Epstein–Barr Virus Nuclear Antigen 3C Binds to BATF/IRF4 Or SPI1/IRF4 Composite Sites and Recruits Sin3a to Repress CDKN2A

Epstein–Barr Virus Nuclear Antigen 3C binds to BATF/IRF4 or SPI1/IRF4 composite sites and recruits Sin3A to repress CDKN2A

Sizun Jianga,b,c,d, Bradford Willoxa,c, Hufeng Zhoua,c, Amy M. Holthausa,c, Anqi Wange, Tommy T. Shia,c, Seiji Maruof, Peter V. Kharchenkod,g, Eric C. Johannsene, Elliott Kieffa,b,c,1, and Bo Zhaoa,c

aDepartment of Medicine, Brigham and Women’s Hospital, bProgram in Virology, cDepartment of Microbiology and Immunobiology, and dCenter for Biomedical Informatics, Harvard Medical School, Boston, MA 02115; eDepartment of Medicine and McArdle Laboratory for Cancer Research, University of Wisconsin–Madison, Madison, WI 53706; fDepartment of Tumor Virology, Institute for Genetic Medicine, Hokkaido University, Sapporo 060-0815, Japan; and gDivision of Hematology/Oncology, Children’s Hospital, Boston, MA 02115

Contributed by Elliott Kieff, November 20, 2013 (sent for review October 3, 2013)

Epstein–Barr virus nuclear antigen 3C (EBNA3C) repression of LCL growth, indicating that EBNA3C is specifically required for CDKN2A p14ARF and p16INK4A is essential for immortal human LCL growth (6). Similar experiments reveal EBNA3C N-termi- B-lymphoblastoid cell line (LCL) growth. EBNA3C ChIP-sequencing nal amino acids 50–400 to be essential for LCL growth (3, 7, 8). identified >13,000 EBNA3C sites in LCL DNA. Most EBNA3C sites EBNA3C also up-regulates EBV LMP1 and cell CXCR4 and were associated with active transcription; 64% were strong CXCL12 gene expression (9–12), which are required for LCL H3K4me1- and H3K27ac-marked enhancers and 16% were active growth in nude mice (13). EBNA3C and EBNA3A joint re- p14ARF p16INK4A promoters marked by H3K4me3 and H3K9ac. Using ENCODE LCL pression of and expression is essential for LCL ARF INK4A p16INK4A transcription factor ChIP-sequencing data, EBNA3C sites coincided growth and knock down of p14 and p16 or null ± mutations allow LCL growth in the absence of EBNA3C, in- ( 250 bp) with RUNX3 (64%), BATF (55%), ATF2 (51%), IRF4 (41%), p14ARF p16INK4A MEF2A (35%), PAX5 (34%), SPI1 (29%), BCL11a (28%), SP1 (26%), dicating that EBNA3C repression of and is an essential EBNA3C function (14, 15). Both EBNA3A and EBNA3C TCF12 (23%), NF-κB (23%), POU2F2 (23%), and RBPJ (16%). EBNA3C have repressive activities that correlate with cell histone mod- sites separated into five distinct clusters: (i)Sin3A,(ii) EBNA2/RBPJ, MICROBIOLOGY iii iv v ifications: EBNA3A induces histone modifications at the CXCL10 ( ) SPI1, and ( )strongor() weak BATF/IRF4. EBNA3C signals and CXCL9 chemokine genes (16), whereas EBNA3C induces ARF were positively affected by RUNX3, BATF/IRF4 (AICE) and SPI1/IRF4 histone modifications which are important for p14 and INK4A (EICE) cooccupancy. Gene set enrichment analyses correlated p16 repression (14, 17). However, the detailed mecha- ARF INK4A EBNA3C/Sin3A promoter sites with transcription down-regulation nism through which EBNA3C represses p14 and p16 (P < 1.6 × 10−4). EBNA3C signals were strongest at BATF/IRF4 and ARF expression is unknown. SPI1/IRF4 composite sites. EBNA3C bound strongly to the p14 In contrast to EBNA2, which is tethered to DNA by RBPJ, promoter through SPI1/IRF4/BATF/RUNX3, establishing RBPJ-, Sin3A-, EBNA3C binding to RBPJ prevents RBPJ binding to DNA in and REST-mediated repression. EBNA3C immune precipitated with electrophoretic mobility-shift assays (EMSAs) and blocks EBNA2 Sin3A and conditional EBNA3C inactivation significantly decreased activation effects (18, 19). EBNA3C affects cell and EBV gene ARF Sin3A binding at the p14 promoter (P < 0.05). These data support expression through cell TFs. Chromatin immunoprecipitation a model in which EBNA3C binds strongly to BATF/IRF4/SPI1/RUNX3 (ChIP) followed by quantitative PCR (qPCR) found EBNA3C sites to enhance transcription and recruits RBPJ/Sin3A- and REST/ boundtovirusandcellpromotersites,includingtheEBV NRSF-repressive complexes to repress p14ARF and p16INK4A expression. LMP1, cell BIM, and ITGA4 promoters (20, 21, 22), whereas

EBV | tumor suppressor | resting B lymphocyte | lymphoma Significance

pstein–Barr virus (EBV) is a highly prevalent γ-herpesvirus Epstein–Barr virus (EBV) is an important causative agent of Ethat causes B and T lymphomas, Hodgkin disease, naso- B-cell lymphomas and Hodgkin disease in immune-deficient pharyngeal carcinoma, and gastric carcinoma. EBV-associated people, including HIV-infected people. The experiments de- B lymphomas and Hodgkin disease are more prevalent in T-cell scribed here were undertaken to determine the mechanisms immune-deficient people and are major causes of mortality in through which the EBV-encoded nuclear protein EBNA3C ARF INK4A HIV-infected people. EBV infection of B lymphocytes, in vitro, blocks the cell p14 and p16 tumor suppressor-mediated results in continuously proliferating lymphoblastoid cell lines inhibition of EBV-infected B-cell growth, thereby unfettering (LCLs). LCLs and EBV-infected cells in T-cell immune-deficient EBV-driven B-cell proliferation. The experiments also identify people express six EBV-encoded nuclear proteins (EBNA1, the molecular basis for diverse EBNA3C enhancer interactions EBNA2, EBNA3A, EBNA3B, EBNA3C, and EBNALP), three with cell DNA-binding proteins and cell DNA to regulate MYC, latent infection-associated membrane proteins (LMP1, LMP2A, pRB, BCL2, and BIM expression. Surprisingly, EBNA3C’s role in and LMP2B), microRNAs, and EBER 1 and 2 RNAs. Genetic enhancer-mediated cell gene transcription up-regulation is analyses indicate that EBNA1, EBNALP, EBNA2, EBNA3A, primarily mediated by combinatorial effects with cell tran- EBNA3C, and LMP1 are essential for LCL outgrowth (1, 2). scription factors, most notably AICEs, EICEs, and RUNX3. EBNA3A, EBNA3B, and EBNA3C are similar proteins, com- posed of ∼1,000 aa. Each has a near N-terminal site that binds Author contributions: S.J., E.C.J., E.K., and B.Z. designed research; B.W., A.M.H., T.T.S., and RBPJ, the cell sequence specific transcription factor (TF) that B.Z. performed research; A.M.H., A.W., S.M., and E.C.J. contributed new reagents/analytic mediates EBNA2 or Notch binding to DNA. EBNA3A and tools; S.J., H.Z., P.V.K., and B.Z. analyzed data; and S.J., E.K., and B.Z. wrote the paper. EBNA3C are both required for LCL growth, whereas EBNA3B is The authors declare no conflict of interest. not (3–5). When conditional, hydoxytamoxifen (HT)-dependent, Data deposition: The ChIP sequence data reported in this paper has been deposited in the EBNA3C (EBNA3CHT)-expressing LCLs are grown in medium GenBank database (accession no. GSE52632). without HT, the cells stop growing (6). HT addition or comple- 1To whom correspondence should be addressed. E-mail: [email protected]. mentation with wild-type (WT) EBNA3C-expression plasmids, This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. but not with EBNA3A- or EBNA3B-expression plasmids, restores 1073/pnas.1321704111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1321704111 PNAS | January 7, 2014 | vol. 111 | no. 1 | 421–426 Downloaded by guest on September 25, 2021 A binding sites, using input DNA controls. Sequencing reads were mapped to the human genome using Bowtie with two mismatches (28). Phantom peak calling was used to evaluate ChIP-seq quality. Both EBNA3C ChIP-seq replicates had a Quality tags >1, in ac- cordance with the Encyclopedia of DNA Elements (ENCODE) high quality data standard (29). The ChIP-seq processing pipeline (SPP) identified over 13,000 EBNA3C peaks with irreproducible discovery rates (IDRs) <0.01 (29, 30). EBNA3C peaks were annotated using the ENCODE GM12878 LCL epigenetic landscape, which divides the genome into seven functional domains defined by unique histone modifications (31). EBNA3C sites were 38% (4,969 sites) at strong enhancers, marked by high H3K4me1 and H3K27ac; 25% (3,287 sites) at weak enhancers, marked by in- termediate H3K4me1 and little H3K27ac; 8% (1,021 sites) at B active promoters, marked by high H3K4me3 and H3K9ac; 8% (1,108 sites) at weak or poised promoters, marked by high H3K4me3 and low H3K27ac; or high H3K27me3, and 16% (2,093 sites) at heterochromatin, marked by the absence of detectable histone modifications (Fig. 1A). EBNA3C peaks were significantly enriched over background controls at weak promoters, poised promoters, and strong and weak enhancers (Fisher’sexacttest: − P < 1 × 10 5)(Fig. S1A), identifying EBNA3C as a TF that significantly affects enhancer and promoter activity.

EBNA3C and EBNA2/RBPJ Peaks Overlap. Reanalysis of previous EBNA2 and RBPJ data (32) using IDR identified the top 5,000 EBNA2 and 10,000 RBPJ sites. Overall, 9% of EBNA3C sites overlapped with EBNA2/RBPJ sites and 7% overlapped with RBPJ sites lacking EBNA2 (Fig. 1B). EBNA3C was 84% at DNA sites without significant RBPJ, whereas most EBNA2 sites coincided with RBPJ. EBNA2-associated RBPJ sites had two- fold higher signals than RBPJ sites that lacked EBNA2, indicating a strong EBNA2 effect on RBPJ association with DNA

Fig. 1. Genome-wide distribution of EBNA3C sites and EBNA3C site overlap with EBNA2/RBPJ, BATF/IRF4, SPI1/IRF4, and RUNX3. (A) Genome-wide distribution of EBNA3C DNA binding by chromatin states (%). EBNA3C distribution calculated based on the chromatin state of GM12878 cells; 12,992 EBNA3C sites within annotated genomic regions were used to identify the genome-wide distribution of EBNA3C sites. (B) Venn diagram showing EBNA3C site overlap with RUNX3, SPI1/IRF4, EBNA2/RBPJ, or BATF/IRF4 (±250bpofEBNA3Csites).

ChIP-sequencing (ChIP-seq) using an antibody that immune precipitates EBNA3A, EBNA3B, and EBNA3C, found an EBNA3 at around 7,000 sites in a Burkitt lymphoma cell line (23), consistent with the hypothesis that EBNA3C may affect transcription through interactions with other cell TFs (11, 23–27). Recently, EBNA3C residues 130–159wereshowntobindtoIRF4orIRF8,TFsim- portant for lymphopoiesis (24). EBNA3C can also coactivate the EBV LMP1 promoter with EBNA2 through a SPI1 site (10, 11). Overall, these findings are consistent with the hypothesis that EBNA3C affects transcription through interactions with cell TFs. We therefore undertook ChIP-seq studies to identify EBNA3C’s genome-wide interactions with LCL DNA and cell TFs. EBNA3C was at around 13,000 genomic sites and was highly colocalized with cell TFs RUNX3, BATF, IRF4, and SPI1. EBNA3C re- ARF cruited Sin3A repressive complexes to the p14 promoter to ARF INK4A mediate p14 and p16 repression. Results Fig. 2. EBNA3C-associated cell TF clusters, chromatin state, and associated motifs at different clusters. Partitioning Around Medoids (PAM) clustering EBNA3C Binding Sites in LCLs. An LCL transformed by a recombi- was used to cluster all nonheterochromatin EBNA3C sites together with mid nant EBV BAC in which EBNA3C was C-terminally tagged EBNA2 and cell TFs into five unique clusters. ChIP-seq signals centered at the with HA and Flag epitopes (EBNA3CFH) was used for ChIP- EBNA3C sites with neighboring ±2 kb were shown on a red (highest binding) seq analyses. The EBNA3CFH LCLs grew similarly to WT LCLs to white (no binding) scale. The chromatin state distribution of EBNA3C sites and expressed similar EBNA3C levels. Two anti-HA ChIP-seq are represented in the column titled “State.” Enriched motifs at each cluster biological replicates identified EBNA3C genome-wide DNA are shown next to the clusters.

422 | www.pnas.org/cgi/doi/10.1073/pnas.1321704111 Jiang et al. Downloaded by guest on September 25, 2021 and weak enhancer marks (Fig. 2, yellow). Cluster 5 was enriched for STAT1 and STAT5 motifs, for which there are no ENCODE data, and had moderate EBNA3C signals, with weak RUNX3, BATF, and MEF2A signals, with weak enhancer marks. RBPJ was at only 16% of EBNA3C sites, which were mostly cluster 2 strong enhancers with high EBNA2, RBPJ, RUNX3, and EBNA3C signals. In contrast, EBNA3C clusters 1 and 4 and especially five sites (Fig. 2) had much lower RBPJ signals than cluster-2 sites, indicating that EBNA3C was associated with DNA at sites that are RBPJ deficient. Anchor plots of Sin3A signals show that Sin3A signals were high in cluster 1 and very low in all other clusters. Anchor plots of EBNA2 or RBPJ signals indicated that EBNA2 or RBPJ signals were much stronger in cluster 2 than in other clusters (Fig. S1B). However, SPI1 signals were only strong at cluster-3 sites, whereas BATF and IRF4 Fig. 3. Anchor plots of EBNA3C (E3C) or RUNX3 centered at E3C or RUNX3 signals were strong at cluster-2, -3, and -4 sites (Fig. S1B). These peaks. E3C (Left) or RUNX3 (Right) was intersected pairwise (±250 bp) with data are consistent with EBNA2 and RBPJ having strong en- HOMER and the anchor plots of E3C without RUNX3, E3C with RUNX3, or hancer effects at EBNA3C cluster-2 sites. RUNX3 without E3C ChIP-seq signals were plotted around a ±2-kb region. EBNA3C Signals Are Affected by Cooccupancy with RUNX, BATF, IRF4, SPI1, or RBPJ. Anchor plots of EBNA3C sites—alone or at sites (32). In contrast, RBPJ sites with or without EBNA3C had similar with RUNX3, BATF, IRF4, or SPI1—had stronger EBNA3C- signals (Fig. S1C). normalized ChIP-seq signals (Fig. 3 and Fig. S1C). Sites cooccupied with RUNX3, BATF, IRF4, or SPI1, had EBNA3C sig- EBNA3C Site-Enriched Cell TF Motifs. Motif analysis using HOMER nals that were increased from 15 to 18 for EBNA3C sites alone ± to 22–25 for EBNA3C sites with RUNX3, BATF, IRF4, or SPI1 identified EBNA3C sites ( 250 bp) to be significantly enriched − for cell TF-binding motifs, including IRF4 (49%), RUNX (44%), (one-tailed paired t test, P < 10 20)(Fig. S1C). Similarly, SPI1-IRF4 (46%), REL (18%), E2A (21%), and EBF1 (12%). RUNX3, BATF, IRF4, or SPI1 had stronger signals at sites Homer known motif enrichment analysis identified SPI1-IRF4 cooccupied with EBNA3C than at sites with these factors alone (50%), IRF4 (30%), AP-1 (37%), RUNX (40%), JUN-AP1 and EBNA3C sites co-occurring with RBPJ or EBNA2 had MICROBIOLOGY (15%), ETS1 (34%), SPI1 (21%), and ISRE (10%). stronger EBNA3C signals than EBNA3C alone. However, RBPJ and EBNA2 sites with or without EBNA3C had similar signals, EBNA3C Sites Are Highly Associated with Cell TFs. ENCODE consistent with EBNA3C having no incremental effect on EBNA2 GM12878 LCL ChIP-seq data were used to elucidate EBNA3C or EBNA2 and RBPJ signals (Fig. S1C). EBNA3C sites with or site cell TF cooccupancies. Cell TFs, including those identified without Sin3A had similar signals, whereas Sin3A sites with −30 by motif (Fig. 2), frequently cooccupied EBNA3C DNA sites EBNA3C had higher signals than Sin3A only sites, (P < 10 ), (±250 bp) (Table S1). EBNA3C site-associated cell TFs in- consistent with EBNA3C recruiting Sin3A to these sites (Fig. cluded RUNX3 (64%), BATF (55%), ATF2 (51%), IRF4 (41%), S1C). Furthermore, RBPJ signals at sites with Sin3A had lower MEF2A (35%), PAX5 (34%), SPI1 (29%), BCL11a (28%), SP1 signals than RBPJ sites alone (Fig. S1D). In contrast, Sin3A (26%), TCF12 (23%), RELA (23%), POU2F2/Oct2 (23%), RBPJ signals at Sin3A and RBPJ sites were stronger than Sin3A signals P < −11 (16%), EBNA2 (9%), RXRA (7%), and Sin3A (1%) (Table S1). without RBPJ ( 10 ), consistent with RBPJ tethering Sin3A D In B and T lymphocytes, IRF4 (or IRF8) frequently bind with ATF to DNA (Fig. S1 ). to composite elements (AICEs) or with ETS to composite ele- Notably, EBNA3C signals at BATF/IRF4 (AICE) sites were P < × −23 ments (EICEs) (33–35). EBNA3C can bind to IRF4 or IRF8 (24). stronger than EBNA3C-alone sites ( 1 10 ), EBNA3C/ P < −25 P < −28 Indeed, AP-1/IRF4 motifs were highly enriched at EBNA3C sites BATF ( 10 ) sites, or EBNA3C/IRF4 sites ( 10 ), (Fig. 2). Furthermore, EBNA3C sites also coincided with RUNX3 consistent with EBNA3C preferential binding to AICE com- (Fig. 1B). Overall, 37% of EBNA3C sites coincided with BATF posite sites. In parallel, EBNA3C signals at SPI1/IRF4 (EICE) sites were similar to EBNA3C/IRF4 sites and higher than EBNA3C and IRF4 and 20% of EBNA3C sites coincided with SPI1 and −23 −27 B alone (P < 10 )orEBNA3C/SPI1sites(P < 10 ), consistent IRF4 (Fig. 1 ). Thus, EBNA3C was highly associated with AICEs ’ or EICEs. with EBNA3C s preferential binding to EICE sites (Fig. 4). Most of EBNA3C/IRF4 sites without SPI1 were associated with BATF B EBNA3C Promoter and Enhancer Site-Associated Cell TFs Divided into (Fig. 1 ), accounting for their similar signal. Five Unique Clusters Using Partitioning Around Medoids. EBNA3C sites coincided with EBNA2, RBPJ, RUNX3, BATF, IRF4, SPI1, MEF2A, REST/NRSF, and Sin3A in five distinct clusters (Fig. 2). Surprisingly, cluster 1 was highly enriched for RUNX3, Sin3A, and REST/NRSF. Sin3A is a repressive scaffold for histone deacetylases 1 and 2 (HDAC 1 and 2) (36), but was mostly at active promoter-associated EBNA3C sites. Repressive HDACs frequently bind to active promoters (37). Sin3A HDAC-repressive complexes may reset chromatin to preactivation levels to prevent effects of prolonged gene activation (37). Sin3A cluster-enriched cell TF motifs included E2F and IRF4. EBNA3C cluster 2 was strongly enriched for AP-1/IRF4 (AICE), RUNX, RBPJ, and EBF1 motifs and had strong EBNA2, RBPJ, RUNX3, BATF, and IRF4 signals, with predominantly strong enhancer marks (Fig. 2, purple). EBNA3C cluster 3 was enriched in IRF4/SPI1 (EICE), SPI1, RUNX and AP-1 motifs and had high SPI1, RUNX3, BATF, and IRF4 signals with strong enhancer marks. Cluster 4 was enriched for AP1/IRF4 (AICE), Fig. 4. Anchor plots at EBNA3C sites, showing the effects of AICEs (BATF RUNX, REL, and ETS1 motifs and had prominent RUNX3, and IRF4, Left) and EICEs (SPI1 and IRF4, Right) on EBNA3C ChIP-seq signals BATF, IRF4, and EBNA3C signals, with minimal SPI1 signals on DNA. ChIP-seq signals were plotted around a ±1-kb region.

Jiang et al. PNAS | January 7, 2014 | vol. 111 | no. 1 | 423 Downloaded by guest on September 25, 2021 which EBNA3C binds to BATF/IRF4/SPI1 AICE or EICE sites ARF at the p14 promoter and recruits RBPJ with its associated repressors, notably including Sin3A (39). ENCODE Sin3A ChIP-seq data identified Sin3A signals that ARF were coincident with EBNA3C signals at the p14 promoter site (Fig. 6A). To test the hypothesis that EBNA3C recruits ARF ARF Sin3A to the p14 promoter to repress p14 expression, Sin3A ChIP-qPCRs were done using conditional EBNA3CHT LCLs. Under permissive conditions for EBNA3C expression, ARF Sin3A was enriched 41-fold at the p14 promoter, normalized to control antibody. Whereas, under nonpermissive conditions for EBNA3C expression, Sin3A was enriched 12-fold compared with control antibody (P < 0.05) (Fig. 6B). These data indicate ARF that Sin3A binding to p14 DNA is EBNA3C dependent, ARF further indication that EBNA3C tethers Sin3A to the p14 ARF promoter to repress p14 expression. Previous immune flo- Fig. 5. GSEA analysis of EBNA3C/Sin3A binding and gene expression. The rescence studies had shown that EBNA3C extensively colocalizes enrichment profile is shown on top, 229 genes with EBNA3C/Sin3A sites at ±2 with Sin3A in LCLs (24). To further evaluate EBNA3C associ- kb of TSS were indicated with blue vertical lines in the center, and 9,193 genes ation with Sin3A in LCLs, agarose beads conjugated to anti-Flag profiled were ranked from EBNA3C induced (left) to EBNA3C repressed (right). antibody were used to pull down EBNA3CHF and associated proteins from EBNA3CHF LCLs. Western blotting showed that EBNA3CHF immune precipitated 0.5–1% of input Sin3A from EBNA3C Promoter Sites Correlated with Effects on Gene Expression. EBNA3CHF LCLs, whereas Sin3A was not detected in immune Transcription profiling of conditional EBNA3CHT LCLs under precipitates from control non–HF-tagged GM12878 LCLs (GM) nonpermissive conditions for EBNA3C expression versus the (Fig. 6C). Together with previous studies, these results indicate ARF same LCLs transcomplemented with WT EBNA3C expression that recruitment of Sin3A complexes to the p14 promoter is identify EBNA3C conditionally regulated cell genes (12). Eight EBNA3C dependent. hundred and ninety one EBNA3C sites localized within ±2kbof the transcription start site (TSS) of an EBNA3C dynamically affected cell gene (Fig. S2). Gene set enrichment analysis (GSEA) negatively correlated EBNA3C promoter binding with − cell gene repressive effects (P < 1 × 10 6, Fig. S2) (38). GSEA A analysis of 409 EBNA3C/Sin3A sites in cluster 1 (Fig. 2) were at 292 affected cell promoters and correlated WT EBNA3C expression levels with normal cell growth and a 28% average lower − expression of EBNA3C/Sin3A-affected genes (P < 1.6 × 10 4, Fig. 5), indicative of repressed cell gene expression at EBNA3C- and Sin3A-bound promoters.

EBNA3C Site-Associated Histone Modifications. To evaluate EBNA3C, EBNA2, and associated cell TF effects on epigenetic modifications indicative of enhanced or repressed transcription, ENCODE LCL histone H3K4me1, H3K27ac, H3K4me3, H3K27me3, and p300 ChIP-seq signals at EBNA3C sites were analyzed (Fig. S3). EBNA3C cluster-2 sites with significant EBNA2/RBPJ signals had highest H3K4me1, H3K27ac, nucleosome depletion, and p300 signals, with low H3K27me3, indicative of high-level enhancer activity. EBNA3C sites with significant Sin3A signals had high H3K4me3, H3K9ac, and nucleosome depletion consistent with promoter activation. However, GSEA analyses indicate that EBNA3C has repressive effects on cluster-1 promoters, com- patible with a model that EBNA3C attenuates promoter over- activation, by recruiting repressive Sin3A complexes to these B C sites (Fig. 5).

EBNA3C Binds to the p14ARF Promoter and Recruits Sin3A. To un- derstand the molecular mechanisms through which EBNA3C ARF INK4A represses p14 and p16 expression, we investigated EBNA3C binding to these loci. By ChIP-seq, EBNA3C localized ARF ARF to the p14 promoter (Fig. 6A). The EBNA3C p14 promoter site coincided with strong BATF, IRF4, and SPI1 signals p14ARF consistent with EBNA3C binding to the promoter Fig. 6. EBNA3C site at the p14ARF promoter and association with Sin3A. (A) through AICES or EICES (Fig. 6A). Notably, EBNA2 was ab- ARF p14ARF EBNA3C binding at the p14 promoter. Normalized EBNA3C, Sin3A, and sent from the promoter site and weak RBPJ signals co- other TF tag density signals at the CDKN2A/B locus are shown. Red arrows incided with EBNA3C, consistent with the hypothesis that indicate the direction of transcription. The peak heights are indicated at the EBNA3C can recruit RBPJ to AICE or EICE sites (Fig. 6A). ARF left of and for each track. (B) EBNA3CHT LCLs were grown under permissive Although RBPJ signals at the p14 promoter site were two- to or nonpermissive conditions for 7 d. Sin3A enrichment at the p14ARF pro- threefold over controls after 10 min of formaldehyde cross- moter over the control antibody was determined by ChIP-qPCR. (C) Anti-Flag linking, 20 min of cross-linking increased RBPJ signals to 13-fold agarose beads were used to immune precipitate EBNA3CFH and associated over control antibody, consistent with RBPJ being more distal to proteins from EBNA3CFH LCL with GM12878 LCLs as control. EBNA3C-asso- DNA than EBNA3C. These data are consistent with a model in ciated Sin3A was detected by Western blotting.

424 | www.pnas.org/cgi/doi/10.1073/pnas.1321704111 Jiang et al. Downloaded by guest on September 25, 2021 is not directly binding DNA, but is tethered to DNA by EBNA3C binding to composite IFR4/BATF/SPI1/RUNX3 sites. EBNA3C is most frequently bound to DNA by AICE (IRF4/ BATF) and EICE (IRF4/SPI1) sites. EBNA3C signals are also positively affected by RUNX3. RUNX3, IRF4, BATF, and SPI1 likely bridge EBNA3C to DNA. ChIP-seq and EMSA experiments implicate cooperative BATF/IRF4 binding to AICE (33–35). Our data detect extensive EBNA3C interactions with AICE and EICE sites to drive LCL proliferation. IRF8 can replace IRF4 in AICE complexes in B- and T-cell development and EBNA3C can bind to IRF8 (24). These data further underscore the extent to which EBV uses intrinsic B-cell developmental pathways to immortal- ize B cells (Fig. 8). INK4A ARF EBNA3C repression of p16 and p14 expression is essential for LCL growth (14, 15). In LCLs this repression is marked by increased H3K27me3 across this locus (14, 17). However, the mechanisms through which EBNA3C mediates repression were unclear. Using ENCODE data, we found Sin3A- and REST/NRSF-repressive complexes associated with EBNA3C ARF at the p14 promoter. EBNA3C bound to Sin3A and EBNA3C- INK4A ARF Fig. 7. EBNA3C sites upstream of MYC. EBNA3C, EBNA2, RBPJ, IRF4, BATF, Sin3A complexes repressed p16 and p14 transcription SPI1, EBF, Pol II, and histone signals at MYC andupto∼600 kb upstream (Fig. 6). Conditional EBNA3C inactivation resulted in loss are shown. of Sin3A at this site, indicating that EBNA3C is required for ARF INK4A Sin3A DNA binding. The CDKN2A (p14 and p16 ) and INK4B CDKN2B (p15 ) sites are highly dynamic in chromatin con- EBNA3C Binds to MYC Enhancers, the pRB Promoter, and BIM INK4B INK4A formation (43). The p15 promoter can loop to the p16 MYC ARF Enhancers. EBNA2 and EBNALP induce expression within promoter to coregulate both genes, looping out the p14 pro- 24 h after EBV infection of resting B lymphocytes. EBNA2 binds ARF INK4A moter (36). Because EBNA3C represses p14 and p16 to multiple enhancer sites 400–500 kb upstream of MYC, causing simultaneously, these repressive effects are likely coregulated by MICROBIOLOGY enhancer looping to the MYC promoter, cell-cycle entry, and cell ARF the p14 promoter, via looping. proliferation (32). After EBNA3C expression is turned on in These experiments further illustrate the extent to which EBV infected B cells, EBNA3C attenuates MYC expression, prevent- has evolved to use preexisting B-cell programs to drive cell-cycle ing MYC overexpression-induced apoptosis (40). LCL growth entry to maintain infected cell persistence or alternatively to requires MYC modulation and expression of antiapoptotic BCL2 enable virus replication. family proteins. However, EBNA3C can also up-regulate MYC when its expression is low. EBNA3C up-regulates MYC 1.3-fold P < × −4 Materials and Methods ( 1 10 ) in EBNA3CHT LCLs (12). EBNA3C localized to mid 18 sites upstream of MYC (Fig. 7), a region deficient in other Cell Lines. LCLs transformed by a recombinant EBV BAC in which EBNA3C annotated genes. These sites have pronounced H3K4me1, was C-terminally tagged with HA and Flag epitopes (EBNA3CFH) were grown in RPMI 1640 medium supplemented with 10% (vol/vol) Fetal Plex H3K27Ac, Pol II, and p300 enhancer marks, implicating EBNA3C in MYC up-regulation. pRb is frequently inactivated in tumors, and human papilloma virus- or polyomavirus-infected cells have low pRb levels, which enable continuous cell proliferation. EBNA3C can associate with and decrease pRb (41). EBNA3C down-regulated pRB 1.21-fold in conditional EBNA3C transcription-profiling experiments (12). EBNA3C and Sin3A peaks at the pRB promoter and enhancer likely mediate pRB repression (Fig. S4). EBNA3C was also significantly enriched at the IRF4 and BATF promoters (Fig. S4). EBNA3C and Sin3A localization at the BIM promoter support ChIP-qPCR studies of EBNA3C-mediated BIM repression (21, 42). Strong EBNA3C signals likely down-regulate BIM by bringing Sin3A to BIM promoter and enhancer sites (Fig. S5B). These findings position EBNA3C at cell genes critical for cell growth and survival. Discussion EBNA3C amino acids 50–400 are essential for continuous LCL growth. This region includes an IRF4-binding domain, amino acids 130–150 (24); an RBPJ-binding domain, amino acids 180– 231 (7, 8, 18); and an SPI1-binding domain, amino acids 181–365 (11). These EBNA3C-associated cell TFs are likely critical for EBNA3C association with DNA and transcription effects. Previously, EBNA3C interaction with DNA could only be assessed at a few sites (11, 19). The EBNA3C ChIP-seq analysis presented here finds EBNA3C at over 13,000 cell DNA sites. < Surprisingly, RBPJ is at only 16% of EBNA3C sites. EBNA3C/ Fig. 8. EBNA3C binds to promoters via RUNX3, IRF4, and E2F to recruit RBPJ DNA signals were similar to EBNA3C IRF4/BATF/SPI1/ Sin3A repressor complexes (Sin3A, HDACs 1 and 2, and RBPJ). This complex RUNX3 DNA signals (Fig. 2 and Fig. S1C). EBNA3C-associ- represses the nearby p14ARF TSS. EBNA3C binds to H3K4me1-enriched ated RBPJ had lower DNA signals than EBNA2-associated enhancers via BATF/IRF4 and/or SPI1, RUNX3 to recruit p300 and drive RBPJ, consistent with the model that EBNA3C-associated RBPJ distal target gene expression.

Jiang et al. PNAS | January 7, 2014 | vol. 111 | no. 1 | 425 Downloaded by guest on September 25, 2021 animal serum complex (Gemini) and 2 mM L-glutamine, and 4HT withdraw SPP was used to call peaks with an IDR < 0.01 (29, 30). EBNA3C peaks were experiments were done as described (6). mapped to the hg18 GM12878 chromatin states (31).

ChIP-Seq. Anti-HA antibody (Abcam, ab9110) was used to immune precipitate Enriched Motif Calling. Sequences ±250 bp of EBNA3C sites per cluster (Fig. 2) EBNA3CFH from formaldehyde cross-linked EBNA3CFH LCLs. After extensive were extracted for motif analysis using HOMER. Random sites with identical washing, Protein-DNA complexes were eluted and cross-linking was reversed (31). chromatin distribution were used as control. HOMER was used to identify The purified DNA was sequenced using a HiSEq 2500 (Illumina). enriched motifs around these EBNA3C sites (44). Additional materials and methods are available SI Materials and Methods. ChIP-Seq Data Analyses. ChIP-seq reads were mapped to hg18 using Bowtie (27), allowing one alignment per read and two mismatches per read. The 39.9 ACKNOWLEDGMENTS. The authors thank Anshul Kundaje, Burak Alver, Richard Sandstrom, and Ben Gewurz for helpful input and discussions. These and 43.2 million unique reads from biologic EBNA3C replicates 1 and 2, of experiments were supported by Grants R01CA047006, R01CA170023, which 37.5 million reads (78.8%) and 35.9 million reads (77.8%) mapped to R01CA085180, and R01CA131354 from the National Cancer Institute, the the hg18 genome. Phantom peak calling was used for quality control (29). National Institutes of Health, and the US Public Health Service.

1. Rickinson A, Kieff E (2007) Epstein-Barr virus. Fields Virology, eds Howley P, Knipe D 22. Jiménez-Ramírez C, et al. (2006) Epstein-Barr virus EBNA-3C is targeted to and reg- (Lippincott Williams & Wilkins, Philadelphia), pp 2655–2700. ulates expression from the bidirectional LMP-1/2B promoter. J Virol 80:11200–11208. 2. Kieff E, Rickinson A (2007) Epstein-Barr virus and its replication. Fields Virology, eds 23. McClellan MJ, et al. (2013) Modulation of enhancer looping and differential gene Howley P, Knipe D (Lippincott Williams & Wilkins, Philadelphia), pp 2603–2654. targeting by Epstein-Barr virus transcription factors directs cellular reprogramming. 3. Tomkinson B, Robertson E, Kieff E (1993) Epstein-Barr virus nuclear proteins EBNA-3A and PLoS Pathog 9(9):e1003636. EBNA-3C are essential for B-lymphocyte growth transformation. JVirol67(4):2014–2025. 24. Banerjee S, et al. (2013) The EBV Latent Antigen 3C Inhibits Apoptosis through Targeted 4. Chen A, Divisconte M, Jiang X, Quink C, Wang F (2005) Epstein-Barr virus with the Regulation of Interferon Regulatory Factors 4 and 8. PLoS Pathog 9(5):e1003314. latent infection nuclear antigen 3B completely deleted is still competent for B-cell 25. Knight JS, Lan K, Subramanian C, Robertson ES (2003) Epstein-Barr virus nuclear an- – growth transformation in vitro. J Virol 79(7):4506 4509. tigen 3C recruits histone deacetylase activity and associates with the corepressors 5. Tomkinson B, Kieff E (1992) Use of second-site homologous recombination to dem- mSin3A and NCoR in human B-cell lines. J Virol 77(7):4261–4272. onstrate that Epstein-Barr virus nuclear protein 3B is not important for lymphocyte 26. Radkov SA, et al. (1999) Epstein-Barr virus nuclear antigen 3C interacts with histone infection or growth transformation in vitro. J Virol 66(5):2893–2903. deacetylase to repress transcription. J Virol 73(7):5688–5697. 6. Maruo S, et al. (2006) Epstein-Barr virus nuclear protein EBNA3C is required for cell 27. Touitou R, Hickabottom M, Parker G, Crook T, Allday MJ (2001) Physical and func- cycle progression and growth maintenance of lymphoblastoid cells. Proc Natl Acad Sci tional interactions between the corepressor CtBP and the Epstein-Barr virus nuclear USA 103(51):19500–19505. antigen EBNA3C. J Virol 75(16):7749–7755. 7. Lee S, et al. (2009) Epstein-Barr virus nuclear protein 3C domains necessary for lym- 28. Langmead B, Trapnell C, Pop M, Salzberg SL (2009) Ultrafast and memory-efficient phoblastoid cell growth: Interaction with RBP-Jkappa regulates TCL1. J Virol 83(23): 12368–12377. alignment of short DNA sequences to the human genome. Genome Biol 10(3):R25. 8. Maruo S, et al. (2009) Epstein-Barr virus nuclear protein EBNA3C residues critical for 29. Landt SG, et al. (2012) ChIP-seq guidelines and practices of the ENCODE and mod- – maintaining lymphoblastoid cell growth. Proc Natl Acad Sci USA 106(11):4419–4424. ENCODE consortia. Genome Res 22(9):1813 1831. 9. Allday MJ, Farrell PJ (1994) Epstein-Barr virus nuclear antigen EBNA3C/6 expression 30. Kharchenko PV, Tolstorukov MY, Park PJ (2008) Design and analysis of ChIP-seq ex- – maintains the level of latent membrane protein 1 in G1-arrested cells. JVirol68:3491–3498. periments for DNA-binding proteins. Nat Biotechnol 26(12):1351 1359. 10. Lin J, Johannsen E, Robertson E, Kieff E (2002) Epstein-Barr virus nuclear antigen 3C 31. Ernst J, et al. (2011) Mapping and analysis of chromatin state dynamics in nine human putative repression domain mediates coactivation of the LMP1 promoter with EBNA- cell types. Nature 473(7345):43–49. 2. J Virol 76(1):232–242. 32. Zhao B, et al. (2011) Epstein-Barr virus exploits intrinsic B-lymphocyte transcription 11. Zhao B, Sample CE (2000) Epstein-Barr virus nuclear antigen 3C activates the latent programs to achieve immortal cell growth. Proc Natl Acad Sci USA 108(36):14902–14907. membrane protein 1 promoter in the presence of Epstein-Barr virus nuclear antigen 33. Tussiwand R, et al. (2012) Compensatory dendritic cell development mediated by 2 through sequences encompassing an spi-1/Spi-B binding site. JVirol74(11):5151–5160. BATF-IRF interactions. Nature 490(7421):502–507. 12. Zhao B, et al. (2011) Epstein-Barr virus nuclear antigen 3C regulated genes in lym- 34. Glasmacher E, et al. (2012) A genomic regulatory element that directs assembly and phoblastoid cell lines. Proc Natl Acad Sci USA 108(1):337–342. function of immune-specific AP-1-IRF complexes. Science 338(6109):975–980. 13. Piovan E, et al. (2005) Chemokine receptor expression in EBV-associated lympho- 35. Ochiai K, et al. (2013) Transcriptional regulation of germinal center B and plasma cell proliferation in hu/SCID mice: Implications for CXCL12/CXCR4 axis in lymphoma fates by dynamical control of IRF4. Immunity 38(5):918–929. generation. Blood 105(3):931–939. 36. Nagy L, et al. (1997) Nuclear receptor repression mediated by a complex containing 14. Maruo S, et al. (2011) Epstein-Barr virus nuclear antigens 3C and 3A maintain lym- SMRT, mSin3A, and histone deacetylase. Cell 89(3):373–380. phoblastoid cell growth by repressing p16INK4A and p14ARF expression. Proc Natl 37. Wang Z, et al. (2009) Genome-wide mapping of HATs and HDACs reveals distinct – Acad Sci USA 108(5):1919 1924. functions in active and inactive genes. Cell 138(5):1019–1031. 15. Skalska L, et al. (2013) Induction of p16(INK4a) is the major barrier to proliferation 38. Subramanian A, et al. (2005) Gene set enrichment analysis: A knowledge-based ap- when Epstein-Barr virus (EBV) transforms primary B cells into lymphoblastoid cell proach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA lines. PLoS Pathog 9(2):e1003187. 102(43):15545–15550. 16. Harth-Hertle ML, et al. (2013) Inactivation of intergenic enhancers by EBNA3A ini- 39. Zhou S, Fujimuro M, Hsieh JJ-D, Chen L, Hayward SD (2000) A role for SKIP in EBNA2 tiates and maintains polycomb signatures across a chromatin domain encoding activation of CBF1-repressed promoters. J Virol 74(4):1939–1947. CXCL10 and CXCL9. PLoS Pathog 9(9):e1003638. 40. Nikitin PA, et al. (2010) An ATM/Chk2-mediated DNA damage-responsive signaling 17. Skalska L, White RE, Franz M, Ruhmann M, Allday MJ (2010) Epigenetic repression of pathway suppresses Epstein-Barr virus transformation of primary human B cells. Cell p16(INK4A) by latent Epstein-Barr virus requires the interaction of EBNA3A and EB- Host Microbe 8(6):510–522. NA3C with CtBP. PLoS Pathog 6(6):e1000951. 41. Knight JS, Sharma N, Robertson ES (2005) Epstein-Barr virus latent antigen 3C can 18. Zhao B, Marshall DR, Sample CE (1996) A conserved domain of the Epstein-Barr virus mediate the degradation of the retinoblastoma protein through an SCF cellular nuclear antigens 3A and 3C binds to a discrete domain of Jkappa. JVirol70(7):4228–4236. – 19. Robertson ES, et al. (1995) Epstein-Barr virus nuclear protein 3C modulates tran- ubiquitin ligase. Proc Natl Acad Sci USA 102(51):18562 18566. scription through interaction with the sequence-specific DNA-binding protein J 42. Paschos K, et al. (2009) Epstein-Barr virus latency in B cells leads to epigenetic repression kappa. J Virol 69(5):3108–3116. and CpG methylation of the tumour suppressor gene Bim. PLoS Pathog 5(6):e1000492. 20. McClellan MJ, et al. (2012) Downregulation of integrin receptor-signaling genes by 43. Kheradmand Kia S, et al. (2009) EZH2-dependent chromatin looping controls INK4a Epstein-Barr virus EBNA 3C via promoter-proximal and -distal binding elements. and INK4b, but not ARF, during human progenitor cell differentiation and cellular J Virol 86(9):5165–5178. senescence. Epigenetics Chromatin 2(1):16. 21. Paschos K, Parker GA, Watanatanasup E, White RE, Allday MJ (2012) BIM promoter 44. Heinz S, et al. (2010) Simple combinations of lineage-determining transcription fac- directly targeted by EBNA3C in polycomb-mediated repression by EBV. Nucleic Acids tors prime cis-regulatory elements required for macrophage and B cell identities. Res 40(15):7233–7246. Mol Cell 38(4):576–589.

426 | www.pnas.org/cgi/doi/10.1073/pnas.1321704111 Jiang et al. Downloaded by guest on September 25, 2021