A Critical Role for REV1 in Regulating the Induction of C:G Transitions and A:T Mutations during Ig Gene Hypermutation

The Ig genes undergo somatic hypermutation (SHM)² at both C:G and A:T base pairs in the germinal centers (GC) B cells (1). SHM is initiated by activation-induced cytidine deaminase (AID), which catalyzes the deamination of cytosine (C) to uracil (U) and generates a U:G mismatch on DNA (2, 3). Accumulating evidence suggests that mutations are introduced during replication and repair of the U:G mismatch (1). Direct replication of the U:G mismatch would generate C to T and G to A transitions because U normally pairs with adenine (A). In addition, U can be excised by the uracil DNA glycosylase (UNG), and replication of the resulting abasic site by the low-fidelity DNA polymerases would generate both transitions and transversions at C:G base pairs. The U:G mismatch and the abasic site can also be processed by a mismatch repair-like and a base excision repair-like pathway, respectively, resulting in the generation of mutations at undamaged A:T base pairs (4, 5, 6, 7). Studies thus far suggest that DNA polymerase η (POLH) is an essential enzyme for the induction of A:T mutations in both pathways (8, 9, 10, 11, 12, 13).

REV1 is a deoxycytidyl transferase that preferentially incorporates deoxycytidines opposite deoxyguanines and abasic sites (14, 15, 16, 17, 18). REV1 contains a BRCA1 C-terminal (BRCT) domain in the N terminus, a central polymerase domain, and a Y-family polymerase contacting region in the C terminus. The BRCT domain serves as a scaffold that mediates protein-protein interactions. Deletion of the BRCT domain of REV1 in ES cells resulted in increased sensitivity to DNA-damaging agents, indicating that this domain is involved in tolerance to DNA damage (19). The SHM of Ig genes in BRCT-mutant mice was reportedly normal, both in terms of mutation frequency and patterns (19). The polymerase domain is essential for the deoxycytidyl transferase activity and mutations of the highly conserved aspartate and glutamate residues to alanines completely abolished the catalytic activity of human and mouse REV1 (15, 18). The C-terminal region of REV1 has been shown to mediate the interaction with Y-family polymerases, including POLK, POLI, and POLH (18, 20). REV1-deficient mice exhibited reduced body sizes and were not fertile after backcrossing for two generations to C57BL/6 mice (21). Analysis of Ig gene SHM in these mice revealed an absence of C to G and a reduction of G to C transversions and a compensatory increase of C to A, A to T, and T to C mutations (21). These observations have led to the conclusion that REV1 participates in Ig gene SHM primarily by inserting C opposite abasic sites generated by UNG.

To investigate the role of the catalytic activity of REV1 in SHM in mammalian cells, we have generated Rev1 knockin mice expressing a catalytically inactive REV1 (REV1AA). REV1AA mice exhibited a great reduction of not only C to G and G to C transversions but also C to T and G to A transitions and A:T mutations. Because REV1 itself does not possess the catalytic property to directly introduce C:G transitions and A:T mutations, these observations suggest that REV1 regulates the induction of these mutations through functional interaction with other polymerases during Ig gene hypermutation.

The highly conserved aspartate and glutamate residues encoded by exon 10 were mutated to alanines by site-directed mutagenesis (Takara Bio). A fragment containing the mutated exon 10 was used as the 5′-arm of the targeting vector. Transfection and selection of the C57BL/6-derived Bruce4 ES cells were performed as described previously (22). Homologous recombinants were identified by long-range genomic PCR using primers s1 (5′-GCTTTGTCTTTGGGGGATGA-3′) and neos (5′-TCGCCTTCTATCGCCTTCTT-3′) for 5′ and as1 (5′-TCAGCCACATCCACATACCC-3′) and neoas (5′-ATAGCCGAATAGCCTCTCCA-3′) for 3′ homology regions. PCR was performed at 95°C for 2 min, followed by 30 cycles of amplification at 95°C for 10 s, 57°C for 20 s, and 68°C for 10 min (for 5′) or 95°C for 10 s, 63°C for 20 s, and 68°C for 13 min (for 3′) using LA-Taq polymerase (Takara Bio). Chimeric mice were bred with C57BL/6 mice to obtain heterozygotes, which were then crossed with CAG-Cre transgenic mice to delete the neo gene. The neo-deleted heterozygous mice were bred to obtain homozygous REV1AA mice. Mouse genotypes were determined by PCR using primers s2 (5′-CGATGTTAATGTCACAATGG-3′) and as2 (5′-CAACGTGTAGAAAGCTAAGC-3′) at 95°C for 2 min, followed by 30 cycles of amplification at 95°C for 10 s, 56°C for 20 s, and 72°C for 1 min using Taq polymerase (Toyobo). To further verify the introduced mutations, a genomic fragment containing exon 10 was amplified using primers s3 (5′-ATGTGTTGGTGTGCTGGTGT-3′) and as3 (5′-ATTTGGGAGAACTTTAGGTC-3′) and directly sequenced. PCR was performed at 95°C for 2 min, followed by 30 cycles of amplification at 95°C for 10 s, 56°C for 20 s, and 68°C for 1 min using KOD plus polymerase (Toyobo). Mice were maintained in specific pathogen-free conditions, and all experiments were approved by the Animal Facility Committee of RIKEN Yokohama Institute (permission no. 20-025).

These experiments were performed as described previously (23).

Five REV1AA mice and four wild-type (WT) littermates were injected i.p. with 100 μg of 4-hydroxy-3-nitrophenyl-acetyl coupled to chicken γ-globulin precipitated with alum. Two weeks later, B220⁺PNA^high GC B cells were sorted from spleen of each mouse, and the genomic DNA was extracted. The J_H4 intronic region was amplified and sequenced as described previously (23). Only unique sequences were analyzed in each mouse.

Purified spleen B cells were seeded in 12-well plates in duplicate at 5 × 10⁵/ml (1 ml/well) in the presence of 10 μg/ml LPS. The cells were then exposed to different doses of methyl methanesulfonate (MMS) or cisplatin or irradiated with UV light and cultured for 2 days. The cells were then stained with Annexin V^FITC and propidium iodide (PI) to detect apoptotic and dead cells, respectively. The percentages of live cells (Annexin V⁻PI⁻) were determined by FACS.

We mutated the highly conserved aspartate and glutamate residues encoded by exon 10 to alanines by gene targeting (Fig. 1, A–C). Northern blot analysis revealed a similar transcript level of Rev1 gene in WT and REV1AA B cells (Fig. 1,D), indicating that the targeted mutations did not affect Rev1 gene transcription. Sequence analysis of a fragment containing exon 10 further verified the designated mutations (Fig. 1 E). Attempts to detect WT and mutant REV1 protein by immunoblot have been unsuccessful due to the poor specificity of the available Abs. However, it has been shown previously that the corresponding mutations in human REV1 did not affect the global protein structure, and the mutant REV1 could be successfully purified like the WT protein (15). These observations collectively suggest that REV1AA mice express normal levels of REV1AA.

REV1AA mice were derived from Bruce4 ES cells and were therefore on a pure C57BL/6 genetic background. The homozygous mice were born at the expected ratio and developed normally with no obvious abnormalities by appearance. FACS analysis of BM cells of WT and REV1AA mice revealed no obvious differences in the percentages of B220⁺CD43⁺IgM⁻ progenitor, B220⁺CD43⁻IgM⁻ precursor, B220^dullIgM⁺ immature, and B220^highIgM⁺ recirculating B cells (Fig. 2^,A, upper and middle panels). The proportion of CD23^highCD21^dull follicular and CD23^dullCD21^high marginal zone B cells in the spleen was also similar between WT and REV1AA mice (Fig. 2^,A, lower panels). T cell development and differentiation were not affected in REV1AA mice as judged by CD4 and CD8 profiles in the thymus and spleen (Fig. 2^,B). In addition, REV1AA B cells exhibited normal proliferative responses to anti-IgM Abs, CD40L and LPS (Fig. 2^,C), and switched normally from IgM to IgG1 upon in vitro stimulation (Fig. 2 D). B cell activation in vivo was also normal as revealed by a similar representation of GC B cells in WT and REV1AA mice after immunization with a T-dependent Ag (supplemental Fig. 1).³ These results collectively suggest that B and T cell differentiation and activation were not affected by the disruption of the deoxycytidyl transferase activity of REV1.

To address the role of the catalytic activity of REV1 in Ig gene SHM, we analyzed the J_H4 intronic region of GC B cells isolated from spleens of immunized mice. We have analyzed four WT and five REV1AA mice, and the results are summarized in Table I (see supplemental Table I for detailed results of individual WT and REV1AA mice).³ Consistent with our previous studies (24), the overall mutation frequency in the J_H4 intronic region was ∼1% in WT mice with roughly half of the mutations occurring at C:G and A:T. The overall mutation frequency was reduced by 35% in REV1AA mice (0.651 vs 0.996% in WT mice). Mutations at C:G and A:T were decreased by 37 and 33%, respectively, compared with WT mice (Table I). To determine more precisely how SHM was affected in REV1AA mice, we calculated the absolute frequency of each type of nucleotide substitutions. For this purpose, we included data of WT mice from our previous studies (24), allowing the analysis of a large number of each type of mutations for comparison. As expected, C to G and G to C transversions were greatly decreased (75 and 71% reduction, respectively) in REV1AA mice, with a significant increase of C to A substitution (Fig. 3 and supplemental Table II).³ Intriguingly, C to T and G to A transitions were also significantly decreased (47 and 49% reduction, respectively) in REV1AA mice as compared with WT mice. Moreover, we found a uniform reduction of each type of nucleotide substitutions at A:T base pairs in REV1AA mice (Fig. 3 and supplemental Table II),³ a phenotype resembling that observed in Polh^+/− mice (24).

REV1 is known to play a critical noncatalytic role in tolerance to DNA damage in yeast and chicken DT40 cells presumably by interacting with other polymerases and coordinating their activities during translesion DNA synthesis (25, 26). Consistently, WT and REV1AA B cells exhibited similar sensitivities to MMS, cisplatin, and UV light (Fig. 4). In addition, cell cycle analysis revealed no differences in the relative proportion of cells at G₁, S, and G₂-M phases between WT and REV1AA B cells treated with these agents (data not shown). These results demonstrate that REV1 catalytic activity per se is not essential for tolerance to DNA damage in mammalian B cells.

The great reduction of C to G and G to C transversions in REV1AA mice is in agreement with the results of REV1-deficient mice (21). In addition, REV1-deficient chicken DT40 cells reconstituted with a catalytically inactive human REV1 (D570A/E571A) also exhibited a dramatic reduction of C to G and G to C transversions (27). These observations are consistent with the catalytic property of REV1 and suggest that REV1 participates in Ig gene SHM by incorporating C opposite UNG-mediated abasic sites. However, unlike REV1-deficient mice in which C to G transversions were virtually absent while G to C transversions were reduced, we found a similar reduction of C to G and G to C transversions in REV1AA mice. It is unclear at this point why REV1-deficient but not REV1AA mice exhibit a strand-biased defect in C:G transversions.

Both REV1-deficient and REV1AA mice exhibited a significant increase of C to A and a moderate increase of G to T transversions. These observations suggest that induction of these mutations not only does not require REV1 but also is not inhibited by the presence of an inactive REV1. Therefore, these mutations are generated by a REV1-independent pathway. The increased C to A and, to a lesser extent, G to T transversions are likely due to compensatory activation of this mutagenic pathway in the absence of a functional REV1.

The reduction of C to T and G to A transitions and A:T mutations observed in REV1 AA mice is unexpected because REV1 does not possess the catalytic property to directly introduce these mutations. According to the DNA deamination model of Ig gene SHM, C:G transitions can be generated by replication of either activation-induced cytidine deaminase-triggered U:G lesion or UNG-mediated abasic site (1). Replication of the U:G lesion is thought to be conducted by replicative polymerases, and it is thus unlikely that REV1 participates in this process. It is more likely that REV1 is involved in the generation of C:G transitions during replication of the abasic site. Mammalian REV1 has been shown to interact with Y-family polymerases, raising the possibility that REV1 may regulate the generation of C:G transitions by recruiting other polymerases. This possibility, however, is unlikely because the interaction of REV1 with Y-family polymerases is mediated by the C-terminal region of REV1, which is intact in REV1AA mice. A more likely possibility is that the REV1AA might be stuck or stabilized at the abasic site, thereby preventing the switching between REV1 and other polymerases involved in the generation of C:G transitions. A similar mechanism may account for the reduction of A:T mutations in REV1AA mice. The induction of A:T mutations is highly dependent on POLH (8, 9, 10, 11, 12, 13), and REV1AA might inhibit the access of POLH to the site of DNA lesions. We have recently found that Polh heterozygous mice exhibit a uniform reduction of each type of nucleotide substitutions at A:T pairs (24). A similar uniform reduction of A:T mutations in REV1AA mice is consistent with the hypothesis that POLH function is compromised in the presence of an inactive REV1. Regardless of the precise mechanism, our results suggest that REV1 and other polymerases, including POLH, do not function independently, and the interaction and switching between these polymerases are important for efficient induction of somatic mutations in Ig genes. Recently, REV1-POLH interaction was shown to be required for the formation of REV1 foci and for suppression of spontaneous mutations in human cells (28). It would be interesting to generate mice expressing a mutant REV1 that has normal catalytic activity but cannot interact with POLH and other Y-family polymerases.

The reduction of C:G transitions and A:T mutations in REV1AA mice was not observed in REV1-deficient mice. If the reduction of these mutations is indeed due to the stabilization of REV1AA at the site of DNA lesions, it is not surprising that the absence of REV1 did not affect the generation of these mutations. Conversely, the increased A to T and T to C substitutions were observed in REV1-deficient but not REV1AA mice. In this case, the absence of REV1 might have resulted in increased access of POLH to the abasic site, leading to the increase of certain A:T mutations. The differences in the mutation patterns between REV1AA and REV1-deficient mice might also be due to the different methods used to analyze SHM. We have analyzed the J_H4 intronic region of GC B cells isolated from immunized mice, and the mutations are induced within 2 wk by acute Ag stimulation. In contrast, Vλ1 genes of memory B cells isolated from 3- to 5-mo-old mice were analyzed for Rev1-deficient mice, and these memory B cells had likely undergone relatively long-term selection by endogenous Ags. Further studies are required to completely reveal how REV1 participates in Ig gene SHM. In conclusion, our results suggest that REV1 is involved in multiple mutagenic pathways and contributes to the generation of a significant portion of both C:G and A:T mutations during Ig gene hypermutation.

We thank C. Stewart for providing the Bruce4 ES cells, Y. Masuda for helpful discussions, the Animal Facility for breeding and maintaining the mice, and the FACS Laboratory for cell sorting.

The authors have no financial conflict of interest.