Enhancer control of miR-155 expression in Epstein-Barr virus infected B cells

The oncogenic microRNA-155 (miR-155) is the most frequently upregulated miRNA in Epstein-Barr virus (EBV)-positive B cell malignancies and is upregulated in other non-viral lymphomas. Both the EBV nuclear antigen 2 (EBNA2), and B cell transcription factor, interferon regulatory factor 4 (IRF4) are known to activate transcription of the host cell gene from which miR-155 is processed (miR-155HG, BIC). EBNA2 also activates IRF4 transcription indicating that EBV may upregulate miR-155 through direct and indirect mechanisms. The mechanism of transcriptional regulation of IRF4 and miR-155HG by EBNA2 however has not been defined. We demonstrate that EBNA2 can activate IRF4 and miR-155HG expression through specific upstream enhancers that are dependent on the Notch signaling transcription factor RBPJ, a known binding partner of EBNA2. We demonstrate that in addition to activation of the miR-155HG promoter, IRF4 can also activate miR-155HG via the upstream enhancer also targeted by EBNA2. Gene editing to remove the EBNA2- and IRF4-responsive miR-155HG enhancer located 60 kb upstream of miR-155HG led to reduced miR155HG expression in EBV-infected cells. Our data therefore demonstrate that specific RBPJ-dependent enhancers regulate the IRF4-miR-155 expression network and play a key role in the maintenance of miR-155 expression in EBV-infected B cells. These findings provide important insights that will improve our understanding of miR-155 control in B cell malignancies. IMPORTANCE MicroRNA-155 (miR-155) is expressed at high level in many human cancers particularly lymphomas. Epstein-Barr virus (EBV) infects human B cells and drives the development of numerous lymphomas. Two EBV-encoded genes (LMP1 and EBNA2) upregulate miR-155 expression and miR-155 expression is required for the growth of EBV-infected B cells. We show that the EBV transcription factor EBNA2 upregulates miR-155 expression by activating an enhancer upstream from the miR-155 host gene (miR-155HG) from which miR-155 is derived. We show that EBNA2 also indirectly activates miR-155 expression through enhancer-mediated activation of IRF4. IRF4 then activates both the miR-155HG promoter and the upstream enhancer, independently of EBNA2. Gene editing to remove the miR-155HG enhancer leads to a reduction in miR-155HG expression. We therefore identify enhancer-mediated activation of miR-155HG as a critical step in promoting B cell growth and a likely driver of lymphoma development.

indirect mechanisms. The mechanism of transcriptional regulation of IRF4 and miR-23 155HG by EBNA2 however has not been defined. We demonstrate that EBNA2 can 24 activate IRF4 and miR-155HG expression through specific upstream enhancers that are 25 dependent on the Notch signaling transcription factor RBPJ, a known binding partner of 26 EBNA2. We demonstrate that in addition to activation of the miR-155HG promoter, IRF4 27 can also activate miR-155HG via the upstream enhancer also targeted by EBNA2. Gene 28 editing to remove the EBNA2-and IRF4-responsive miR-155HG enhancer located 60 kb 29 upstream of miR-155HG led to reduced miR155HG expression in EBV-infected cells. Our 30 data therefore demonstrate that specific RBPJ-dependent enhancers regulate the IRF4-31 miR-155 expression network and play a key role in the maintenance of miR-155 expression 32 in EBV-infected B cells. These findings provide important insights that will improve our 33 understanding of miR-155 control in B cell malignancies. MicroRNAs (miRNAs) are a class of highly conserved, non-coding RNA molecules of 18-51 25 nucleotides in length that play an important role in post-transcriptional gene control. 52 MiRNAs hybridize to target mRNAs, often in the 3' untranslated region, and promote their 53 degradation and/or inhibit their translation. MiRNAs can be transcribed from specific 54 promoters or processed from coding or non-coding gene transcripts. Deregulation of 55 miRNA expression is implicated in the pathogenesis of many diseases, including a diverse 56 range of human cancers and the term oncomiR is used to describe miRNAs with tumor-57 promoting properties (1). 58

59
The miR-155 oncomiR was originally discovered as a non-coding RNA within the B cell 60 integration cluster (BIC) gene (2). Bic was previously identified as a proto-oncogene 61 activated by proviral insertion in avian leucosis virus-induced lymphomas (3,4). The miR-62 155 locus is highly conserved across species and in humans lies within the third exon of 63 BIC (miR-155 host gene; miR-155HG). MiR-155 appears to play a key role in the 64 regulation of B lymphocyte function. Transcription of miR-155HG is activated upon B cell 65 receptor signaling and in murine models dysfunction or loss of miR-155 in B lymphocytes 66 causes a severe decrease in antibody-induced signaling (5, 6). Overexpression of miR-155 67 in mice results in the development of precursor B lymphoproliferative disorders and B cell 68 lymphomas (7). MiR-155 expression is highly upregulated in a number of human 69 lymphomas including Hodgkin's and diffuse large cell B-cell lymphoma (5,8,9). The basis 70 of the oncogenic activity of miR-155 has not been fully elucidated however a number of 71 target genes that regulate B cell proliferation and survival have been identified. These 72 include transcription regulators, receptors and signaling pathway components e.g. HDAC4,73 PIK3R1,SMAD5,SHIP1,PU.1,BCL2 and C/EBPβ (10,11 demonstrated that EBNA2 had no effect on the miR-155HG promoter but activated 168 transcription up to 7.3-fold when a region encompassing both E1 and E2 was inserted 169 upstream of the promoter (Figure 2A). The level of activation was similar to that observed 170 for the EBNA2 responsive EBV C promoter ( Figure 2B). When testing each enhancer 171 separately, we found that the presence of E1 alone did not convey EBNA2 responsiveness, 172 but it increased basal transcription levels compared to the promoter alone by approximately 173 2-fold ( Figure 2A). This indicates that E1 has EBNA2-independent enhancer function. 174 EBNA2 activated transcription via E2 alone up to 10.8-fold indicating that E2 is an 175 EBNA2-responsive enhancer ( Figure 2A). Interestingly, the presence of E2 decreased 176 basal transcription levels approximately 5-fold compared to the promoter alone ( Figure  177 2A). This is consistent with the presence of repressive elements in the enhancer that can 178 limit basal transcription activity, a feature we observed previously for the main EBNA2-179 responsive enhancer at RUNX3 (24). As a result, the overall level of transcription in the 180 presence of E2 was lower than that in the presence of E1 and E2 combined ( Figure 2A). 181 Since EBNA2 upregulates IRF4 and IRF4 is a known activator of miR-155HG, we 182 investigated whether the effects of EBNA2 in these reporter assays may be indirect and the 183 result of increased endogenous IRF4 expression. We found that transient expression of 184 EBNA2 did not increase endogenous IRF4 expression ( Figure 2A). We conclude that 185 EBNA2 independent and EBNA2-dependent enhancers regulate miR-155HG transcription 186 in EBV-infected cells and that EBNA2 activates transcription directly via association with 187 a specific miR-155HG enhancer. 188 189 EBNA2 binds to many target gene enhancers through the cell transcription factor RBPJ 190 (CBF1)(22). We investigated whether EBNA2 activation of miR-155HG E2 was mediated 191 via RBPJ. ChIP-QPCR analysis of RBPJ binding in the GM12878 LCL detected RBPJ 192 binding at E2 and not E1, consistent with a role for RBPJ in EBNA2 activation of E2 193 ( Figure 2C). To confirm this, we carried out reporter assays in a DG75 RBPJ knock-out 194 cell line (33). This cell line was derived from a different parental DG75 cell line that also 195 lacks IRF4 expression, so for comparison we also carried out reporter assays in the parental 196 DG75 wild type cell line ( Figure 2D). Our data demonstrated that EBNA2 activated 197 transcription of the miR-155HG E1 and E2 containing reporter construct in the wild type 198 DG75 cell line to the same extent as the EBV C promoter control, confirming our previous 199 results ( Figure 2D). However in DG75 RBPJ knock-out cells, the activation of this reporter 200 construct by EBNA2 was almost completely abolished ( Figure 2D). This mirrored the loss 201 of EBNA2 activation observed for the RBPJ-dependent viral C promoter ( Figure 2E). 202 These data also provide further evidence that EBNA2 activation of miR-155HG E2 is not 203 an indirect effect mediated by IRF4 upregulation and we confirmed that IRF4 expression 204 is not induced by EBNA2 in this cell background ( Figure 2D). 205 206 Interestingly, EBNA2 binding sites often coincide with binding sites for IRF4 or IRF4-207 containing transcription complexes, indicating that IRF4 may be involved in EBNA2 208 binding to DNA (25,34). However, our results indicate that IRF4 is not required for 209 EBNA2 targeting of miR-155HG E2 enhancer element since EBNA2 activation was 210 efficient in the absence of IRF4 ( Figure 1D). We conclude that EBNA2 can directly 211 upregulate miR-155HG transcription through a distal RBPJ-dependent enhancer (E2) 212 independently of IRF4. 213

IRF4 independently activates miR-155HG via promoter and enhancer elements 215
Our data demonstrate that IRF4 is not required for the effects of EBNA2 on miR-155HG 216 transcription. However, IRF4 can independently activate the miR-155HG promoter 217 through an ISRE (29). It is not known whether IRF4 can also activate miR-155HG 218 transcription through enhancer elements. We therefore tested whether exogenous 219 expression of IRF4 in DG75 cells can activate miR-155HG transcription via upstream 220 enhancers. Because IRF4 activates the control plasmid (pRL-TK), Firefly reporter activity 221 was normalized to actin expression as a previously described alternative in these assays 222 (35). Consistent with published data, we found that exogenous expression of IRF4 resulted 223 in a 4-fold increase in miR-155HG promoter activity (29). The presence of E1 did not result 224 in any further increase in miR-155HG transcription by IRF4 ( Figure 3). However, the 225 additional presence of E2 increased the activation of the miR-155HG reporter to 10-fold. 226 These data demonstrate that miR-155HG E2 is IRF4-responsive and contributes to IRF4 227 activation of miR-155HG transcription. 228 229 Taken together our results indicate that miR-155HG promoter activation by IRF4 and the 230 independent effects of IRF4 and EBNA2 on a specific miR-155HG enhancer contribute to 231 the high level expression of miR-155HG and miR-155 in EBV-infected B cells. 232 233

An IRF4 upstream enhancer is activated by EBNA2 through RBPJ 234
Our data support a role for IRF4 as a key regulator of miR-155HG expression in EBV 235 infected cells. IRF4 is also an EBNA2 target gene, but the mechanism of IRF4 upregulation 236 by EBNA2 has not been defined (29, 31). RNA pol II ChiA-PET analysis recently 237 identified a number of upstream regions that interact with IRF4 in the GM12878 LCL (28). 238 These include the transcription unit of DUSP22, an intergenic region upstream from 239 DUSP22 predicted to be a super-enhancer and intergenic regions between IRF4 and 240 DUSP22. The upstream super-enhancer linked to both DUSP22 and IRF4, so likely 241 represents an important regulatory region (28). EBNA2 ChIP-sequencing data that we 242 obtained using EBV infected cells derived from a Burkitt's lymphoma cell line additionally 243 identified two large EBNA2 binding peaks within the region 35 kb directly upstream of 244 IRF4 ( Figure 4A). We investigated the potential role of these regions in EBNA2 activation 245 of IRF4. These putative proximal and distal EBNA2-bound enhancer regions are referred 246 to as IRF4 enhancer 1 (E1) and IRF4 enhancer 2 (E2), respectively ( Figure 4A). Luciferase 247 reporter assays carried out in the two different DG75 cell line clones in the absence or 248 presence of transient EBNA2 expression demonstrated that EBNA2 had a small activating 249 effect on the IRF4 promoter ( Figure 4B and D). The presence of IRF4 E1 reduced basal 250 transcription by 2-fold and increased EBNA2 activation to up to 6.6-fold similar to the 251 level of EBNA2 activation observed for the EBV C promoter ( Figure 4B). The additional 252 inclusion of IRF4 E2 alongside IRF4 E1 had little further effect on EBNA2 activation 253 ( Figure 4B). These data indicate that IRF4 E1 acts as an EBNA2-responsive enhancer. 254 Consistent with EBNA2 activation through RBPJ, ChIP-QPCR detected RBPJ binding at 255 IRF4 E1 and not E2 ( Figure 4C). Accordingly, EBNA2 activation of the IRF4 enhancer 256 construct was decreased from 5.7-fold to 1.9 fold in RBPJ knock out cells. Our data 257 therefore demonstrate that EBNA2 can activate IRF4 transcription through an RBPJ-258 dependent enhancer (E1) located 13 kb upstream from the transcription start site (TSS). 259 260

Deletion of miR-155HG E2 from the B cell genome reduces mIR-155HG expression 261
Since EBNA2 and IRF4 can activate transcription through miR-155HG E2 in reporter 262 assays, we next tested the role of this enhancer in the regulation of mIR-155HG in EBV 263 infected B cells. To do this, we used CRISPR/Cas9 gene editing to remove the region 264 encompassing E2 ( Figure 5A) from the genome of the EBV immortalized LCL IB4. We 265 designed two small guide RNAs (sgRNAs), one targeting a region 5' to the enhancer and 266 one targeting a region 3' to the enhancer, so that DNA repair following Cas9 cleavage 267 would generate an E2 deletion ( Figure 5A). Both sgRNAs comprised 20 nucleotide 268 sequences that target the genomic region adjacent to a protospacer adjacent motif (PAM) 269 required for Cas9 cleavage ( Figure 5C). sgRNAs were transfected into IB4 cells alongside 270 Cas9 protein and single cell clones were generated by limiting dilution. PCR screening was 271 used to identify cell line clones containing E2 deletions using a forward primer located 5' 272 of the E2 region and a reverse primer located 3' of E2 to amplify a 180 bp DNA product 273 across the deletion site ( Figure 5A  We next used real-time PCR analysis to determine whether deletion of miR-155HG E2 287 affected the levels of endogenous miR-155HG RNA in IB4 cells. We found that all three 288 deletion mutant cell line clones had reduced levels of miR-155HG transcripts compared to 289 parental IB4 cells or the non-deleted C4A cell line ( Figure 5D). miR-155HG RNA 290 expression was reduced by 47%, 63% and 78% in cell line clones C4D, C2B and C5B, 291 respectively ( Figure 5D). This indicates that the RBPJ-dependent EBNA2 responsive 292 enhancer (E2) located 60 kb upstream of miR-155HG plays an important role in 293 maintaining miR-155HG expression in EBV infected cells. Given that miR-155 is derived 294 by processing of the miR-155HG transcript, our data indicate that this enhancer would be 295 important in controlling miR-155 expression. 296 297 In summary we have identified and characterized new enhancer elements that play a key 298 role in the direct and indirect upregulation of miR-155 expression in EBV infected cells by 299 the EBV transcription factor EBNA2 ( Figure 6). Importantly, we show that an EBNA2 and 300 IRF4 responsive enhancer element located 60 kb upstream from the miR-155HG TSS is 301 essential to maintain high level miR-155HG RNA expression. 302

DISCUSSION 304
We have characterized an enhancer 60 kb upstream of the miR-155-encoding gene miR-305 155HG that is bound by EBNA2, the key transcriptional regulator encoded by Epstein-Barr 306 virus. We have shown that the presence of this enhancer in the B cell genome is required 307 to maintain high level miR-155HG expression in an EBV-infected B cell line, indicating 308 that enhancer control is critical for miR-155 upregulation by the virus. This enhancer 309 (enhancer 2) was responsive to EBNA2 in reporter assays and EBNA2 activation was 310 dependent on the expression of host cell protein RBPJ. Since EBNA2 cannot bind DNA 311 directly, this is in line with EBNA2 binding via its interaction with RBPJ (36, 37). MiR-312 155HG enhancer 2 also contains binding sites for a number of other B cell transcription 313 factors (e.g. SPI1 (PU.1), RUNX3, NF-B rel A, BATF and SRF) that likely play a role in 314 regulating its activity in uninfected B cells. It is also possible that some of these 315 transcription factors may help to stabilize EBNA2 or EBNA2-RBPJ binding in the context 316 of B cell chromatin, a scenario that we cannot examine in reporter assays. PU.1 for example 317 has been shown to bind EBNA2 (38). However, in reporter assays loss of RBPJ alone 318 severely diminishes EBNA2 responsiveness indicating that RBPJ is the major mediator of 319 EBNA2 activation of miR-155HG enhancer 2. 320 321 MiR-155HG enhancer 2 is located within a region upstream of miR-155HG that is detected 322 by CHi-C and RNA pol II ChIA-PET to associate with the miR-155HG promoter. 323 Although, another putative enhancer bound by EBNA2 in the GM12878 LCL (enhancer 1) 324 is also present in this region, we found that enhancer 1 was not EBNA2 responsive but did 325 upregulate transcription from the miR-155HG promoter in reporter assays. This indicates 326 that this region possesses EBNA2-independent enhancer function. The detected EBNA2 327 binding at enhancer 1 may therefore be the consequence of looping between enhancer 1 328 and enhancer 2 that would lead to the precipitation of this region of DNA in EBNA2 ChIP 329 experiments. Interestingly, binding at enhancer 1 is not detected by EBNA2 ChIP-seq in a 330 BL cell background (23), so its activity and looping interactions may be cell-type 331 dependent. Two further upstream regions also interact with the miR-155HG promoter by 332 CHi-C and RNA pol II ChIA-PET in LCLs (one intergenic and one proximal to the 333 LINC00158 promoter). This is consistent with the presence of an active enhancer-promoter 334 hub formed between two intergenic enhancer regions (one of which encompasses enhancer 335 2) and the promoter-proximal regions of miR-155HG and LINC00158. In two EBV infected 336 LCL backgrounds (GM12878 and IB4), maximal EBNA2 (and RBPJ) binding at the miR-337 155HG locus is detected in the intergenic interacting region encompassing miR-155HG 338 enhancer 2 (22,24,28). This is despite the classification of the remaining intergenic region 339 and the LINC00158 promoter proximal region as EBV super-enhancers based on their 340 chromatin and TF landscape profiles (28). It is therefore possible that EBNA2 accesses the 341 miR-155HG enhancer hub and upregulates miR-155 expression through its RBPJ-342 dependent association with miR-155HG enhancer 2. Our observations highlight the 343 importance of testing the EBNA2 responsiveness of EBNA2-bound regions rather than 344 relying on binding profiles alone to assign EBNA2 enhancer function. 345

346
The constitutively active EBV membrane protein LMP1 also activates miR-155HG 347 transcription. NF-B and AP-1 sites in the miR-155HG promoter have been shown to be 348 important to maintain miR-155HG promoter activity in LCLs and two NF-B sites and the 349 AP-1 site mediate LMP1 responsiveness in transiently transfected EBV negative cells (18, 350 19). NF-B RelA also binds to the miR-155HG enhancer 2 region and the putative 351 upstream super-enhancer, so it is also possible that LMP1 activation of the NF-B and AP-352 1 pathways also activates miR-155HG enhancers. Thus promoter (and possibly enhancer) 353 activation by LMP1 and enhancer activation by EBNA2 may all contribute to the high-354 level miR-155 expression observed in EBV-infected cells. 355

356
Our results also revealed that the B cell transcription factor IRF4 can also activate miR-357 155HG transcription via enhancer 2 in addition to its known effects on the miR-155HG 358 promoter. IRF4 activates the miR-155HG promoter via an ISRE. There are no ISREs within 359 miR-155HG enhancer 2, but the 5' sequence of the PU.1 binding site partially matches a 360 reverse ETS-IRF composite element (EICE), so IRF4 could bind in combination with PU.1. 361 In addition to direct control of miR-155 expression through the miR-155HG enhancer, 362 EBNA2 also indirectly influences miR-155 expression through the transcriptional 363 upregulation of IRF4. We demonstrate that again enhancer control by EBNA2 plays an 364 important role in IRF4 activation. In addition to the presence of an EBNA2-bound super-365 enhancer upstream of the neighboring DUSP22 gene (28), we found that EBNA2 can also 366 upregulate IRF4 transcription through an RBPJ dependent enhancer located in an 367 intergenic region 35 kb upstream from IRF4. At IRF4 and DUSP22, EBNA2 therefore 368 likely targets multiple enhancers and super-enhancers. 369 370 MiR-155 is overexpressed in many tumor contexts, including hematological malignancies 371 and is implicated in cancer therapy resistance (11). It therefore represents an important 372 therapeutic target. The first in human phase I trial of a synthetic locked nucleic acid anti-373 miR to miR-155 has been initiated and preliminary results show that the inhibitor is well 374 tolerated in patients with cutaneous T cell lymphoma when injected intratumorally (39). The miR-155HG promoter sequence from -616 to +515 (Human GRCh37/hg19 chr 21 413 26933842-26934972) was synthesized by GeneArt Strings® (Invitrogen) to include XhoI 414 and HindIII restriction enzyme sites and cloned into pGL3 basic (Promega) to generate the 415 pGL3 miR155HG promoter construct. The pGL3miR-155HG enhancer 1 (E1) construct 416 was generated in a similar way by synthesis of the promoter and upstream E1 region (chr21 417 26884583-26885197) as a single DNA fragment that was then cloned into pGL3 basic. To 418 generate the miR-155HG promoter E1 + E2 construct, the promoter and E1 and E2 regions 419 (chr21 26873921-26875152) were synthesized as a single DNA fragment and cloned into 420 pGL3 basic. The pGL3-miR-155HG promoter E2 construct was generated using sequence 421 and ligation independent cloning. The E2 region was amplified by PCR from the miR-422 155HG promoter E1 + E2 construct using primers containing vector and insert sequences 423 The IRF4 promoter sequence from -739 to +359 (Human GRCh37/hg19 chr6 391024-433 392121) was synthesized by GeneArt (Invitrogen) and the promoter fragment was 434 amplified from the supplied vector (pMK-RQ) using primers to introduce XhoI restriction 435 sites at each end (forward 5' GTCTCGAGATTACAGGCTTGAGCCACA 3', reverse 436 5'GACTCGAGCTGGACTCGGAGCTGAGG 3'). The promoter was then cloned into the 437 XhoI site of pGL3 basic (Promega) to generate the pGL3 IRF4 promoter construct. Pulser II) using 0.4cm cuvettes and luciferase assays carried out as described previously 452 with some modifications (48). Briefly, DG75 cells were diluted 1:2 into fresh medium 24 453 hours prior to electroporation. For transfection, cells were pelleted and conditioned media 454 reserved for later use. Cells were then resuspended in serum-free media to a density of 455 2x10 7 cells/ml. 500 µl of cell suspension was pre-mixed with DNA and then added to the 456 cuvette and immediately electroporated. Transfected cells were then transferred to 10 ml 457 of pre-warmed conditioned media, and cultured for 48 hours in a humidified incubator at 458 37°C, with 5% CO2. 459 Cells were transfected with 2µg of the pGL3 luciferase reporter plasmids and 0.5µg pRL-460 TK (Promega) as a transfection control where indicated. Transfection reactions also 461 included 10 or 20 µg of the EBNA2 expressing plasmid (pSG5 EBNA2), 5 or 10 µg of 462 IRF4 expressing plasmid (pCMV6XL5-IRF4, Cambridge Biosciences) or empty vector 463 control. One tenth of each transfection was processed for Western blotting to analyse 464 EBNA2, IRF4 and actin protein expression levels. The remaining cells were lysed and 465 firefly and Renilla luciferase activity measured using the dual luciferase assay (Promega) 466 and a Glowmax multi detection system (Promega). For transfections where IRF4 was 467 expressed, firefly luciferase signals were normalized to actin expression. ChIP-QPCR for RBPJ was carried out as described previously (24). MiR-155HG locus 504 primers were located in the miR-155HG promoter (forward 5' 505 AGCTGTAGGTTCCAAGAACAGG 3' and reverse 5' 506 GACTCATAACCGACCAGGCG 3', miR-155HG enhancer 1 (forward 5' 507 ACCTGTTGACTTGCCTAGAGAC 3' and reverse 5' TTCTGGTCTGTCTTCGCCAT 508 3'), a 'trough' region between miR-155HG enhancer 1 and enhancer 2 (forward 5' 509 TATTCAGCTATTCCAGGAGGCAG 510 3' and reverse 5' GTGACATTATCTGCACAGCGAG 3'), and miR-155HG enhancer 2 511 (forward 5' CCTAGTCTCTCTTCTCCATGAGC 3' and reverse 5' 512 AGTTGATTCCTGTGGACCATGA 3'). IRF4 locus primers were located in the IRF4 513 promoter (forward 5' TCCGTTACACGCTCTGCAA 3' and reverse 5' 514 CCTCAGGAGGCCAGTCAATC 3'), a 'trough' region between the IRF4 promoter and 515 enhancer 1 (forward 5' TGTGACAAGTGACGGTATGCT 3' and reverse 5' 516 TTGTAACAGCGCCTAATGTTGG 3'), IRF4 enhancer 1 (forward 5' 517 TTACCACCTGGGTACCTGTCT 3' and reverse 5' ACAGTAGCATGCAGCACTCTC 518 3') and IRF4 enhancer 2 (forward 5' AGTGAGACGTGTGTCAGAGG 3' and reverse 5' 519 AAGCAGGCACTGTGATTCCA 3'). 520

RT-QPCR 522
Total RNA was extracted using TriReagent (Sigma) and RNA samples then purified using 523 the RNeasy kit (Qiagen). RNA concentrations were determined using a Nanodrop 2000 524 (Thermo Scientific) and 1 μg was used to prepare cDNA using the ImProm II reverse 525 transcription kit with random primers (Promega). Quantitative PCR was performed in 526 duplicate using the standard curve absolute quantification method on an Applied 527 Biosystems 7500 real-time PCR machine as described previously (23)