Activation-induced cytidine deaminase (AID) generates U:G mismatches in immunoglobulin genes that can be converted into untemplated mutations during somatic hypermutation (SHM) or DNA double-strand breaks (DSB) during class switch recombination (CSR). Null mutations in UNG and MSH2 demonstrate the complementary roles of the base excision repair (BER) and mismatch repair (MMR) pathways, respectively, in CSR. Phosphorylation of AID at serine-38 (pS38-AID) was previously hypothesized to regulate BER during CSR, as the AID phosphorylation mutant, AID(S38A), cannot interact with APE1, a BER protein. Consistent with these findings, we observe a complete block in CSR in AID S38A/S38A MSH2 −/− mouse B cells that correlates with an impaired mutation frequency at 5'Sμ. Similarly, SHM is almost negligible at the JH4 intron in AID S38A/S38A MSH2 −/− mouse B cells and, consistent with this, NP-specific affinity maturation in AID S38A/S38A MSH2 −/− mice is not significantly elevated in response to NP-CGG immunization. Surprisingly, AID S38A/S38A UNG −/− mouse B cells also cannot complete CSR or affinity maturation despite accumulating significant mutations in 5'Sμ as well as the JH4 intron. These data identify a novel role for pS38-AID in MMR-dependent CSR and affinity maturation.

Processing of AID (activation-induced cytidine deaminase)-dependent dU:dG mismatches by the base excision repair (BER) and mismatch repair (MMR) pathways generates untemplated mutations during somatic hypermuation (SHM) or DNA breaks during class switch recombination (CSR). Genetic deletion of UNG or MSH2, which are essential for BER or MMR, respectively, impairs CSR and alters the spectrum of mutations in SHM. Ablation of both UNG and MSH2 completely blocks CSR and limits SHM to GC transition mutations. Phosphorylation of AID at Ser38 (pS38-AID) is required for wild-type levels of CSR and SHM. B cells with a homozygous knock-in mutation of the AID phosphorylation site (AIDS38A/S38A) have intermediate levels of CSR and SHM as compared to wild-type and AID−/− B cells. AID phosphorylation is hypothesized to regulate BER through the interaction of pS38-AID with APE1, which functions downstream of UNG. Consistent with this hypothesis, AIDS38A/S38AMSH2−/− B cells have a complete block in CSR and significantly impaired SHM of Sμ and the JH4 intron. Surprisingly, CSR is also completely blocked in AIDS38A/S38AUNG−/− B cells; however, these cells generate significant mutations in Sμ and JH4 intron. These data demonstrate a critical role for pS38-AID in regulating BER and MMR and uncover a novel pS38-AID-dependent mechanism of mediating MMR-dependent DNA break formation and/or ligation during CSR.

Class-switch recombination (CSR) produces secondary Ig isotypes and requires activation-induced cytidine deaminase (AID)–dependent DNA deamination of intronic switch regions within the IgH (Igh) gene locus. Noncanonical repair of deaminated DNA by mismatch repair (MMR) or base excision repair (BER) creates DNA breaks that permit recombination between distal switch regions. Ataxia telangiectasia mutated (ATM)–dependent phosphorylation of AID at serine 38 (pS38-AID) promotes its interaction with apurinic/apyrimidinic endonuclease 1 (APE1), a BER protein, suggesting that ATM regulates CSR through BER. However, pS38-AID may also function in MMR during CSR, although the mechanism remains unknown. To examine whether ATM modulates BER- and/or MMR-dependent CSR, Atm−/− mice were bred to mice deficient for the MMR gene mutS homolog 2 (Msh2). Surprisingly, the predicted Mendelian frequencies of Atm−/−Msh2−/− adult mice were not obtained. To generate ATM and MSH2-deficient B cells, Atm was conditionally deleted on an Msh2−/− background using a floxed ATM allele (Atmf) and B cell–specific Cre recombinase expression (CD23-cre) to produce a deleted ATM allele (AtmD). As compared with AtmD/D and Msh2−/− mice and B cells, AtmD/DMsh2−/− mice and B cells display a reduced CSR phenotype. Interestingly, Sμ–Sγ1 junctions from AtmD/DMsh2−/− B cells that were induced to switch to IgG1 in vitro showed a significant loss of blunt end joins and an increase in insertions as compared with wild-type, AtmD/D, or Msh2−/− B cells. These data indicate that the absence of both ATM and MSH2 blocks nonhomologous end joining, leading to inefficient CSR. We propose a model whereby ATM and MSH2 function cooperatively to regulate end joining during CSR through pS38-AID.