Histone H2A and H2B Are Monoubiquitinated at AID-Targeted Loci

Background Somatic hypermutation introduces base substitutions into the rearranged and expressed immunoglobulin (Ig) variable regions to promote immunity. This pathway requires and is initiated by the Activation Induced Deaminase (AID) protein, which deaminates cytidine to produce uracils and UG mismatches at the Ig genes. Subsequent processing of uracil by mismatch repair and base excision repair factors contributes to mutagenesis. While selective for certain genomic targets, the chromatin modifications which distinguish hypermutating from non-hypermutating loci are not defined. Methodology/Principal Findings Here, we show that AID-targeted loci in mammalian B cells contain ubiquitinated chromatin. Chromatin immunoprecipitation (ChIP) analysis of a constitutively hypermutating Burkitt's B cell line, Ramos, revealed the presence of monoubiquitinated forms of both histone H2A and H2B at two AID-associated loci, but not at control loci which are expressed but not hypermutated. Similar analysis using LPS activated primary murine splenocytes showed enrichment of the expressed VH and Sγ3 switch regions upon ChIP with antibody specific to AID and to monoubiquitinated H2A and H2B. In the mechanism of mammalian hypermutation, AID may interact with ubiquitinated chromatin because confocal immunofluorescence microscopy visualized AID colocalized with monoubiquitinated H2B within discrete nuclear foci. Conclusions/Significance Our results indicate that monoubiquitinated histones accompany active somatic hypermutation, revealing part of the histone code marking AID-targeted loci. This expands the current view of the chromatin state during hypermutation by identifying a specific nucleosome architecture associated with somatic hypermutation.


Introduction
The immunoglobulin (Ig) genes in activated B cells are diversified by somatic hypermutation and class switch recombination to promote immunity. Somatic hypermutation introduces point mutations into the rearranged and expressed Ig variable regions while class switch recombination coordinates the exchange of one Ig constant region for a downstream region, deleting the intervening DNA. Both pathways require the Activation Induced cytidine Deaminase (AID) protein [1,2]. AID is active on transcribed DNA and functions to convert single-stranded cytidines into uracil and uracil-guanine mismatches (reviewed in [3,4]). Canonical mismatch repair and base excision repair are able correct genomic uracil [5][6][7], but in somatic hypermutation these normally faithful repair pathways become mutagenic [8]. The diversion to mutagenesis is promoted by the synthesis activities of low-fidelity DNA polymerases, such as polymerase eta (reviewed by [9,10]), and their participation in hypermutation may be regulated by PCNA monoubiquitination (reviewed by [11]). While hypermutation is largely confined to the rearranged and expressed Ig genes, other B cell loci are prone to aberrant hypermutation, leading to lymphoma [12,13]. The molecular mechanisms responsible for targeting AID to certain sites in the genome and the regulation of subsequent mutagenesis have not been established.
Monoubiquitination pathways may be important for regulating Ig gene diversification. DT40 B cells disrupted for PCNA monoubiquitination by a K164R substitution showed decreased AID-initiated Ig gene diversification [14]. Similarly, PCNA K164R knock-in mice had an altered spectrum of Ig gene mutagenesis, paralleling mismatch repair defective mice and suggesting that monoubiquitination of PCNA influences uracil repair outcomes in mammals [15,16]. PCNA modification may be facilitated in part by the RAD6 pathway because inactivation of a RAD6 associating E3 ligase, RAD18, in DT40 B cells resulted in reduced levels of PCNA monoubiquitination and decreases in hypermutation [14,17], however residual PCNA monoubiquitination in rad18 mutant DT40 suggests the involvement of more than one E3 ligase [18]. Further evidence for protein ubiquitinations in Ig gene diversification pathways comes from recent studies on the RNF8 and RNF168 E3 ligases. Silencing of these two proteins decreased class switch recombination efficiency in a murine model B cell line [19], and RNF168 was identified as the RIDDLE syndrome protein, a disease characterized by immunodeficiencies and DNA repair defects [20]. Recently, RNF8 knock-out mice were shown to have impaired class switch recombination and defects in DNA damage responses [21,22], further connecting histone monoubiquitination with Ig gene diversification. Together, it is likely that multiple proteins are monoubiquitinated to promote Ig gene diversification and hypermutation specific histone E3 ligases could serve as signals for locus-specific mutagenesis.
In light of the DNA repair machinery-driven nature of hypermutation at the Ig loci, characterized roles for chromatin monoubiquitination in facilitating normal DNA repair throughout the genome (reviewed by [23]), and the involvement of RNF8 in class switching [19,[21][22], we speculated that histone monoubiquitination could accompany hypermutation. We therefore assayed the nucleosomes at AID-targeted and actively hypermutating loci for ubiquitin modification. Here, we show by Chromatin Immunoprecipitations (ChIPs) that both histone H2A and H2B are monoubiquitinated at two different AID-targeted loci in the constitutively hypermutating B cell line Ramos. ChIP analysis of primary LPS activated murine splenocytes showed similar levels of enrichments for hypermutating template upon immunoprecipitation with antibody specific to AID, pol II, and monoubiquitinated forms of H2A and H2B. Furthermore, we find by immunofluorescence microscopy that AID colocalizes with monoubiquitinated H2B in discrete nuclear foci, suggesting that both are concomitant with hypermutation. Our results identify a considerable and previously uncharacterized chromatin modification associated with AID-induced somatic hypermutation.

AID and hypermutation target two discrete loci in Ramos
Ramos is a human Burkitt's lymphoma cell line with both functional and non-functionally rearranged Ig heavy chain (IgH) loci. One IgH allele contains a productive VDJ rearrangement (re-V H ), which is a gene that is expressed and constitutively hypermutated [24]. The other IgH allele has participated in a reciprocal translocation between the ends of chromosomes 8 and 14, replacing the antibody variable region on chromosome 14 with an intact c-MYC allele from chromosome 8 (MYC 14 ), and this locus is also hypermutated [25]. The partially rearranged V region on chromosome 8 and the mu constant region downstream of re-V H are not diversified by hypermutation. Using EST data from the NCBI database, and previously reported c-MYC sequences [25], we constructed a map of the rearranged and expressed heavy chain loci in Ramos (Fig. 1). The documented constitutive hypermutation of Ramos re-V H and MYC 14 predict that AID selectively targets these two distinct loci for deamination.
We first established active somatic hypermutation of the re-V H and MYC 14 in our Ramos subclone using standard surface IgM loss assays and direct sequencing of MYC 14 . Constitutive sequence diversification of the expressed IgM locus results in mutations that inactivate surface display, providing a marker for hypermutation rate in Ramos. Starting from a single Ramos sIgM positive isolate, 90 days of continuous culture resulted in 6.83% sIgM loss, as measured by FACS with FITC-conjugated anti-IgM antibody ( Fig. 2A). This compares well with the sIgM loss rate reported for Ramos, showing 4% sIgM loss for after 4 weeks of in vitro culture [26], and ,17% loss after 6 months [24]. We conclude that the re-V H is actively hypermutated. To ensure ongoing hypermutation of the other IgH locus, individual expressed MYC 14 clones of from our Ramos population were sequenced, and 20% of our sequence reads contained one or more novel mutations (6/30, mean length = 748 nt) (Fig. 2B). Only the translocated c-MYC is expressed in Ramos, and MYC 14 contains a single nucleotide polymorphism A523T [25], distinguishing it from the nontranslocated allele. This base substitution was present in all sequence reads we obtained. The MYC 14 point mutations we identified are novel, and do not correspond to previously identified c-MYC hypermutations [25]. We conclude that both the re-V H and MYC 14 in our Ramos subclone are expressed and subject to hypermutation, as expected.
AID activity is required for initiating somatic hypermutation, predicting that the protein should be in physical contact with both the Ramos re-V H and c-MYC genes. To test this, we used chromatin immunoprecipitations (ChIPs) to test for AID association with hypermutating loci in Ramos. ChIPs rely on specific antibodies to precipitate crosslinked protein-DNA complexes, followed by quantitative PCR (qPCR) analysis of  (Fig. 3A). To further test hypermutation specificity ChIPs with PCNA and pol eta antibodies enriched for the re-V H (Fig. S1), and AID antibody failed to enrich for the c-MYC locus in AID negative HEK293 and A549 cells (Fig. S2), and westerns with this same antibody did not detect AID in HEK293 extracts, but identified a single 24 kDa band in Ramos, as expected (Fig. S3). In addition, IPs using this AID antibody and Ramos extract precipitated native AID, as detected by western with an independent AID antibody (Fig. S4).
Primers designed to amplify MYC 14 could potentially amplify the other c-MYC allele if precipitated during ChIPs; however, only the translocated c-MYC contains the single nucleotide polymorphism A523T [25]. Further, ChIP results are suggestive of AID association with MYC 14 because AID tracks with pol II to deaminate transcribed DNA (reviewed by [27]) and the nontranslocated allele of c-MYC is not expressed ( [25] yet pol II ChIPs enriched for exons 1-3 c-MYC. Most importantly, 100% (8/8) of different sequence reads from PCR product amplified from anti-AID ChIPed template corresponded to the translocated MYC 14 allele (not shown). Together, our results indicate that AID is physically associated with the Ramos re-V H and with exons 1-3 of MYC 14 , supporting the model that both loci are subject to AIDinitiated somatic hypermutation.
The chromatin at the re-V H and c-MYC gene in Ramos B cells is monoubiquitinated Having confirmed AID associations at both re-V H and c-MYC in our Ramos subculture, we next examined the chromatin state at these loci to identify histone markers of hypermutation. Based on roles for histone ubiquitination in normal DNA repair [23] and apparent involvement of RNF8/168 ubiquitin E3 ligases in class switch recombination [19][20][21][22], we focused our analysis on ubiquitin-modified chromatin. We used ChIPs and antibodies specific to monoubiquitinated histones to ask if the re-V H and . Also, mUb-H2A and mUb-H2B ChIPs with AID negative HEK293 and A549 cells did not enrich for the c-MYC loci (Fig. S2). We conclude that the re-V H and c-MYC loci in Ramos are occupied by nucleosomes that contain monoubiquitinated forms of H2A and H2B.
Previous studies connect histone monoubiquitination with transcriptional silencing and/or polymerase elongation (reviewed by [28]) so we next asked if histone ubiquitination is a broad consequence of high RNA polymerase II (pol II) driven expression of the heavy chain loci in Ramos, or associated with mutagenesis. ChIPs using RNA pol II antibody and Ramos chromatin showed comparable enrichment via normalized qPCR for the re-V H , exons 1-3 of c-MYC, b-actin, and V H -Cm. We did not observe enrichment of the unrearranged V H locus (V 8 ) or the region downstream of MYC 14 , MYC4, which was anticipated because these regions are not expressed. In contrast, b-actin, and V H -Cm were all enriched by pol II ChIPs (Fig. 3D), but not after ChIP using monoubiquitinated histone antibody (Figs. 3B, C). These results mirror our AID analyses, and suggest that pol II-dependent gene expression alone is insufficient for establishing sustained chromatin monoubiquitination in Ramos.
If monoubiquitinated chromatin is associated with Ig gene mutagenesis, we also anticipate other hypermutation factors will be associated with these same loci. AID-induced uracil is processed by mismatch repair factors at the Ig loci, and substitutions are introduced opposite A and T bases through PCNA-dependent recruitment of pol eta synthesis in place of pol delta (reviewed by [11]). ChIP analysis suggests that the re-V H , but not the unrearranged allele (V 8 ), is in physical contact with PCNA and pol eta (Fig. S1), supporting the model that the monoubiquitinated loci we identify here also harbor proteins known to support mutagenesis in response to uracil in DNA. The qPCR enrichments obtained for Ramos ChIPs were independently verified by qualitative, traditional PCR amplification and ethidium bromide visualization (Fig. 3E) and westerns verified specificity for AID and monoubiquitinated histone antibodies (Fig. 3F). The chromatin of Ig genes in activated primary murine B cells is ubiquitinated It is possible that chromatin monoubiquitination is a unique feature of Ramos hypermutation. To address this, we tested histone monoubiquitination of an AID-targeted locus in LPS activated murine B cells. Typically, hypermutation is difficult to quantify by PCR in primary cells because of the sequence variability inherent to the rearranged Ig genes. To overcome this technical limitation, we used a quasi-monoclonal mouse, which was engineered to contain a functional IgH chain allele generated by targeted replacement of heavy chain locus variable region with a defined VDJ rearrangement (17.2.25) [29] (Fig. 4A). The 17.2.25 VDJ region in these mice is subject to diversification by somatic hypermutation [30].
Primary splenocytes were isolated from six healthy quasimonoclonal mice and cultured with lipopolysaccharide (LPS) to activate primary B cells. ELISAs for secreted IgG confirmed activation of Ig gene diversification (data not shown). In chromatin prepared from these activated splenocytes, normalized enrichment of 17.2.25 after ChIP with mUb-H2A and mUb-H2B antibody showed robust enrichment, 16.2-fold and 21.8-fold for each antibody, respectively (Fig. 4B). Similarly, pulldowns using these same antibodies were also enriched for the Sc3 switch region (Fig. 4B). AID appears to target the 17.2.25 and Sc3 loci after activation because anti-AID ChIPs resulted in 10.4 and 7.1 fold enrichments of these regions respectively (Figs. 4B, C). The murine triose phosphate isomerase gene (TPI) and unexpressed neural specific gene (STEP) were used as controls, and only the TPI locus showed modest enrichment (6.9-fold) after pol II ChIP, and neither locus was precipitated upon IP with anti-AID, anti-mUb-H2A, or anti-mUb-H2B. Due to the limited antibody repertoire encoded by the quasi-monoclonal mice, selective pressure for Ig gene diversity is high, causing both hypermutation and frequent V gene replacements [30]. Because it is unlikely that 17.2.25 gene rearrangements would be amplified by our primer sets, enrichments we obtained for the 17.2.25 locus after AID ChIP (Fig. 4B) are likely to be an under-representation of the actual level of AID association. Regardless, we conclude that the 17.2.25 and Sc3 loci from primary LPS activated quasimonoclonal murine splenocytes are expressed by RNA pol II, targeted by AID, and occupied by monoubiquitinated histone H2A and histone H2B.
AID and mUb-H2B colocalize in discrete nuclear foci ChIPs examine total cellular populations, and hypermutation events may occur transiently and at unsynchronized time points. Therefore, we used immunofluorescence confocal microscopy to ask if AID and ubiquitinated chromatin co-occupy the DNA. Both mUb-H2B and AID antibody produced clear robust signals in Ramos cell preparations. Figure 5 shows a representative image of AID and mUb-H2B foci in Ramos, and overlay of these images showed clear colocalization of AID and mUb-H2B signal in discrete nuclear foci. We also found multiple examples where more than one AID and monoubiquitinated H2B are colocalized within a single cell nucleus (Fig. S5), supporting the notion that at least two and possibly more loci are diversified by somatic hypermutation in Ramos cells. We observe multiple AID foci in individual Ramos cells, highly consistent with the behavior of endogenous AID [31][32][33]. It is possible that non-Ig genes are bound and deaminated by AID in Ramos. Antibodies to mUb-H2A and mUb-H2B antibody are specific for the monoubiquitinated forms of histones, and our westerns probing Ramos whole cell extract clearly show single bands at ,25-30 kDa (Fig. 3F) corresponding to the size of a single histone ubiquitination (,28 kDa). Even though mUb-H2A, and mUb-H2B antibodies were competent for western analyses (Fig. 3F) and ChIPs (Figs. 3 and 4), we were unable to clearly detect mUb-H2A by immunofluorescence microscopy. Nevertheless, our imaging results suggest that AID is spatially and temporally associated with ubiquitinated nucleosomes.

Discussion
Our results show that genomic loci that are targeted for somatic hypermutation are also occupied by monoubiquitinated chromatin. Histone H2A and H2B monoubiquitination was coincident with transcription, but not due to pol II catalyzed gene expression alone because anti-mUb-H2A and mUb-H2B ChIPs failed to enrich for the highly expressed Ramos b actin locus or the TPI gene in primary mouse splenocytes, even though pol II ChIP enriched for these regions at levels comparable to the re-V H , c-MYC Exons 1-3, and 17.2.25 (Figs. 3, 4). Furthermore, the only genomic regions enriched by mUb-H2A and mUb-H2B ChIP were those that were also enriched upon anti-AID ChIP (Fig. 3A); and these same loci also showed significant enrichments after ChIP with antibodies to the DNA repair factors PCNA and pol eta (Fig. S1), which are two proteins involved in somatic hypermutation opposite A and T bases (recently reviewed by [11]). Histone monoubiquitination may occur in response to AID activity, but could also be a signal for AID recruitment because AID colocalized with mUb-H2B as visualized by immunofluorescence microscopy (Fig. 5). Either way, our results identify the presence of both histone H2A and H2B tail monoubiquitinations at genes that are targeted selectively for mutagenesis, identifying this specific set of chromatin modifications with genomic loci that are deaminated by AID and mutated by responding DNA repair pathways. As such, co-occupancy of monoubiquitinated H2A and H2B within nucleosomes at the Ig genes may in part characterize the chromatin architecture associated with somatic hypermutation.
Histone termini extend away from the core histone octamer as tails that serve as acceptors for a wide range of post-transcriptional modifications. The combination of these modifications or ''histone code'' regulates a wide range of functions, although the specific DNA level activities triggered by various combinatorial signals have not been completely defined. In hypermutation, some small molecule histone modifications have been described. Acetylation of histone H3 and H4 at the IgH variable region, but not the constant region, results from activation of hypermutation in BL2 cells, suggesting a role in targeting somatic hypermutation machinery [34]. However, in primary mouse B cells changes in histone acetylation at hypermutating loci did not follow with B cell activation [35]. Nevertheless, H4 acetylation may be important for some types of AID-initiated diversification because disruption of the E2A transcription factor in DT40 leads to decreases in H4 acetylation and gene conversion [36]. One unique signal for V region hypermutation may be phosphorylation of histone H2A at serine 14 [35], and it is possible that this signal may operate in concert or in response to histone monoubiquitination to promote hypermutation. More than likely, locus-specific mutagenesis in B cells depends upon the combinatorial signal generated through multiple histone tail modifications, and further studies are required to decipher the specific histone code assigned to somatic hypermutation.
H2A and H2B monoubiquitination may have active roles in the hypermutation pathway, and immunofluorescence results confirm proximity of AID and monoubiquitinated H2B (Figs. 5,  S5). Chromatin modification may occur as a consequence of AID activity to promote mutagenesis at some loci, or histone monoubiquitinations could help recruit AID. These models are not necessarily mutually exclusive. If chromatin ubiquitination is important for regulating the balance between mutagenic and faithful DNA repair in Ig gene hypermutation, one might predict that AID could deaminate multiple genes but the resulting uracils are faithfully repaired at the non-Ig loci. Indeed, we and others have observed multiple AID foci by immunofluorescence microscopy (Figs. 5, S5) [31][32][33], and genome sequencing of germinal center B cells reveals that AID acts widely at transcribed genes, but genomic instability is largely suppressed by faithful DNA repair [12]. There are likely to be signals specific to only some genes (like the Ig loci) that regulate the switch from faithful to mutagenic uracil repair, and because mUb-H2B appears to occupy only a small subset of the AID-targeted loci (Figs. 5, S5), chromatin ubiquitination is not an automatic result of AID binding but rather a locus-specific event. It is conceivable that the conspicuous presence of ubiquitinated H2A and H2B at the Ig genes represents a signal that marks these loci for mutagenesis. Consistent with a model whereby histone monoubiquitination facilitates the repair responses to DNA deamination, recent studies have identified histone-specific E3 ubiquitin ligases in class switch recombination. Starting with histone H2A modification, ubiquitination cascades appear critical for establishing assembly of recombination factors at sites of AID activity. The loss of the RNF168 H2A E3 ligase disrupts histone ubiquitination and causes DNA repair defects and immunodeficiencies associated with RIDDLE syndrome [20]; and RNF8 works in concert with RNF168 to initiate and expand histone ubiquitinations at the Ig loci that resulting in 53BP1 activity and proper class switching [20][21][22]. In normal break repair, chromatin ubiquitinations promote assembly of DNA damage response factors at sites of ionizing radiation damage [37]. Furthermore, the replication cofactor PCNA, which we found physically associated with the re-V H (Fig. S1) and may be a lynchpin for regulating A/T biased hypermutation [11], colocalizes with monoubiquitinated H2A [38]. Upon AID activity, chromatin ubiquitination may signal assembly of hypermutation factors and/or ubiquitination cascades that regulate the shift between faithful uracil repair and mutagenesis. However, further studies are required in order to identify the ubiquitin E3 ligases functioning in normal and aberrant somatic hypermutation.
Histone monoubiquitination is a substantial chromatin modification, and suggests the DNA at AID-targeted loci is accessible for metabolism. Unlike small molecule modifiers like phosphates or acetyl groups, ubiquitin is an 11 kDa protein.
The addition of a single ubiquitin on both H2A and H2B would increase nucleosome mass by nearly 40%, which may greatly influence the association between DNA and the core histones or create an assembly site for repair factors. The significant enrichments we obtained by ChIP of the re-V H and MYC 14 with mUb-H2A and H2B antibodies (Figs. 3B, C) suggest the presence of a highly ubiquitinated locus, with multiple antibody binding targets at the site of hypermutation, but apparently not extending into the IgH constant region (V H 1 amplicon, Figs. 3B, C). Thus, histone monoubiquitination is localized to the DNA regions that are also bound by AID. It follows that V H 1, and likely the proximal S region, were not enriched by AID ChIP (Figs. 3A-C), which is expected because Ramos does not class switch. Because AID is active on single-stranded DNA and associated with the re-V H (Fig. 3A), it is conceivable that histone monoubiquitination creates a chromatin architecture that is permissible to AID attack or downstream uracil repair. Indeed, nucleosomal DNA is protected from AID in the absence of transcription [39], suggesting that deamination by AID may require histone remodeling and chromatin-level signaling. Importantly, our results identify a new chromatin modification associated with hypermutation and future studies will be required to next define functional roles in mechanisms of somatic hypermutation.

Cells and Cell lines
Ramos cells were purchased from the ATCC and were cultured in suspension at 37uC in RPMI (10% (v/v) FBS, 1% (v/v) PS, 2 mM L-glutamine) (Life Technologies, Grand Island, NY) in 75 cm 2 flasks. Following limiting dilution cloning, a sIgM positive isolate was expanded for 90 days, and then analyzed for sIgM loss by flow cytometry. Ramos cells were collected by centrifugation at 3006g for 10 minutes, washed once with ice-cold PBS, and then resuspended in 100 mL PBS and incubated with 10 mL of antihuman-IgM antibody (Sigma, St Louis, MO) for 10 minutes at 4uC. Cells were again collected, washed twice with PBS, resuspended in 100 mL PBS and incubated with 3 mL of FITCconjugated anti-goat Ig secondary for 10 minutes at 4uC in the dark. Cells were washed twice with PBS then resuspended in 400 mL PBS. Samples included 10% normal rat serum to prevent non-specific binding. Cells were analyzed immediately on a Becton Dickinson FACS Calibur flow cytometer using CellQuest Pro software (San Jose, CA) with 1610 4 events collected.

PCR analyses
Oligonucleotide sequences from all analyses are detailed in Table S1). Sequencing of Ramos c-MYC was performed as described previously [25]. Briefly, total Ramos RNA was converted to cDNA via RT-PCRs using random 20-mers and Protoscript Reverse Transcriptase (New England BioLabs (NEB), Ipswich, MA). Unless otherwise indicated, PCR amplifications mUb-H2A/H2B and Hypermutation were performed in 40 ml reactions at standard concentrations (1.5 mM MgCl 2 , 0.2 mM dNTP, 16NEB PCR buffer, 0.5 U Taq (NEB), 0.5 uM each primer) and then cloned into Topo TA PCR 2.1 (Invitrogen, Carlsbad, CA, K450001) for sequencing. MYC 14 was distinguished from c-MYC on chromosome 8 by the presence of an A at position 523 to T polymorphism found within MYC 14 [25]. Traditional ChIP amplifications from recovered DNA was quantified by standard SybrGreen quantification (Invitrogen), then diluted to 0.1 ng/ml. PCR reactions were performed for 25, 30 or 35 cycles (as described above). Starting material was standardized across all reactions (0.1 ng eluted DNA), excluding inputs, which were titrated from 0.5 ng to 10 ng per reaction. Following amplification, products were resolved by 1.5% agarose gel electrophoresis. The identities of all amplicons were verified by Topo TA PCR 2.1 cloning and sequencing. Quantitative PCR (qPCR) used a series of primer pairs generated for each genomic locus using the Integrated DNA Technology design tool, PrimerQuest (https://www.idtdna.com). Individual amplicons were each evaluated by standard PCR ethidium bromide visualization, dissociation curve determination, and direct sequencing. A final validated primer pair was then selected for each locus (Table S1). Recovered DNA was diluted 1:100 with ddH20 and 20 ml reactions prepared using DyNAmo SYBR Green qPCR 26 master mix (NEB, F400L) and ROX Reference Dye for Quantitative PCR (Sigma, R4526) were assembled in triplicate. Reactions were performed using an Applied Biosystems 7300 Real-Time PCR thermocycler (ABI, Foster City, CA) and DNA enrichments calculated by standard delta-delta Ct.

Quasi-monoclonal mouse primary B cells
Spleens from six healthy quasi-monoclonal mice (created by [29]) were removed aseptically, placed in PBS and gently ground between frosted slides to produce a single-cell suspension. The suspension was centrifuged at 3006g for 5 minutes, the pellet was resuspended in ammonium-tris-chloride buffer to hypotonically lyse erythrocytes, and the remaining cells were washed with PBS by centrifugation at 3006g for 5 minutes. Splenocytes were cultured at 37uC/5% CO 2 /95% humidity in 75 cm 2 flasks in RPMI 1640 (Life Technologies) supplemented with 100 U/ml of penicillin, 100 mg/ml of streptomycin and 10% FBS. Cultures were allowed to recover for 1 hour then supplemented with 25 mg/ml LPS (Sigma, L2143) and incubated for 72 hours to induce proliferation. Active Ig gene diversification was monitored by ELISA of the RPMI media for the presence of IgG, a result of LPS induced class switch recombination, as described [40,43] and by ChIPs to show AID association with the 17.2.25 locus ( Figure 4B). Cells were collected after 72 hours and immediately crosslinked with formaldehyde for ChIP analysis as described.

Confocal immunofluorescence microscopy
Hypermutating Ramos cells were cultured in suspension at 37uC in RPMI (10% (v/v) FBS, 1% (v/v) PS, 2 mM L-glutamine) in 75 cm 2 flasks. At ,1610 6 cells/ml, 750 mL of Ramos suspensions were transferred into individual poly-lysine/ConA-coated chambered slide wells and incubated at 37uC for 1 hour to allow adhesion. Cells were rinsed twice with PBS then fixed with 2% paraformaldehyde for 20 minutes. Cells were again rinsed twice with PBS then blocked with for 30 minutes (PBS, 1% Goat serum, 0.2% TritonX-100). Cells were next incubated with primary antibodies diluted 1:1000 in PBTG (PBS, 1% Goat serum, 0.2% TritonX-100) for 2 hours at room temperature. Cells were rinsed 3 times with PBTG then incubated in the dark for 2 hours at room temperature with secondary antibodies diluted 1:1000 in PBTG. Cells were again rinsed 3 times with PBTG then mounted with vectashield (Vector Laboratories, Burlingame, CA, H1000). Confocal images were obtained using a Leica TCS SP2 Confocal Microscope and contrast processed with Adobe Photoshop 7.0 software. Primary antibodies included: anti-mUb-H2B (Millipore, no. 05-1312) and anti-AID (Santa Cruz, sc-25620). Anti-mUb-H2A (Millipore, 05-678) was also examined but produced no discernable immunofluorescence microscopy signal. Secondary Abs were goat anti-mouse conjugated to Alexa 555 and goat antirabbit conjugated to Alexa 488 (1:1000) (Invitrogen). DAPI (1:1000) Alexa 633-wheat germ agglutinin (1:1000) was used to stain nuclear envelopes (Invitrogen). Figure S1 PCNA and polymerase eta are associated with the Ramos re-VH region. Ramos cells were crosslinked, and sheared chromatin subjected to immunoprecipitation using antibody specific to PCNA, MSH2 and polymerase eta, and non-specific IgG antibody. Precipitated template DNA was amplified by qPCR using primers specific to a control glutamine tRNA locus (tRNA), unrearranged VH (V8), and the rearranged VH (VH2)(see Figure 1). Enrichment using poly-specific IgG for each locus was subtracted to normalize for non-specific IP. Fold enrichment relative to input DNA is shown. Found at: doi:10.1371/journal.pone.0011641.s001 (3.87 MB TIF) Figure S2 HEK293 and A549 ChIPs using AID, mUb-H2A and mUb-H2B antibodies find no enrichment of sequences undergoing hypermutation in Ramos cells. Normalized template enrichment upon anti-AID, anti-mUb-H2A and anti-mUb-H2B ChIPs. Template obtained from IPs was amplified and analyzed by qPCR using primer sets diagramed in Figure 1. Results were normalized to non-specific IgG IP and relative to input. Each bar indicates the mean of the values obtained in triplicate with standard deviation.