLA1 on centrosomes was 16 hours, and the beads were washed with lysis buffer. Samples therefore analyzed in the BRCA1-deficient breast cancer cell line were subjected to SDS–PAGE and immunoblotted using anti-HA HCC1937 (Fig. 1D). Although transfection of HA-OLA1-WT (HA.11, Covance or Y-11, Santa Cruz Biotechnology), anti-b-actin alone increased the number of cells with extra centrosomes, (Sigma), anti-GST, anti-His (GE Healthcare), anti-BARD1 (Santa cotransfection with HA-BRCA1-WT markedly enhanced the cen- Cruz Biotechnology, Bethyl), anti-BRCA1 (Ab-1, Calbiochem), trosome amplification, while cotransfection with HA-BRCA1- anti-Myc, anti-FLAG antibodies, or polyclonal antibodies specific I42V, which reduces the binding activity to OLA1, did slightly. to OLA1 residues 1–171. These suggest that overexpression of OLA1 induces centrosome amplification via an interaction with BRCA1 protein. GST pulldown assay His-tagged proteins were expressed in BL21-competent cells. Abnormal OLA1 expression causes centrosome amplification in The bacterial cell pellets were lysed with extraction buffer (50 cell lines derived from mammary tissue mmol/L phosphate buffer pH 8.0, 0.3 mol/L NaCl, 0.1% NP-40). Centrosome amplification by BRCA1 inhibition is observed Purification of His-tagged protein was performed using Ni-NTA- only in cell lines derived from mammary tissue (11). We have agarose (Qiagen). GST-tagged proteins were expressed in BL21 reported that OLA1 knockdown results in centrosome amplifica- cells and the bacterial cell pellets were lysed with sonication buffer tion in cell lines derived from mammary tissue (15). We analyzed (20 mmol/L Tris-OAc pH 7.9, 120 mmol/L KoAc, 10% glycerol, the effect of overexpression of OLA1 on centrosomes in three cell 1% Triton X-100, 1 mmol/L EDTA, 1 mmol/L DTT). The super- lines derived from mammary tissue and four cell lines derived natant was incubated with glutathione-sepharose beads (GE from other tissue (Fig. 1E; Supplementary Fig. S1C). MCF7 and Healthcare) at 4 C for 2 hours. The beads were washed and T47D are breast cancer cell lines and MCF10A is a cell line derived incubated at 4 C for 2–16 hours with purified His-tagged pro- from normal human mammary epithelia. HeLa and PC3 are teins. The beads were washed and samples were immunoblotted. derived from cervical cancer and prostate cancer, respectively; U2OS and Saos-2 are derived from osteosarcoma. Transfection Three-dimensional structures and computational docking with HA-OLA1 caused centrosome amplification in cell lines model derived from mammary tissue, but not in those derived from Three-dimensional structures of OLA1 (PDB ID 2OHF) and other tissues. BARD1 C-terminal domain (PDB ID 2NTE) were downloaded We analyzed the effect of OLA1 knockdown on centrosome in from Protein Data Bank Japan. Missing residues in the OLA1 cell lines, not derived from mammary tissue (Supplementary Fig. structure were modeled using MODELLERv9.10 (33). Rigid body S2A and S2B). In these cell lines, knockdown of OLA1 did not docking of OLA1 and BARD1 was performed using ZDOCK3.0.2 cause centrosome amplification. (34), generating 2,000 docked poses. Hydrogen atoms were added to these complex structures, and they were reranked using Overexpression of OLA1 causes overduplication of centrioles ZRANK (35). The final docking pose was selected by the ZRANK Centrin localizes to both mother and daughter centrioles and is ranking and the interface complementarity. used as a marker for both centrioles. Each animal cell normally has two or four centrin-positive spots distributed in pairs. Cep170 is Results used as a marker for mother centrioles only and, therefore, Overexpression of OLA1 causes centrosome amplification via double-immunofluorescence staining for Cep170 with centrin interaction with BRCA1 could discriminate centriole overduplication from centrosome OLA1 is overexpressed in many malignancies (22). To analyze accumulation (36). Cells with extra centrioles due to centriole whether overexpression of OLA1-WT affects centrosomes in breast overduplication have only one or two Cep170-positive centrioles. cancer cells, Hs578T cells were transfected with either control By contrast, cells with extra centrioles due to aborted cell divisions (HA) or the HA-tagged OLA1-WT (HA-OLA1-WT) plasmid, and accumulate at least two and often more than two Cep170-positive immunostained with anti-HA and g-tubulin antibodies to detect centrioles. exogenous OLA1 and centrosomes, respectively (Fig. 1A). Cells Inhibition of BRCA1 and knockdown of OLA1 increase the with amplified centrosomes (more than two) were counted and number of cells with extra centrioles, which are found frequently the percentage was calculated (Fig. 1B). Of the cells transfected not paired and have only one or two Cep170-positive centrioles, with HA-OLA1-WT, only cells stained with the anti-HA antibody suggesting centrosome fragmentation and centriole overduplica- were analyzed (Supplementary Fig. S1A). Overexpression of tion (15, 27). Therefore, we analyzed the effect of OLA1 over- OLA1-WT caused centrosome amplification (Fig. 1A and B). expression on centrioles. To visualize centriole and identify the Expression of HA-OLA1-WT was confirmed by Western blotting transfected cells, Hs578T cells were transfected with control vector with an anti-OLA1 antibody (Supplementary Fig. S1B). or HA-OLA1 along with GFP-centrin and then immunostained

www.aacrjournals.org Mol Cancer Res; 2018 OF3

Yoshino et al.

** A HA g-tubulin Merge B ** % **

60 Control

20 HA-OLA1 -WT Cellsextra withcentrosomes 0

Control

HA-OLA1 -E168Q C

Vector Control IB: HA OLA1

** IB: b-Actin b-Actin D ** ** ** % ** % ** E 20 Control 60 HA-OLA1-WT 15 ** 40 **

10 ** 20

Cellsextra with centrosomes 5 0 GFP-OLA1-WT - + + + Cellsextra withcentrosomes HA-BRCA1-WT - --+ 0 HA-BRCA1-I42V - - - + MCF7 T47D MCF10A HeLa U2OS PC3 Saos-2

Figure 1. Overexpression of OLA1 causes centrosome amplification via an interaction with BRCA1 in cell lines derived from mammary tissue. A, Hs578T cells were transfected with control vector (HA), HA-OLA1-WT, or HA-OLA1-E168Q expression vector. Thirty-six hours after transfection, cells were fixed and stained with anti-HA and anti–g-tubulin antibodies and DAPI. Representative images of the cells are shown. Scale bar, 10 mm. B, The cells were counted, and the percentages of cells with amplified centrosomes (more than two) were scored. C, Cell lysates were immunoblotted (IB) with anti-HA and anti–b-actin antibodies. D, HCC1937 cells were transfected with GFP-OLA1-WT and HA-BRCA1-WT or HA-BRCA1-I42V. Cells were stained with anti-HA and anti–g-tubulin antibodies. The cells were counted, and the percentages of cells with amplified centrosomes were scored. E, MCF7, T47D, MCF10A, HeLa, U2OS, PC3, and Saos-2 were transfected with control vector (HA) or HA-OLA1-WT vector. The cells were coimmunostained with anti-HA and g-tubulin antibodies and stained with DAPI. Of the cells transfected with HA-OLA1-WT, only cells stained with the anti-HA antibody were analyzed. The percentages of cells with amplified centrosomes were scored. For B, D, and E, at least 100 cells were counted in each experiment. Histograms show the mean and SEM of three independent experiments. Statistical significance is indicated by asterisks (, P < 0.01) as calculated by the Student t test.

with an anti-Cep170 antibody. We found that overexpression of to the mother centrioles, and centrosome fragmentation was HA-OLA1 increased the number of cells with extra centrioles unclear (Fig. 2A). (more than four; Fig. 2A and B) but only one or two Cep170- positive centrioles (Fig. 2C); these centrioles were mostly Overexpression of missense mutants of OLA1 fails to cause unpaired (Fig. 2A). These suggest that overexpression of OLA1 centrosome ampliﬁcation causes centriole overduplication. However, most of the extra OLA1-E168Q mutant does not rescue OLA1 knockdown– daughter centrioles in OLA1-overexpressing cells were close induced centrosome ampliﬁcation (15). This suggests that

OF4 Mol Cancer Res; 2018 Molecular Cancer Research

OLA1 Requires BARD1 to Regulate Centrosome Number

A B % Cep170 GFP-Centrin Merge **

20 Control Figure 2. Centrosome amplification induced by OLA1 overexpression is due to overduplication of centrioles. A, 10 Hs578T cells were transfected with control vector (HA) or HA-OLA1 expression vector along with GFP- HA-OLA1 centrin expression vector. Cells were Cells with extra centrioles fixed and stained with anti-Cep170 0 antibody and DAPI. Representative images of the cells are shown. Scale bar, 10 mm. B, The cells were counted, Control and the percentages of cells with amplified centrioles (more than four) C were scored. C, Quantification of the % >4 centrioles control % >4 centrioles HA-OLA1 numbers of Cep170-positive centrioles in the cells with more than 40 40 four centrioles from the experiments shown in A. For B and C, at least 100 cells were counted in each 30 30 experiment. Histograms show the mean and SEM of three independent experiments. Statistical significance 20 20 is indicated by asterisks (, P < 0.01) t

as calculated by the Student test. % of Cells % of Cells 10 10

0 0 1 2 43 >4 1 2 43 >4 Cep170-positive centrioles Cep170-positive centrioles

OLA1-E168Q is deficient in the regulation of centrosome num- We analyzed whether these five mutants could rescue the ber. The proportion of cells overexpressing OLA1-E168Q and centrosome amplification induced by knockdown of OLA1. showing centrosome amplification was significantly lower than Hs578T cells were transfected with OLA1 siRNA, OLA1-1 against that in cells overexpressing OLA1-WT (Fig. 1B); we hypothesized the 30UTR of OLA1, and the expression vectors, HA-OLA1-WT or that overexpression of OLA1 with normal function efficiently mutants, and stained with anti–g-tubulin and anti-HA antibodies. induces centrosome amplification, but that overexpression of Cotransfection of OLA1-WT partially, but significantly, rescued OLA1 mutants, which are deficient at regulating centrosome the centrosome amplification induced by knockdown of OLA1, number, causes centrosome amplification in significantly fewer which were not observed with these five mutants similar to E168Q cells. To identify the residues of OLA1 that are important for mutant (Fig. 3F). These suggest that these five mutants are also centrosome regulation, we analyzed the efficiency of overex- deficient in the regulation of centrosome number (Supplemen- pressed OLA1 missense mutants at inducing the centrosome tary Table S2). amplification in breast cancer cells. The structure of OLA1 has been determined (19); S36, T124, Nine candidate phosphorylation residues, S36, S45, T124, F127, and K242 exist in the G domain, and E168 and T325 are T320, T325, T336, Y350, S358, and Y375, were chosen by located in the coiled-coil domain and TGS domain, respectively using NetPhos 2.0 Server (http://www.cbs.dtu.dk/services/Net (Fig. 3A and G). The F127 side-chain interacts with the adenine Phos/), and three residues modified with lysine acetylation, base of AMPPCP (an ATP analogue), whereas S36 hydrogen K216, K242, and K296, and two ATP-binding residues, F127 and bonds to a phosphate group of AMPPCP. N230, which have been previously identified (19, 37), were substituted alanine or arginine (Fig. 3A; Supplementary Table The OLA1–BARD1 interaction is an important regulator of S1). Compared with Hs578T cells transfected with OLA1-WT, in centrosome number cells transfected with five mutants, HA-OLA1-S36A, T124A, Because OLA1 binds to the C-terminal region of BARD1 F127A, K242R, or T325A, the proportion with centrosome ampli- (BARD1-546–777), g-tubulin, and the N-terminal region of fication was reduced to a level similar to that seen with HA-OLA1- BRCA1 (BRCA1-1–304) (Fig. 4A), we analyzed the binding ability E168Q overexpression (Fig. 3B; Supplementary Table S1). Expres- of OLA1 mutants. We prepared purified GST-tagged OLA1 (GST- sion of the HA-OLA1 mutants was confirmed by Western blot (Fig. OLA1) mutants and His-tagged BARD1-546–777 (His-BARD1- 1C, Fig. 3C–E). 546–777), g-tubulin, and BRCA1-1–304 and performed

www.aacrjournals.org Mol Cancer Res; 2018 OF5

Yoshino et al.

A (29-36) (54-57) (92-95) (230-233) (263-265) 1 G1 G2 G3 133 197 G4 G5 305 396 aa coiled-coil TGS

S36A S45A K242R K294R Y375A T124A F127A E168Q K216R T320A S358A N230A T325A T336A Y350A ** B ** ** ** % ** 80 **

Cells with extra extra Cells with centrosomes 0 WT S36A S45A F127A T325A T124A T336A T320A Y350A K216R E168Q K242R K294R Y375A N230A S358A Control C Vector T320A WT Y375A Control S358A S45A T325A Y350A T336A S36A T124A IB: HA OLA1

IB: b-Actin b-Actin

D E

Vector Vector WT N230A F127A K294R control control K242R WT K216R IB: HA OLA1 IB: HA OLA1 IB: b-Actin b-Actin IB: b-Actin b-Actin

F G AMPPCP F127 ** ** ** % G domain K242 ** S36 60 ** ** ** **

40 T124

Coiled-coil Cells with extra withCells centrosomes 0 domain siRNA - ++ + + + +++ T325

Vector - TGS domain WT WT S36A

T325A E168 F127A E168Q K242R T124A

Figure 3. Centrosome amplification induced by the overexpression of OLA1 missense mutants. A, Diagram of the structure and the locations of missense mutations of OLA1. OLA1 is composed of an N-terminal G domain, a coiled-coil domain, and a C-terminal TGS domain. The G domain has characteristic sequence motifs, G1–G5. B, Hs578T cells were transfected with control vector (HA) or vector expressing HA-OLA1-WT or mutants. Cells were coimmunostained with anti-HA and anti–g-tubulin antibodies, and stained with DAPI. The cells were counted, and the percentages of cells with amplified centrosomes were scored. C–E, Cell lysates were immunoblotted. F, Hs578T cells were transfected with control or OLA1 siRNA, OLA1-1, and vectors for HA-OLA1-WT, S36A, T124A, F127A, E168Q, K242R, or T325A. Cells were stained with anti–g-tubulin and anti-HA antibodies. The cells with amplified centrosomes were counted, and the percentages of cells were scored. For B and F, at least 100 cells were counted in each experiment. Histograms show the mean and SEM of three independent experiments. Statistical significance is indicated by asterisks (, P < 0.01) as calculated by the Student t test. G, Mapping of the residues apparently important for the regulation of centrosome number. The G domain, coiled-coil domain, and TGS domain are shown in red, yellow, and green, respectively. Residues that are deficient in the regulation of centrosome number are indicated with space filling. The ATP analogue AMPPCP is shown in sticks.

OF6 Mol Cancer Res; 2018 Molecular Cancer Research

OLA1 Requires BARD1 to Regulate Centrosome Number

A B Pull-down GST-OLA1 GST F127A Input E168Q WT K242R T124A T325A S36A

IB: His BARD1-546-777 kD 73 GST-OLA1 48

IB: GST

27 GST

C D HA-BARD1-546-777 - + + + + + + + +

FLAG-OLA1 - WT F127A WT S36A K242R T124A E168Q T325A

Input / IB: HA BARD1

IP: FLAG / IB: HA BARD1

Input / IB: FLAG OLA1

IP: FLAG / IB: FLAG OLA1

E F HA-γ-tubulin -+++ +++++ HA-BRCA1-1-304 - ++ + +++++ - FLAG-OLA1 - FLAG-OLA1 WT WT WT T325A S36A WT K242R F127A K242R T325A E168Q E168Q T124A F127A S36A T124A

Input / IB: HA γ-tubulin Input / IB: HA BRCA1

IP: FLAG / IB: HA γ-tubulin IP: FLAG / IB: HA BRCA1

Input / IB: FLAG OLA1 Input / IB: FLAG OLA1

IP: FLAG / IB: FLAG OLA1 IP: FLAG / IB: FLAG OLA1

Figure 4. The OLA1–BARD1 interaction is important for the regulation of centrosome number. A, Model of the OLA1-containing complex (15). OLA1 binds to the N-terminal region of BRCA1, the C-terminal region of BARD1, and g-tubulin. The N-terminal region of BRCA1 binds to the N-terminal region of BARD1. The middle portion of BRCA1 is associated with OLA1 via g-tubulin. The C-terminal region of BRCA1 are likely to be associated with OLA1 via the unknown proteins Protein X. Protein X is an unknown protein. "N," the N-terminal region. "C", the C-terminal region. B, Glutathione-sepharose beads bound with GST, GST-OLA1-WT, or mutants were incubated with His-BARD1-546–777. The bound proteins were immunoblotted. C, Model of the conformation changes of the OLA1-containing complex induced by mutations of OLA1. OLA1 with the mutations (M; S36A, F127A, and T325A) do not bind to the C-terminal region of BARD1. D–F, HEK-293T cells were transfected with FLAG-OLA1-WT or mutants and HA-BARD1-546–777, HA-g-tubulin, or HA-BRCA1-1-304. Cell lysates were immunoprecipitated (IP) using an anti-FLAG antibody.

pulldown assays. GST-tagged OLA1-WT, T124A, E168Q, and 546–777, HA-g-tubulin, or HA-BRCA1-1–304. Cell lysates were K242R bound directly to BARD1, whereas S36A, F127A, and immunoprecipitated with an anti-FLAG antibody (Fig. 4D–F). T325A mutants did not (Fig. 4B and C; Supplementary Tables S36A and F127A mutants markedly, and E168Q mutation slightly S1 and S2). By contrast, all OLA1 mutants bound to g-tubulin and decreased the binding to BARD1, g-tubulin, and BRCA1. The to BRCA1-1–304 (Supplementary Fig. S3A–S3B). We previously T325A mutation reduced the binding to BARD1 and g-tubulin. found that GST-BRCA1-1–304 did not bind to His-OLA1-E168Q These results suggest that the binding of OLA1 to the C-terminal (15); in the current study, GST-OLA1-E168Q showed weaker region of BARD1 is important for the formation of the BRCA1/ binding to His-BRCA1-1–304 than GST-OLA1-WT. The exchange BARD1/OLA1/g-tubulin complex in cells. of tags might affect BRCA1–OLA1 interaction. All mutants bound to g-tubulin in a manner similar to OLA1-WT. S36, T124, and T325 in OLA1 are the likely phosphorylated Next, we analyzed the association in cells of the OLA1 mutants residues with the C-terminal region of BARD1, g-tubulin, and the N- As S36, T124, and T325 are candidate residues of phosphor- terminal region of BRCA1 by immunoprecipitation. HEK-293T ylation, we compared the phosphorylated and nonphosphory- cells were transfected with FLAG-OLA1 mutants and HA-BARD1- lated mutants for their effect on centrosome ampliﬁcation. We

www.aacrjournals.org Mol Cancer Res; 2018 OF7

Yoshino et al.

A % ** ** ** 60 ** **

40 Figure 5. 20 Phosphorylation of OLA1 seems to be involved in the regulation of centrosome number. A, Hs578T cells 0

Cells withextra centrosomes were transfected with control

WT vector (HA) or vector expressing S36C S36D S36A T325A T325E T124E T124A E168Q

Control HA-OLA1-WT or mutants. Cells were coimmunostained with anti-HA and B anti–g-tubulin antibodies and stained Vector with DAPI. The cells were counted, and WT S36A S36C T325E T124A T124E T325A S36D E168Q Control the percentages of cells with ampliﬁed IB: HA OLA1 centrosomes were scored. At least 100 cells were counted in each experiment. IB: b-Actin b-Actin Histograms show the mean and SEM of three independent experiments. ﬁ Pull-down Pull-down Statistical signi cance is indicated by C D asterisks ( , P < 0.01) as calculated by GST-OLA1 GST-OLA1 the Student t test. B, Cell lysates were immunoblotted. C and D, Glutathione- Input GST WT WT S36D Input GST S36A S36C T325E T325A sepharose beads bound with GST, IB: BARD1 BARD1-546-777 IB: BARD1 BARD1-546-777 GST-OLA1-WT, or mutants were kDA kDA incubated with His-BARD1-546–777. 73 73 GST-OLA1 GST-OLA1 The bound proteins were immunoblotted. 48 48 IB: GST IB: GST 34 34 GST 27 27 GST

constructed plasmids to express the phosphomimetic mutants, expression of BARD1 (Fig. 6A). Knockdown of BARD1 caused OLA1-S36D, T124E, and T325E. These mutants caused centro- centrosome amplification among T47D and Hs578T cells, in a some amplification to a significantly higher extent than nonpho- similar manner to that seen with the knockdown of OLA1 (Fig. 6B sphorylated mutants (Fig. 5A). Expression of the HA-OLA1 and C). mutants was confirmed by Western blot (Fig. 5B). These results Three mutations, C645R, V695L, and S761N, are reported in suggest that S36, T124, and T325 are the likely phosphorylated the C-terminal region of BARD1 (Fig. 6D). C645R has been residues (Supplementary Table S2), and that these phosphoryla- identified in breast and ovarian cancers, V695L is reported in tions are involved in the regulation of centrosome number. breast cancer, and S761N is found in breast and uterine cancers Because nonphosphorylated S36A and T325A mutants failed to (38, 39). We analyzed the association in cells of HA-BARD1-546– bind to the C-terminal region of BARD1, we prepared purified 777 bearing each of these mutations with FLAG-OLA1 by immu- GST-OLA1-S36D and T325E and analyzed binding to the C- noprecipitation. HEK-293T cells were transfected with HA- terminal region of BARD1. GST-OLA1-S36D and T325E bound BARD1-546–777-C645R, -V695L or -S761N, and/or FLAG-OLA1. to BARD1 (Fig. 5C). Cell lysates were immunoprecipitated with an anti-FLAG anti- S36C has been reported as a mutation observed in cervical body (Fig. 6E). HA-BARD1-full-length with each of the mutations cancer [Catalogue of Somatic Mutations in Cancer (COSMIC); was also analyzed (Fig. 6F). The association of the C-terminal http://cancer.sanger.ac.uk/cosmic); we therefore tested its role in region of BARD1 with the V695L or the S761N mutation with centrosome amplification. Overexpression of OLA1-S36C failed OLA1 reduced markedly. In the case of the full-length BARD1, the to cause efficient centrosome amplification and this mutant did presence of each of the three mutations decreased the association not bind to BARD1 (Fig. 5A and D; Supplementary Table S1). with OLA1. Because the V695L mutation showed the most severe effect on BARD1 is involved in the regulation of centrosome number the OLA1–BARD1 association, we prepared purified His-BARD1- Because the three of six OLA1 mutants that were deficient in the 546–777 with the V695L mutation and analyzed the binding to regulation of centrosome number lost the binding activity to OLA1. BARD1-546–777-V695L failed to bind to OLA1 (Fig. 6G). BARD1 (Supplementary Tables S1 and S2), we presumed that Furthermore, we analyzed whether HA-BARD1-V695L could res- the OLA1–BARD1 interaction is important for the regulation of cue the centrosome amplification induced by BARD1 knockdown. centrosome number. We therefore analyzed the effect of knocking Hs578T cells were cotransfected with BARD1 siRNA, BARD1-3, down BARD1 on centrosome. We used two different siRNAs: against the 30 UTR of BARD1, and the expression vector, HA- BARD1-1 and BARD1-2. Both siRNAs efficiently reduced the BARD1-WT or -V695L, and stained with anti-g-tubulin and

OF8 Mol Cancer Res; 2018 Molecular Cancer Research

OLA1 Requires BARD1 to Regulate Centrosome Number

siRNA A -2 B Hs578T T47D

siRNA BARD1 OLA1 BARD1-1 Control IB: BARD1 Control

IB: OLA1

IB: b-Actin

** BARD1 C % ** 30 ** Control BARD1-2 BARD1-1 OLA1

20 ** ** D RING ANK BRCTs ** BARD1-FL 777 1 10 C645R S761N V695L BARD1-546-777 Cellsextra withcentrosomes 0 546 777 Hs578T T47D

E F WT V695L C645R WT V695L HA-BARD1-546-777 HA-BARD1-FL WT S761N WT C645R S761N FLAG-OLA1 - + + + + FLAG-OLA1 - + + + + Input / IB: HA BARD1-546-777 Input / IB: HA BARD1-FL

IP: FLAG / IB: HA BARD1-546-777 IP: FLAG / IB: HA BARD1-FL

Input / IB: FLAG OLA1 Input / IB: FLAG OLA1

IP: FLAG / IB: FLAG OLA1 IP: FLAG / IB: FLAG OLA1

Pull-down % G ** GST H ** Input GST -OLA1 30 His-BARD1-WT + - + - + - His-BARD1-V695L - + - + - + centrosomes 20 IB: His BARD1 kD 73 GST-OLA1 10 48

IB: GST Cellsextra with 0 34 siRNA Control BARD1-3 BARD1-3 BARD1-3 BARD1 BARD1 27 GST Vector Control control -WT -V695L

Figure 6. BARD1 is involved in the regulation of centrosome number via the interaction with OLA1. A, Hs578T cells were transfected with control siRNA, BARD1 siRNA, BARD1-1, or BARD1-2, or OLA1 siRNA, OLA1-2. Whole cell lysates were immunoblotted with anti-BARD1, OLA1, and b-actin antibodies. B, Hs578T and T47D cells were transfected with control, BARD1, or OLA1 siRNAs. Cells were stained with anti–g-tubulin antibody and DAPI. Representative images of H578T and T47D cells transfected with control or BARD1-1 siRNA are shown. Scale bar, 10 mm. C, The cells with amplified centrosomes were counted, and the percentages of cells were scored. D, Diagram of the location of missense mutations in BARD1-full-length (FL) and BARD1-546–777. RING, RING domain. ANK, Ankyrin repeats. BRCTs, BRCT domains. E, HEK-293T cells were transfected with HA-BARD1-545–777-WT or mutants and FLAG-OLA1. Cell lysates were immunoprecipitated using anti-FLAG antibody. F, HEK-293T cells were transfected with HA-BARD1-FL-WT or mutants and FLAG-OLA1. Cell lysates were immunoprecipitated using anti-FLAG antibody. G, Glutathione-sepharose beads bound with GST or GST-OLA1 were incubated with His-BARD1-546–777-WT or -V695L. The bound proteins were immunoblotted. H, Hs578T cells were transfected with control or BARD1-3 siRNA and vectors for HA-BARD1-WT or V695L. Cells were stained with anti–g-tubulin and anti-HA antibodies. The cells with amplified centrosomes were counted, and the percentages of cells were scored. For C and H, at least 100 cells were counted in each experiment. Histograms show the mean and SEM of three independent experiments. Statistical significance is indicated by asterisks (, P < 0.01) as calculated by the Student t test.

www.aacrjournals.org Mol Cancer Res; 2018 OF9

Yoshino et al.

anti-HA antibodies. Cotransfection with BARD1-WT partially, regulating centrosome number (Supplementary Tables S1 and but significantly, rescued the BARD1 knockdown–induced cen- S2). S36A, F127A, and T325A mutants failed to bind to the C- trosome amplification, but BARD1-V695L did not (Fig. 6H). terminal region of BARD1. S36C, a mutation reported in cancer, Knockdown of BARD1 and expression of HA-BARD1-WT and also failed to bind to BARD1 and cause efficient centrosome V695L in Hs578T cells were confirmed by Western blot (Supple- amplification by its overexpression. Thus, analyzing centrosome mentary Fig. S4). amplification induced by the overexpression of OLA1 variants is useful assay of OLA1. Overexpression of BRCA1-I42V, which is The BARD1–OLA1 interaction is important for BARD1 deficient in the regulation of centrosome number (13) and centrosomal localization and centrosome amplification binding to OLA1, also caused centrosome amplification, whereas induced by BARD1 overexpression the percentage of cells with extra centrosomes was significantly Next, we analyzed the centrosomal localizations and the effects lower than that of the cells transfected with BRCA1-WT (Supple- of overexpression of HA-BARD1-WT and the V695L mutant. HA- mentary Fig. S5A). Furthermore, overexpression of BARD1-V695L BARD1-WT or -V695L was overexpressed in Hs578T and T47D also failed to cause efficient centrosome amplification. Therefore, cells. BARD1-WT localized to the centrosome and caused cen- centrosome amplification by the overexpression of BRCA1 and trosomal amplification in these breast cancer cells (Fig. 7A–C). BARD1 might also be useful for the evaluation of their function in HA-BARD1 colocalized to only half of centrosomes in amplified the regulation of centrosome number. centrosomes. The V695L mutation significantly reduced both That overexpression of the phosphomimetic mutants S36D, centrosomal localization of BARD1 and centrosome amplifica- T124E, and T325E recovered centrosome amplification activity tion. In HeLa cells, the V695L mutation slightly decreased the suggests that phosphorylation of S36, T124, and T325 is impor- centrosomal localization of BARD1 protein and overexpression of tant for the regulation of centrosome number (Supplementary BARD1 did not cause centrosome amplification. Tables S1 and S2). S36D and T325E recovered the binding activity Finally, we confirmed the association in cells of Myc-tagged to the C-terminal region of BARD1 that contains BRCT domains BARD1-WT (Myc-BARD1-WT) and -V695L with endogenous that bind to phosphorylated proteins (40). This suggests that S36 BRCA1 by immunoprecipitation. HEK-293T cells were transfected and T325 of OLA1 are first phosphorylated and then OLA1 binds with Myc-BARD1-WT or -V695L. Cell lysates were immunopre- to BARD1 to regulate the centrosome. In Escherichia coli, the S16 of cipitated with an anti-Myc antibody (Fig. 7D). The association of the OLA1 homolog, YchF, which corresponds to S36 of OLA1, has BARD1-WT and -V695L with endogenous BRCA1 was clearly been identified as a phosphorylation site in a phosphoproteome detected. These suggest that the association of BARD1 with BRCA1 study (41). Phosphorylation of S16 in YchF affects the ATPase is not involved in the centrosomal localization and the effect of activity and is important for the inhibitory effect of YchF on the overexpression of BARD1 on centrosome number. oxidative stress response (42). F127 is important for nucleotide binding site and F127A mutation abolishes ATP binding drastically (19). In the tertiary Discussion structure, S36 locates close to F127 and the ATP analogue in the G Here, we showed that overexpression of OLA1 causes centro- domain. By contrast, T325 locates to the TGS domain in the C- some amplification in cell lines derived from mammary tissue. terminal region, distant from S36 and F127. Consistent with this, Suppression of BRCA1 and/or OLA1 causes centrosome amplifi- S36A and F127A similarly abolished the association with BRCA1, cation in cell lines derived from mammary tissue (11, 15). We BARD1, and g-tubulin, whereas T325A diminished the associa- found that overexpression of BRCA1 also caused centrosome tion with BARD1 and g-tubulin, but not BRCA1. The C-terminal amplification in breast cancer lines, Hs578T and MCF7 (Supple- region of OLA1 (300–396 aa) binds to the C-terminal region of mentary Fig. S5). Furthermore, overexpression and knockdown of BARD1 (15). The structure of the C-terminal region of BARD1 has BARD1 caused centrosome amplification in breast cancer cell also been determined (43, 44). Consistent with a role in direct lines. These suggest that adequately regulated expression levels physical interaction, when the structure of OLA1 is computation- of the components of the OLA1/BRCA1/BARD1 complex are ally docked to the C-terminal region of BARD1, energetically important for the regulation of centrosome number in cell lines favorable predicted interactions show that T325 of OLA1 locates derived from mammary tissue. Abnormal expression levels of a close to the binding surface to BARD1 (Fig. 7E). component of the complex might affect the formation and the BARD1 mutations, C645R, V695L, and S761N, decreased the function of the complex in centrosome. association with OLA1. C645R and V695L destabilize the BRCT Normally, cells with one or two centrosomes have two or four domain, and while normal S761 is exposed to the surface, the paired centrioles. Inhibition of BRCA1 or OLA1 increases the S761N mutation reduces affinity of BARD1 for its binding partner number of cells with extra centrioles and which are frequently (43). When the structure of OLA1 is computationally docked to found not to be paired, suggesting centriole overduplication and the C-terminal region of BARD1, energetically favorable predicted centrosome fragmentation (11, 15, 27). We found that over- interactions show that S761 locates at the binding surface to OLA1 expression of OLA1 also caused centriole overduplication. How- (Fig. 7E). The V695L mutation, which did not affect the interac- ever, overduplicated daughter centrioles were located close to the tion with BRCA1, abolished the binding to OLA1, the localization mother centriole. The clear separation of the increased centrioles, to the centrosome, and the regulation of centrosome number. which suggests centrosome fragmentation, was rarely observed. These suggest that interaction with OLA1 via its C-terminal region We hypothesized that centrosome amplification by overexpres- is important for the localization and function in centrosome of sion of OLA1 in breast cancer cells might be an easy and useful BARD1. Consistent with these findings, the C-terminal region of assay for evaluating the function of OLA1 mutants in the centro- BARD1 is required for the centrosomal localization of BARD1, but some. In addition to E168Q, we identified a further five mutants, not the RING domain, which is responsible for the interaction S36A, T124A, F127A, K242R, and T325A, which were deficient at with BRCA1 (45). We previously proposed the existence of a

OF10 Mol Cancer Res; 2018 Molecular Cancer Research

OLA1 Requires BARD1 to Regulate Centrosome Number

B A HA g-tubulin Merge % ** 80 HA-BARD1-WT

-tubulin HA-BARD1-V695L g Control 60

40 **

WT 20

0 Colocalization of BARD1 with BARD1 Colocalization of Hs578T T47D HeLa

V695L D Myc-BARD1-WT - + - Myc-BARD1-V695L - - +

* Input / IB: BRCA1 BRCA1 C * ** Control % IP: Myc / IB: BRCA1 BRCA1 HA-BARD1-WT Input / IB: Myc BARD1 20 HA-BARD1-V695L IP: Myc / IB: Myc BARD1

** 10 **

F BRCA1 Cells with extra centrosomesCells with

g-tubulin 0 Hs578T T47D HeLa N OLA1 Protein X C

N C E ; Direct binding OLA1 BARD1 G domain OLA1 Coiled-coil domain OLA1 mutations BARD1 mutation

BRCA1 BRCA1

g-tubulin g-tubulin N N OLA1 C T325 S761 OLA1 Protein X C Protein X -M N N BARD1- BARD1 C C V695L OLA1 TGS domain The C-terminal region of BARD1

Figure 7. The BARD1–OLA1 interaction is important for BARD1 centrosomal localization and centrosome amplification induced by BARD1 overexpression. A, Hs578T, T47D, and HeLa cells were transfected with control vector (HA) or HA-BARD1-FL-WT or -V695L expression vectors. Thirty-six hours after transfection, cells were fixed and stained with anti-HA and anti–g-tubulin antibodies and DAPI. Representative images of Hs578T cells are shown. Scale bar, 10 mm. B, The percentages of cells showing colocalization of HA-BARD1 with g-tubulin were scored. C, The cells with amplified centrosomes were counted, and the percentages of cells were scored. For B and C at least 100 cells were counted in each experiment. Histograms show the mean and SEM of three independent experiments. Statistical significance is indicated by asterisks (, P < 0.05; , P < 0.01) as calculated by the Student t test. D, HEK-293T cells were transfected with Myc-BARD1-FL-WT or -V695L. Cell lysates were immunoprecipitated using anti-Myc antibody. E, Computational docking model of OLA1 and the BARD1. T325 of OLA1 (color coded as in Fig. 3G) and S761 of the C-terminal region of BARD1 (blue) are indicated with space filling. F, Model of the conformation changes of the BRCA1/BARD1/OLA1/g-tubulin complex induced by OLA1 and BARD1 mutations. OLA1 mutations (M), S36A, S36C, F127A, and T325A abolish the binding to the C-terminal region of BARD1. BARD1 mutation, V695L, abolishes binding to OLA1. These missense substitutions cause similar alterations in the conformation of the protein complex. Dissociation of OLA1 with BARD1 might reduce the association of OLA1 with BRCA1 and g-tubulin.

www.aacrjournals.org Mol Cancer Res; 2018 OF11

Yoshino et al.

BRCA1/BARD1/OLA1/g-tubulin complex, which is important for Acquisition of data (provided animals, acquired and managed patients, the regulation of centrosome number. Both E168Q mutant of provided facilities, etc.): Y. Yoshino, H. Qi, H. Fujita, M. Shirota, S. Abe, OLA1 and I42V variant of BRCA1 impair the BRCA1/OLA1 Y. Komiyama, K. Shindo, M. Nakayama, A. Matsuzawa, A. Kobayashi, H. Ogoh, fi N. Chiba interaction and are de cient at regulating centrosome number Analysis and interpretation of data (e.g., statistical analysis, biostatistics, (15). In this study, we demonstrated that the V695L mutation computational analysis): Y. Yoshino, T. Watanabe, N. Chiba could cause alterations in the conformation of the protein com- Writing, review, and/or revision of the manuscript: Y. Yoshino, T. Watanabe, plex similar to those caused by the OLA1 mutations S36A, S36C, N. Chiba F127A, and T325A, which abolish the binding to the C-terminal Study supervision: C. Ishioka, N. Chiba region of BARD1 (Fig. 7F). The BRCA1/BARD1 ubiquitinates centrosomal proteins, including g-tubulin; it is thus feasible that Acknowledgments OLA1 regulates the BRCA1/BARD1 E3 ubiquitin ligase. This study was supported by grants-in-aid from JSPS KAKENHI grant num- BRCA1/BARD1 functions in homology-directed DNA repair, bers JP24300327 to N. Chiba and T. Watanabe, JP25640086 to N. Chiba, whereas mutations of V695L and S761N in BARD1 do not affect JP16H04690 to N. Chiba, T. Watanabe, and Y. Yoshino, Project for Develop- homology-directed DNA repair (46, 47). In BRCA mutation ment of Innovative Research on Cancer Therapeutics (P-DIRECT) to C. Ishioka fi and N. Chiba, Program for interdisciplinary research in Frontier Research carriers, centrosome ampli cation is observed even in normal Institute for Interdisciplinary Sciences (FRIS), Tohoku University to N. Chiba, mammary tissue (48). Therefore, centrosomal regulation could Astellas Foundation for Research on Metabolic Disorders to N. Chiba, the contribute to the function of the BRCA1/BARD1 as a tumor Yasuda Medical Foundation to N. Chiba, the Princess Takamatsu Cancer suppressor together with OLA1 in mammary tissue. On the other Research Fund (14-24621) to N. Chiba, Daiwa Securities Health Foundation hand, OLA1 is overexpressed in many other malignancies, includ- to N. Chiba, Kobayashi Foundation for Cancer Research to N. Chiba, Kobayashi ing of the colon and lung, but not in breast cancer (22). Both lower International Scholarship Foundation to N. Chiba, Friends of Leukemia Research Fund to N. Chiba, the Cooperative Research Project Program of Joint expression of OLA1 in breast cancer and higher expression of Usage/Research Center at the Institute of Development, Aging and Cancer, OLA1 in lung adenocarcinoma are associated with poor prognosis Tohoku University to T. Watanabe, and Research Program of the Smart-Aging (23, 25). OLA1 might therefore play a cancer type–specific role in Research Center, Tohoku University to N. Chiba. carcinogenesis and progression. Further study is needed to clarify The authors are grateful to the members of the Chiba lab for advice and the function of the BRCA1/BARD1/OLA1 complex in tumor technical support throughout the course of this study. development. The costs of publication of this article were defrayed in part by the Disclosure of Potential Conflicts of Interest payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate No potential conflicts of interest were disclosed. this fact. Authors' Contributions Conception and design: Y. Yoshino, N. Chiba Received March 20, 2018; revised April 28, 2018; accepted May 18, 2018; Development of methodology: Y. Yoshino published first June 1, 2018.

References 1. Miki Y, Swensen J, Shattuck-Eidens D, Futreal PA, Harshman K, Tavtigian S, instability in BRCA1 exon 11 isoform-deficient cells. Mol Cell 1999;3: et al. A strong candidate for the breast and ovarian cancer susceptibility 389–95. gene BRCA1. Science 1994;266:66–71. 13. Kais Z, Chiba N, Ishioka C, Parvin JD. Functional differences among BRCA1 2. Wu LC, Wang ZW, Tsan JT, Spillman MA, Phung A, Xu XL, et al. Identi- missense mutations in the control of centrosome duplication. Oncogene fication of a RING protein that can interact in vivo with the BRCA1 gene 2012;31:799–804. product. Nat Genet 1996;14:430–40. 14. Shimomura A, Miyoshi Y, Taguchi T, Tamaki Y, Noguchi S. Association of 3. Parvin JD. The BRCA1-dependent ubiquitin ligase, gamma-tubulin, and loss of BRCA1 expression with centrosome aberration in human breast centrosomes. Environ Mol Mutagen 2009;50:649–53. cancer. J Cancer Res Clin Oncol 2009;135:421–30. 4. Leung CC, Glover JN. BRCT domains: easy as one, two, three. Cell Cycle 15. Matsuzawa A, Kanno S, Nakayama M, Mochiduki H, Wei L, Shimaoka T, 2011;10:2461–70. et al. The BRCA1/BARD1-interacting protein OLA1 functions in centro- 5. Parvin JD.Overview of history and progress in BRCA1 research: the first some regulation. Mol Cell 2014;53:101–14. BRCA1 decade. Cancer Biol Ther 2004;3:505–8. 16. Verstraeten N, Fauvart M, Versees W, Michiels J. The universally conserved 6. Fujita H, Yoshino Y, Chiba N. Regulation of the centrosome cycle. Mol Cell prokaryotic GTPases. Microbiol Mol Biol Rev 2011;75:507–42. Oncol 2016;3:e1075643. 17. Leipe DD, Wolf YI, Koonin EV, Aravind L. Classification and evolution of P- 7. Bettencourt-Dias M, Hildebrandt F, Pellman D, Woods G, Godinho SA. loop GTPases and related ATPases. J Mol Biol 2002;317:41–72. Centrosomes and cilia in human disease. Trends Genet 2011;27:307–15. 18. Teplyakov A, Obmolova G, Chu SY, Toedt J, Eisenstein E, Howard AJ, et al. 8. Sankaran S, Starita LM, Simons AM, Parvin JD. Identification of domains of Crystal structure of the YchF protein reveals binding sites for GTP and BRCA1 critical for the ubiquitin-dependent inhibition of centrosome nucleic acid. J Bacteriol 2003;185:4031–7. function. Cancer Res 2006;66:4100–7. 19. Koller-Eichhorn R, Marquardt T, Gail R, Wittinghofer A, Kostrewa D, 9. Sankaran S, Starita LM, Groen AC, Ko MJ, Parvin JD. Centrosomal micro- Kutay U, et al. Human OLA1 defines an ATPase subfamily in tubule nucleation activity is inhibited by BRCA1-dependent ubiquitina- the Obg family of GTP-binding proteins. J Biol Chem 2007;282: tion. Mol Cell Biol 2005;25:8656–68. 19928–37. 10. Hsu LC, Doan TP, White RL. Identification of a gamma-tubulin-binding 20. Kira Y, Nishikawa M. The identification and characterization of a domain in BRCA1. Cancer Res 2001;61:7713–8. new GTP-binding protein (Gbp45) involved in cell proliferation and 11. Starita LM, Machida Y, Sankaran S, Elias JE, Griffin K, Schlegel BP, et al. death related to mitochondrial function. Cell Mol Biol Lett 2008; BRCA1-dependent ubiquitination of gamma-tubulin regulates centro- 13:570–84. some number. Mol Cell Biol 2004;24:8457–66. 21. Gradia DF, Rau K, Umaki AC, de Souza FS, Probst CM, Correa A, et al. 12. Xu X, Weaver Z, Linke SP, Li C, Gotay J, Wang XW, et al. Centrosome Characterization of a novel Obg-like ATPase in the protozoan Trypano- amplification and a defective G2-M cell cycle checkpoint induce genetic soma cruzi. Int J Parasitol 2009;39:49–58.

OF12 Mol Cancer Res; 2018 Molecular Cancer Research

OLA1 Requires BARD1 to Regulate Centrosome Number

22. Sun H, Luo X, Montalbano J, Jin W, Shi J, Sheikh MS, et al. DOC45, a novel like kinase 1 and serves as a marker for mature centrioles. Mol Biol Cell DNA damage-regulated nucleocytoplasmic ATPase that is overexpressed in 2005;16:1095–107. multiple human malignancies. Mol Cancer Res 2010;8:57–66. 37. Choudhary C, Kumar C, Gnad F, Nielsen ML, Rehman M, Walther TC, et al. 23. Chen H, Song R, Wang G, Ding Z, Yang C, Zhang J, et al. OLA1 regulates Lysine acetylation targets protein complexes and co-regulates major protein synthesis and integrated stress response by inhibiting eIF2 ternary cellular functions. Science 2009;325:834–40. complex formation. Sci Rep 2015;5:13241. 38. Thai TH, Du F, Tsan JT, Jin Y, Phung A, Spillman MA, et al. Mutations in the 24. Sprang SR.G protein mechanisms: insights from structural analysis. Annu BRCA1-associated RING domain (BARD1) gene in primary breast, ovarian Rev Biochem 1997;66:639–78. and uterine cancers. Hum Mol Genet 1998;7:195–202. 25. Bai L, Yu Z, Zhang J, Yuan S, Liao C, Jeyabal PV, et al. OLA1 contributes to 39. Sauer MK, Andrulis IL. Identification and characterization of missense epithelial-mesenchymal transition in lung cancer by modulating the alterations in the BRCA1 associated RING domain (BARD1) gene in breast GSK3beta/snail/E-cadherin signaling. Oncotarget 2016;7:10402–13. and ovarian cancer. J Med Genet 2005;42:633–8. 26. Ding Z, Liu Y, Rubio V, He J, Minze LJ, Shi ZZ. OLA1, a Translational 40. Glover JN, Williams RS, Lee MS. Interactions between BRCT repeats and Regulator of p21, maintains optimal cell proliferation necessary for devel- phosphoproteins: tangled up in two. Trends Biochem Sci 2004;29:579–85. opmental progression. Mol Cell Biol 2016;36:2568–82. 41. Macek B, Gnad F, Soufi B, Kumar C, Olsen JV, Mijakovic I, et al. Phos- 27. Ko MJ, Murata K, Hwang DS, Parvin JD. Inhibition of BRCA1 in breast cell phoproteome analysis of E. coli reveals evolutionary conservation of lines causes the centrosome duplication cycle to be disconnected from the bacterial Ser/Thr/Tyr phosphorylation. Mol Cell Proteomics 2008;7: cell cycle. Oncogene 2006;25:298–303. 299–307. 28. Okano S, Kanno S, Nakajima S, Yasui A. Cellular responses and repair of 42. Wenk M, Ba Q, Erichsen V, Macinnes K, Wiese H, Warscheid B, et al. A single-strand breaks introduced by UV damage endonuclease in mamma- universally conserved ATPase regulates the oxidative stress response in lian cells. J Biol Chem 2000;275:32635–41. escherichia coli. J Biol Chem 2012;287:43585–98. 29. Ransburgh DJ, Chiba N, Ishioka C, Toland AE, Parvin JD. Identification of 43. Birrane G, Varma AK, Soni A, Ladias JA. Crystal structure of the BARD1 breast tumor mutations in BRCA1 that abolish its function in homologous BRCT domains. Biochemistry 2007;46:7706–12. DNA recombination. Cancer Res 2010;70:988–95. 44. Edwards RA, Lee MS, Tsutakawa SE, Williams RS, Nazeer I, Kleiman FE, 30. Wei L, Lan L, Hong Z, Yasui A, Ishioka C, Chiba N. Rapid recruitment of et al. The BARD1 C-terminal domain structure and interactions with BRCA1 to DNA double-strand breaks is dependent on its association with polyadenylation factor CstF-50. Biochemistry 2008;47:11446–56. Ku80. Mol Cell Biol 2008;28:7380–93. 45. Brodie KM, Mok MT, Henderson BR. Characterization of BARD1 targeting 31. Li L, Ryser S, Dizin E, Pils D, Krainer M, Jefford CE, et al. Oncogenic BARD1 and dynamics at the centrosome: the role of CRM1, BRCA1 and the Q564H isoforms expressed in gynecological cancers. Cancer Res 2007;67:11876–85. mutation. Cell Signal 2012;24:451–9. 32. Li M, Yu X. Function of BRCA1 in the DNA damage response is mediated by 46. Laufer M, Nandula SV, Modi AP, Wang S, Jasin M, Murty VV, et al. ADP-ribosylation. Cancer Cell 2013;23:693–704. Structural requirements for the BARD1 tumor suppressor in chromo- 33. Sali A, Blundell TL. Comparative protein modelling by satisfaction of somal stability and homology-directed DNA repair. J Biol Chem 2007; spatial restraints. J Mol Biol 1993;234:779–815. 282:34325–33. 34. Pierce BG, Hourai Y, Weng Z. Accelerating protein docking in ZDOCK using 47. Lee C, Banerjee T, Gillespie J, Ceravolo A, Parvinsmith MR, Starita LM, et al. an advanced 3D convolution library. PLoS One 2011;6:e24657. Functional analysis of BARD1 missense variants in homology-directed 35. Pierce B, Weng Z. ZRANK: reranking protein docking predictions with an repair of DNA double strand breaks. Hum Mutat 2015;36:1205–14. optimized energy function. Proteins 2007;67:1078–86. 48. Martins FC, De S, Almendro V, Gonen M, Park SY, Blum JL, et al. Evolu- 36. Guarguaglini G, Duncan PI, Stierhof YD, Holmstrom T, Duensing S, Nigg tionary pathways in BRCA1-associated breast tumors. Cancer Discov EA. The forkhead-associated domain protein Cep170 interacts with Polo- 2012;2:503–11.

www.aacrjournals.org Mol Cancer Res; 2018 OF13

BRCA1-Interacting Protein OLA1 Requires Interaction with BARD1 to Regulate Centrosome Number

Yuki Yoshino, Huicheng Qi, Hiroki Fujita, et al.

Mol Cancer Res Published OnlineFirst June 1, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-18-0269

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2018/06/01/1541-7786.MCR-18-0269.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/early/2018/08/16/1541-7786.MCR-18-0269. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.