BRD1-Mediated Acetylation Promotes Integrin Av Gene Expression Via

Published OnlineFirst February 16, 2018; DOI: 10.1158/1541-7786.MCR-17-0527

Chromatin, Epigenetics and RNA Regulation Molecular Cancer Research BRD1-Mediated Acetylation Promotes Integrin aV Gene Expression Via Interaction with Sulfatide Qian Qian Cai1, Yi Wei Dong1, Bing Qi1, Xiao-Ting Shao2, Rong Wang1, Zhong Yi Chen1, Bao Mei He2, and Xing Zhong Wu1

Abstract

Integrin aV gene expression is often dysregulated in cancers found to induce BRD1, monocytic leukemia zinc ﬁnger (MOZ) especially in hepatocellular carcinoma (HCC); however, the and histone acetyltransferase binding to ORC1 (HBO1) acetyl- mechanism of regulation is poorly understood. Here, it is dem- transferase multiprotein complex recruitment to the integrin onstrated that sulfatide activated integrin aV gene transcription, aV promoter, which is responsible for histone H3K9/14 acetyla- through histone H3K9/14 acetylation at the promoter, and high tion. Finally, knockdown of BRD1 limited sulfatide-induced integrin aV expression are closely associated with poor prognosis. H3K9/14 acetylation and occupancy of MOZ or HBO1 on integrin To elucidate the mechanism of regulation of acetylation, sulfa- aV gene promoter. tide-bound proteins were screened by mass spectrometry (MS), and bromodomain containing protein 1 (BRD1) was identiﬁed as Implications: This study demonstrates that sulfatide inter- an interacting protein that also colocalized with sulfatide in HCC action with BRD1 mediates acetylation and is important for cells. BRD1 was also formed a complex with Sp1, which was regulation of integrin aV gene expression. Mol Cancer Res; 1–13. recruited to the integrin aV gene promoter. Sulfatide was also 2018 AACR.

Introduction During tumor progression integrins interact with extracellular matrix molecules such as vitronectin, collagen, or glycolipids to Angiogenesis and metastasis are the major cause of tumor trigger signal transduction involved in cell migration and tumor progression and tumor-related mortality. More than 90% of metastasis. Cerebroside sulfotransferase (CST) is the key enzyme cancer-associated deaths are due to metastatic disease, usually for the synthesis of sulfatide in the human body. CST recognizes resistant to and incurable by current therapies (1). Thus, regulators galactocerebroside (Gal-Cer) and transfers a sulfate group from in metastatic cascade, draw priority attention for extensive inves- 0 0 0 3 -phosphoadenosine-5 -phosphosulfate to the 3 hydroxyl tigation, especially those driving cancer cells to leave primary foci group of the galactose in cerebroside to produce sulfatide. and migrate to distant tissues, Cancer cells acquiring migration Sulfatide is enriched in the brain, myelin sheath, kidney, and ability is then considered as an important malignant trait to leave liver. We previously showed that overexpressed sulfatide in HCC primary tumor. Often expressed in tumor stem cells, integrin promoted aVb3 on HCC cells and metastasis (6), and this is aVb3 initiates complex with oncogene and oncogene homologue associated with the enhanced expression of the integrin aV to promote tumor cell motility and activate metastasis process (2). subunit through Sp1 (7), although the precise molecular regu- High integrin aV expression was more often seen in the colorectal lation mechanism of sulfatide remains unknown. cancer patients with later TNM stage from the statistical analysis in Gene transcription initiation is largely controlled by chro- 198 samples (3), and antagonists against integrin aVb3oraVb5 mosome remodeling and histone acetylation which is catalyzed have shown robust activities in suppression of tumor progression by histone acetyltransferases (HAT; ref. 8), a family that (4). Suppression of integrin aV expression by microRNA-25 contains many members sharing conserved domains and func- subsequently reduced the invasive cytoskeleton, migration ability, tions. Bromodomain containing protein 1 (BRD1) has been and metastasis in highly metastatic prostate cancer stem cells (5). identified in the monocytic leukemia zinc finger protein (MOZ)/MOZ paralog (MORF) complex that has intrinsic acetyltransferase activity specific for histone 3 (H3; ref. 9). 1 Department of Biochemistry and Molecular Biology, School of Basic Medical BRD1, also known as bromodomain and PHD finger- Sciences, Fudan University, Key Lab of Glycoconjugate Research, Ministry of containing protein 2 (BRPF2), belongs to the BRPF protein Public Health, Shanghai, China. 2Yu Ying Hospital, Wenzhou Medical University, Wenzhou, China. family that includes BRPF1, BRD1/BPF2, and BRPF3, and contains a bromodomain, two plant homeodomain (PHD) Note: Supplementary data for this article are available at Molecular Cancer zinc fingers, and a proline–tryptophan–tryptophan–proline Research Online (http://mcr.aacrjournals.org/). (PWWP) domain. The bromodomain can selectively recognize Q.Q. Cai and Y.W. Dong contributed equally to this article. acetylated lysine residues, particularly on histones (10). In this Corresponding Author: Xing Zhong Wu, School of Basic Medical Sciences, study, we investigated the roles of BRD1 in sulfatide-induced Fudan University, 138 Yi Xue Yuan Road, Shanghai 200032, PR China. Phone: transcriptional activation of the integrin aV subunit gene in 54237697; Fax: 54237697; E-mail: [email protected] human HCC cells. We demonstrated that BRD1 formed a novel doi: 10.1158/1541-7786.MCR-17-0527 complex that was involved in the regulation of integrin aV 2018 American Association for Cancer Research. subunit expression by sulfatide.

www.aacrjournals.org OF1

Cai et al.

Materials and Methods sion vector. Plasmids of CST were constructed as in our previous study (7). Cell culture and treatment SMMC-7721 and BEL-7404 HCC cell lines were obtained from Polymerase chain reaction the Shanghai Institute of Biochemistry & Cellular Biology, Chi- PCR procedures were performed as previously described (7). nese Academy of Science, Shanghai, China, and their morpho- Quantitative PCR was performed with Bestar SybrGreen qPCR fi logical characteristics were identi ed every 3 years to be consistent mastermix (DBI Bioscience) in the iQ5 PCR detection system with the report of establishment (11). Human embryonic kidney (Bio-Rad). The specific primers are listed in Supplementary cells (HEK293T) were obtained from the ATCC and cultured not Table S2. more than 20 passages each time after thawing. These cells were grown in monolayer in the medium of DMEM supplemented Immunoprecipitation and Western blot analysis with 10% newborn calf serum (Biosera, France) or FBS (Gibco), Immunoprecipitation assays were performed as previously with penicillin(100 U/mL)/streptomycin (100 mg/mL) under a described (13). Briefly, whole cell lysates were incubated with 5% CO atmosphere at 37C. All cells were not contaminated by 2 indicated antibodies and rocked at 4C with protein A/G agarose mycoplasma as tested by the mycoplasma PCR Detection Kit beads overnight. The beads were then washed with ice-cold (Sigma), and also not infected by bacteria or fungi. For treatment, lysate buffer (50 mmol/L Tris-HCl pH 8.0, 120 mmol/L NaCl, cells were incubated with Gal-Cer (galactocerebroside), sulfatide 10 mg/mL aprotinin, 1 mmol/L PMSF, 1 mmol/L sodium vana- (2 mmol/L, 24 hours for RNA analysis, 48 hours for protein date, 100 mmol/L NaF, and 0.5% NP-40). The proteins obtained assays),the histone deacetylase (HDAC) inhibitor trichostatin A were separated on an 8% SDS-PAGE gel and transferred to a (TSA; 300 nmol/L, 24 hours for RNA analysis, 48 hours for protein PVDF membrane. The blot was probed with the anti-integrin aV, assays), or nicotianamine (NA; 20 mmol/L, 48 hours; all from HDAC, MOZ, HBO1, Sp1, BRD1, SIN3B, ING4, and acetylated Sigma-Aldrich). histone antibodies (Santa Cruz Biotechnology), followed by a horseradish peroxidase-conjugated secondary antibody. After Tissue microarray and IHC visualizing, the density of protein bands were analyzed using A total of 80 patients with HCC from Shanghai Oriental TotalLab v2.01 (Nonlinear Dynamics Ltd.). Hepatobiliary Hospital, Zhongshan University Affiliated 3rd Hos- fi pital, Wenzhou Medical University Af liated 1st Hospital, and Nuclear and cytoplasmic extracts Shanghai 10th People's Hospital underwent curative surgery and Nuclear and cytoplasmic extracts were prepared using the fi the diagnosis of HCC was con rmed by histological examination. Nuclear Extraction Kit (Chemicon International Inc.) according fi Tissue blocks embedded in paraf n and the corresponding hema- to manufacturer's instructions. Briefly, the cells were disrupted by toxylin- and eosin-stained sections were overlaid for tissue micro- cytoplasmic lysis buffer, and the cytosolic portion was collected fi array preparation. The paraf n-embedded tissue microarray was after centrifugation. Then the nuclear pellets were further dis- constructed from triplicate pairs of tumor and nontumor tissue rupted by nuclear extraction buffer. The cytoplasmic and nuclear fi from each patient. Paraf n sections were stained with an integrin proteins were then analyzed by Western blotting. aV mouse monoclonal antibody and BRD1 rabbit antibody (Santa Cruz Biotechnology) using the DouSP Double-Stain Kit Mass spectrometry (Maxim-Biomedical). Scoring of immunohistochemical staining Epoxy magnetic particles were first reacted with 4-hydroxy- was performed independently by two senior pathologists, benzhydrazide to prepare the hydroxyl-coated magnetic parti- blinded to all samples, using a widely accepted German semi- cles that were then covalently coupled with sulfatide (14). To quantitative scoring system (12). Scoring was based on the enrich the sulfatide coating on the surface, the magnetic par- staining intensity of the cytoplasm, nucleus, and/or membrane ticles (1 mg) were further mixed with sulfatide or Gal-Cer (none ¼ 0, weak ¼ 1, moderate ¼ 2, strong ¼ 3) and the extent of solution (0.5 mg in 0.5 mL chloroform, respectively) and the stained cells (0% ¼ 0, 1%–24% ¼ 1, 25%–49% ¼ 2, 50%–74% ¼ chloroform was evaporated. After blocking with 1% BSA, the 3, 75%–100% ¼ 4). The final immunoreactivity score was particles were incubated with 500 mL cell lysate overnight at achieved by multiplying the intensity score and the extent score, 4 C, and washed with lysate buffer containing 0.05% NP-40, and ranged from 0 to 12. The clinicopathologic characteristics of followed by PBS. After SDS-PAGE, the gel was stained with the patients and the IHC are summarized in Supplementary Table Coomassie Brilliant Blue (CBB) and the bands specificfor S1. This study was approved by the Ethics Committee, Fudan sulfatidecomparedtoGal-CerwereexcisedforMSanalysis. University Shanghai Medical College, with the registration number 14000000020000024 and permit number 97. Immunofluorescent staining After culture, the cells were washed, fixed with 4% paraformal- Plasmids and transfections dehyde, and permeabilized with 0.5% Triton X-100. After block- To construct interference plasmids, specific interference ing, the cells were successively incubated with primary (sulfatide, sequences (Supplementary Table S2) targeting BRD1, MOZ, BRD1, SIN3B) and FITC- or rhodamine-tagged secondary anti- HBO1, and integrin aV were cloned into pSilencer 4.1. The bodies before mounting. Images were obtained using a fluores- promoter (1295 to þ207) of the integrin aV subunit gene cence confocal microscope (Nikon Eclipse TE 2000-U) and the was cloned into the pGL3 basic reporter. BRD1 and its mutant fluorescent signals were analyzed using Image-Pro Plus software constructs were generous gifts from Professor Atsushi Lwama, (V6.0, Media Cybernetics). Fluorescence intensity was measured Chiba University, Japan. The MOZ plasmid was kindly provid- in four random fields in more than 25 cells of each field in each ed by Professor David M. Heery, The University of Nottingham, group. The Pearson correlation coefficient was calculated to Nottingham, UK. Sp1 and HDAC1 were in the pcDNA3 expres- quantify colocalization.

OF2 Mol Cancer Res; 2018 Molecular Cancer Research

Sulfatide Promotes ITGAV Expression

Luciferase assays Results The cells were seeded into 96-well plates and transfected Integrin aV gene expression is promoted by sulfatide with 0.2 mg plasmids for 48 hours. After washing once with To investigate the expression of integrin aV in hepatocellular PBS, the cells were harvested in passive lysis buffer (Promega). carcinoma, we measured integrin aV in 80 patient samples by The activities of firefly luciferase and Renilla luciferase were IHC. We observed that 82.25% of the HCC patients showed measured using the dual luciferase assay system (Promega) by stronger integrin aV staining in tumors than in nontumor tissues. a luminometer (Lumat LB 9507). Fold change of luciferase The total immunoreactivity score for integrin aV was significantly activity was calculated by normalizing luciferase activity of the higher in tumors than in adjacent nontumor tissues (Fig. 1A). We test groups to the control. Experiments were repeated in also noted that integrin aV expression was significantly associated triplicate. with the low survival rate in these patients (Fig. 1B, top left); 30 subjects had strong integrin staining (þþþ, immunoreactivity Chromatin immunoprecipitation score 9–12) and had significantly poorer survival than those with Chromatin immunoprecipitation (ChIP) was performed with weak integrin aV staining (þ, immunoreactivity score 1–4), the EZ-ChIP Kit (Merck Millipore) according to the manufac- despite all having received surgical resection. The integrin aV turer's instructions. Genomic DNAs of HCC cells, after chemically immunoreactivity scores were also significantly higher in the cross-linking with 1% formaldehyde, were disrupted to an average 22 patients with metastasis (including vessel invasion, portal vein – size of 200 750 bp in SDS lysis buffer by Bioruptor sonicator and cancer embolus, lymph node invasion, and remote metastasis) incubated with specific p300, H3K9/14, HBO1, MOZ, BRD1, than in the 58 patients without metastasis (Fig. 1B, top right). HDAC, and BRG1 antibodies (Santa Cruz Biotechnology) at 4 C Notably the scores of integrin aV immunoreactivity increased overnight. After eluting from the chromatin complexes, the puri- as the TNM stage advanced (Supplementary Table S1). Further fi a ed DNAs were used to detect occupancy of the integrin V gene analysis in another cohort of 383 patients with HCC from TCGA promoter by quantitative PCR. The primer sequences are given in mRNA microarray database confirmed that patients with high Supplementary Table S2. integrin aV expression levels survived significantly shorter than those with low expression levels of integrin aV (Fig. 1B, bottom). Sulfatide and BRD1 binding assays Also the levels of integrin aV mRNA in 118 patients with Cells were incubated with 5 mmol/L Bodipy-labeled sulfatide, metastasis were significantly higher than 228 patients without Bodipy-labeled Gal-Cer, or free Bodipy for 6 hours, and then cells metastasis. were lysed with lysate buffer containing 1% Triton X-100 and In the HCC cell lines, SMMC-7721 or BEL-7404, integrin aV centrifuged to remove insoluble material. Next, the lysis solution subunit transcription and protein levels increased significantly was immunoprecipitated with anti-BRD1 antibody and protein after exposure to exogenous sulfatide, but not galactocerebroside A/G agarose at 4 C overnight. After centrifugation, the pellets were (Fig. 1C). Similarly, expression of integrin aV subunit protein was mixed with 50 mL 1% SDS and measured by a fluorescent also enhanced significantly in response to endogenous sulfatide microtiter plate reader at 530 nm wavelength after excitation produced by CST transfection, whereas silencing of CST through at 480 nm. RNA interference (shCST-2 or shCST-5) robustly reduced integrin aV expression levels (Fig. 1C, bottom). No significant differences Migration assays were observed following transfection with mock or scrambled Cell migration was performed using Millipore Millicell hanging control siRNA. These suggest that sulfatide has the potential to a inserts with an 8.0 mm pore size. Cells were seeded on the inserts transactivate integrin V subunit gene transcription. coated with ECM and allowed to migrate. The cells that migrated through the membrane were counted after incubation (15). Sulfatide regulates acetylation The state of histone acetylation plays an important role in Cell adhesion assay regulating transcription. Interestingly, we noted that sulfatide The adhesion assay was performed in a 96-well plate pre-coated treatment resulted in significant enhancement of global histone with 100 mL/well of 40 mg/mL collagen type I, 20 mg/mL fibrin- H3K9/14 and transcription factor Sp1 acetylation (Fig. 1D). ogen, 10 mg/mL fibronectin, or 2 mg/mL vitronectin (7). However, sulfatide had no effect on the levels of total acetylated histone H4k12 levels. The ChIP results showed that sulfatide significantly enhanced the occupancy of acetylated histone Molecular modeling H3K9/14 on the integrin aV gene promoter (Fig. 1D right). The BRD1 protein model was created using the Build Module of HDACs mediate the deacetylation of histones. We interestingly Discovery Studio (DS) version 3.1 Client (Biovia), with the crystal noted that the HDAC inhibitor TSA robustly strengthened integrin structure of PHD1 domain (Protein Data Bank entry 2KU3), aV expression at both the mRNA and protein levels (Fig. 2A). We PHD2 domain (Protein Data Bank entry 2LQ6), Bromo domain further observed that ectopic expression of HDAC1 significantly (Protein Data Bank entry 3RCW), and PWWP domain (Protein reduced integrin aV subunit transcription (Fig. 2A, right). TSA also Data Bank entry 3LYI). enhanced global histone H3K9/14, H3K9, and Sp1 acetylation (Fig. 2B). Furthermore, the recruitment of Sp1 and acetylated Quantitative analysis H3K9/14 to the integrin aV promoter was significantly enhanced Quantitative data were expressed as mean SEM of at least after TSA treatment (Fig. 2C). Therefore, sulfatide induced the three independent replicates. Statistical analysis was performed acetylation of H3K9/14 and recruitment of Sp1 to the promoter of using SPSS software v22 (IBM) and P values less than 0.05 were the integrin aV gene, which was associated with an induction of considered significant. integrin aV expression in HCC cells.

www.aacrjournals.org Mol Cancer Res; 2018 OF3

Cai et al.

Figure 1. Sulfatide in HCC enhances integrin aV expression and acetylation. A, Representative immunohistochemical staining of integrin aV in tumor tissues (T) and corresponding adjacent nontumor (NT) liver tissues from 80 HCC patients (400). Immunoreactivity scores for integrin aVexpressiononthemembrane for individual HCC specimens (T) and the adjacent non-tumor (NT) tissue (lower panel). B, The overall survival rates of 80 patients with HCC were analyzed using Kaplan–Meier survival stratified by weak (þ,score1–4), moderate, and strong (þþþ,score9–12) integrin aV expression (left). Analysis of integrin aV staining scores in patients (n ¼ 80) with and without metastasis (right). The overall survival rates of 383 patients with HCC from TCGA database and the analysis of integrin aV expression in these patients with and without metastasis (bottom). C, Integrin aV expression was analyzed by PCR and Western blot analysis. Relative expression (right) was quantified based on the loading control (top). Integrin aV expression in two clones of CST transfectants or shCST transfectants was determined by Western blot analysis and quantified in SMMC-7721 cells (bottom). D, Histone 3/4 and Sp1 acetylation was analyzed by Western blot analysis and quantified. Recruitment of AcH3K9/14 to the integrin aV promoter was analyzed by ChIP and quantified (right). Data are representative from three independent experiments. , P < 0.05; , P < 0.01.

OF4 Mol Cancer Res; 2018 Molecular Cancer Research

Sulfatide Promotes ITGAV Expression

Figure 2. Sulfatide interaction with BRD1. A, qPCR analysis of HDAC1 and integrin aV mRNA expression (left). Integrin aV expression levels were determined by Western blot analysis and quantified in the indicated treatment (right). B, Acetylation of histone and immunoprecipitated Sp1 was determined by Western blot analysis and quantitative analysis. C, Recruitment of Sp1 and AcH3K9/14 to the integrin aV gene promoter was analyzed by ChIP and quantification analysis. D, Binding proteins captured by immobilized sulfatide were identified as BRD1 by mass spectrometry (top left). BEL-7404 cells that were pre-incubated with Bodipy-tagged sulfatide (B-sulfatide) were lysed and immunoprecipitated with a BRD1 or SIN3B antibody. The fluorescence intensity in the pellet complex was analyzed (top right). Colocalization analysis of sulfatide and BRD1 by immunostaining of SMMC-7721 cells followed by confocal microscopic analysis (bottom left). Quantification of the Pearson correlation coefficient for quantifying colocalization (bottom right). Cells from four random fields of view were counted in each condition and data are representative of two independent experiments (F test). E, Representative immunofluorescence micrographs of BRD1 intracellular distribution as examined by confocal microscopy(400). F, Representative of sulfatide immunofluorescence micrographs in SMMC-7721 cells treated with exogenous sulfatide(400). Data are representative from three independent experiments. , P < 0.05; , P < 0.01.

www.aacrjournals.org Mol Cancer Res; 2018 OF5

Cai et al.

Sulfatide forms a complex with BRD1 transfection than the control (Fig. 4A). In a ChIP assay, we Sulfatide is a sphingolipid that has not been previously found that BRD1 was bound to the integrin aV gene promoter, associated with acetylation. Therefore, we investigated the and sulfatide promoted the binding activity of BRD1, but not mechanism by which sulfatide influences acetylation. We cou- SIN3B (Fig. 4A). Afterwards, in reciprocal co-immunoprecipi- pled sulfatide to magnetic beads and incubated these beads tation experiments, the results showed that BRD1 and SIN3B with lysates from SMMC-7721 cells, and the bound proteins existed in the complex immunoprecipitated using an Sp1 anti- were separated by SDS-PAGE and stained with Coomassie body (Fig. 4B). After sulfatide treatment, more BRD1 was Brilliant Blue. The unique band at 130 kDa was noted com- present in the Sp1 complex than in the control, although pared to the Gal-Cer control. Then it was isolated and analyzed sulfatide did not significantly affect the total level of BRD1. by MS. A total of 43 proteins were identified in the band as Conversely, less SIN3B was present in the complex following compared to the control (Supplementary Tables S3 and S4). sulfatide treatment than in the control. Following immunopre- Among them BRD1 had the highest grading values and are cipitation with an anti-BRD1 antibody, Sp1 was also present involved in acetylation regulation (Fig. 2D). Next, to examine and significantly enhanced by sulfatide treatment (Fig. 4C). In the sulfatide–BRD1 interaction, BEL-7404 cells were incubated further immunoprecipitation with deletion mutants of BRD1, with 10 mmol/L Bodipy-tagged sulfatide or control Bodipy for 4 the results showed that the BRD1 PHD fragment (192-538) hours, and the cell lysates were incubated and immunopreci- interacted with Sp1 as strongly as the full-length BRD1, whereas pitated with BRD1 -specific antibodies. Following BRD1 immu- fragments lacking PHD or PWWP did not (Fig. 3B, right). noprecipitation, the fluorescence signal was significantly higher Interestingly, the fragment dPWWP, which contained PHD but in the Bodipy-sulfatide group than in the Bodipy control group not PWWP, also showed no interaction with Sp1. Thus, the PHD (Fig. 2D, right). Immunofluorescence staining showed that fragment was necessary and sufficient for the physical interac- intracellular BRD1 and sulfatide colocalized in SMMC-7721 tion with Sp1, but the PHD fragment within the complete BRD1 cells (Fig. 2D, bottom). The Pearson correlation coefficient protein required PWWP to maintain its binding affinity. These value, a statistic to quantify the extent of colocalization of two data suggested that BRD1 interacting with sulfatide was signals, was significantly higher in sulfatide-treated cells than recruited to the Sp1-transcriptional complex on the integrin aV following Gal-Cer treatment (Fig. 2D, bottom right). Further gene promoter and promoted transcriptional activation. confocal microscopic observation indicated that BRD1 became To further investigate the role of BRD1 in HCC, we mea- more abundant in the nucleus (Fig. 2E) after sulfatide treat- sured BRD1 expression in 80 patient samples by IHC. The ment and exogenous sulfatide enhanced nuclear distribution of BRD1 immunoreactivity scores were significantly higher in the sulfatide (Fig. 2F). Western blot analysis confirmed the nuclear nuclei of tumor tissue than in adjacent nontumor tissues, accumulation of BRD1 in the cells treated with sulfatide (Fig. although staining was stronger in the cytoplasm of adjacent 3A). To determine the binding domain that mediated the BRD1 nontumor tissues than in tumor tissues (Fig. 4D, left). The and sulfatide interaction, a number of recombinant constructs BRD1 immunoreactivity scores in the nuclei of tumor tissue expressing either full-length BRD1 or forms with domains were significantly higher in the 22 patients with metastasis deleted (N, PHD, dPWWP, dN, Br-PWWP) were used (Fig. (including vessel invasion and remote metastasis) than in the 3B). Fragment dN showed binding activity to sulfatide as 58 patients without metastasis (Fig. 4D, right top).There was a similar as the full-length BRD1 (Fig. 3B, bottom left), whereas significant correlation between integrin aV and BRD1 immu- others did not show binding activity as strong as full-length noreactivity scores in the nuclei (Fig. 4D, middle right). BRD1 BRD1. We further analyzed sulfatide–BRD1 interaction through mRNA expression levels in 423 patients with HCC from the molecular dynamics simulation and found that sulfatide TCGA database also correlated well with integrin aV expres- formed a more stable complex with BRD1 in lower potential sion levels (Fig. 4D, bottom right). energy than cerebroside-BRD1 (Fig. 3C, top). After sulfatide binding, the whole 3D conformation of BRD1 altered. By Acetylation promotes integrin aV gene expression overlapping analysis of sulfatide-BRD1 and cerebroside-BRD1 BRD1 is responsible for substrate recognition in the acetyl- conformations, the most significant conformation alteration transferase complex. Therefore, we further investigated the was noted in the PHD2 domain of BRD1 because its a-helix in acetylation status of histone H3 in SMMC-7721 cells, which Leu370-Thr375 turned into an b-sheet structure when sulfatide were transfected with the full-length BRD1 construct. Ectopic was bound (Fig. 3C, bottom). To explain the basis of the overexpression of BRD1 in the cells efficiently induced acety- affinity and specificity of the sulfatide interacting BRD1, we lation of H3K9/14 and Sp1, but not H4K12 (Fig. 5A). BRD1 constructed molecular models and superimposed the energy- was more abundant in the complex immunoprecipitated by the minimized structures of PHD1 and PHD2 domain in BRD1– Sp1 antibody following BRD1 transfection than in the mock sulfatide complex using backbone atoms and compared the control. We further investigated whether the acetyltransferases conformations of the active sites in both complexes, paying MOZ and HBO1 or HDAC1 were recruited to the BRD1 com- particular attention to the hydrophobic environment for sulfa- plex by sulfatide stimulation. Although only MOZ protein tide binding. The results showed that Arg322–Trp323–Lys324– upregulation was induced by sulfatide treatment, BRD1 immu- Leu325, loop region in PHD2 domain was closest to the sulfate noprecipitation from extracts isolated from SMMC-7721 cells group among all the residues nearby. treated with sulfatide followed by Western blot analysis showed that HBO1, MOZ, and HDAC1 existed in the BRD1 BRD1 recruitment to integrin aV gene promoter complex (Fig. 5B). However, co-immunoprecipitation experi- Next, we investigated the contribution of BRD1 to integrin aV ments showed that the recruitment of HDAC1, HBO1, and transcription. The integrin aV subunit gene promoter reporter ING4 except MOZ was not different among sulfatide, Gal-Cer, showed a significantly stronger response to sulfatide and BRD1 and control groups. Interestingly, HBO1 and MOZ occupancy

OF6 Mol Cancer Res; 2018 Molecular Cancer Research

Sulfatide Promotes ITGAV Expression

Figure 3. BRD1 binds to Sp1 and is recruited to integrin aV promoter. A, BRD1 and SIN3B in nuclear and cytoplasmic fractions were analyzed by Western blot analysis and quantified. B, Schematic representation of BRD1 structure and the deletion mutation constructs (top). Interaction between BRD1 and Bodipy-tagged sulfatide was examined by measuring the fluorescence intensity in the complex following immunoprecipitation with the HA or Flag antibody (left). The interaction between Sp1 and BRD1 deletion mutants was examined by immunoprecipitation with HA or Flag antibody and blotting with the Sp1 antibody (right). C, Molecular modeling of BRD1, cerebroside-BRD1, sulfatide–BRD1 complex (top). Superimposed modeling of BRD1 in cerebroside-BRD1 complex and sulfatide–BRD1 complex showing the conformational changes induced by sulfatide (bottom). The local conformation of the Leu370-Thr375 PHD2 domain turning from a-helix to b-sheet is framed and the amplification is shown on the right. Data are representative from three independent experiments. N.S., not significant; , P < 0.05; , P < 0.01.

www.aacrjournals.org Mol Cancer Res; 2018 OF7

Cai et al.

Figure 4. Correlation between integrin aV and BRD1. A, Integrin aV promoter reporter assay in cells cotransfected with BRD1 (left). BRD1 occupancy on the integrin aV promoter was determined by ChIP using a BRD1 antibody (middle) and quantification analysis (right). B, The BRD1 and Sp1 interaction was analyzed by co- immunoprecipitation and quantification analysis (top). Total expression of BRD1 and SIN3B was analyzed by Western blot analysis and quantified (middle) in the indicated samples. C, SIN3B and Sp1 proteins were measured by Western blot analysis in the complex precipitated by the BRD1 antibody (bottom). D, Representative immunohistochemical staining of BRD1 staining in tumor tissues (T) and corresponding adjacent nontumor (NT) liver tissues from 80 HCC patients (400). Immunoreactivity scores for BRD1 in the cytoplasm and the nucleus for individual HCC specimens (T) and the adjacent NT tissue (left). Nuclear BRD1 expression was analyzed with Image-Pro Plus software. The scores are presented as box plots, with horizontal lines for the median and range (middle). Analysis of cytoplasmic and nuclear BRD1 staining scores in patients (n ¼ 80) with and without metastasis (right top). Pearson's correlation analysis between integrin aV and BRD1 (right middle and bottom). Data are representative from three independent experiments. , P < 0.05; , P < 0.01.

OF8 Mol Cancer Res; 2018 Molecular Cancer Research

Sulfatide Promotes ITGAV Expression

on the integrin aV subunit gene promoter was significantly sulfatide, the integrin aV subunit gene promoter activity was still enhanced (Fig. 5C) by sulfatide treatment, whereas HDAC1 significantly suppressed by MOZ or HBO1 silencing (Fig. 7C, occupancy was attenuated although sulfatide was not recruited bottom). These data suggest that the HBO1 or MOZ was the to integrin aVgenepromoter. major HAT required for H3K9/14 acetylation at the integrin aV subunit gene promoter, which was induced by BRD1 and sulfa- BRD1 is crucial in sulfatide-induced transcription of the tide stimulation. integrin aV gene To confirm the role of BRD1 in sulfatide regulation, BRD1 was knocked down in SMMC-7721 cells (Fig. 6A). In these Discussion cells, BRD1 knockdown induced an attenuation of integrin aV Histone acetylation is critical for chromatin remodeling expression. Even in the cells stimulated with sulfatide, silenc- and the formation of a transcriptional complex (16). Histone ing of BRD1 efficiently suppressed integrin aV expression at acetylation is maintained by acetyltransferases, particularly the mRNA and protein levels (Fig. 6A). Importantly, depletion members of the MYST family (MOZ, MORF, and HBO1; of BRD1 profoundly decreased the H3K9/14 acetylation level, ref. 17). MOZ and HBO1 acetylate H3 at K9 and K14. whereas H4K12 acetylation was not significantly altered Although sulfatide is abnormally elevated in HCC, its signif- (Fig.6A).TheattenuationofintegrinaV expression by siBRD1 icance remains largely unknown (18). Our previous study wasalsoobservedbyimmunofluorescence(Fig.6B).Asinteg- indicated that the elevated sulfatide stimulated the expression rin aV was associated with cell migration potential, BRD1 of the integrin aV subunit gene (7) and induced metastatic knockdown significantly reduced the number of sulfatide-stim- behaviors. Here, we identified the mechanism by which sulfa- ulated migrating cells compared to control (Fig. 6B, bottom). tide regulates integrin aV subunit expression. In this study, we The adhesion results showed that silencing of BRD1 signifi- observed that sulfatide robustly increased the levels of acet- cantly attenuated sulfatide-promoted adhesion of SMMC-7721 ylated histone H3K9/14 in SMMC-7721 and BEL-7404 cells. cells to vitronectin and fibronectin (Fig. 6C). Although exog- Although acetylation is generally catalyzed by HATs, sulfatide enous sulfatide strongly increased integrin aVgenepromoter does not have intrinsic enzymatic activity and has not been transcription activity (Fig. 4A), silencing of BRD1 significantly associated with acetylation. Therefore, we proceeded to inves- suppressed the response of the integrin aV subunit gene pro- tigate whether this acetylation modification was related to the moter to sulfatide (Fig. 6D, top). Furthermore, sulfatide could protein compounded with sulfatide. By immunoprecipitation not rescue the activity of the integrin aV promoter suppressed and mass spectrometry, we identified 43 proteins bound to by BRD1 knockdown. BRD1 recruitment to integrin aVsubunit sulfatide, including BRD1. BRD1 has three major domains: gene promoter (Fig. 4A) was significantly enhanced by sulfatide the PHD, bromodomain, and PWWP domain. Bromodomains treatment, but knockdown of BRD1 attenuated the sulfatide- recognize acetylated lysine side chains mainly in histones induced MOZ and HBO1 recruitment to the integrin aVgene and are involved in transcriptional regulation (19). The sec- promoter (Fig. 6D, middle left). Although BRD1 was able to ond PHD domain binds directly to DNA (20).The PHD-zinc induce integrin aV subunit transcription, overexpression of knuckle-PHD module in BRD1 interacts with the unmodified BRD1 in cells with integrin aV knockdown could not stimulate N-terminus of histone H3 in a manner sensitive to lysine integrin aV expression (Fig. 6D, right). The integrin aVknock- modification (21). The N-terminal 198 amino acids of BRD1 down significantly reduced cell migration ability (Fig. 6E), and is the domain that interacts with HBO1 (22). BRD1 was the adhesion to vitronectin or fibronectin (Fig. 6F) although mainly distributed in cytoplasm in our observation, but its BRD1 was overexpressed in these cells. These suggest that BRD1 nuclear accumulation could be seen after sulfatide exposure. is crucial for the transcription of the integrin aV gene after The fluorescence-tagged sulfatide was recruited in BRD1 com- treatment with sulfatide. plex. The interaction between BRD1 and sulfatide may occur in the cytoplasm where BRD1 recognized the membrane- Acetyltransferase complex for H3K9/14 acetylation bound sulfatide at cytosolic side. The motif constituted by Transcription initiation requires the acetylation of local his- RWKL(322–324) was closest to the sulfate group of sulfatide tones. Because the acetyltransferases MOZ and HBO1 were and might provide positive charge for the electrostatic inter- recruited to the integrin aV subunit gene promoter by sulfatide action with sulfatide. Interestingly, this motif is located in the stimulation (Fig. 5C), we further investigated integrin aV expres- PHD2 domain in BRD1 dN fragment, which was coincident sion levels in the cells with MOZ or HBO1 knockdown to with the binding region of sulfatide. Besides electrostatic evaluate their roles in sulfatide regulation. Both siMOZ and attraction, hydrophobic interaction would be also important siHBO1 decreased integrin aV transcription and protein expres- since the PHD fragment alone was not sufficient for the sion levels (Fig. 7A). Interestingly, after MOZ or HBO1 knock- binding of sulfatide. BRD1 was also found to interact with down, the sulfatide stimulatory effects on integrin aV expression transcription factor Runx1 (23). Our study found that the at both the mRNA and protein levels (Fig. 7B) were almost fragment (192-538 amino acids) which includes PHD completely abrogated. Notably, the acetylation of histone H3K9/ domain, of BRD1 was necessary and sufficient for physical 14 was significantly decreased in MOZ or HBO1 knockdown interaction with Sp1 especially in the presence of sulfatide. cells, even after stimulation with sulfatide (Fig. 7B), although This suggested that sulfatide-induced Leu370-Thr375 con- there was no significant alteration in histone H4K12 acetylation. formational change enhanced BRD1 binding with Sp1 to form Accordingly, the reporter activity of the integrin aV subunit gene a complex that binds the promoter of the integrin aVgene. promoter that was stimulated by sulfatide was significantly Moreover, HBO1 and MOZ/MORF MYST HAT complexes suppressed by MOZ or HBO1 knockdown (Fig. 7C). Even in target chromatin via interactions with histone H3 tails in the the cells that overexpressed BRD1 and were stimulated with presence of multiple PHD finger domains (24). In a ChIP

www.aacrjournals.org Mol Cancer Res; 2018 OF9

Cai et al.

Figure 5. Acetylation promotes integrin aV expression. A, Histone acetylation was analyzed by Western blot analysis (left) and quantified (middle). BRD1 was quantified following Sp1 immunoprecipitation (right). B, HDAC1, HBO1, MOZ, and ING4 were analyzed by Western blot analysis in the BRD1 immunoprecipitation complex in the treated cells (top). The total expression levels of MOZ and HBO1 in the indicated cells were analyzed by Western blot analysis (middle and bottom). C, Occupancy of HDAC1, P300, HDAC9, HDAC10, BRG1,HBO1, MOZ, and sulfatide on the integrin aV promoter was determined by ChIP in untreated cells (left), or treated cells (right) and quantified by qPCR (bottom). Data are representative from three independent experiments. , P < 0.05; , P < 0.01.

assay, we observed the occupancy of BRD1 and AcH3K9/14 on action between BRD1, SIN3B, and Sp1. Following sulfatide the integrin aV subunit gene promoter. This interaction was treatment, the interaction of Sp1 with BRD1 was enhanced, signiﬁcantly enhanced by sulfatide. We also noted the inter- but the interaction with SIN3B was attenuated, suggesting that

OF10 Mol Cancer Res; 2018 Molecular Cancer Research

Sulfatide Promotes ITGAV Expression

Figure 6. BRD1 knockdown decreases the promotion effect of sulfatide. A, Knockdown efficiency was validated by Western blot analysis. Integrin aV and BRD1 expression and histone acetylation were analyzed by Western blot analysis in the indicated cells. B, Representative images of integrin aV fluorescent staining in the cells with BRD1 knockdown and sulfatide treatment (top). Representative images of cell migration in the presence of sulfatide or after BRD1 knockdown and quantitative analysis (bottom). C, Adhesion to collagen type I, fibrinogen, fibronectin, and vitronectin was analyzed in the cells with BRD1 knockdown. D, Luciferase activity of the reporter containing the integrin aV gene promoter in the cells transfected along with BRD1 siRNA or treated with sulfatide (top). ChIP analysis of BRD1 at the integrin aV gene promoter in the cells with BRD1 knockdown and sulfatide treatment (middle left). Integrin aV knockdown and BRD1 overexpression were confirmed by Western blot analysis and quantified (middle right and bottom). E, Representative images of cell migration and quantitative analysis in cells with integrin aV knockdown and BRD1 overexpression. F, The adhesion to collagen type I, fibrinogen, fibronectin, and vitronectin was analyzed in cells with BRD1 overexpression and integrin aV knockdown. Data are representative from three independent experiments. , P < 0.05; , P < 0.01.

www.aacrjournals.org Mol Cancer Res; 2018 OF11

Cai et al.

Figure 7. MOZ and HBO1 are important for the regulation of integrin aVexpression.A, Integrin aV expression was measured by RT-qPCR and Western blot analysis in cells with MOZ or HBO1 knockdown. B, RT-qPCR (top) measurement of integrin aV expression and Western blot (bottom) analysis of integrin aV and acetylated histones in cells treated with sulfatide and transfected with MOZ or HBO1 siRNA. C, Integrin aV promoter activity was determined by a reporter assay in cells with MOZ or HBO1 knockdown, treated with sulfatide, or transfected with the BRD1 construct. D, Summary of BRD1 roles in sulfatide regulation of integrin aV gene transcription.

OF12 Mol Cancer Res; 2018 Molecular Cancer Research

Sulfatide Promotes ITGAV Expression

regulation of protein complexes by sulfatide for transcription- Authors' Contributions al activation of the integrin aV gene occurs via acetylation. Conception and design: X.Z. Wu BRD1 joins the HAT complex with MOZ or HBO1 and thus Development of methodology: Q.Q. Cai, Y.W. Dong, X.Z. Wu mediates the acetylation of histone H3K9/14. In SMMC-7721 Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Q.Q. Cai, Y.W. Dong, B. Qi, X.-T. Shao, R. Wang, cells, we observed that overexpression of BRD1 enhanced the level Z.Y. Chen, B.M. He of acetylated histone H3K9/14. Conversely, silencing of BRD1 Analysis and interpretation of data (e.g., statistical analysis, biostatistics, signiﬁcantly inhibited the acetylation of H3K9/14 and the tran- computational analysis): Q.Q. Cai, Y.W. Dong scription induction of the integrin aV gene by sulfatide. BRD1 Writing, review, and/or revision of the manuscript: Q.Q. Cai, Y.W. Dong, failed to enhance integrin aV expression in cells transfected with X.Z. Wu integrin aV siRNA, suggesting that BRD1 could not affect post- Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Q.Q. Cai, B. Qi, X.-T. Shao, Z.Y. Chen, B.M. He transcription regulation. We also noted that the occupancy of Study supervision: X.Z. Wu MOZ and HBO1, which catalyze the H3K9/14 acetylation at the integrin aV gene promoter, was enhanced by sulfatide. Knock- ﬁ down of BRD1 signi cantly suppressed sulfatide-stimulated Acknowledgments a HBO1 or MOZ recruitment to the integrin V promoter. Further- This work was supported by grants from the National Natural Science more, knockdown of either MOZ or HBO1 inhibited the sulfatide Foundation of China (31570800, 31400689). stimulation and the BRD1 enhancement of integrin aV gene The authors thank Professors Atsushi Lwama, Chiba University, Japan, transcription. and David M. Heery, The University of Nottingham, for BRD1 and MOZ Taken together, this study demonstrated that enhancement of plasmids. integrin aV gene expression was associated with metastasis and the poor prognosis in patients with HCC. Interaction of sulfatide The costs of publication of this article were defrayed in part by the a payment of page charges. This article must therefore be hereby marked with BRD1 induced acetylation and promoted integrin V gene advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate transcription in HCC cells (Fig. 7D). this fact.

Disclosure of Potential Conﬂicts of Interest Received September 22, 2017; revised December 16, 2017; accepted January No potential conﬂicts of interest were disclosed. 26, 2018; published OnlineFirst February 16, 2018.

References 1. Lambert AW, Pattabiraman DR, Weinberg RA. Emerging biological prin- 13.HuP,ShiB,GengF,ZhangC,WuW,WuXZ.E-cadherincore ciples of metastasis. Cell 2017;168:670–91. fucosylation regulates nuclear beta-catenin accumulation in lung 2. Seguin L, Kato S, Franovic A, Camargo MF, Lesperance J, Elliott cancer cells. Glycoconj J 2008;25:843–50. KC, et al. An integrin beta(3)-KRAS-RalB complex drives tumour 14. Sun X, Yang G, Sun S, Quan R, Dai W, Li B, et al. The hydroxyl-functio- stemness and resistance to EGFR inhibition. Nat Cell Biol 2014; nalized magnetic particles for purification of glycan-binding proteins. 16:457–68. Curr Pharm Biotechnol 2009;10:753–60. 3. Ha SY, Shin J, Kim JH, Kang MS, Yoo HY, Kim HH, et al. Overexpression of 15. Wu LH, Shi BZ, Zhao QL, Wu XZ. Fucosylated glycan inhibition of integrin aV correlates with poor prognosis in colorectal cancer. J Clin human hepatocellular carcinoma cell migration through binding to Pathol 2014;67:576–81. chemokine receptors. Glycobiology 2010;20:215–23. 4. Desgrosellier JS, Cheresh DA. Integrins in cancer: biological 16. Kurdistani SK, Tavazoie S, Grunstein M. Mapping global histone acet- implications and therapeutic opportunities. Nat Rev Cancer 2010; ylation patterns to gene expression. Cell 2004;117:721–33. 10:9–22. 17. Yang XJ. MOZ and MORF acetyltransferases: Molecular interaction, animal 5. Zoni E, van der Horst G, van de Merbel AF, Chen L, Rane JK, Pelger development and human disease. Biochim Biophys Acta 2015;1853: RC, et al. miR-25 modulates invasiveness and dissemination of 1818–26. human prostate cancer cells via regulation of aV- and a6-integrin 18. Takahashi T, Suzuki T. Role of sulfatide in normal and pathological cells expression. Cancer Res 2015;75:2326–36. and tissues. J Lipid Res 2012;53:1437–50. 6. Wu XZ, Honke K, Zhang YL, Zha XL, Taniguchi N. Lactosylsulfatide 19. Zeng L, Zhou MM. Bromodomain: an acetyl-lysine binding domain. FEBS expression in hepatocellular carcinoma cells enhances cell adhesion to Lett 2002;513:124–8. vitronectin and intrahepatic metastasis in nude mice. Int J Cancer 20. LiuL,QinS,ZhangJ,JiP,ShiY,WuJ.SolutionstructureofanatypicalPHD 2004;110:504–10. finger in BRPF2 and its interaction with DNA. J Struct Biol 2012;180:165–73. 7.WuW,DongYW,ShiPC,YuM,FuD,ZhangCY,etal.Regulation 21. Lalonde ME, Avvakumov N, Glass KC, Joncas FH, Saksouk N, of integrin aV subunit expression by sulfatide in hepatocellular carci- Holliday M, et al. Exchange of associated factors directs a switch noma cells. J Lipid Res 2013;54:936–52. in HBO1 acetyltransferase histone tail specificity. Genes Dev 2013; 8. Brown CE, Lechner T, Howe L, Workman JL. The many HATs of 27:2009–24. transcription coactivators. Trends Biochem Sci 2000;25:15–9. 22. Mishima Y, Miyagi S, Saraya A, Negishi M, Endoh M, Endo TA, et al. 9. Doyon Y, Cayrou C, Ullah M, Landry AJ, Cote V, Selleck W, et al. ING tu- The Hbo1-Brd1/Brpf2 complex is responsible for global acetylation of mor suppressor proteins are critical regulators of chromatin acetylation re- H3K14 and required for fetal liver erythropoiesis. Blood 2011;118: quired for genome expression and perpetuation. Mol Cell 2006;21:51–64. 2443–53. 10. Filippakopoulos P, Knapp S. The bromodomain interaction module. FEBS 23. Mishima Y, Wang C, Miyagi S, Saraya A, Hosokawa H, Mochizuki-Kashio Lett 2012;586:2692–704. M, et al. Histone acetylation mediated by Brd1 is crucial for Cd8 gene 11. Dong RC, Zhou RH, Lv FD, Tao FD. Establishment and biological obser- activation during early thymocyte development. Nat Commun 2014; vation of human hepatocellular carcinoma cell line SMMC-7721. Bull 5:5872–82. Second Millitary Med Univ 1980;1:5–9. 24. Saksouk N, Avvakumov N, Champagne KS, Hung T, Doyon Y, 12. Pan X, Zhou T, Tai YH, Wang C, Zhao J, Cao Y, et al. Elevated expression of Cayrou C, et al. HBO1 HAT complexes target chromatin throughout CUEDC2 protein confers endocrine resistance in breast cancer. Nat Med gene coding regions via multiple PHD finger interactions with 2011;17:708–14. histone H3 tail. Mol Cell 2009;33:257–65.

www.aacrjournals.org Mol Cancer Res; 2018 OF13

BRD1-Mediated Acetylation Promotes Integrin αV Gene Expression Via Interaction with Sulfatide

Qian Qian Cai, Yi Wei Dong, Bing Qi, et al.

Mol Cancer Res Published OnlineFirst February 16, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-17-0527

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2018/02/16/1541-7786.MCR-17-0527.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/2018/02/27/1541-7786.MCR-17-0527. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.