AID Phosphorylation Regulates Mismatch Repair–Dependent Class Switch Recombination and Affinity Maturation

B cells undergo three genetically programmed somatic DNA recombination and mutation events to diversify their Ig genes: V(D)J recombination, class switch recombination (CSR), and somatic hypermutation (SHM). V(D)J recombination assembles the exons encoding the V regions of the IgH and L chain genes (1). CSR generates secondary Ig isotypes by replacing the IgH constant coding exon (C_H) through a DNA deletional recombination reaction. The generation and subsequent ligation of DNA double-strand breaks (DSBs) in the recombining switch (S) regions that precede each C_H exon exchanges the default Cμ, which codes for the IgM constant region, for another downstream C_H exon (2–6). SHM introduces untemplated substitution mutations into V genes of both IgH and IgL chains (10⁻³–10⁻⁴ per bp per generation) to promote Ig affinity maturation (7).

CSR and SHM require activation-induced cytidine deaminase (AID) (8, 9), which generates deoxyuridines (U) in S or V regions that are processed either into DSBs or mutations by the base excision repair (BER) or mismatch repair (MMR) pathway (10). During BER, uracil DNA glycosylase (UNG) recognizes and removes the uracil base to create an abasic site, which is cleaved by apurinic/apyrimidinic endonuclease 1 (APE1) to generate a single-strand break (SSB). In MMR, a heterodimer of mutS homolog 2 (MSH2) and mutS homolog 6 (MSH6) recognizes the U:G mismatch and recruits a complex of mutL homology 1 (MLH1) and PMS1 homolog 2 (PMS2) to generate a nick 5′ of the U:G mismatch. Exonuclease I (EXO1) subsequently degrades the nicked DNA to produce SSBs. Staggered SSBs generated by MMR or BER constitute the DSBs necessary for CSR. In the absence of MSH2 and UNG (MSH2^−/−UNG^−/−), both MMR and BER pathways are disrupted and CSR is completely blocked (11).

In V regions undergoing SHM, AID-generated U:G mismatches can be replicated to generate C:G to T:A transition mutations, which are observed in MSH2^−/−UNG^−/− B cells (11). Removal of the uracil base from the U:G mismatch prior to DNA replication leaves an abasic site, which can be converted to transition or transversion mutations at C:G bp following replication (11). Alternatively, mutagenic repair of the U:G mismatch by MMR generates mutations at A:T bp. Mice deficient in both MSH2 and Polη, the polymerase used in MMR, have a complete loss of mutations at A:T bp (12).

AID activity is regulated by phosphorylation at serine 38 (S38) by PKA (cAMP-dependent protein kinase A) (13–15). Mice with a homozygous knock-in mutation of the phosphorylation site (AID^S38A/S38A), which changes S38 to an alanine, have significantly reduced CSR and SHM (16, 17), even though the mutant AID(S38A) enzyme retains wild-type (WT) deaminase activity (14, 18). Similarly, B cells from mice carrying a hypomorphic mutation in the PKA regulatory subunit (RIα) display significantly reduced CSR (18). Phosphorylation of AID mediates its interaction with APE1 (19), suggesting that phosphorylation of AID at serine 38 (p-AID) promotes BER-dependent CSR. To determine if p-AID regulates MMR- or BER-dependent CSR and SHM, we bred AID^S38A/S38A mice onto either an MSH2^−/− or UNG^−/− background. As expected, AID^S38A/S38AMSH2^−/− B cells have a complete block in CSR and reduced mutation frequency in 5′Sμ due to the absence of p-AID and MSH2. Similarly, almost no mutations are observed in the JH4 intron of AID^S38A/S38AMSH2^−/− B cells, which correlates with a loss of affinity maturation. Interestingly, although AID^S38A/S38AUNG^−/− B cells can generate mutations in S regions, AID^S38A/S38AUNG^−/− mice cannot produce secondary Ig isotypes, indicating the absence of CSR. Although JH4 intron mutations are observed in AID^S38A/S38AUNG^−/− B cells, no significant affinity maturation is observed. These results implicate a role for p-AID in MMR-dependent CSR and affinity maturation.

C57BL/6 mice were purchased from The Jackson Laboratory. AID^S38A/S38A mice were generated as described (16), UNG^−/− mice were a gift from T. Lindahl (20), MSH2^−/− were a gift from H. te Riele (21), and AID^−/− were a gift from T. Honjo (8). All mutant mice were maintained on a C57BL/6 background. Age-matched male and female mice were used for all experiments. Husbandry of and experiments with mice were conducted according to protocols approved by The City College of New York Institutional Animal Care and Use Committee.

B cells were purified from the spleens of mice (aged 2–5 mo) through negative selection with anti-CD43 magnetic beads (Miltenyi Biotec) and cultured in RPMI 1640 media (Life Technologies) supplemented with 10% FBS (Atlanta Biologicals), 2 mM l-glutamine, 1× penicillin/streptomycin (Corning), and 47.3 μM 2-mercaptoethanol. Cells were stimulated in vitro with 25 μg/ml LPS (catalog no. L7261; Sigma-Aldrich) and 12.5 ng/ml IL-4 (R&D Systems) for class switching to IgG1; 25 μg/ml LPS for class switching to IgG3; or 300 ng/ml anti–IgD-dextran (Fina Biosolutions), 2 ng/ml TGF-β1 (R&D Systems), and 10 μg/ml LPS for class switching to IgA. At 48 and 72 h poststimulation, the cells were passaged 1:2 with fresh stimulation media. After 96 h, the stimulated cells were analyzed by flow cytometry using anti-IgG1 conjugated to allophycocyanin (Becton Dickinson), anti-IgG3 conjugated to FITC (Becton Dickinson), or anti-IgA conjugated to FITC (Becton Dickinson). DAPI was used to distinguish live and dead cells.

Mice (aged 2–4 mo) were injected i.p. with 50 μg of 4-hydroxy-3-nitrophenylacetyl conjugated to chicken gamma globulin (NP-CGG) (Biosearch Technologies) in Imject Alum (Thermo Fisher Scientific) (22). Immunized females were housed together, immunized male littermates were housed together, and immunized nonlittermate males were housed individually. Mice were given a boost of NP-CGG in Imject Alum 10 d post–primary immunization (day 10). Sera were obtained from blood that was collected on day 0 (preimmunization) and day 7 and day 21 postimmunization through cheek bleeds or cardiac puncture (day 21). Total Ig titers were analyzed by ELISA on anti-IgM (2 μg/ml, 1020-01; SouthernBiotech), anti-IgG1 (2 μg/ml, 1070-01; SouthernBiotech), or anti-IgG3 (2 μg/ml, 1100-01; SouthernBiotech) coated plates, which were detected using anti-IgM-HRP (1:5000, 1020-05; SouthernBiotech), anti-IgG1-HRP (1:5000, 1070-05; SouthernBiotech), or anti-IgG3-HRP (1:5000, 1100-05; SouthernBiotech), respectively. Coating Abs were diluted in PBS pH 8. Sera and detection Abs were diluted in 1% BSA in borate buffer (100 mM boric acid, 25 mM sodium borate, and 75 mM sodium chloride). NP-specific IgM titers were analyzed on plates coated with NP7-BSA or NP20-BSA (3 μg/ml) in borate buffer and detected with anti-IgM-HRP. ELISA plates were read on SpectraMAX190 at 450 nm. To determine affinity maturation, the ratio of NP7- and NP20-specific IgM concentrations was calculated, and this ratio was normalized to the day 0 NP20 absorbance of each animal to account for variations in basal IgM titers (23).

To analyze mutations in 5′Sμ and the JH4 intron, germinal center (GC) B cells (GCBCs) were sorted from the Peyer's patches of mice (aged 2–3 mo) at day 10 postimmunization with 50 μg of NP-CGG in Imject Alum. GCBCs were sorted after staining with PNA-biotin (biotinylated peanut agglutinin; catalog no. B-1075-5; Vector Labs) followed by strepavidin-allophycocyanin-Cy7 (catalog no. 47-4317-82; Thermo Fisher Scientific) and anti-B220-allophycocyanin (catalog no. 553092; BD Pharmingen). DAPI was used for live/dead cell exclusion. Sorted cells were lysed in lysis buffer (10 mM Tris, 0.1 M EDTA, and 0.5% SDS), and genomic DNA was isolated by isopropanol precipitation. 5′Sμ was amplified using the following primers: Su5, 5′-GCGGCCCAGCTCATTCCAGTTCATTACAG-3′; and Su3, 5′-AATGGATACCTCAGTGGTTTTTAATGGTGGGTTTA-3′. JH4 intron was amplified by a nested PCR (16, 17). The primers for the first PCR were as follows: J558FR3Fw, 5′-GCCTGACATCTGAGGACTCTGC-3′; and VHJ558.2, 5′-CTGGACTTTCGGTTTGGTG-3′. The primers for the second PCR were as follows: VHJ558.3, 5′-GGTCAAGGAACCTCAGTCA-3′; and VHJ558.4, 5′-TCTCTAGACAGCAACTAC-3′. To analyze mutations within V_H186.2, RNA was extracted from Peyer's patch cells of day 10 NP-CGG–immunized mice (aged 2–3 mo). cDNA was synthesized using the SuperScript III first-strand synthesis kit (Invitrogen). V_H186.2 was amplified using a nested PCR as previously described (23). All PCRs were performed with Q5 high-fidelity DNA polymerase (New England Biolabs). Amplicons were purified, ligated into pJET (Thermo Fisher Scientific), and analyzed by Sanger sequencing (Genewiz). Sequences were analyzed with MegAlign (Lasergene). Reference sequences were as follows: National Center for Biotechnology Information Reference No. NG_005838 for 5′Sμ and JH4 intron and musIGHV057 on VBASE2 for V_H186.2. Mutation hotspots were identified as RGYW/WRCY motifs (24, 25).

Unless otherwise noted, a two-tailed Student t test was used for statistical analysis.

Phosphorylation of AID at serine 38 (S38) is required for WT levels of CSR. AID^−/− B cells cannot complete CSR in vitro or in vivo (8), whereas AID^S38A/S38A B cells, which carry a serine 38 to alanine knock-in mutation of AID, complete CSR at ∼25% of WT levels (16, 17) (Fig. 1A, 1B). Phosphorylated AID promotes CSR in part through its interaction with APE1 (19), suggesting that the residual CSR observed in AID^S38A/S38A B cells is mediated by the MMR pathway (11). To confirm this, we generated AID^S38A/S38AMSH2^−/− mice and examined CSR to IgG1, IgG3, or IgA (Fig. 1A, 1B, Supplemental Fig. 1). As previously shown, AID^S38A/S38A and MSH2^−/− B cells displayed a hypomorphic CSR phenotype (11, 16, 17), whereas the double-mutant AID^S38A/S38AMSH2^−/− B cells had a complete block in CSR in vitro and in vivo that was comparable to AID^−/− and MSH2^−/−UNG^−/− B cells (Fig. 1A, 1B). The absence of CSR was not due to reduced AID(S38A) protein expression, μ or γ1 germline transcription, cellular proliferation, or GCBC formation (B220⁺PNA^Hi) (Supplemental Fig. 2A–D). AID^S38A/S38AMSH2^−/− mice produced IgM titers comparable to WT mice postimmunization (Fig. 1B), suggesting that Ig secretion and plasma cell development were not altered. These data demonstrate that p-AID regulates BER-dependent CSR, as the loss of MMR and AID phosphorylation ablates CSR.

To examine whether p-AID modulates MMR-dependent CSR, AID^S38A/S38AUNG^−/− mice were immunized with NP-CGG, and serum Igs were analyzed by ELISA. UNG^−/− B cells do not exhibit measurable levels of CSR in vitro (11), and, consistent with this, AID^S38A/S38AUNG^−/− B cells could not complete CSR in vitro (Fig. 1A). However, UNG^−/− mice have detectable levels of serum IgG1 and IgG3 before and after immunization, indicating that UNG^−/− B cells can undergo CSR in vivo, likely because of compensation by the MMR pathway (11). Surprisingly, AID^S38A/S38AUNG^−/− mice had no detectable serum IgG1 and IgG3 before and after immunization (Fig. 1B), indicating that AID^S38A/S38AUNG^−/− B cells have a block in CSR in vivo, comparable to AID^S38A/S38AMSH2^−/− and MSH2^−/−UNG^−/− B cells. The AID^S38A/S38AUNG^−/− mice produced serum IgM that increased upon immunization with NP-CGG, suggesting that Ig secretion and plasma cell development are not impaired (Fig. 1B, p = 0.0032 for day 0 versus day 21 IgM titers in AID^S38A/S38AUNG^−/− mice). AID(S38A) protein expression, μ or γ1 germline transcription, cellular proliferation, and GCBC formation (B220⁺PNA^Hi) were not significantly different between WT and AID^S38A/S38AUNG^−/− B cells (Supplemental Fig. 2A–D). Interestingly, basal serum IgM titers in AID^S38A/S38AUNG^−/− mice were increased 5-fold relative to WT mice (Fig. 1B, p = 0.0209). Similarly, basal serum IgM titers in AID^S38A/S38AMSH2^−/− mice were increased 3-fold relative to WT mice (Fig. 1B, p = 0.0111). This hyper-IgM phenotype is also observed in MSH2^−/−UNG^−/− mice and suggests that IgM titers increase when CSR is blocked from the lack of MMR and BER. Therefore, the hyper-IgM and absence of CSR in AID^S38A/S38AMSH2^−/− and AID^S38A/S38AUNG^−/− mice demonstrates that BER and MMR, respectively, were inhibited because of the loss of phosphorylated AID.

To determine if the ablation of CSR in AID^S38A/S38AMSH2^−/− and AID^S38A/S38AUNG^−/− B cells resulted from a lack of AID(S38A)-induced mutation of S regions, we sequenced 5′Sμ in B220⁺PNA^hi Peyer's patch GCBCs that were sorted from AID^S38A/S38AMSH2^−/−, AID^S38A/S38AUNG^−/−, or control mice. As previously described (11), WT B cells generated transition and transversion mutations at all bp, MSH2^−/− B cells displayed a reduction in mutations at A/T bp, and UNG^−/− B cells showed an increase in transition mutations at G/C bp (Fig. 2A). Mutations in 5′Sμ of AID^S38A/S38A B cells were reduced as compared with WT B cells, as previously reported (16, 17). Similarly, the number and frequency of mutations in 5′Sμ were lower in AID^S38A/S38AMSH2^−/− and AID^S38A/S38AUNG^−/− B cells as compared with MSH2^−/− and UNG^−/− B cells, respectively (Fig. 2A, 2C). AID^S38A/S38AMSH2^−/− B cells had a detectable number of transition mutations and very few transversion mutations; however, the reduced mutation frequency suggests that the mutations observed in MSH2^−/− B cells are p-AID–dependent and may be mediated by the interaction of AID with RPA or APE1 (19, 26). AID^S38A/S38AUNG^−/− B cells had a higher mutation frequency at 5′Sμ as compared with AID^S38A/S38A B cells (Fig. 2A) but, unlike AID^S38A/S38A B cells, AID^S38A/S38AUNG^−/− B cells could not complete CSR in vivo (Fig. 1B). Thus, the lack of CSR in AID^S38A/S38AUNG^−/− B cells is not due to the inability of AID(S38A) to deaminate and mutate S regions, suggesting that phosphorylation of AID may regulate the generation of DSBs by MMR, the synapsis of S regions, or the ligation of recombining S regions.

GCBCs complete SHM to generate untemplated mutations within recombined V(D)J coding segments to promote Ig affinity maturation. AID initiates SHM by deaminating deoxycytidine (C) residues in variable regions, which can be replicated across to form transition mutations at C:G bp (10). Alternatively, removal of the uracil base by UNG creates an abasic site that can generate both transition and transversion mutations at C:G bp following DNA replication. Mutations at A/T bp are generated by MMR through the activity of the error prone DNA polymerase η (27). To examine the role of AID phosphorylation in BER and MMR during affinity maturation, we analyzed mutations in the JH4 intron in B220⁺PNA^Hi Peyer's patch GCBCs sorted from AID^S38A/S38AMSH2^−/−, AID^S38A/S38AUNG^−/−, or control mice (8, 28). WT B cells had transition and transversion mutations at all bp, which were lost in AID^−/− B cells and reduced in AID^S38A/S38A B cells (Fig. 3A) (8, 16). As previously shown, UNG^−/− B cells displayed increased transition mutations at G/C bp, MSH2^−/− B cells had reduced mutations at A/T bp, and MSH2^−/−UNG^−/− B cells had a complete loss of mutations at A/T bp (20). AID^S38A/S38AMSH2^−/− B cells showed mutations in the JH4 intron at the same frequency as AID^−/− B cells (Fig. 3A–C), which is the error rate of the DNA polymerase that was used to amplify the DNA (8, 9). The absence of mutations in the JH4 intron of AID^S38A/S38AMSH2^−/− B cells contrasts with the observed mutations in 5′Sμ (Fig. 2A–C). This difference may result from the inability of AID(S38A) to deaminate V regions without RPA (26) or increased deamination within S regions, which form R-loops that expose the ssDNA substrates for AID (10). AID^S38A/S38AMSH2^−/− B cells display a V_H186.2 mutation spectrum and frequency comparable to MSH2^−/− B cells (Supplemental Fig. 3), suggesting that mutations 3′ of V coding sequences in MSH2^−/− B cells may be dependent on phosphorylated AID and RPA, which stabilizes V sequences for AID-mediated DNA deamination (26, 29).

In contrast, AID^S38A/S38AUNG^−/− B cells can generate mutations in the JH4 intron (Fig. 3A). As compared with AID^S38A/S38A B cells, the mutation spectra in AID^S38A/S38AUNG^−/− B cells shifted to increased G/C transition mutations (Fig. 3A), which is consistent with the observed increase in G/C transition mutations found in UNG^−/− B cells (20). However, the mutations per sequence and the mutation frequency of AID^S38A/S38AUNG^−/− B cells were lower than WT and UNG^−/− B cells (Fig. 3B, 3C), which highlights the previously described role of p-AID in SHM (16, 17). Notably, the mutation spectra, the number of mutations per sequence, and the mutation frequency per bp in AID^S38A/S38AUNG^−/− B cells mirrored those of MSH2^−/− B cells (Figs. 2, 3). The mutation frequency at 5′Sμ in AID^S38A/S38AUNG^−/− and MSH2^−/− B cells was 1.35 × 10⁻³ and 1.48 × 10⁻³ mutations/bp, respectively (Fig. 2C), and the mutation frequency at the JH4 intron in AID^S38A/S38AUNG^−/− and MSH2^−/− B cells was 1.23 × 10⁻³ and 1.05 × 10⁻³ mutations/bp, respectively (Fig. 3C). Furthermore, the frequency of mutations at A/T bases in both AID^S38A/S38AUNG^−/− and MSH2^−/− B cells was similarly shifted. AID^S38A/S38AUNG^−/− and MSH2^−/− B cells showed a loss of mutations at T bases in 5′Sμ (Fig. 2A) and comparable percentages of transition and transversion mutations at A bases at the JH4 intron (Fig. 3A). AID^S38A/S38AUNG^−/− B cells also mutated V_H186.2 at frequencies comparable to MSH2^−/− B cells (Supplemental Fig. 3A). These data show that the SHM phenotype of the AID^S38A/S38AUNG^−/− B cells closely resembles the SHM phenotype of MSH2^−/− B cells and suggest that the altered SHM pattern in AID^S38A/S38AUNG^−/− B cells results from a disruption of the MMR pathway due to the absence of phosphorylated AID.

To determine if the observed JH4 and V_H186.2 mutations correlate with affinity maturation, we performed NP-specific IgM ELISAs with sera from NP-CGG–immunized AID^S38A/S38AMSH2^−/−, AID^S38A/S38AUNG^−/−, and control mice. To quantify the change in affinity maturation, we first calculated the ratio of IgM that bound to NP7 (high affinity) and NP20 (high and low affinity) preimmunization (day 0) and postimmunization (day 21). Subsequently, the NP7/NP20 ratio was normalized to the preimmunization NP20 IgM titer to account for variations in basal IgM titers, which are elevated in AID^−/−, AID^S38A/S38A, MSH2^−/−UNG^−/−, AID^S38A/S38AMSH2^−/−, and AID^S38A/S38AUNG^−/− mice relative to WT mice (Fig. 1B; p < 0.05) (8, 11). As expected, WT mice, unlike AID^−/− mice, showed NP-specific IgM affinity maturation upon immunization with NP-CGG (Fig. 4; p = 0.006). Affinity maturation also occurred in AID^S38A/S38A, MSH2^−/−, and UNG^−/− mice (Fig. 4; p = 0.04, 0.007, and 0.005, respectively) (30). Consistent with the observed reduction in mutations at JH4 and V_H186.2, AID^S38A/S38AMSH2^−/− did not demonstrate significant NP-specific IgM affinity maturation (p = 0.086). Surprisingly, the moderate mutation frequency in the JH4 intron and V_H186.2 of AID^S38A/S38AUNG^−/− mice did not translate into significant affinity maturation (p = 0.105), even though these mice had a JH4 and V_H186.2 mutation frequency and pattern that was comparable to MSH2^−/− mice, which exhibit affinity maturation (Figs. 3, 4, Supplemental Fig. 3). This observation is similar to the high mutation frequency in the 5′Sμ of AID^S38A/S38AUNG^−/− B cells that did not lead to CSR, suggesting that p-AID is required for the recruitment of pathways downstream of AID-dependent deamination or mutation to drive affinity maturation or CSR.

AID frequently deaminates C within RGYW/WRCY hotspot motifs in V and S regions (31). To determine if the reduced affinity maturation in AID^S38A/S38AMSH2^−/− and AID^S38A/S38AUNG^−/− mice may be due to altered mutations at hotspot or nonhotspot motifs by AID(S38A), we cataloged the location and frequency of mutations at hotspot motifs in the JH4 intron and 5′Sμ for each genotype (Fig. 5). Compared with WT B cells, the percentage of hotspot mutations was not reduced in the genotypes that show mutations above the background levels observed in AID^−/− B cells (i.e., AID^S38A/S38A, MSH2^−/−, UNG^−/−, MSH2^−/−UNG^−/−, and AID^S38A/S38AUNG^−/−). Interestingly, MSH2^−/−, MSH2^−/−UNG^−/−, and AID^S38A/S38AMSH2^−/− B cells have more hotspot than nonhotspot mutations compared with WT, UNG^−/−, and AID^S38A/S38A B cells, respectively (Fig. 5C), consistent with the known role for MMR in generating nonhotspot mutations downstream of AID-induced deamination (32, 33). A similar shift toward hotspot mutations was observed for UNG^−/− and AID^S38A/S38AUNG^−/− B cells in 5′Sμ (Fig. 5C). However, at the JH4 intron, AID^S38A/S38AUNG^−/− and WT B cells had comparable percentages of hotspot and nonhotspot mutations, consistent with the known role for MSH2-dependent MMR in generating nonhotspot mutations at the V region (32, 33). Furthermore, the absence of affinity maturation in AID^S38A/S38AUNG^−/− B cells was likely independent of DNA deamination by AID(S38A) because WT, AID^S38A/S38A, and AID^S38A/S38AUNG^−/− B cells showed comparable percentages of JH4 hotspot mutations (Fig. 5C). Thus, the decreased frequency of affinity maturation and CSR in AID^S38A/S38AUNG^−/− B cells is likely due to altered processing of the AID-generated mutations by BER or MMR rather than altered AID localization to V or S regions or reduced AID(S38A) deamination activity in vivo.

In this study, we describe additional data that implicate a role for phosphorylated AID in directing BER via APE1 toward CSR and affinity maturation (19). In support of our hypothesis, mice deficient for AID phosphorylation and MMR (AID^S38A/S38AMSH2^−/−) cannot complete CSR both in vitro and in vivo and have a severe decrease in SHM. Furthermore, we have uncovered a previously unknown role for AID phosphorylation in promoting MMR-dependent CSR. B cells in mice deficient for AID phosphorylation and BER (AID^S38A/S38AUNG^−/−) can mutate V and S regions but cannot complete affinity maturation or CSR. These data suggest that, in the absence of UNG, p-AID is required for MMR-mediated processing of deaminated DNA in SHM and CSR.

We propose a model in which phosphorylation of AID directs BER- or MMR-dependent CSR or SHM in the absence of the complementary pathway (11). The lack of CSR and reduced SHM in AID^S38A/S38AMSH2^−/− B cells suggests that p-AID and its interaction with APE1 complement MMR-dependent CSR and SHM (Figs. 1, 2) (19). Similarly, the absence of AID phosphorylation and BER in AID^S38A/S38AUNG^−/− B cells completely blocks CSR, suggesting that p-AID regulates the recruitment of MMR proteins to generate DSBs, synapsis of recombining S regions, or ligation of DSBs between S regions. The phosphorylated form of AID may be required to recruit PMS2 or EXO1 to S regions to generate the DNA breaks for CSR (10). Alternatively, p-AID may be interacting with ATM, RPA, or another protein to tether distal, recombining S regions together and promote their subsequent ligation (34).

The lack of DSB formation in the S regions may account for the loss of CSR in AID^S38A/S38AUNG^−/− B cells. MSH2^−/−UNG^−/− B cells cannot complete CSR, likely because of the absence of DSBs in S regions, and, consequently, can only produce G/C transition mutations (35–39). AID^S38A/S38AUNG^−/− B cells generate mutations at all bases in 5′Sμ (Fig. 2), suggesting that SSBs are formed at sites of AID(S38A) deamination. However, the absence of CSR suggests that these SSBs do not form a sufficient number of DSBs to complete CSR. Alternatively, altered S–S junction formation may account for the lack of CSR despite the presence of 5′Sμ mutations in the AID^S38A/S38AUNG^−/− B cells. AID^S38A/S38A B cells have disrupted interaction between AID and APE1 (19), and APE1-deficient CH12 cells have reduced Sμ–Sα synapsis and CSR without affecting mutations in S regions (40). Additionally, ATM^−/− B cells have severely impaired interaction between p-AID and APE1 (19), and B cells from ATM^−/− patients demonstrate altered S–S junctions (41). Thus, the absence of AID phosphorylation and recruitment of APE1 may alter S–S synapsis or ligation, preventing AID^S38A/S38AUNG^−/− B cells from completing CSR despite the moderate frequency of 5′Sμ mutations.

Recently, APE1 and APE2 were shown to be differentially expressed in pre-GC BCs and GCBCs. APE1 is highly expressed in pre-GCBCs and downregulated in GCBCs, whereas APE2 expression is low in pre-GCBCs and upregulated in GCBCs (42, 43). Furthermore, GLTs were highly expressed in pre-GCBCs and reduced in GCBCs, indicating that CSR occurred prior to GC formation (42). Phylogenetic analysis of GC V regions indicated that CSR occurred prior to SHM of V coding sequences (42). In our studies, the absence of CSR despite the generation of mutations in Sμ and the JH4 intron of AID^S38A/S38AUNG^−/− B cells (Figs. 2C, 3C) implicates differential roles for p-AID during CSR and SHM in the absence of UNG. Outside of GCs, when APE1 and GLT are high, p-AID may promote the generation of a high density of DSBs at S regions by recruiting both APE1 and MMR proteins, which drives CSR (19). The hypomorphic CSR in UNG^−/− mice suggests that p-AID and MMR partially compensate for the absence of BER. Thus, the absence of either p-AID or MMR completely inhibits CSR in AID^S38A/S38AUNG^−/− and MSH2^−/−UNG^−/− mice and B cells (Fig. 1) (11). In GCBCs undergoing SHM, p-AID cannot recruit APE1 to V regions to create DSBs because APE1 expression is low (42, 43). In this case, APE2 promotes mutations rather than DSB formation, possibly through a p-AID–dependent mechanism (Fig. 3) (43). Additional biochemical studies examining the interaction of p-AID with APE2 and other DNA repair proteins in pre-GC and GCBCs will clarify the differential function of AID phosphorylation in CSR and SHM.

The presence of mutations in V_H186.2 and the absence of mutations in the JH4 intron were observed exclusively in AID^S38A/S38AMSH2^−/− B cells (Fig. 3A, Supplemental Fig. 3A). We observed a decrease in mutation frequency at the JH4 intron as compared with V_H186.2 in AID^S38A/S38AMSH2^−/− B cells (Fig. 3, Supplemental Fig. 3). This phenotype suggests p-AID and MSH2 are in complementary pathways for the propagation of V region mutations, which spread 3′ from the V promoter for 2 kb (29). We hypothesize that, in the absence of MMR, p-AID recruits RPA to stabilize the transcription bubbles in the V region that would permit the generation of mutations downstream of the V promoter (26). In the presence of MMR, SSBs in V sequences may recruit RPA independently of p-AID to propagate mutations 3′ of the V region, which is observed in AID^S38A/S38A B cells (Fig. 3, Supplemental Fig. 3) (44). In the absence of both MMR and p-AID, transcription of the V gene may compensate for the lack of RPA to allow AID(S38A) to generate the limited mutations observed in the V region of AID^S38A/S38AMSH2^−/− B cells (i.e., mutations in V_H186.2 but not the JH4 intron) (Fig. 3, Supplemental Fig. 3).

The absence of CSR and the resulting hyper-IgM in AID^S38A/S38AUNG^−/− mice resemble the CSR phenotype in MSH2^−/−UNG^−/− mice (11), suggesting a disruption of the MMR pathway due to the absence of phosphorylated AID. The similarity in the shift in SHM spectra between AID^S38A/S38AUNG^−/− and MSH2^−/− B cells (Figs. 2, 3, Supplemental Fig. 3) further implicates a perturbation of MMR in AID^S38A/S38AUNG^−/− B cells. Whether the disruption in MMR occurs as a result of altered interactions between MMR proteins and AID(S38A), as compared with WT AID, remains unclear. Binding assays using anti-MSH2 Abs coimmunoprecipitated MSH6 but not AID in WT B cell lysates (Supplemental Fig. 4), suggesting that p-AID does not actively recruit MSH2/MSH6 to sites of deamination and that the disruption of MMR in AID^S38A/S38AUNG^−/− mice may be downstream of MSH2/MSH6 recognition of U:G mismatches. Phosphorylated AID may promote the formation of DSBs that are required to complete CSR by promoting the recruitment or stable binding of PMS2, MLH1, or EXO1 (45). Alternatively, the role of phosphorylated AID in MMR may be mediated by RPA, which promotes AID deamination of transcribed ssDNA (26, 39). RPA has been shown to facilitate MMR (46, 47), and AID phosphorylation mutants, including AID^S38A/S38A, display a significantly reduced recruitment of RPA to the IgH locus (34). Thus, the absence of RPA-dependent recruitment of MMR proteins in AID^S38A/S38AUNG^−/− B cells may account for the observed block in CSR and reduced SHM.

Unlike AID^S38A/S38AUNG^−/− B cells, which generate mutations at both 5′Sμ and the JH4 intron, AID^S38A/S38AMSH2^−/− B cells generate few mutations in 5′Sμ and no mutations at the JH4 intron (Figs. 2, 3). The mutations at 5′Sμ in the AID^S38A/S38AMSH2^−/− B cells may be the result of replication over abasic sites, which generates the G/C mutations, or residual BER, which generates the A/T mutations. In support of the latter hypothesis, ubiquitinated PCNA has been shown to be responsible for the generation of A/T mutations through BER in MSH2^−/− B cells (48). We speculate that the lack of mutation in the JH4 intron is due to canonical DNA repair, which may be unable to correct the high density of deaminated DNA within the GC-rich Sμ region. Interestingly, APE1 interacts with nonhomologous end joining proteins, such as XRCC1 (49), to direct canonical DNA repair. Phosphorylation of AID and the subsequent interaction of p-AID with APE1 may disrupt the APE1/XRCC1 interaction at S regions, thereby preventing canonical repair mediated by XRCC1 and promoting CSR. Consistent with this hypothesis, XRCC1^+/− B cells have increased SHM (50), and deletion of XRCC1 in CH12 cells increases CSR (51). Thus, the disruption of the AID and APE1 interaction in AID^S38A/S38AMSH2^−/− B cells may direct canonical DNA repair at the JH4 intron and S regions, which manifests as reduced mutation frequency.

The C-terminal domain of AID has been proposed to promote the recruitment of UNG and MSH2/MSH6 to S regions during CSR (52). Deletion of the C-terminal 10 aa of AID significantly reduces CSR to 10% of WT levels without altering 5′Sμ mutations (53), which almost resembles the CSR and SHM phenotype of the AID^S38A/S38AUNG^−/− B cells. The similar phenotype observed in these studies suggests that the phosphorylation of AID and the C-terminal domain of AID may function cooperatively to recruit or stabilize MMR proteins at V or S regions during SHM or CSR, respectively. Alternatively, the C terminus may bind to an unidentified factor that regulates MMR processing of deaminated DNA into DSBs. Loss of this factor could allow for mutational repair but severely limits CSR, which is consistent with the observed 5′Sμ mutations and loss of CSR in B cells expressing the AID C-terminal deletion mutant (53). Additional biochemical studies characterizing the interaction between the C-terminal domain, p-AID, and MMR proteins will provide insight into the molecular pathways by which phosphorylated AID regulates CSR and SHM through MMR.

We thank Jayanta Chaudhuri for the anti-AID Ab, Tomas Lindahl for the UNG^−/− mice, Hein te Riele for the MSH2^−/− mice, Tasuku Honjo for the AID^−/− mice, Simin Zheng, Montse Cols, and Emily Sible for critical reading of the manuscript, and Bharat Vaidyanathan for scientific discussions.

The authors have no financial conflicts of interest.