DNA replication stress-induced transcriptome of Human Burkitt’s lymphoma identifies MBD1 as a novel suppressor of BCL6 rearrangements in germinal center derived B-lymphomagenesis

BCL6 is a master transcriptional regulator of germinal center (GC) B cells. BCL6 is 31 frequently translocated at the major translocation cluster (MTC) within intron 1 of the BCL6 locus, 32 a hotspot commonly rearranged in diffuse large B cell lymphomas (DLBCLs). BCL6 33 amplifications are associated with therap eutic resistance and poor survival outcomes in 34 hematological and solid cancers. However the mechanisms suppressing genome instability at the 35 BCL6-MTC preventing BCL6 rearragements remain unclear. Here, transcriptome analysis and 36 genome-wide mapping of histone H3 lysine 4 trimethylation (H3K4me3) in hydroxyurea (HU) -37 treated Raji cells (a Burkitt’s lymphoma model) revealed the induced expression of MBD1, 38 encoding the DNA CpG methylation-binding protein. Functional studies using shRNA silencing 39 and ectopic overexpression demonstrated that MBD1 suppresses BCL6 transcription whose 40 promoter harbours conserved CpG methylation sites, suggesting a DNA methylation-dependent 41 regulation of BCL6 trasncription by MBD1 . Conversely, BCL6 repressed MBD1 expression by 42 binding to its promoter. MBD1-depleted Raji cells exhibited increased genomic instability at the 43 BCL6-MTC upon HU treatment, heightened sensitivity to DNA replication inhibitors (HU, 44 gemcitabine, and etoposide), and reduced tumorigenicity in xenograft mouse models. We propose 45 that MBD1 prevents genomic instability at the BCL6-MTC to suppress DLBCL formation . 46 Moreover, MBD1 promotes genomic stability and cell viability during DNA replication stress. 47 MBD1 thus represents a potential therapeutic target for cancers exhibiting resistance to 48 chemotherapies targeting DNA replication. 49 50

51 Dark zones (DZs) and Light zones (LZs) are two main anatomical compartments (Phan, 2005 #36) 52 within GCs [1]. DZs accommodate GC B cells undergoing activation induced cytosine deaminase 53 (AID)-induced somatic hypermutation (SHM) [2], and LZs are proposed anatomical sites favoring 54 AID induced class switch recombination (CSR) [3, 4]. GC B cells remain at high risk of genetic 55 rearrangement in both DZs and LZs because dysregulated AID activity during SHM and CSR can 56 initiate oncogenic chromosomal translocations including those fusing the Immunoglobulin locus 57 (IG) with trans loci such as MYC and BCL6 [5, 6]. GC B cells with these translocations develop 58 into Burkitt’s lymphoma (IG-MYC) and DLBCL s (IG-BCL6) [5]. The hijacked regulation of 59 differentiation pathways toward plasma and memory B cells in Germinal center derived B -60 lymphomas (GCDBL) can lead to Multiple Myeloma (MM) and B cell chronic lymphocytic 61 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 3 leukemia (B -CLL), respectively [7]. Additionally, DLBCL s overexpressing CXCR4, which 62 encodes G -protein coupled receptors (GPCRs), can develop into Waldenström 63 macroglobulinemia, another DLBCL subtype marked by BCL6 and CXCR4 co-amplification [1, 64 8-11]. Thus, controlling genetic rearrangements in GC B cells, along with regulated differentiation 65 and affinity maturation, is important for GCDBL elimination. 66 67 The BCL6 is a master transcriptional regulator of GC B cells belonging to the zinc finger and BTB 68 (ZBTB) family and is comprised of an N-terminal BTB/POZ (Broad-complex, Tramtrack, Bric-a-69 brac/Poxvirus and zinc fingers) domain and Krüppel-type zinc fingers at the C terminus [12, 13]. 70 The zinc finger domain of BCL6 is required for DNA binding to recruit the 71 SMRT/mSIN3A/histone deacetylase for transcriptional suppression of genes in cell cycle arrest, 72 DNA damage repair and p53 response [12, 13] . Thus, BCL6 drives the transcription program 73 promoting GC B cell activation and survival by inhibiting the DNA damage response, apoptosis 74 in GC B cells undergoing the SHM and CSR [14, 15]. 75 76 BCL6 overexpression is frequently observed in DLBCL [5, 16] . The breakpoint analysis of 77 hematological malignancies suggests that 31% of DLBCLs harbor BCL6 translocations on intron 78 1, a 10502 base pair sequence also known as major translocation cluster (MTC) [5, 16]. The start 79 codon of BCL6 is located on exon 3, thus translocations involving intron 1 encodes for full length 80 BCL6 when juxtaposed with an active locus in trans. 52% of BCL6 translocations involve the 81 Immunoglobulin heavy chain (IGH) locus, 10% with immunoglobulin light chain (IGL), and 38% 82 with non-IG loci [5, 16, 17], suggesting random nature of BCL6-translocation to sites other than 83 IG locus. BCL6 also acts as a proto -oncogene in the pathogenesis of breakpoint cluster region 84 (BCLR)–v-abl Abelson murine leukemia viral oncogene homolog 1(ABL1)-driven acute 85 lymphoblastic leukemia (ALL). BCL6 translocations have been also observed in glioblastoma and 86 elevated BCL6 activity—along with its corepressor NcoR—is associated with AXL signaling and 87 therapy resistance in glioblastoma [18]. BCL6 overexpression enables survival of Philadelphia 88 chromosome-positive ALL (Ph + ALL) cells upon BC LR–ABL1–kinase inhibition through 89 repression of p53 expression [19]. BCL6 also drives therapy escape in non-small cell lung cancers, 90 breast cancer, colorectal and gall bladder cancer [20, 21]. 91 92 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 4 GC B cell differentiation is driven by dynamic changes in chromatin state including the DNA 93 methylation and EZH2 driven histone H3K27me3 of the genes regulating the plasma B cell 94 differentiation such as IRF4 and PRDM1 [22, 23]. The DNA CpG methylation binding protein 1 95 (MBD1) exhibits binding to methylated and unmethylated DNA via its CXXC domains ({Fujita, 96 2003 #78 ) and regulates gene expression by recruiting transcriptional suppressors such as 97 SETDB2, HP-1a, SUV39H, histone-deacetylases-3 and PML-RARA {Fujita, 2003 #78;Clouaire, 98 2010 #79[24]. However, it is not yet clear if MBD1 regulates BCL6 transcription through DNA 99 methylation-dependent or -independent manner. During the plasma B cell differentiation, DNA 100 hypomethylation of the promoters of PRDM1, XBP1, IRF8, SPIB genes assists plasma B cell 101 differentiation [22, 25]. DNA methyltransferase activity was reported to be decreased in GC B 102 cells than naïve B cells and Dnmt1 hypomorphic mice had a defective GC reaction [22], suggesting 103 a role for DNA methylation axis in GC B cell activation and differentiation. 104 105 Intron 1 of BCL6 locus harbors four major CpG rich domains [16] [5], suggesting a role of DNA 106 methylation in BCL6 transcription. CTCF, an insulator which binds to unmethylated DNA, also 107 binds on intron 1 and suppresses BCL6 transcription, suggesting this transcriptional suppression is 108 methylation-independent [26]. CpG methylation is catalyzed by the ten eleven translocation (TET) 109 enzymes and is recognized by methylation bi nding proteins (MBDs), suggest ing a complex 110 regulation involving DNA demethylation, TET enzymes, MBD and CTCF binding in the 111 regulation of BCL6 transcription [4, 27 -30]. The mechanistic link between DNA methylation-112 dependent suppression of BCL6 transcription and whether MBD proteins are involved in BCL6 113 transcription and remain unknown. 114 115 In the current study, we have characterized the roles of MBD1 in regulation of key genes involved 116 in GC B cell fate determination utilizing a recently described approach for identification of new 117 GC regulators. Functional characterization of MBD1 revealed its roles in suppression of BCL6 118 under DNA replication stress. Moreover, MBD1 knockdown resulted in increased DNA damage 119 on the BCL6-MTC region while MBD1-depelted Raji cells exhibited higher sensitivity to DNA 120 replication inhibitors. These place MBD1 as a novel regulator of BCL6 translocation and 121 expression of BCL6. This role of MBD1 in GC B cells undergoing dynamic chromatin 122 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 5 modification and differentiation could be essential for allowing the differentiation of GC B cells 123 assuring the healthy immune response and suppression of the GCDBLs. 124 125

126 Inverse correlation between MBD1 and BCL6 expression in GC B and cancers 127 We recently defined a novel approach using Raji cells to define the molecular mechanisms 128 regulating the fate of GC B cells and GCDBLs [31]. Briefly, we exposed Raji cells to 4mM HU 129 for 12 hours to induce the genotoxic-stress as experienced by GC B cells experience during rapid 130 proliferation coupled with SHM and CSR [31]. We hypothesized that factors induced during this 131 treatment could be important regulators on GC B cells. Based on this screening, we centered our 132 investigation on MBD1 due to its significant upregulation as one the the top 50 genes identified in 133 our study (Figure 1A). We first confirmed whether MBD1 expression is induced in cancer cells 134 treated with DNA replication inhibitors and the DNA damage inducers Hydroxyurea (HU), 135 Gemcitabine, and Etoposide (Supplementary Figures 1A, B). We performed short hairpin RNA 136 (shRNA) mediated knockdown of MBD1 in Raji cells (shMBD1 -Raji) and in breast cancer cells 137 (shMBD1-SUM149PT), and control cells with scrambled shRNA (shSCR -Raji and shSCR -138 SUM149PT) (Supplementary Figures 1A, B). We confimed that MBD1 expression was 139 significantly induced in scramble-knockdown cells, shSCR -Raji cells treated with HU, 140 Gemcitabine or Etoposide (Supplementary Figures 1A). While we confirmed that shMBD1 cells 141 reduced MBD1 expression by ~75% compared to shSCR-Raji cells (Supplementary Figure 1A), 142 shMBD1-Raji cells treated with these agents also exhibited an trend of induced MBD1 expression 143 (Supplementary Figure 1A). Similarly, breast cancer origin cells, shSCR-SUM149PT also induced 144 MBD1 expression with HU -treatment (Supplementary Figure 2B ). These results suggest that 145 MBD1 expression is induced in DNA replication stress-dependent manner. In addit ion, we 146 observed histone H3K4me3 enrichment within -2 kilobase pairs of its TSS in HU -treated cells 147 (Figure 1B), suggesting MBD1 expression is partially dependent on histone H3K4me3 under HU 148 stress (Figure 1B). 149 To examine the expression of MBD1 in GC B cells, we analyzed the real-time transcriptome data 150 from activated human tonsil GC B cells (AGCBs) (Figure 1 C) [32]. We simultaneously plotted 151 the log2 fold change (Log2FC) values of MBD1 transcripts against BCL6 transcripts across 152 subpopulations, including DZa, DZb, DZc, and intermediate (INT) populations (INTa, INTb, 153 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 6 INTc, INTd, INTe), as well as LZa, LZb, LZc, pre -memory (preM), and plasmablasts A and B 154 (Figure 1C). These subpopulations were clustered based on genes specifically enriched in each 155 compartment [32]. Log2FC values for MBD1 were relatively low in DZa, DZb, INTa, INTb, INTc, 156 INTd, INTe, LZa, and LZb, with values ranging from -0.34 to 0.163 (Figure 1C). However, log2FC 157 values of MBD1 were higher in pre-memory (0.625) and plasmablast a cells (1.320) (Figure 1C). 158 In contrast, BCL6 expression sharply declined in pre-memory and plasmablast stages (Figure 1C), 159 suggesting a reciprocal relationship between MBD1 and BCL6 expression in GC B cells. 160 Given the reciprocal expression of BCL6 and MBD1 in human AGCBs, we checked the possibility 161 of mututal suppression of MBD1 and BCL6. Among the tumor samples avai lable in the cancer 162 genome atlas (TCGA), we first categorized BCL6 and MBD1 altered tumors based on their genetic 163 status and quantified their mRNA level in each group (Supplementary Figure 1B, C). We found 164 that MBD1 mRNA levels positively correlate with genetic alteration (Supplementary Figure 1C). 165 We also observed a similar relationhip of BCL6 mRNA expression with its copy number aleration 166 in tumors (Supplementary Figure 1D). 167 We then examined MBD1-mutated tumors for alterations in BCL6 expression. Tumors with 168 MBD1-shallow deletions exhibited higher BCL6 expression than MBD1-diploid tumors (Figure 169 1D). On the other hand, BCL6 mRNA levels were significantly lower in MBD1-gain tumors 170 (Figure 1 D). Similarly, MBD1-amplified tumors exhibited reduced trend of BCL6 mRNA 171 expression while tumors with MBD1-deep-deletion status exhibited higher BCL6 mRNA levels 172 than MBD1-diploid tumors, although this trend was not significant in these groups (Figure 1D). 173 Further, DLBCL samples with MBD1-gain status exhibited lower BCL6 expression than MBD1-174 diploid tumors (Figure 1E). Taken together, these results indicate an inverse relationship between 175 MBD1 and BCL6 expression. To determine if BCL6 levels also inversely correlate with MBD1 176 mRNA levels, we sorted tumors with MBD1-diploid samples and classified them as BCL6-diploid, 177 BCL6-deep-deletion, BCL6-shallow-deletion, BCL6-gain and BCL6-amplification (Figure 1 F). 178 We found that MBD1 mRNA was significantly lower in BCL6-gain and BCL6-amplified cancers, 179 while MBD1-mRNA was higher in BCL6-deleted samples (Figure 1F). These results indiciate that 180 BCL6 levels are inversely corealted with the MBD1 mRNA levels, sugesting the mutual suppresion 181 of BCL6 and MBD1 by each others. 182 183 MBD1 suppresses BCL6 expression during DNA replication stress 184 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 7 Given the abnormal expression of BCL6 in breast cancer cells , we investigated whether MBD1 185 knockdown also impacts BCL6 levels in breast cancer cell lines MCF7, MDA-MB-231, SUM-186 149PT, and MDA -MB-436 (Figure 2 A). Notably, BCL6 expression was induced in shMBD1 -187 MCF7 and shMBD1 -MDA-MB-231 cells compared to control cells (Figure 2 A). These results 188 confirm that MBD1 can suppress BCL6 expression (Figure 2A). However, MBD1 depletion in 189 SUM149PT and MDA-MB-436 cells did not induce the BCL6 suppression, suggesting additional 190 mechanisms may regulate BCL6 transcription in these cells. Overall, these results confirm that 191 MBD1 loss is associated with the induced BCL6 expression not only in B-lymphoma cells but also 192 in breast cancer. 193 194 To further confirm that MBD1 suppresses BCL6 expression in B cells, we transiently transfected 195 Raji cells with pcDNA-MBD1 (Figure 2B, C). Compared to pcDNA3.1 (empty vector : EV) 196 transfected cells, pcDNA-MBD1 cells induced MBD1 expression in Raji cells (Figure 2B). While 197 pcDNA-MBD1-transfected cells did not have significantly reduced BCL6 mRNA, treatment with 198 gemcitabine caused a significant reduction in BCL6 levels in pcDNA-MBD1 transfected cells 199 (Figure 2C). In addition, 24 and 48 hours of gemcitabine treatment yielded significant reduction 200 in BCL6 mRNA since EV-transfected Raji cells exhibited 86% and 71% of BCL6 levels after 24 201 and 48 hours of gemcitabine treatment , respectively, while pcDNA-MBD1-transfected Raji cells 202 showed 71% and 61% of BCL6 levels (Figure 2C). Exposure to HU induced a similar trend: after 203 24 hours of HU treatment, BCL6 levels were 90% in EV-transfected Raji cells and 86% in pcDNA-204 MBD1-transfected cells (Figure 2C). Though not significant, this trend persisted after 48 hours of 205 HU treatment, with pcDNA-MBD1 showing 43% BCL6 mRNA levels compared to 51% in EV-206 transfected cells (Figure 2C). These results further support a role for MBD1 in suppressing BCL6 207 transcription during replication stress. This is distinct from breast cancer cells, where exogenous 208 DNA replication stress was not required for MBD 1-mediated suppression of BCL6 expression 209 (Figure 2A). 210 211 MBD1 recognizes DNA methylation on CpG sequences[24, 33, 34] , we investigated whether 212 MBD1 might bind to methylated DNA sequences on the BCL6 locus and suppress BCL6 213 transcription. Using the ENCODE database, we identified four dispersed CpG methylation sites 214 within intron 1 of the BCL6 locus, where the BCL6-MTC resides (Figure 2D). In addition, by 215 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 8 analyzing ChIP-atlas data and MBD1 ChIP-seq results from HCT116 cells, we confirmed MBD1 216 binding along the BCL6 gene body including intron 1 (Figure 2 E). These results suggest that 217 MBD1 binds to the BCL6 locus and the repression of BCL6 transcription could be influenced by 218 DNA methylation and DNA replication stress. 219 220 MBD1 suppresses DNA break formation on the BCL6-MTC in Raji cells- 221 Mutations increasing MBD1 expression negatively correlate with BCL6 transcription (Figure 1D), 222 however the relationship between MBD1 and BCL6 mutations in tumors is unclear. To determine 223 if MBD1 and BCL6 mutations co-occur, we analyzed their co -occurrence in 27 different tumor 224 types using TCGA (Figure 3A). We found that MBD1 deep-deletion or MBD1-shallow deletion 225 samples (commonly referred as MBD1-deleted hereafter) exhibited a higher percentage of BCL6 226 alterations, suggesting that genetic alterations in MBD1 and BCL6 are positively associated (Figure 227 3A). Further, DLBCL tumors with MBD1-deleted status exhibited a significantly higher frequency 228 of BCL6 alterations, suggesting a strong correlation between MBD1 and BCL6-mutated B 229 lymphomas (Figure 3B). Based on these results, we hypothesized that MBD1 may also influence 230 BCL6 translocation by suppressing DNA breaks on BCL6-MTC preventing its rearrangement. To 231 investigate this hypothesis, we analyzed gH2AX signal by chromatin immunoprecipitation (ChIP) 232 on BCL6-MTC after HU-induced DNA replication stress in shSCR-Raji and shSCR-MBD1 cells 233 (Figure 3C). We analyzed gH2AX abundance at four primer locations G, J, M and P on BCL6 234 intron 1, which spans 10.5 kb (Figure 3C) [26]. The level of gH2AX signal was higher in HU-235 treated shMBD1-Raji cells than HU-treated shSCR-Raji cells (Figure 3D), suggesting that MBD1 236 prevents HU-induced DNA damage at BCL6-MTC (Figure 3D). This role of MBD1 in suppressing 237 DNA damage at BCL6-MTC suggests that it could suppress BCL6 rearrangements in GC B cells 238 and prevent DLBCL formation. Since CTCF binds to intron 1 of BCL6-MTC and suppresses BCL6 239 transcription in MM cells [26], we hypothesized that MBD1 may prevent BCL6 transcription by 240 collaborating with CTCF. We compared CTCF ChIP signal in HU -treated shMBD1-Raji and 241 shSCR-Raji cells on the BCL6-MTC (Figure 3E). Compared to HU-treated shSCR-Raji cells, HU-242 treated shMBD1-Raji cells exhibited decreased CTCF occupancy at BCL6-MTC compared to HU-243 treated shSCR-Raji cells (Figure 3E). This result suggests that MBD1 promotes CTCF recruitment 244 to BCL6-MTC during replication stress (Figure 3E). 245 246 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 9 BCL6 negatively regulates MBD1 transcription 247 BCL6 suppresses plasma B cell differentiation by inhibiting expression of IRF4 and PRDM1 [35]. 248 Given the inverse correlation of BCL6 and MBD1 expression in AGCBs (Figure 1 C, D, F ), we 249 hypothesized that BCL6 may suppress MBD1 expression in GC B cells undergoing plasma B cell 250 differentiation. To test the effect of BCL6 inhibition on MBD1 expression, we treated Raji cells 251 with FX1, a potent BCL6 inhibitor which binds to the BCL6 lateral groove and inhibits the 252 formation of the BCL6 repression complex [36]. We found that FX1 treatment significantly 253 induced MBD1 expression in Raji cells (Figure 4A), suggesting that inhibition of BCL6 binding 254 to its DNA target sequences suppresses MBD1 transcription (Figure 4A). These results suggest 255 that MBD1 and BCL6 can regulate each other’s expression in GB B cells and is suggestive of 256 mutual negative feedback. In addition, shBCL6-Raji cells exhibited a trend of increased MBD1 257 expression compared to shSCR-Raji cells, however this difference was not significant (Figure 4B). 258 We confirmed that BCL6 knockdown was significantly achieved in shBCL6-Raji cells (Figure 259 4B), and IRF4 levels were enhanced in shBCL6 -Raji cells confirming the previous reports on 260 BCL6 of the IRF4 [37]. 261 262 To explore how BCL6 suppresses MBD1 transcription, we performed in-silico binding analysis of 263 BCL6 on the MBD1 promoter (Figure 4C). We observed two BCL6 binding motifs on the MBD1 264 promoter present at -235 and -900 bp upstream of MBD1-TSS (Figure 4 C, D). To determine if 265 BCL6 binding occurs at these sites, we cloned 1 kb of the MBD1 promoter region, and inserted 266 into the luciferase reporter pGL4 vector then performed site -directed mutagenesis on DNA 267 sequences flanking these two binding sites (Figure 4D) . 293T cells transfected with pGL4-pr-268 MBD1 induced luciferase signal nearly 38-fold compared to 293T cells transfected with empty 269 pGL4 (Figure 4D). Interestingly, co-transfection of pGL4-pr-MBD1 and pcDNA-BCL6 reduced 270 the luciferase signal to around 15%, suggesting that BCL6 suppresses MBD1 expression by 271 binding its promoter (Figure 4D). Deletion of either BCL6 binding motif significantly increased 272 luciferase signal, suggesting that BCL6 binding to the predicted binding motifs negatively regulate 273 transcription (Figure 4D). Of note, pcDNA-BCL6 expression had a smaller impact on luciferase 274 signal from the pGL4-pr-MBD1 Δ235-BCL6 or pGL4-pr-MBD1 Δ900-BCL6 reporters than the 275 wild-type promoter sequence (Figure 4D). Particularly, deletion of the site located -900 upstream 276 of MBD1-TSS (Figure 4D) had a greater rescue from BCL6-mediated suppression, suggesting that 277 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 10 BCL6 binding to this motif plays a crucial role in regulating MBD1 expression (Figure 4D). We 278 next examined whether BCL6 binding is present on the endogenous MBD1 promoter. Using BCL6 279 ChIP-seq datasets from ChIP-atlas, we confirmed BCL6 binding on MBD1 promoter in multiple 280 cell lines including SUDHL4 (DLBCL), OCI-LY1 (DLBCL), OCI-LY3 (DLBCL) and in B-cell 281 acute lymphoblastic leukemia (B-ALL) cell lines (Figure 4E). These results suggest that BCL6 282 binds to the MBD1 promoter and suppresses its transcription (Figure 4D, E). 283 284 To investigate whether BCL6 regulation is associated with Mbd1 regulation in mouse GC B cells, 285 we examined the DNA methylation status of the Mbd1 promoter in wild-type and Sca1-Bcl6Δ mice 286 (Supplementary Figure 2A[38]. We noted that the Mbd1 promoter in activated GC B cells of wild 287 type mice showed two distinct DNA methylation peaks, and one peak was lost in Sca1-Bcl6Δ mice 288 (Supplementary Figure 2A). These results suggest that BCL6 suppression of Mbd1 could be 289 associated with DNA methylation on the Mbd1 promoter in mouse GC B cells. To determine if 290 methylation impacts MBD1 expression, we analyzed human B lymphomas treated with 5 -291 Azacytidine (AZA), a DNA methyltransferase inhibitor (Supplementary Figure 2B), and observed 292 that AZA treatment increased transcripts per million (TPM) counts of MBD1 in OCI -LY1, 293 SUDHL2, and OCI -LY19 cells. Together these results suggest that MBD1 expression in GC B 294 cells is regulated by BCL6 and DNA methylation (Supplementary Figure 2B). 295 296 GC B cells are regulated by BCL6 and AID, both of which also act as oncoproteins. Since BCL6 297 suppresses MBD1 (Figure 4A, D), it is possible that MBD1 may function as a tumor suppressor in 298 GC B cells. AID is also involved in B-lymphomagenesis due to its mutagenic roles in GC B cells 299 [10, 39], therefore we explored if AID is associated with MBD1 mutagenesis. We looked for AID 300 signatures in MBD1-mutant cancers in the Catalogue of somatic mutation in cancer cells 301 (COSMIC) (Figure 4F) . Interestingly, AID mutation signatures (C>T) , which are initiated by 302 cysteine deamination to uracil in non-replicating B cells, were frequent (38.9%) along the MBD1 303 gene in MBD1 mutant tumors (Figure 4F). Furthermore, C>G and C>A signatures, which are not 304 as strongly associated with AID activity and could arise due to translesion synthesis and mismatch 305 repair, were 3.4 % and 5.5%, respectively (Figure 4 F). Among non -AID signature mutations in 306 the MBD1 gene, G>A were the most frequent (27.9%) among all samples (Figure 4 F). These 307 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 11

suggest that MBD1 is mutated by AID and underscore MBD1 as a pivotal target for both 308 BCL6 and AID in promoting B lymphomagenesis. 309 310 Increased sensitivity of MBD1-depleted cells to chemotherapeutics targeting DNA 311 replication and reduced tumorigenicity in mouse xenograft model 312 The increased levels of gH2AX signal at the BCL6-MTC in HU-treated shMBD1 cells indicates 313 increased susceptibility of MBD1-depleted cells to chemotherapeutic agents inducing replication 314 stress (Figure 3D). To determine if shMBD1 -Raji cells exhibit higher susceptibility to other 315 chemotherapeutic agents , we treated shSCR -Raji and shMBD1 -Raji cells with gemcitabine or 316 etoposide (Figure 5A). Control ShSCR-Raji cells showed modest sensitivity to HU and 317 Gemcitabine, and greater sensitivity to etoposide (Figure 5A). However, depletion of MBD1 led 318 to significantly greater sensitivity to all drugs tested : the viability in shMBD1 -Raji cells was 319 further reduced to 60 %, 8 1.5% and 40.5% compared to shSCR-Raji cells treated with HU, 320 gemcitabine or etoposide (Figure 5A). These results suggest that MBD1-deficient cells are more 321 sensitive to genotoxic agents than shSCR-Raji and indicate a role of MBD1 in promoting cell 322 viability in response to DNA replication stress/genotoxic stress. 323 324 Finally, to test whether tumorigenicity is altered in MBD1-silenced cells, we performed xenograft 325 experiments by injecting shSCR-Raji and shMBD1-Raji cells into nude mice (Figure 5B-D). Mice 326 injected with shSCR-Raji cells exhibited mean tumor weight of 0.62 gm (Figure 5D). In contrast, 327 the mean weight of shMBD1-Raji tumors was 0.15 gm (Figure 5D). This reduction in tumor weight 328 of shMBD1-Raji xenograft was significantly lower than those in shSCR-Raji cells (Figure 5C, D). 329 This result suggests that MBD1 is required for the tumorgenicity of Raji cells in vivo (Figure 5E, 330 C, D). It is possible that accumulative genotoxic stress may surpass the threshold cells can tolerate 331 in shMBD1-Raji cells therefore perturbing the cellular proliferation , tumorgenicity and tumor 332 growth (Figure 5B,C, D). 333 334 Correlation of MBD1 and IRF4 expression in B-Lymphoma 335 Human AGCBs induce MBD1 expression in GC B cells undergoing differentiation into plasma 336 and memory B cells while BCL6 expression was reduced at similar stages, suggesting that the 337 upregulation of MBD1 in differentiating GC B cells is correlated with BCL6 downregulation 338 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 12 (Figure 1C). We hypothesized that MBD1 could regulate the expression of genes in GC B cells 339 undergoing terminal differentiation. To define the relationship between MBD1 mutation status and 340 IRF4 expression, we explored their correlation in samples from TCGA datasets and the Gene 341 Expression Profiling Interactive Analysis (GEPIA) (Supplementary Figure 3A, B ). IRF4 342 expression was highest in DLBCL samples exhibiting MDB1:gain status, suggesting a positive 343 correlation between MBD1 and IRF4 expression (Supplementary Figure 3A). Moreover, DLBCL 344 tumors from GEPIA also exhibited a positive correlation in MBD1 and IRF4 (Supplementary 345 Figure 3B). Since IRF4 expression is induced in plasma B cells, we examined whether mouse 346 plasma B cells exhibit higher Mbd1 expression than activated GC B cells (Supplementary Figure 347 3C). Interestingly, Mbd1 expression was increased in mouse plasma B cells , when IRF4 is high 348 (Supplementary Figure 3C) [40], suggesting that Mbd1 induction in plasma B cells is coordinated 349 with Irf4 expression. We also measured the effect of MBD1 knockdown on the expression of two 350 key genes determining GC B cell fate, IRF4 and PRDM1 (Supplementary Figure 3D). MBD1 351 depletion in Raji cells resulted in increased expression of IRF4 and PRDM1, indicating a negative 352 regulatory role of MBD1 in their transcription (Supplementary Figure 3D). Conversely, 353 overexpression of MBD1 led to a reduction in IRF4 expression (Supplementary Figure 3E) , 354 suggesting that MBD1 plays a critical role in IRF4 regulation. 355 356 Finally, we analyzed if MBD1 depletion affects the expression of key factors regulating GC B cell 357 activation and survival including essential homing receptors (CXCR4, CXCR5, CCR7, CXCR4, 358 CXCR5, S1PR2, P2RY8), genes encoding the components of migration machinery (RAC1, RHOA), 359 BCL6 targets, (IRF7, IFNGR1, STAT1), or genes encoding factor s regulating BCL6 turnover, 360 (FBXO11, XBP1) (Supplementary Figure 3F). Among the GC B cell receptors tested , shMBD1-361 Raji cells exhibited increased expression of CCR7, P2RY8 and STAT1 but reduced expression of 362 RAC1 compared to shSCR-Raji cells (Supplementary Figure 3F). The expression of BCL6 targets 363 or regulators S1PR2, RHOA, FBXO11, XBP1, IRF7 and IGNGR1 was not alter ed upon in the 364 shMBD1-Raji cells (Supplementary Figure 3 F). Overall, these observations suggest that MBD1 365 can affect the position of B cells within the GC compartment by regulating the expression levels 366 of CCR7, CXCR5, and CXCR4 in cells belonging to T-B border (T-B), light zone (LZ), and dark 367 zone (DZ), respectively. MBD1’s regulation of RAC1 and P2RY8 may also affect the dynamic 368 migration of GC B cells, given a role of RAC1 and P2RY8 in GC B cell LZ-migration [41, 42]. 369 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 13 370

371 Mechanism of MBD1 induction in GC B cells undergoing Plasma B Cell Differentiation – 372 The increased MBD1 expression we observed in HU-treated Raji cells correlates with elevated 373 MBD1 levels in human AGCBs classified as pre -memory and plasma B cells [32] (Figure 1 C). 374 Similarly, mouse GC B cells undergoing PC differentiation also exhibit increased Mbd1 levels 375 (Supplementary Figure 3C). This evidence highlights MBD1’s role in GC B cell differentiation in 376 both humans and mice (Figures 1C, supplementary Figure 3C). The transition from GC B cells to 377 plasma cells necessitates dynamic DNA hypomethylation in genes involved GC B differentiation 378 [25, 43], suggesting that MBD1 expression could be induced by hypomethylation of its promoter. 379 This is supported by our observations of DNA methylation loss in the Mbd1 promoter in Sca1-380 BCL6Δ mice (Supplementary Figure 2A) , and the induction of MBD1 expression in the presence 381 of the methylation inhibitor AZA in DLBCLs (Supplementary Figure 2B) [38]. MBD1 induction 382 during plasma B cell differentiation could also be related to BCL6 downregulation since we 383 provide evidence that BCL6 suppresses MBD1 expression by binding to its promoter (Figures 4A, 384 D). The increased MBD1 levels in plasma B cells might also result from reduced EZH2-dependent 385 silencing of the MBD1 promoter, as this suppresion is removed in GC B cell undergoing 386 differentiation [23]. In contrast, induction of MBD1 in cancer cells undergoing chemotherapy with 387 DNA replication inbibito rs suggest s a pr ogrammed activation of MBD1 transcription which 388 potentially regulates cell survival (Supplementary Figure 1A, B , Figure 5A -D). The detailed 389 mechanims regulating MBD1 -mediated cell survival should be explored in future studies to 390 rationally design effective combination therapy for MBD1-inactivated cancers. MBD1 alteration 391 may also be an effective biomarker for tumor exhibiting higher BCL6 and IRF4 expression which 392 also correlates with increased chemotherapeutic resistance [19, 44]. 393 394 BCL6 translocations and reduced tumorigenicity in MBD1-depleted cancer cells–The BCL6-395 MTC contains highly conserved CpG methylated sites on it intron 1 (Figure 2D) [26], therefore 396 MBD1 may bind to the BCL6-MTC in a DNA methylation-dependent manner (Figure 2D, E). 397 MBD1 binding can help recruit histone deacetylases like SUV39H for transcriptional repression 398 at the BCL6 locus [5, 24, 33, 45]. MBD1-dependent BCL6 suppression may take place in GC B 399 cells undergoing terminal differentiation, correlating with reduced BCL6 levels in plasma B cells 400 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 14 [46]. MBD1 binding to BCL6 promoter is observed in HCT116 cells (ChIP atlas:DRX021118 401 ;https://chip-atlas.org/view?id=DRX021118; Figure 2 E). MBD1 binding may contribute to 402 reduced genomic instability at the BCL6 locus preventing BCL6 rearregements as well as global 403 rearrangements in GC B cells. On the other hand, shMBD1-Raji cells exhibited higher sensitivity 404 to HU, gemcitabine, and etoposide as well as reduced tumorogencity in xenograft mo dels than 405 shSCR-Raji cells (Figure 5A). These results suggest that MBD1 promotes survival in cancer cells, 406 making MBD1 an attractive therpautic target. The reduced tumorigenicity observed in the 407 shMBD1-Raji xenograft could be due to the increased accumulation of genomic instabilit y in 408 shMBD1-Raji cells compared to shSCR-Raji cells (Figure 5C, D). 409 410 AID mutation signatures in MBD1-mutant somatic cancers – AID-mediated mutagenesis 411 drives B lymphomagenesis by mutating tumor suppresor genes and superenhancers, altering the 412 binding of transcription factors and chromatin remodelers [10]. Interestingly, about 39% of single 413 base-pair alterations in MBD1 were C>T transition mutations (Figure 4F), a classic signature of 414 AID-dependent mutagenesis (Figure 4F). Thus AID-dependent MBD1 mutagenesis could serve as 415 an important adapation for B lymphoma selection during the GC reaction. 416 417 MBD1 and IRF4 expression in differentiating GC B cells and B-lymphomas- We found that 418 MBD1 and IRF4 expression is positively correlated in DLBCL (Supplementary Figure 3A,B). On 419 the other hand, Mbd1 and Irf4 were induced in mouse and human GC B cells undergoing plasma 420 B differentiation (Figure 1C, Supplementary Figure 3C), indicating a synchronous upregulation of 421 MBD1 and IRF4 in GC B cells undergoing differe ntiation. The mechanims of MBD1 and IRF4 422 co-expression in GC B cells is not yet clear. It is possible that MBD1 and IRF4 could be induced 423 by decreased BCL6 levels in differetiating GC B cells. Also, MBD1 in differentiating GC B cells 424 could suppreses BCL6, allowing for higher IRF4 expression and plasma B cell differe ntiation. 425 Importantly, IRF4 suppression is mediated by EZH2 -dependent H3K27me3 [23]. Is is po ssible 426 that MBD1 may counteract EZH2 by recruiting histone deacetylases and transcriptional repressors 427 to the EZH2 promoter [5, 24, 33, 45]. This positions MBD1 as a novel regulator of GC B cell 428 differentiation regulating the plasma B cell differe ntiation [47, 48]. Another po ssibility is that 429 MBD1 may work alongside BCL6 and BACH2, both of which negatively regulate IRF4 and 430 PRDM1 [46], al lowing for homestasis of IRF4 expression. This is consistent with our finding 431 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 15 where MBD1 depletion in Raji cells induced IRF4 expression (Supplementary Figure 3 D). 432 However MBD1-mediated suppression of IRF4 in Raji lymphoma cells may be distinct from WT 433 B cells residing in the GC comp artment, therefore we can not rule out the possibility that MBD1 434 promotes plasma B cell differentiation. 435 436 MBD1-mediated regulation of IRF4 and PRDM1 and its significance in the suppression and/or 437 survival of B-lymphomas remains to be elucidated. We propose two potential regulatory scenarios 438 for MBD1 -dependent IRF4 regulation depending on the stage of tumorigenesis . In the first 439 scenario, GC B cells with high -affinity B cell receptors (BCR) might experience higher MBD1 440 expression, facilitating normal GC B cell differentiation. In the second, B lymphoma precursors—441 unproductive GC B cells with damaged BCRs —may experience reduced MBD1 expression 442 blocking the expression of plasma B cell differentiation genes including IRF4 and mitigating the 443 risk of malignant transformation into multiple myeloma (MM). Thus, MBD1 may differentially 444 regulate plasma B cell differentiation in normal plasma B cells versus B lymphoma precursors. 445 446 In summary, our findings suggest that MBD1 is a key regulator of GC B cell differentiation, acting 447 independently of both BCL6 and BACH2. Additionally, MBD1 may prevent BCL6 rearrangement 448 in GC B cells, thereby inhibiting the formation of GCDBL. Finally, considering MBD1’s role in 449 cell viability in cells treated with DNA-replication inhibitors, it could serve as a novel therapeutic 450 target for relapsed and refractory B cell cancers that are resistant to chemotherapy. 451 452 Declarations 453 Ethical Approval – The use of human/animal samples (wherever applicable) was approved as per 454 Institutional ethical board. 455 Competing interests- The authors declare no competing interest financially. 456 Authors' contributions - SKG designed the original hypothesis, performed experiments, and 457 analyzed data. KO analyzed Mbd1 transcripts in mouse plasma B cells and provided constructive 458 feedback. SKG wrote the manuscript. JHB critically read the manuscript and provided constructive 459 feedback. All authors read and agreed to manuscript. 460 461 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 16 Funding- This work is supported grant number 21K16142 from Japan Society for the Promotion 462 sciences (JSPS) to SKG. 463 464 Availability of data and materials- All data needed to evaluate the conclusions in the paper are 465 present in the main text and the supplementary materials. 466 467

468 Human B lymphoma cultures , treatment with inhibitors and plasmid overexpression : 469 Human B lymphoma Cell lines Raji were obtained from Department of Hematology, Kyoto 470 university hospital, RIKEN cell bioresources, Japan. MCF7, MDA-MB-231, SUM149PT, MDA-471 MB-436 were available at the Cancer Research Institute, Kanazawa University. Raji and 472 HEK293T cells were cultured in RPMI medium (30264-85, Nakalai) including the 10% fetal 473 bovine serum ( FBS), 1% Penicillin -Streptomycin & Amphotericin B solution (Waken# 161-474 23181) at 37°C maintaining the 5% CO2 concentration. Breast cancer cell lines MDA -MB-231, 475 MDA-MB-436, MCF-7 cells were cultured DMEM, supplemented with 10% fetal bovine serum 476 (FBS), 2 mM L-glutamine and 1 × Penicillin-Streptomycin (Life Technologies). SUM149PT cell 477 were cultivated in Ham’s F -12 medium containing 5% FBS, 10 μg/ml insulin, 1% Penicillin -478 Streptomycin & Amphotericin B solution (Waken# 161-23181), and supplemented with 0.5 μg/ml 479 hydrocortisone. Cells were propagated at 37 °C in a 5% CO 2 atmosphere. FX1 (S8591, Selleck 480 Japan), Gemcitabine ( S1714, Selleck, Japan), Hydroxyurea ( 085-06653, Wako, Japan) and 481 etoposide (055-0843, Wako, Japan) treatments were performed as described in figure legends. The 482 preparation of these chemical was performed as per the manufacturer’s recommendation. For the 483 overexpression of pcDNA and pcDNA-BCL6 plasmids in Raji cells, the cells were cultured for 2 484 hours in RPMI medium containing 1% FBS, followed by transfection with Lipofectamine 3000 485 according to the manufacturer's instructions (Thermo Fisher, Japan). Twelve hours post -486 transfection, the total FBS concentration was adjusted to 10% by adding additional FBS. Cells 487 were collected for analysis at the indicated time points. 488 489 shRNA preparation and len tivirus induced knockdown- HEK293T cells were culture d at the 490 70% confluency. For transfection mixture and lentivirus preparation, viral vector PAX2, pVSVG 491 and pLKO.1 plasmids (1 µg of each) cloned with respective shRNA sequences were mixed in 200 492 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 17 µl of warm Optimem medium followed by addition of 3 µg of PEI reagent. The mix incubated for 493 20 minutes at room temperature follwed by addition to the H3K293T cells. Fresh medium DMEM 494 medium was replaced to transfected cells after 24 hours. The H3K293T cells were then cultured 495 for additional 48 hours. The culture medium containing the lentivirus particles was then filtered 496 using the 0.45-micron filters and the filtrate was aliquoted and stocked in -80 °C until used. For 497 stable transfection of Raji cells, 1 million cells were transduced with 400 µl of filtered lentiviral 498 supernatant follwed by selection with puromycin (2000 ng/ml) for 2 days . The 2 days selected 499 cells were again washed and selected for additional 2 days with similar dose of Puromycin. 500 501 RNA isolation, cDNA preparation, qPCR analysis and RNA-seq analysis - Total RNA was 502 isolated with NucleoSpin® RNA Plus (#740984.50, Takara) and cDNA was prepared with the RT-503 PCR mix (FSQ#201, Toyobo Japan) . qPCR was performed using the SYBR mix as per 504 manufacturer’s instructions (Thunderbolt QPS-201, Toyobo Japan). Data were analyzed using the 505 delta-delta Ct method indicating relative transcript levels to the respective control samples. ACTB 506 gene signals were used as a housekeeping gene. Computational analyses of RNA -seq were 507 performed as described (Reference; transcriptome paper). The RNA-seq and ChIP data set are 508 under GEO accession numbers GSE242375 and GSE242936 respectively. 509 510 Chromatin Immunoprecipitation (ChIP) , ChIP-seq Library Preparation, Computational 511 Analysis of ChIP -seq Data, and MBD1 Binding to BCL6 locus withChIP-Atlas- ChIP was 512 performed as previously described [6]. The ChIP -seq library was prepared as per the 513 manufacturer’s instruction (Thruplex DNA -seq Kit, R400675, Takarabio, Japan). ChIP-seq was 514 performed as described [31]. The ChIP-seq files were deposited to NCBI under accession number 515 GSE242936. DNA CpG methylation status of mouse Mbd1 promoter in B cells were reanalyzed 516 using the wild type and Sca1-Bcl6Δ mice data [25, 43]. Binding of MDB1 on BCL6 promoter was 517 examined using the dataset # DRX021118 available on ChIP-atlas (https://chip-518 atlas.org/view?id=DRX021118). 519 520 BCL6 binding motif, luciferase assay and COSMIC datasets : The luciferase plasmids 521 depleting the BCL6 binding motifs on the MBD1 promoter were constructed using the Q5 site -522 directed mutagenesis kit (NEB# E0554S) and as per the manufacturer’s instructions. Primers used 523 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 18 for the site -directed mutagenesis are listed in Table 1. The plasmids were transfected in the 524 HEK293T cells for 24 h ours using lipofectamine -3000 (ThermoFisher# L3000001). After 24 525 hours, cell lysates were prepared using Promega luciferase assay kit. The Catalogue of somatic 526 mutation in cancers (COSMIC) datasets were searched for cancers with MBD1 mutation and the 527 mutation signatures were classified and filtered for the AID signatures. 528 529 Tumor xenograft in nude mouse and estimation of cell viability- For tumor xenograft, and 530 injection of Raji into nude mice, 0.6 million cells were suspended in 200 µl of Matrigel and 531 medium mix, suspended well and subcutaneously injected into the right arm of the nude mice. The 532 tumor growth was observed for 55 days and mice were sacrificed, followed by isolation of tumor 533 and weighing. Treatment with chemotherapeutics and cell viability was tested by the cell viability 534 kit (Dojindo Japan) as per the manufacturer’s instructions. 535 536 Supplementary Table 1 for primer sequences- 537 Transcript name Primer sequences for qPCR (5’ to 3’) ACTB-FP GATGATGATATCGCCGCGCT ACTB-RP GAATCCTTCTGACCCATGCC CXCR5-FP AGATTCTCTTCGCCAAAGTC CXCR5-RP (GP48) CACCAGCATGGGCAGCAGGAAT CCR7-FP GTACTCCATCATTTGTTTCGTG CCR7-RP CAGAAGGGAAGGGTCAGGAG S1PR2-FP CAACAGAGGGAGACTCCATCTC S1PR2-RP CTTGTACTCGGAGTACCTGAAC P2RY8-FP CATCATCACCTGCTTCGACGT P2RY8-RP CAACAGCTTGAGGATGGTGGC BCL6-FP CATGCAGAGATGTGCCTCCACA BCL6-RP TCAGAGAAGCGGCAGTCACACT PRDM1-FP CAGTTCCTAAGAACGCCAACAGG PRDM1-RP GTGCTGGATTCACATAGCGCATC XBP1-FP CTGCCAGAGATCGAAAGAAGGC XBP1-RP CTCCTGGTTCTCAACTACAAGGC FBXO11-FP ATCATGGACGTGATGTTGGTGTG .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 19 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 Supplementary Table-2 558 Antibody name Purpose Product/Manufacturer BCL6 WB GeneTex#GTC101338 Histone H3K4-trimethyl ChIP CST#C42D8 ACTB WB Novusbio (NB100-56874) Histone H3 WB Biolegend (819411) gH2AX ChIP Abcam#ab20669 CTCF ChIP, WB, IP Active Motif#61932 MBD1 WB Abeomics (ABM15H2) 559 FBXO11-RP CCACTGTAGGGTTAGCATAGGC IRF4-FP GAACGAGGAGAAGAGCATCTTCC IRF4-RP CGATGCCTTCTCGGAACTTTCC Primers for shRNA-Knockdown (5’ to 3’) shRNA-MBD1 CCGGGAACAGAGAATGTTTAA shRNA-CTCF ACGTGTCCACGGCGTTCAAAT shRNA-BCL6 CCCATGATGTAGTGCCTCTTT Primer sequences for ChIP analysis Primers Primer sequence BCL6-G_Forward TGGATTCGTGCGGCTGTG BCL6-G_Reverse GGGAAGGGAAGGAAGAAGAGG BCL6-J_Forward GGCGACTTGAAGGAGACAGC BCL6-J_Reverse CCCATCCCTATCCACCAAACC BCL6-M_Forward GCTGAAGGGTGTGGGTCTC BCL6-M_Reverse TTCTCGCCAGGCTACTATGC BCL6-P_Forward AAATAAACTTCGGAATCGGACAAC BCL6-P_Reverse CCCACCTCTCACCCACAAC Primer for BCL6-binding motif deletion on MBD1-promoter BCL6-D235-FP GCTGTCGTTGAACGTC BCL6-D235-RP AACCTGAAGGGAAGC BCL6-D875-FP TTCTGGATAAGATGGGG BCL6-D875-FP GCAAGGATGAGTGGAC .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 20 Supplementary Table 3- List of tumor acronym from TCGA study datasets 560 561 Study Abbreviation Study Name LAML Acute Myeloid Leukemia ACC Adrenocortical carcinoma BLCA Bladder Urothelial Carcinoma BRCA Breast invasive carcinoma CESC Cervical squamous cell carcinoma and endocervical adenocarcinoma CHOL Cholangiocarcinoma COAD Colon adenocarcinoma ESCA Esophageal carcinoma GBM Glioblastoma multiforme HNSC Head and Neck squamous cell carcinoma KICH Kidney Chromophobe KIRC Kidney renal clear cell carcinoma KIRP Kidney renal papillary cell carcinoma LIHC Liver hepatocellular carcinoma LUAD Lung adenocarcinoma LUSC Lung squamous cell carcinoma DLBC Lymphoid Neoplasm Diffuse Large B cell Lymphoma PAAD Pancreatic adenocarcinoma PCPG Pheochromocytoma and Paraganglioma PRAD Prostate adenocarcinoma READ Rectum adenocarcinoma SKCM Skin Cutaneous Melanoma STAD Stomach adenocarcinoma TGCT Testicular Germ Cell Tumors THYM Thymoma UCS Uterine Carcinosarcoma UCEC Uterine Corpus Endometrial Carcinoma UVM Uveal Melanoma 562 563 Figure legends 564 Figure 1: Identification of the DNA-replication stress-induced transcriptome in Raji cells. 565 (A) Differentially expressed genes in control and HU-treated Raji cells (n=3) x axis indicated 566 log2FC for mRNA fold change and y axis indicated -log10 of adjusted p values. data is from three 567 independent samples (B) Analysis of histone H3K4me3 trimethylation ChIP -seq on the MBD1 568 locus in vehicle -treated or HU -treated Raji cells (C) Real-time kinetics of MBD1 and BCL6 569 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 21 expression in activated germinal center B cells (AGCBs) from human tonsil (D) BCL6-diploid 570 diploid tumors were classified into MBD1-amplification (n=6) , MBD1-deep deletion (n=19), 571 MBD1: diploid (n=4230), MBD1:gain (n=298) and MBD1-shallow deletion (n=1397) exhibiting 572 mean BCL6-mRNA values of 1.110, 1.375,1.103, 0.8686 and 1.309 in each group respectively. 573 p=<0.0001 MBD1:diploids vs MBD1:gain and MBD1-diploid vs MBD1 -shallow deletion 574 respectively. Dunnett’s multiple comparison test One way ANOVA. Data are presented as mean 575 ± SEM . (E) Comparison of BCL6 mRNA levels in the DLBC L samples (TCGA database) 576 exhibiting either MBD1:diploid or MBD1:gain status. Normalization of BCL6 mRNA in MBD1 577 altered groups was calculated by normalizing the RSEM values of BCL6 mRNA by RSEM values 578 of MBD1 mRNA in each sample. p= 0.0.0307 for MBD1:diploid vs MBD1:gain. T-test, Mann-579 Whitney test (F) MBD1-diploid tumors were classified into BCL6-amplification, BCL6-deep 580 deletion, BCL6-diploid, and BCL6-gain and BCL6-shallow deletion groups. Values on y -axis 581 indicates normalized RSEM values of MBD1 mRNA with the RSEM values of BCL6 mRNA. The 582 mean of normalized MBD1 mRNA was 0.8332 in BCL6-amplification (n=182), 2.201 for BCL6-583 deep deletion (n=15), 1.57 in BCL6-diploid (n=4230), 1.171 in BCL6-gain (n=972), and 2.027 in 584 BCL6-shallow deletion (n=390) group. p< 0.0001 for BCL6-amplification vs BCL6-diploid, p< 585 0.0001 for BCL6-diploid vs BCL6-gain. Dunnett’s multiple comparison test, One way ANOVA. 586 587 Figure 2: MBD1 suppresses BCL6 expression under DNA replication stress. 588 (A) shMBD1-MCF7, and shMBD1-MDA-MB-231 cells exhibit higher levels of BCL6 expression 589 than shSCR -MCF7 and shSCR -MDA-MB-231 cells. For BCL6 group: p<0.0001 for shSCR -590 MCF7 vs shMBD1 -MCF7 cells., p0.9999 for shSCR -592 SUM149PT vs shMBD1 -SUM149PT. For MBD1: p= 0.0051 for shSCR -MCF7 vs shMBD1 -593 MCF7, p<0.0001 for shSCR -MDAMB231 vs shSCR -MDAMB231, p=0.0339 for shSCR -594 MDAMB436 vs shMBD1 -MDAMB436, p=0.0001 for shSCR -SUM149PT vs shMBD1 -595 SUM149PT. Two-way ANOVA. Tukey’s multiple comparison test (B) pcDNA (1 μg) or pcDNA-596 MBD1 (1 μg and 5 μg) were transfected into 0.5 million Raji cells seeded in 12-well plates. RNA 597 was extracted 24 hours post-transfection, and cDNA preparation and qPCR were conducted for 598 MBD1. p=0.0070 for pcDNA (1 μg) vs pcDNA-MBD1 (1 μg) and p<0.0001 for pcDNA (1 μg) vs 599 pcDNA-MBD1 (5 μg). Sidak’s multiple comparison test (ordinary one -way Anova). Data are 600 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 22 presented as mean ± SEM from three independent experiments (C) MBD1 overexpression was 601 performed in Raji cells and cells were treated with HU and g emcitabine for 24 and 48 hour s 602 followed by estimation of BCL6 mRNA levels by qRT -PCR. For 24 hour -treated samples: 603 p=0.9990 for pcDNA (DMSO) vs pcDNA-MBD1 (DMSO), p=0.6897 for pcDNA (HU) vs pcDNA-604 MBD1 (HU), and p=0.0009 for pcDNA (Gemcitabine) vs pcDNA-MBD1 (Gemcitabine). For 48 605 hour-treated samples: p=0.9748 for pcDNA (DMSO) vs pcDNA-MBD1 (DMSO), p=0.2108 for 606 pcDNA (HU) vs pcDNA-MBD1 (HU), and p=0.0394 for pcDNA (Gemcitabine) vs pcDNA-MBD1 607 (Gemcitabine). Two -way ANO VA. Tukey’s multiple comparison test (D) BCL6 promoter 608 harbours CpG Methylation sites. Encode datasets of CpG methylation on BCL6 gene among 609 several cells lines is shown using the UCSD datasets (E) MBD1 binds to the BCL6 promoters in 610 HCT116 cells. Region highlighted in the above panel indicates the CpG methylation-rich regions 611 exhibiting MBD1 peaks in HCT116. The lower panel shows BCL6 genetic locus with MBD1 peaks 612 through the gene body. MDB1 binding on BCL6 locus was examined using the dataset 613 #DRX021118 from the ChIP-atlas. (https://chip-atlas.org/view?id=DRX021118). 614 615 Figure 3: Increased genomic instability at BCL6-MTC in MBD1-depleted cells 616 (A) TCGA dataset of 27 tumor types exhibiting the MBD1-diploid of MBD1-deletion status. 617 Deletion and shallow deletion samples are counted as deletion samples. Percentage of BCL6 618 alteration is compared in each tumor type between MBD1:diploid or MBD1:deletion status. BCL6 619 alteration are the samples exhibiting with BCL6:deep deletion, BCL6:shallow deletion, BCL6:gain 620 or BCL6:amplification. Tumor samples exhibiting BCL6:diploid status are not included as altered 621 BCL6 status (B) The percentage of BCL6 alterations in DLBCL samples from TCGA database was 622 quantified in MBD1-diploid and MBD1-deleted tumors and presented in the graph. MBD1-deleted 623 tumors include samples with MBD1 deep and shallow -deletions. BCL6 alterations encompass 624 tumor samples with BCL6 deep-deletion, shallow-deletion, gain, and amplification, while BCL6 625 diploid samples were not considered altered (C) Schematic representation of the major 626 translocation cluster of the human BCL6 gene indicating exons and introns. Primer pair G, J, M, P 627 location analyzed for ChIP-qPCR is indicated (D) γH2AX ChIP analysis in shSCR-Raji cells and 628 shMBD1-Raji cells treated with 10 mM HU . qPCR analysis was performed using four primer 629 pairs, G, J, M, and P, targeting intron 1 of the BCL6 locus. Primer pair locations are not to scale. 630 For primer pair G, p = 0.1563 for shSCR-Raji (10 mM HU) vs. shMBD1 -Raji (10 mM HU); for 631 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 23 J, p = 0.0006 for shSCR-Raji (10 mM HU) vs. shMBD1-Raji (10 mM HU); for M, p = 0.0259 for 632 shSCR-Raji (10 mM HU) vs. shMBD1 -Raji (10 mM HU); for P, p = 0.1209 for shSCR-Raji (10 633 mM HU) vs. shMBD1 -Raji (10 mM HU) . Unpaired t -test. Data are presented as mean ± SEM 634 (n=3) (E) CTCF ChIP analysis in vehicle - or HU-treated shSCR-Raji cells and HU -treated Raji 635 cells. For primer pair G, p = 0.0002 for shSCR-Raji (10 mM HU) vs. shMBD1-Raji (10 mM HU); 636 for J, p = 0.0093 for shSCR-Raji (10 mM HU) vs. shMBD1-Raji (10 mM HU); for M, p = 0.0022 637 for shSCR-Raji (10 mM HU) vs. shMBD1 -Raji (10 mM HU); for P, p <0.0001 for shSCR-Raji 638 (10 mM HU) vs. shMBD1-Raji (10 mM HU) Unpaired t-test. Data are presented as mean ± SEM 639 (n=3). 640 641 Figure 4: BCL6 and AID inactivate MBD1 at transcriptional and genetic levels. 642 (A) Treatment of Raji cells with the BCL6 inhibitor, FX1 (10 μM and 50 μM), induces MBD1 643 transcript levels in a dose -dependent manner. Data are presented as mean ± SEM from three 644 independent experiments. Dunnett's multiple comparison test: p = 0.0223 for Raji (DMSO) vs. 645 Raji (FX1-50 μM) treatment groups (B) qPCR analysis of BCL6, MBD1, and IRF4 transcript levels 646 in shSCR-Raji and shBCL6-Raji cells. Sidak’s multiple comparison test: two-way ANNOVA, For 647 MBD1, p =0,8412 for shSCR-Raji vs shBCL6-Raji groups; For BCL6, p=0.0030 for shSCR-Raji 648 vs. shBCL6-Raji. For IRF4 group, p <0.0001 for shSCR-Raji vs. shBCL6-Raji groups (C) In silico 649 prediction of the BCL6 binding motif at the MBD1 promoter using JASPAR CORE 2018 650 Vertebrates, available in the Eukaryotic Promoter Database (EPD). The sequences of two BCL6 651 binding motif located on -235 and -875 base pair upstream of MBD1 TSS are indicated. The given 652 sequence represents the regions deleted in the pGL4-pr-MBD1 plasmids (D) Schematic 653 representation of the MBD1 promoter cloned upstream of the luciferase gene in pGL4-luc2 and the 654 BCL6 binding motif. Location of BCL6 binding motif and deletion are shown. HEK293T cells 655 were transfected with pGL4-luc2, pGL4-luc2 + pcDNA -BCL6, pGL4-luc2-pr-MBD1, pGL4-pr-656 MBD1 + pcDNA-BCL6, pGL4-luc2-pr-MBD1Δ235 + pcDNA-BCL6, and pGL4-luc2-pr-657 MBD1Δ900 + pcDNA-BCL6 plasmids and examined for luciferase activity after 24 hours. Data 658 are presented as mean ± SEM from three independent experiments. Tukey’s multiple comparison 659 test: p < 0.0001 for pGL4-luc2 vs. pGL4-luc2-pr-MBD1; p < 0.0001 for pGL4-luc2-pr-MBD1 vs. 660 pGL4-luc2-pr-MBD1 + pcDNA-BCL6 groups; p < 0.0001 for pGL4-luc2-pr-MBD1 vs. pGL4-661 luc2-pr-MBD1Δ235 + pcDNA-BCL6; p = 0. 0416 for pGL4-luc2-pr-MBD1 vs. pGL4-luc2-pr-662 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 24 MBD1Δ900 + pcDNA-BCL6 (E) BCL6 binds to MBD1 locus in SUDHL4, OCI -LY1 (DLBCL), 663 B-ALL and HepG2 cells cells. The data are analyzed from ChIP -atlas database. Datasets : 664 SUDHL4:SRX4609168; OCI-LY1:SRX689470, B-ALL:SRX18259603, HepG2:2636278 665 (https://chip-atlas.dbcls.jp/data/hg38/target/BCL6.10.html) (F) The COSMIC mutation landscape 666 of the MBD1 gene, showing the frequencies of A>C, A>G, A>T, C>A, C>T, C>G, G>A, G>C, 667 G>T, T>A, T>C, and T>G mutations. 668 669 Figure 5: Reduced tumorigenicity of MBD1 depleted Raji cells in mouse xenograft model- 670 (A) Cell viability of shSCR-Raji and shMBD1 -Raji cells treated with DMSO, HU (10 mM), 671 Gemcitabine (20 mM), and Etoposide (50 μM) for 24 hours. Sidak’s multiple comparison test: p 672 < 0.001 for shSCR-Raji (HU) vs. shMBD1-Raji (HU); p < 0.0001 for shSCR-Raji (Gemcitabine) 673 vs. shMBD1 -Raji (Gemcitabine); p = 0.0005 for shSCR -Raji (Etoposide) vs. shMBD1 -Raji 674 (Etoposide) (B) Tumor xenografts were established by subcutaneously injecting 1 million shSCR-675 Raji or shMBD1-Raji cells mixed with 200 μl of Matrigel above the right front limb of nude mice 676 to reach the subcutaneous pocket. Mice were monitored for tumor growth over a period of 55 days. 677 The sample size consisted of n=4 for the shSCR -Raji group (Upper panel) and n=5 for the 678 shMBD1-Raji group (lower panel) (C) Tumor xenografts from shSCR -Raji and shMBD1 -Raji 679 cells injected into nude mice (n=4 for shSCR-Raji; n=5 for shMBD1-Raji) (D) Violin plot showing 680 the weight range of tumors in shSCR-Raji and shMBD1-Raji groups. Unpaired t-test: p = 0.0086. 681 682 Supplementary Figure 1 (A) Quantitative PCR (qPCR) analysis of MBD1 mRNA levels in 683 shSCR-Raji and shMBD1-Raji cells treated with DMSO, hydroxyurea (HU, 10 mM), gemcitabine 684 (20 mM), and etoposide (50 μM) for 24 hours. Data are presented as mean ± SEM, n = 3. Tukey’s 685 multiple comparison test results: p < 0.0001 for shSCR-Raji (DMSO) vs. shMBD1-Raji (DMSO) 686 groups, p < 0.0001 for shSCR-Raji (DMSO) vs. shSCR-Raji (gemcitabine) groups, p < 0.0001 for 687 shSCR-Raji (DMSO) vs. shSCR-Raji (etoposide) groups, p = 0.8068 for shMBD1 -Raji (DMSO) 688 vs. shMBD1 -Raji (HU) groups, p=0.0398 for shMBD1 -Raji (DMSO) vs. shMBD1 -Raji 689 (gemcitabine) groups, and p < 0.0010 for shMBD1-Raji (DMSO) vs. shMBD1 -Raji (etoposide) 690 groups. Western blot image of MBD1 using cell lysates from shSCR-Raji and shMBD1-Raji cells 691 (B) qPCR analysis of MBD1 mRNA in shSCR -SUM149PT and shMBD1 -SUM149PT cells 692 treated with or without HU (10 mM) for 24 hours. Tukey’s multiple comparison test: p < 0.001 693 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 25 for shSCR -SUM149PT ( -HU) vs. shSCR -SUM149PT (+10 mM HU) cells (C) MBD1 mRNA 694 expression profile among the TCGA tumor samples. Mean MBD1 mRNA RSEM (log2+1) value 695 was 850.3 in MBD1-shallow deletion (n=3208) , 646 in MBD1-deletions (n=56), 1171 in MBD1-696 diploid (n=5789), 1465 in MBD1-gain (n=819), 1979 in MBD1-amplification (n=17) groups 697 respectively. p< 0.0001 for MBD1-diploid vs MBD1-amplification, p< 0.0001 for MBD1-diploid 698 vs MBD1 Deep-deletion, p< 0.0001 for MBD1-diploid vs MBD1-gain, p< 0.0001 for MBD1-699 diploid vs MBD1-shallow deletion. Dunnett’s multiple comparison test, One way ANOVA. (D) 700 BCL6 mRNA expression profile among the TCGA tumor samples. Mean BCL6 mRNA value 701 (RSEM:log2+1) was 2006 for BCL6-amplification (n=488), 726.3 for BCL6-deep deletions 702 (n=21), 1198 for BCL6-diploid (n=5950), 1417 for BCL6-gain (n=2587) and 949.6 for BCL6-703 shallow deletion (n=864) group. p< 0.0001 for BCL6-diploid vs BCL6-amplification, p= 0.08 for 704 BCL6-diploid vs BCL6 Deep-deletion, p< 0.0001 for BCL6-diploid vs BCL6-gain, p< 0.0001 for 705 BCL6-diploid vs BCL6- shallow deletion. Dunnett’s multiple comparison test, One way ANOVA. 706 707 Supplementary Figure 2 - MBD1 expression is associated with DNA methylation and Bcl6 708 status in mouse GC B cells (A) CpG methylation status of activated mouse GC B cell in wild 709 type and Sca1-Bcl6Δ mice were analyzed from published study [25, 43]. The CpG methylation 710 signal on position 4’4’ of wild type Mbd1 locus are lost in Sca1-Bcl6Δ (B) AZA treatment of OCI-711 LY1, SUDHL2 and OCI-LY19 induced MBD1 mRNA expression (GSE190319). Y-axis indicated 712 values of MBD1 mRNA in counts of Transcription per million. 713 714 Supplementary Figure 3 (A) Correlation of MBD1 and IRF4 expression in DLBCL tumors 715 available at TCGA database. DLBCL tumors exhibiting BACH2-diploid and IRF4-diploid status 716 were sorted and catergorised into MBD1-diploid (n=23), MBD1-gain (n=12) and MBD1-shallow 717 deletion (n=1) groups. The Y -axis displays normalized values of IRF4 mRNA (RSEM counts) 718 with the MBD1 mRNA (RSEM counts). p=0.0051 for MBD1-diploid vs MBD1-gain tumors. Mann 719 Whiteny t-test. (B) Correlation of MBD1 and IRF4 co-expression in DLBCL samples obtained 720 from GEPIA. R=0.7, p=3.8e -08. (C) Mbd1 expression is induced in activated mouse germinal 721 center B cells at the stage of plasma cell differentiation (96 h ours) compared to unstimulated (0 722 hours) and germinal center B cells undergoing peak class switch recombination (CSR) and somatic 723 hypermutation (SHM). (D) qPCR analysis of IRF4, and PRDM1 in shSCR-Raji and shMBD1-Raji 724 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 26 cells. Data are presented as mean ± SEM, n = 3. Unpaired t-test. For IRF4, p <0.0001 between 725 shSCR-Raji vs. shMBD1-Raji, for PRDM1 p =0.0018 between shSCR-Raji vs. shMBD1-Raji cells 726 (E) Raji cells were transfected with either the control pcDNA vector or the pcDNA-MBD1 plasmid 727 for 24 hours, followed by quantification of IRF4 mRNA levels using qRT-PCR. Unpaired t-test. p 728 <0.0001 for pcDNA (1 μg) vs. pcDNA-MBD1 (1 μg); p = 0. 3360 for pcDNA-MBD1 (1 μg) vs. 729 pcDNA-MBD1 (5 μg). Data are presented as mean ± SEM, n=3 (F) qPCR analysis of GC marker 730 genes in shSCR-Raji and shMBD1-Raji cells. MBD1; p <0.0001 for shSCR-Raji vs shMBD1-Raji 731 cells. CCR7; p <0.0001 for shSCR-Raji vs shMBD1-Raji cells. P2RY8; p =0.0030 for shSCR-Raji 732 vs shMBD1-Raji cells. S1PR2; p =0.7833 for shSCR-Raji vs shMBD1-Raji cells. RAC1; p 0.9999 for shSCR-Raji vs shMBD1-Raji cells. XBP1; p =0.6600 for shSCR-Raji vs 735 shMBD1-Raji cells. IRF7; p =0.0724 for shSCR-Raji vs shMBD1-Raji cells. IFNGR1; p =0.3715 736 for shSCR-Raji vs shMBD1-Raji cells. STAT1; p =0.0017 for shSCR-Raji vs shMBD1-Raji cells. 737 Sidak's multiple comparisons test. Data are presented as mean ± SEM, n=3. 738 739 740

741 1. Allen, C.D., et al., Germinal center dark and light zone organization is mediated by 742 CXCR4 and CXCR5. Nat Immunol, 2004. 5(9): p. 943-52. 743 2. Victora, G.D. and M.C. Nussenzweig, Germinal Centers. Annu Rev Immunol, 2022. 40: 744 p. 413-442. 745 3. Muramatsu, M., et al., Class switch recombination and hypermutation require activation-746 induced cytidine deaminase (AID), a potential RNA editing enzyme. Cell, 2000. 102(5): p. 747 553-63. 748 4. Roco, J.A., et al., Class-Switch Recombination Occurs Infrequently in Germinal Centers. 749 Immunity, 2019. 51(2): p. 337-350.e7. 750 5. Lu, Z., et al., BCL6 breaks occur at different AID sequence motifs in Ig-BCL6 and non-751 Ig-BCL6 rearrangements. Blood, 2013. 121(22): p. 4551-4. 752 6. Gothwal, S.K., et al., BRD2 promotes antibody class switch recombination by facilitating 753 DNA repair in collaboration with NIPBL. Nucleic Acids Res, 2024. 754 7. Shaffer, A.L., et al., IRF4 addiction in multiple myeloma. Nature, 2008. 454(7201): p. 755 226-31. 756 8. Kaiser, L.M., et al., CXCR4 in Waldenström’s Macroglobulinema: chances and 757 challenges. Leukemia, 2021. 35(2): p. 333-345. 758 9. Kurtova, A.V., et al., Mantle cell lymphoma cells express high levels of CXCR4, CXCR5, 759 and VLA-4 (CD49d): importance for interactions with the stromal microenvironment and 760 specific targeting. Blood, 2009. 113(19): p. 4604-13. 761 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 27 10. Bal, E., et al., Super-enhancer hypermutation alters oncogene expression in B cell 762 lymphoma. Nature, 2022. 607(7920): p. 808-815. 763 11. Calpe, E., et al., ZAP-70 promotes the infiltration of malignant B-lymphocytes into the 764 bone marrow by enhancing signaling and migration after CXCR4 stimulation. PLoS One, 765 2013. 8(12): p. e81221. 766 12. Dhordain, P., et al., The LAZ3(BCL-6) oncoprotein recruits a SMRT/mSIN3A/histone 767 deacetylase containing complex to mediate transcriptional repression. Nucleic Acids 768 Res, 1998. 26(20): p. 4645-51. 769 13. Chang, C.C., et al., BCL-6, a POZ/zinc-finger protein, is a sequence-specific 770 transcriptional repressor. Proc Natl Acad Sci U S A, 1996. 93(14): p. 6947-52. 771 14. Phan, R.T., et al., BCL6 interacts with the transcription factor Miz-1 to suppress the 772 cyclin-dependent kinase inhibitor p21 and cell cycle arrest in germinal center B cells. 773 Nat Immunol, 2005. 6(10): p. 1054-60. 774 15. Akkaya, M., K. Kwak, and S.K. Pierce, B cell memory: building two walls of protection 775 against pathogens. Nature Reviews Immunology, 2020. 20(4): p. 229-238. 776 16. Ye, B.H., et al., Alterations of a zinc finger-encoding gene, BCL-6, in diffuse large-cell 777 lymphoma. Science, 1993. 262(5134): p. 747-50. 778 17. Ci, W., J.M. Polo, and A. Melnick, B-cell lymphoma 6 and the molecular pathogenesis of 779 diffuse large B-cell lymphoma. Curr Opin Hematol, 2008. 15(4): p. 381-90. 780 18. Ruggieri, S., et al., Translocation of the proto-oncogene Bcl-6 in human glioblastoma 781 multiforme. Cancer Letters, 2014. 353(1): p. 41-51. 782 19. Duy, C., et al., BCL6 enables Ph+ acute lymphoblastic leukaemia cells to survive BCR-783 ABL1 kinase inhibition. Nature, 2011. 473(7347): p. 384-8. 784 20. Walker, S.R., et al., The transcriptional modulator BCL6 as a molecular target for breast 785 cancer therapy. Oncogene, 2015. 34(9): p. 1073-82. 786 21. Xu, L., et al., BCL6 promotes glioma and serves as a therapeutic target. Proc Natl Acad 787 Sci U S A, 2017. 114(15): p. 3981-3986. 788 22. Shaknovich, R., et al., DNA methyltransferase 1 and DNA methylation patterning 789 contribute to germinal center B-cell differentiation. Blood, 2011. 118(13): p. 3559-69. 790 23. Béguelin, W., et al., EZH2 is required for germinal center formation and somatic EZH2 791 mutations promote lymphoid transformation. Cancer Cell, 2013. 23(5): p. 677-92. 792 24. Villa, R., et al., The methyl-CpG binding protein MBD1 is required for PML-RARalpha 793 function. Proc Natl Acad Sci U S A, 2006. 103(5): p. 1400-5. 794 25. Caron, G., et al., Cell-Cycle-Dependent Reconfiguration of the DNA Methylome during 795 Terminal Differentiation of Human B Cells into Plasma Cells. Cell Reports, 2015. 13(5): 796 p. 1059-1071. 797 26. Lai, A.Y., et al., DNA methylation prevents CTCF-mediated silencing of the oncogene 798 BCL6 in B cell lymphomas. Journal of Experimental Medicine, 2010. 207(9): p. 1939-799 1950. 800 27. Bell, A.C. and G. Felsenfeld, Methylation of a CTCF-dependent boundary controls 801 imprinted expression of the Igf2 gene. Nature, 2000. 405(6785): p. 482-5. 802 28. Hark, A.T., et al., CTCF mediates methylation-sensitive enhancer-blocking activity at the 803 H19/Igf2 locus. Nature, 2000. 405(6785): p. 486-9. 804 29. Kosak, S.T., et al., Subnuclear compartmentalization of immunoglobulin loci during 805 lymphocyte development. Science, 2002. 296(5565): p. 158-62. 806 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint 28 30. Jhunjhunwala, S., et al., The 3D structure of the immunoglobulin heavy-chain locus: 807 implications for long-range genomic interactions. Cell, 2008. 133(2): p. 265-79. 808 31. Gothwal, S.K. and J.H. Barlow, Stress Induced Signaling Pathways in Burkitt Lymphoma 809 Play Novel Mechanisms in Fate Determination and Pathogenesis of Germinal Center-810 Derived B-Lymphomas. bioRxiv, 2024: p. 2024.12.19.628635. 811 32. Holmes, A.B., et al., Single-cell analysis of germinal-center B cells informs on lymphoma 812 cell of origin and outcome. Journal of Experimental Medicine, 2020. 217(10). 813 33. Fujita, N., et al., Methyl-CpG binding domain 1 (MBD1) interacts with the Suv39h1-HP1 814 heterochromatic complex for DNA methylation-based transcriptional repression. J Biol 815 Chem, 2003. 278(26): p. 24132-8. 816 34. Clouaire, T., et al., Recruitment of MBD1 to target genes requires sequence-specific 817 interaction of the MBD domain with methylated DNA. Nucleic Acids Res, 2010. 38(14): 818 p. 4620-34. 819 35. Ci, W., et al., The BCL6 transcriptional program features repression of multiple 820 oncogenes in primary B cells and is deregulated in DLBCL. Blood, 2009. 113(22): p. 821 5536-48. 822 36. Cardenas, M.G., et al., Rationally designed BCL6 inhibitors target activated B cell diffuse 823 large B cell lymphoma. J Clin Invest, 2016. 126(9): p. 3351-62. 824 37. Saito, M., et al., A signaling pathway mediating downregulation of BCL6 in germinal 825 center B cells is blocked by BCL6 gene alterations in B cell lymphoma. Cancer Cell, 826 2007. 12(3): p. 280-92. 827 38. Green, M.R., et al., Transient expression of Bcl6 is sufficient for oncogenic function and 828 induction of mature B-cell lymphoma. Nature Communications, 2014. 5(1): p. 3904. 829 39. Oprea, M. and A.S. Perelson, Somatic mutation leads to efficient affinity maturation 830 when centrocytes recycle back to centroblasts. J Immunol, 1997. 158(11): p. 5155-62. 831 40. Ochiai, K., et al., Accelerated plasma-cell differentiation in Bach2-deficient 832 mouse B cells is caused by altered IRF4 functions. The EMBO Journal, 2024. 43(10): p. 833 1947-1964-1964. 834 41. Margiotta, A. and C. Bucci, Coordination between Rac1 and Rab Proteins: Functional 835 Implications in Health and Disease. Cells, 2019. 8(5). 836 42. Green, J.A. and J.G. Cyster, S1PR2 links germinal center confinement and growth 837 regulation. Immunol Rev, 2012. 247(1): p. 36-51. 838 43. Kulis, M., et al., Whole-genome fingerprint of the DNA methylome during human B cell 839 differentiation. Nat Genet, 2015. 47(7): p. 746-56. 840 44. Amanda, S., et al., IRF4 drives clonal evolution and lineage choice in a zebrafish model 841 of T-cell lymphoma. Nat Commun, 2022. 13(1): p. 2420. 842 45. Fujita, N., et al., Mechanism of transcriptional regulation by methyl-CpG binding protein 843 MBD1. Mol Cell Biol, 2000. 20(14): p. 5107-18. 844 46. Huang, C., et al., Cooperative transcriptional repression by BCL6 and BACH2 in 845 germinal center B-cell differentiation. Blood, 2014. 123(7): p. 1012-20. 846 47. Kuo, T.C., et al., Repression of BCL-6 is required for the formation of human memory B 847 cells in vitro. J Exp Med, 2007. 204(4): p. 819-30. 848 48. Lin, K.I., C. Tunyaplin, and K. Calame, Transcriptional regulatory cascades controlling 849 plasma cell differentiation. Immunol Rev, 2003. 194: p. 19-28. 850 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint Figure- 1 Identification of MBD1 as a novel regulator of GC reaction -1.500 -1.000 -0.500 0.000 0.500 1.000 1.500 DZ a DZ b DZ c INT a INT b INT c INT d INT e LZ a LZ b Pre M Plasmablast a Plasmablast b MBD1-FC BCL6-FC A C B E F MBD1: Diploid MBD1: Gain 0 2 4 6 8 BCL6 mRNA normalized with MBD1 mRNA Fig 1 E ( all bcl6 diploids) D 5 kb Raji-Control Raji- HU MBD1 Refseq .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint A C E D Figure- 2 MBD1 suppreses BCL6 transcription under DNA-replication stress B HCT116 Refseq .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint B A MTC C Exon 1 BCL6 NP_001697.2NM_001706.5 Genes, MANE Project (release v1.3) LOC100131635 BCL6 XP_011511364.1XM_011513062.4 XP_047304611.1XM_047448655.1 XP_005247751.1XM_005247694.5 NP_001128210.1NM_001134738.1 NP_001697.2NM_001706.5 NP_001124317.1NM_001130845.2 LOC122526776 NR_173091.1 NCBI RefSeq Annotation GCF_000001405.40-RS_2023_10 enhancerenhancer enhancer enhancer silencer Biological regions, aggregate, NCBI RefSeq Annotation GCF_000001405.40-RS_2023_10 ENSG00000228804 ENST00000449623.5ENST00000437407.1 ENSG00000113916 [+18] ENSG00000285938 ENST00000648485.1ENSP00000498467.1 ENSG00000290021 ENST00000702512.1 ENSG00000289331 ENST00000691617.2 Genes, Ensembl release 110 rs587778090GC/AT rs1474326C/A/T rs1523475T/A/C/G rs3774309A/C/G/T rs2229362C/A/T rs137878288G/A/T rs541016998G/A rs150656653T/C rs1056932G/A/C rs377059215G/A rs587778091T/C rs61752081C/A/Grs200263685G/A rs1880099C/A/G/T rs3774304C/A/G/T rs3733018A/C/G rs3733017T/A/G rs3774303C/A/G/T rs3172469T/C/G rs1523474T/A/C rs80031434C/A/G/T rs16862537T/C rs17797517T/C Cited Variations, dbSNP b156 v2 Live RefSNPs, dbSNP b156 v2 133668AT 729093C 2607821A 2260739T 734312A 2458095T 133676T 2481684T 133675A 133674A 781975G 2540535A 2521673A 2245665C 2596525T 2228659A 2355053T 729094T 2461434A 133670C 2606843T 2335230A 133672A 133671C 2375099A 2612944T 2559906G 717149A 2462878T 133673A 2519975A 2515439G 2463647T 2294551C 725713A 787262T 133669A 2237756A ClinVar variants with precise endpoints 22201 256 22201 256 22201 256 22201 256 RNA-seq exon coverage, aggregate (filtered), NCBI Homo sapiens Annotation Release 110 - log 2 scaled 13177 256 13177 256 13177 256 13177 256 RNA-seq intron-spanning reads, aggregate (filtered), NCBI Homo sapiens Annotation Release 110 - log 2 scaled 5517 52 1156 651 11375102778074 9139 2627 11120 RNA-seq intron features, aggregate (filtered), NCBI Homo sapiens Annotation Release 110 Exon 10ATG D E G J M P Figure 3- Increased genomic instablity at BCL6-MTC in MBD1-depleted Raji cells LAML ACC BLCA BRCA CESC CHOL COAD ESCA GBM HNSC KICH KIRC KIRP LIHC LUAD LUSC DLBC PAAD PCPG PRAD READ SKCM STAD THYM UCS UCEC UVM 0 50 100 150BCL6 alteration (%) Fig 4 A BCL6 alteration in all tumors MBD1: Diploid MBD1: Deletion LAML ACC BLCA BRCA CESC CHOL COAD ESCA GBM HNSC KICH KIRC KIRP LIHC LUAD LUSC DLBC PAAD PCPG PRAD READ SKCM STAD THYM UCS UCEC UVM 0 50 100 150BCL6 alteration (%) Fig 4 A BCL6 alteration in all tumors MBD1: Diploid MBD1: Deletion .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint Figure 4- BCL6 suppreses MBD1 transcription in multiple B-lymphomas C F B D SUDHL4 OCI-LY1 Refseq B-ALL HepG2 E Deleted sequence flanking to BCL6 binding on -875 bp upstream of MBD1 TSS (-906 to -821 deleted) 5’CTCCTCTCCCTGGGTTTTGTT AA T AAAA TTTTGAAG AAACCAAGGAAGCTGTCTCCACATTGCTGCGGTTGC AACTGTTCCAGAC Deleted sequence flanking to BCL6 binding on -235 bp upstream of MBD1 TSS (-299 to -225 bp deleted) 5”GAGGCTGGAAAGCGCATGCGCCAGCTAGATGGGC AGCGAGGAGAGCCGCAACTGCCAGTCCCTCGAAGGG GTTA SUDHL4 OCI-LY1 Refseq B-ALL HepG2 A .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint Figure 5 Increased Sensitivity to DNA-Replication Inhibitors and Reduced Tumorigenicity of MBD1-Depleted Cells A C D B shSCR-Raji shMBD1-Raji-0.5 0.0 0.5 1.0 1.5 Tumor weight (g) .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint A Supplementary Figure 1- MBD1 expression is induced by DNA replication stress C D shSCR-Raji shMBD1-Raji MBD1 β-actin B .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint Supplementary Figure 2- Association among MBD1 expression, DNA methylation and DNA replication stress A B 0 5 10 15 20 25 30 35 OCI-LY1 SUDHL2 OCI-LY19 PBS AZA Transcripts per million Normalized Lost peak Sca1-Bcl6Δ Wild type Refseq Sca1-Bcl6Δ Wild type Refseq .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint Supplementary Figure 3- MBD1 expression in B-lymphomas and mouse GC B cells B 0h 60h 96hSyn+ 0 10 20 30 40 50 Mbd1 C Mbd1 FPKM values Time in hours after BCR stimulation (primary Mouse B cells) unstimulated CSR SHM Plasma A D F E .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 15, 2025. ; https://doi.org/10.1101/2025.03.13.643172doi: bioRxiv preprint