Binding of AID to DNA Does Not Correlate with Mutator Activity

This article is featured in In This Issue, p.1

Activation-induced cytidine deaminase (AID) is essential for somatic hypermutation (SHM) and class switch recombination (CSR) (1, 2). During SHM, AID deaminates deoxycytidines to deoxyuridines at the V region exons (V gene) of the Ig H and L chains (3). Engagement of base excision repair and mismatch repair pathways, along with DNA synthesis by error-prone DNA polymerases at the deoxycytidine:deoxyuridine mismatch, mutates the V genes at a high rate (∼10⁻²–10⁻³ mutations/bp/generation), leading to selection of B cells with increased Ag affinity (4). CSR exchanges the initially expressed Igh C region Cμ for an alternative set of downstream C_H exons (or genes), such as Cγ, Cε, or Cα, altering the B cell expression from IgM to a secondary Ab isotype (IgG, IgE, IgA) with distinct effector function (5). CSR is a deletional-recombination reaction that is initiated by AID-mediated deamination of transcribed, repetitive switch (S) region DNA elements that precede each C_H gene (5). End-joining of double strand breaks (DSBs) between two distinct S regions deletes the intervening DNA as an extrachromosomal circle and juxtaposes a new C_H gene downstream of the rearranged VDJ segment. Thus, CSR allows for the generation of Ig molecules with the same affinity for Ag but with new effector function.

AID is a general mutator and can mutate and induce DSBs at many non-Ig genes (6–11). In fact, aberrant AID activity at oncogenes is a major contributing factor in the ontogeny of a large number of mature B cell lymphomas (12). Despite the ability of AID to target non-Ig genes, the V genes and S region DNA serve as major AID targets, with the efficiency of AID association at the Ig loci being several fold higher than at non-Ig genes (7, 8). In addition to specificity of AID for the Ig loci, there is evidence for intra-Ig locus specificity, because B cells undergoing CSR in culture do not mutate their variable regions (13, 14). Thus, mechanisms must exist to actively recruit AID to V genes and S regions during SHM and CSR, respectively. Several factors, including Spt5, polypyrimidine tract binding protein-2 (Ptbp2), RNA exosome subunits and 14-3-3 adapter proteins, have been implicated in the recruitment of AID to S regions (7, 15–17), although the precise role of these proteins in CSR is yet to be fully elucidated. The mechanism by which AID is specifically recruited to V genes is even more enigmatic. Unlike S regions that are unique in their G:C richness and in their ability to form RNA:DNA hybrid structures (R-loops) upon transcription (18, 19), V genes do not present a recognizable primary or predicted secondary structure that could explain specificity for AID binding. The RGYW (R = A/G, Y = C/T, W = A/T) tetranucleotide does serve as an SHM hot-spot motif, and E2A-transcription factor binding sites promote SHM (6, 20); however, the ubiquitous nature of these sequences at almost all transcribed genes fails to explain AID specificity.

Previously, we identified Ptbp2 as an AID interactor (15). Depletion of Ptbp2 significantly impaired CSR as a result of a defect in the recruitment of AID to S regions. In this study, we used the B lymphoma cell line CH12 to show that when AID recruitment to S regions is impaired through Ptbp2 depletion, association of AID with the expressed V gene is remarkably promoted. Surprisingly, despite the binding of AID to V genes, SHM is not induced. Therefore, AID binding does not correlate with mutation activity, suggesting that SHM requires specific factors and/or subversion of DNA repair pathways that operate downstream of AID binding.

CH12 cells (21) were stimulated at a density of 0.25 × 10⁶ cells/ml for 96 h with anti-CD40 Ab (CIT) (1 μg/ml; HM40-3; eBioscience, San Diego, CA), IL-4 (12.5 μg/ml; 404-ML; R&D Systems, Minneapolis, MN), and TGF-β1 (0.1 ng/ml; 240-B; R&D Systems). IgA⁺ cells were generated from CIT-stimulated CH12 cells by negative selection with anti-IgM MicroBeads (Miltenyi Biotec, San Diego, CA), followed by positive selection using anti-IgA biotin Ab (eBioscience) and Streptavidin MicroBeads (Miltenyi Biotec). The predominantly IgA⁺ cell population was subcloned by serial dilution.

Cells were stained with IgM–PE-Cy 7 (R6-60.2; BD Biosciences, San Jose, CA) and IgA-FITC (C10-3; BD Biosciences) and acquired on an LSR II (BD Biosciences). Data were analyzed using FlowJo software (TreeStar, Ashland, OR). DAPI (Invitrogen, Grand Island, NY) was used for exclusion of dead cells. Whole-cell lysates were prepared in Nonidet P-40 lysis buffer (20 mM Tris [pH 7.5], 5% [v/v] glycerol, 150 mM NaCl, 5 mM 2-ME, and 0.5% [v/v] Nonidet P-40). The following primary Abs were used: anti-Ptbp2 (ab57619; Abcam, Cambridge, MA), anti-AID (22), and anti-Gapdh (loading control; 6C5; Millipore, Billerica, MA).

Knockdown in CH12 cells was described previously (15). Chromatin immunoprecipitation (ChIP) assays were performed as described previously (15, 23). PCR amplification and detection were carried out on a Bio-Rad CFX96 system, and threshold cycle (Ct) values were calculated using CFX Manager software by setting the threshold within the linear phase using a log-amplification plot. For detection of specific DNA sequences in ChIP samples, iQ SYBR Green Supermix (Bio-Rad, Hercules, CA) was used for Sμ and p53 amplifications, whereas FastStart Universal Probe Mix was used for VDJ amplification along with FAM-labeled Universal Probe #45 (both from Roche, Indianapolis, IN). Melt-curve analysis was used for SYBR reactions to verify the presence of a single amplicon of the correct size. ChIP “relative units” were calculated by first normalizing the Ct value of each Ab sample by the Ct value of the input sample (Ct_Ab/Ct_input). The inverse of this normalized Ab Ct value was then taken (1/NormCt_Ab). The resulting value for the nonspecific IgG Ab sample was then subtracted from each of the specific Ab (SA) samples to obtain the final “relative units” [(1/NormCt_SA) – (1/NormCt_IgG)]. Primer sequences are listed in Supplemental Table II, and primer positions within the Igh locus are depicted in Supplemental Fig. 1A.

Genomic DNA was isolated from 1.5–3 × 10⁶ CH12 cells, cultured with or without CIT for 96 h. Amplification of sequences for mutation analysis was done using Phusion DNA polymerase (New England Biolabs, Ipswich, MA). PCR products were cloned into a Zero Blunt TOPO cloning kit (Invitrogen). Individual clones were sequenced with vector-derived M13 primers, and mutation frequency was calculated by dividing the total number of unique mutations by the total number of bases sequenced. Mutation data are plotted as a pie chart showing the portion of total sequences containing 0, 1, or 2 mutations. The number of individual clones sequenced is indicated at the center of each chart, and the mutation frequency (unique mutations/bases sequenced) and the number of sequences containing deletions, if applicable, are indicated below. The VDJ sequence expressed by CH12 cells is GenBank M34581.1. Primer sequences are listed in Supplemental Table II, and primer positions within the Igh locus are depicted in Supplemental Fig. 1A.

A two-tailed paired Student t test was used to assess the significance of ChIP data; p ≤ 0.05 was considered significant. A z-test for comparing proportions of two independent groups was used to assess the significance of mutation frequencies; p ≤ 0.05 was considered significant.

Upon stimulation with CIT, the mouse B lymphoma line CH12 expresses AID and undergoes robust CSR to IgA (21, 24) (Fig. 1A, 1B). Given that CSR in CH12 cells is dependent on several of the known factors implicated in CSR of primary B lymphocytes (7, 15, 25), these cells serve as a bona fide model system to elucidate CSR. Despite the high frequency of CSR, CIT-stimulated CH12 cells do not undergo endogenous SHM (26).

To test whether the failure to undergo SHM is due to preferred recruitment of AID to the S regions as opposed to the V gene, we carried out ChIP experiments in CH12 cells with AID Ab, using histone H3 Ab as a positive control. We analyzed the DNA–protein complexes for the presence of the Sμ, Igh V gene (VDJ), and Trp53 (which encodes p53; non-Ig control) genomic sequence using sequence-specific primers (Supplemental Fig. 1A, Supplemental Table II). AID did not associate with Igh regions in unstimulated cells or with p53 under any condition (Supplemental Fig. 1B, 1C, 1F). As expected, AID binding to Sμ was readily observed in stimulated cells; however, no AID was detected at the Igh V gene segment (Fig. 1C, Supplemental Fig. 1F).

One plausible explanation for the observed specificity is that S regions, being rich in RGYW sequence motifs and having the ability to form ssDNA in the context of R-loops, are significantly better than V genes at recruiting AID (18, 19, 27, 28). Therefore, we hypothesized that in “terminally switched” IgA⁺ CH12 cells, which cannot undergo further CSR, the Igh V gene would have a better chance of competing favorably against the smaller recombined S region for AID binding. To test this notion, we purified IgA⁺ cells from CIT-stimulated CH12 cells and subcloned an IgA⁺ clone by serial dilution (Fig. 1A). When grown in the absence of cytokines, the IgA⁺ clone expressed no detectable AID, whereas robust AID expression was induced upon stimulation with CIT (Fig. 1B). As in CH12 cells, AID did not associate with Igh regions in unstimulated IgA⁺ cells or with the Trp53 (p53; non-Ig control) genomic sequence under any condition (Supplemental Fig. 1B, 1C, 1F). However, contrary to our predictions, AID binding to the Igh V gene segment was not detected in CIT-stimulated IgA⁺ cells (Fig. 1C, Supplemental Fig. 1F). Instead, AID was found associated with the recombined Sμ–Sα DNA (Fig. 1C, Supplemental Fig. 1F). Thus, AID exhibits intra-Igh locus specificity with a bias for binding to S regions, not only in bulk CH12 cells undergoing CSR, but also in IgA⁺ cells in which the Igh V gene might be expected to have a better chance of competing favorably against the remaining S region for AID binding.

To test the possibility that the failure to detect AID at the V gene is not due to the transient nature of the interaction, we isolated genomic DNA from IgA⁺ cells and sequenced the expressed Igh V gene using sequence-specific primers (Supplemental Fig. 1A, Supplemental Table II). The rate of mutation of the Igh V gene (VDJ) in CIT-stimulated IgA⁺ cells was similar to that observed for unstimulated IgA⁺ cells, suggesting that SHM was not induced (Fig. 2A, Supplemental Table IA). In contrast, the recombined Sμ–Sα sequence in CIT-stimulated IgA⁺ cells had significant levels of mutations (Fig. 2B, Supplemental Table IB). Furthermore, the original recombined Sμ–Sα junction present in the unstimulated IgA⁺ cell line underwent deletions to create novel S junctions (Fig. 2B, 2C). Thus, although AID in IgA⁺ cells bound to and mutated the remaining S region, it neither associated with nor induced SHM at the Igh V gene.

The RNA-binding protein Ptbp2 interacts with AID and facilitates the recruitment of AID to S regions (15). We investigated the possibility that, in Ptbp2-depleted cells, AID that fails to bind to S regions can be “retargeted” to the V gene segment. Using a short hairpin RNA (shRNA) directed against the 3′ untranslated region of Ptbp2 mRNA, we knocked down Ptbp2 expression in CH12 cells, with “scrambled” shRNA serving as a control (Fig. 3A). As expected (15), AID binding to S regions was significantly reduced (∼2-fold, p = 0.003), and CSR was impaired (Fig. 3B, 3C, Supplemental Fig. 1G). Strikingly, the reduction in AID at S regions was accompanied by a significant increase (>6-fold, p = 0.01) in AID binding specifically to the Igh V gene of stimulated Ptbp2-depleted cells (Fig. 3C, Supplemental Fig. 1D, 1G). AID was not substantially enriched at other regions of the Igh locus, including 1 kb upstream and downstream of the rearranged VDJ (5′Vh1-53 and 3′Jh2, respectively), Cμ, and the Iμ promoter (Fig. 3D). Surprisingly, AID was not detected at the expressed VJκ L chain locus (Fig. 3D). This could be due to occlusion of the Ab-binding epitope while AID is bound to the L chain locus or, more interestingly, a suggestion that the level of Ptbp2 preferentially effects AID relocalization within the Igh locus during SHM; thus, only the binding of AID to the S region and Igh V gene exons is altered upon Ptbp2 depletion. Finally, AID was not found to be associated with control genomic sequences, such as Trp53 (p53) or other non-Ig genes that were shown to be upregulated in B cells undergoing CSR (23) (Fig. 3E). Thus, when the association of AID with S regions is impaired, its interaction with the expressed Igh V gene segment is specifically and significantly promoted.

AID is phosphorylated at serine 38 (S38), and mutation of S38 to alanine impairs the ability of AID to mediate SHM and CSR (23, 29–33). To determine whether AID bound to the V gene was phosphorylated at S38 (pS38-AID), we carried out ChIP experiments using pS38-AID–specific Ab (23). In accordance with AID bound to Sμ, the amount of pS38-AID localized to Sμ was reduced in Ptbp2-depleted CH12 cells (Fig. 4A). This reduction was accompanied by a significant increase (∼7-fold, p = 0.05) in pS38-AID levels specifically associated with the Igh V gene (Fig. 4A). Localization of pS38-AID to the Trp53 (p53; non-Ig control) genomic sequence was not amplified from pS38-AID samples (Supplemental Fig. 1E). Thus, AID is not only specifically targeted to the V gene segment in Ptbp2-depleted CH12 cells, it is phosphorylated at S38.

AID phosphorylated at S38 interacts with the ssDNA-binding protein replication protein A (RPA) (29, 30). To test whether RPA localization also is altered in Ptbp2-depleted cells, we carried out ChIP analyses using an Ab specific for the 32-kDa subunit of RPA. RPA levels at Sμ were significantly reduced (∼2.5-fold, p = 0.03) in Ptbp2-depleted CH12 cells (Fig. 4A). In contrast, there was a significant increase (∼3-fold, p = 0.01) in RPA localization specifically to the Igh V gene segment (Fig. 4A, Supplemental Fig. 1E).

Phosphorylation of AID at S38 is mediated by protein kinase A (PKA) (23, 30, 31, 34), and it is believed that AID is phosphorylated by S region–bound PKA to activate the CSR cascade (23, 35). In keeping with AID-independent recruitment of PKA to S regions (23), the catalytic subunit of PKA (PKA-Cα) was detected at Sμ in CIT-stimulated Ptbp2-depleted cells (Fig. 4B). However, PKA-Cα was not detected at the Igh V gene segment (Fig. 4B), even though phosphorylated AID was readily detectable (Fig. 4A). Thus, not all proteins known to bind S regions are retargeted to the V gene in the absence of Ptbp2, thereby ruling out the possibility that the binding of AID to the Igh V gene is due to general deregulation of protein–DNA associations in Ptbp2-depleted CH12 cells.

Our findings that PKA-Cα is associated with S regions, but not with the Igh V gene, regardless of Ptbp2 expression, lends credence to the proposal (23) that AID phosphorylation has a different role in SHM than in CSR. It is generally believed that transcribed S regions form R-loops, allowing AID to access S regions independent of its phosphorylation status (36). Phosphorylation of AID at S38 is still required to promote formation of DSBs at S regions through interaction with APE1 (35), as well as for the repair of DSBs through recruitment of RPA to S regions (8, 23). In contrast, transcribed V genes do not readily reveal ssDNA in the context of R-loops, and it is likely that AID only binds V genes in the context of a pS38–AID–RPA complex, because of the ability of RPA to bind and stabilize ssDNA within transcription bubbles (29). Our findings that pS38-AID and RPA, but not PKA-Cα, are detected at the Igh V gene support the notion that AID binding to V genes requires prior phosphorylation. However, we cannot exclude the possibilities that PKA-Cα was not detected at the V gene segment as a result of the transient nature of the interaction or that other proteins bound to the V gene segment mask the Ab binding site of V gene–bound PKA-Cα. Additionally, we cannot rule out the possibility that AID bound to the V region exons is not phosphorylated by PKA but is instead modified by an unidentified kinase.

To determine whether AID binding induced SHM, we sequenced the Igh V gene (VDJ). Surprisingly, the mutation frequency in Ptbp2-depleted cells stimulated for 96 h was similar to that in scrambled cells (Fig. 5A, Supplemental Table IC). Moreover, Ptbp2-depleted cells stimulated continuously for 3 wk did not accumulate additional mutations (Supplemental Table IC). Although the mutation frequency was higher than in unstimulated cells (Supplemental Table IC), it was still considerably lower than typical SHM (32, 33, 37). The absence of SHM activity in Ptbp2-depleted cells cannot be attributed to a deficiency in Igh V gene (VDJ) transcription, because transcription through this region was not markedly altered (<2-fold change) upon Ptbp2 depletion (Supplemental Fig. 2A). Most importantly, there was no correlation between the levels of AID bound to the V gene and SHM frequency (Figs. 3C, 5A). In contrast, the region 5′ of the Sμ-repetitive region (34) was highly mutated (Fig. 5B, Supplemental Table ID). A majority of the mutations in this region were transition mutations, and all mutations occurred at C:G bp (Fig. 5C). This spectrum suggests that mutations in CH12 cells arise mainly from replication across deaminated residues, with a small portion repaired via base-excision repair engagement. The lack of mutation at A:T bases could indicate that the mismatch repair pathway is either not active or is not recruited to the Igh locus in CH12 cells (38). Additionally, there was a marked lack of G to T mutations in scrambled cells, whereas Ptbp2-depleted cells showed both G to A and G to T mutations. The reason behind this intriguing observation is not clear, but it is possible that alterations in AID and/or Ptbp2 at S regions could modulate the engagement of error-prone polymerases during CSR.

Many of the mutations identified in the region 5′ of the Sμ-repetitive region and V gene of stimulated scrambled and Ptbp2-depleted cells were in RGYW hot-spot motifs (Supplemental Fig. 2B, Supplemental Table IE). Of the total bases sequenced for mutation analysis, the amount of available RGYW motifs in the Igh V gene (VDJ) is ∼61% more than the amount available in the region 5′ of the Sμ-repetitive region (based on numbers listed in Supplemental Table IC–E); however, the percentage of RGYW motifs mutated in the region 5′ of the Sμ-repetitive region is much higher than in the Igh V gene (VDJ) (Supplemental Fig. 2B, Supplemental Table IE). Consequently, if the RGYW motifs found in the Igh V gene were being mutated at the same frequency as the RGYW motifs in the region 5′ of the Sμ-repetitive region, the number of RGYW motifs expected to be mutated at the Igh V gene would be significantly higher (Supplemental Fig. 2C, Supplemental Table IE). Thus, the lack of detectable SHM in Ptbp2-depleted CH12 cells is not due to a deficiency in the amount of available hot-spot motifs in this region; rather, it is likely due to the absence in expression of required repair mediators or uncharacterized SHM-specific factors in CH12 cells (Fig. 6). In conclusion, although known mediators of SHM, namely AID and RPA, were abundant at the Igh V gene, and the bound AID was phosphorylated at S38, SHM was not induced.

Our results clearly demonstrate that S regions are preferred targets for AID binding. Even in cells that have undergone CSR to the last C_H gene, AID is efficiently recruited to the remaining S region in lieu of the transcribed Igh V gene. This strong preference for S regions might represent a physiological mechanism for AID regulation, wherein S regions act as a “sink” to prevent AID from interacting with non-Ig genomic sequences. Being noncoding, S regions could sustain mutations and deletions without having any deleterious effect on cell viability. This is evident from the abundant mutations and deletions observed at Sμ–Sα junctions upon re-expression of AID in IgA⁺ cells. Thus, active recruitment of AID to S regions probably functions as a default pathway to sequester AID activity away from other genomic targets.

If AID expressed in a B cell is indeed actively recruited to the S regions, the question that follows is how AID is targeted to the V genes to initiate SHM. Our studies clearly demonstrate that reduction in Ptbp2 levels promotes binding of AID to Igh V gene segments. Thus, in a B cell undergoing SHM, modulating Ptbp2 expression or perturbing the interaction between AID and Ptbp2 through posttranslational modifications of AID and/or Ptbp2 could increase the amount of non-S region–bound AID (Fig. 6). The relative expression of Ptbp2 in germinal center B cells undergoing SHM versus CSR will shed further light onto the differential role of this protein during an immune response.

The most striking finding is that, despite binding to the Igh V gene, AID cannot induce SHM. Over the past few years, several genome-wide association analyses of AID reported that it has the potential to bind to other genomic regions (7, 11, 39). Our findings clearly enforce the notion that binding of AID is not synonymous with mutations. For SHM to take place, at least in the context of CIT-stimulated CH12 cells, additional steps need to occur beyond AID binding. It is feasible that CH12 cells lack a factor(s) required for SHM and/or that CIT induction is not sufficient to induce the expression of all required proteins. Another likely possibility is that AID targeted to the Igh V gene in Ptbp2-depleted cells is indeed actively deaminating, but high-fidelity repair (6), as opposed to error-prone repair required for SHM, “fixes” the lesions in a nonmutagenic fashion. Under physiological conditions, SHM is associated with germinal center structures found in secondary lymphoid organs, where the interaction of B cells with CD4⁺ Th cells (40) could induce SHM-promoting factors or subversion of high-fidelity repair pathways that are absent from the in vitro culture system described in this study.

In conclusion, the data presented in this article intriguingly suggest that targeting of AID to S regions is the default mechanism in a CSR-sufficient B cell line and may indicate a role for S regions as an AID “sink,” sequestering AID away from the rest of the genome. Our findings identify Ptbp2 as a crucial mediator between S region and Igh V gene AID targeting and show that AID localization does not determine AID deamination activity or the induction of mutation. Additionally, we describe a cell line–based model system that will be invaluable in further elucidating factors/conditions required for SHM versus CSR.

We thank Tasaku Honjo (Kyoto University, Kyoto, Japan) for providing the CH12 cell line and David Schatz (Yale University, New Haven, CT) and members of the Chaudhuri laboratory for helpful discussions and suggestions.

The authors have no financial conflicts of interest.