|Home | About | Journals | Submit | Contact Us | Français|
To test the hypothesis that DNA polymerase ζ participates in immunoglobulin hypermutation, we generated two mouse models of Pol ζ function: a B-cell specific conditional knock-out and a knock-in strain with a Pol ζ mutagenesis-enhancing mutation. Pol ζ-deficient B-cells had a reduction in mutation frequency at immunoglobulin loci in the spleen and in Peyer’s Patches, while knock-in mice with a mutagenic Pol ζ, displayed a marked increase in mutation frequency in Peyer’s Patches revealing a pattern that was similar to mutations in yeast strains with a homologous mutation in the gene encoding the catalytic subunit of Pol ζ. Combined, these data are best explained by a direct role for DNA polymerase ζ in immunoglobulin hypermutation.
The genes encoding the antigen-interacting portions of antibodies and immunoglobulin (Ig) receptors are subjected to a process of deliberate hypermutation during immune responses leading to enhanced affinity of antibodies to a specific antigen (1, 2). This mechanism, termed somatichypermutation (SHM), is triggered by the activation-induced deaminase (AID), a molecule expressed when B lymphocytes are activated by foreign antigen (3, 4). AID deaminates cytosines in the DNA encoding the Ig variable (V) regions (5). Mice and humans defective in AID lack SHM and class switch recombination (CSR), as AID is also required to generate switched antibodies such as IgG, IgA, etc. (3, 4). A subset of Hyper IgM syndrome patients are defective in AID and these patients lack CSR, SHM, and suffer from lymph node hyperplasia (6).
Despite the fact that AID is a cytosine deaminase, mice deficient in AID lack mutations at both G-C and A:T base pairs (3). This paradox is in part explained by the hypothesis that AID-mediated deamination of cytosines in Ig V regions triggers the recruitment of translesion synthesis DNA polymerases to Ig loci (7). These DNA polymerases have relaxed geometric requirements, and thus are prone to inserting incorrect bases during DNA synthesis (8). One such polymerase is DNA polymerase η, which has been shown to play a role in the misinsertion of bases at A:T sites during SHM (9, 10). Other DNA polymerases have also been implicated in SHM but, because they have important roles in other cellular functions such as cell division and DNA repair, discerning whether they play a direct or indirect role in SHM has been difficult (11–13).
The mutations made during SHM of Ig V genes are predominantly base substitutions. The pattern of hypermutation suggests that a putative error-prone DNA polymerase must not only insert the incorrect base but also extend from a mismatched terminus, a very difficult task for most DNA polymerases. Confounding this is the fact that some of the base substitutions in SHM occur in tandem, suggesting multiple misinsertion and mismatch extension events during a single DNA transaction (14). This lead to the hypothesis that DNA polymerase ζ plays a direct role in SHM, since it is a robust mismatch extender, alone and in conjunction with other translesion synthesis DNA polymerases, including Pol η (9, 14–16). However, demonstrating this has been difficult because mice deficient in DNA polymerase ζ are early embryonic lethals (17–19). A mouse expressing antisense RNA against Rev3 (encoding the catalytic subunit of Pol ζ) experienced decreased SHM frequency and severely impaired affinity maturation (20). However, because all cells, not just B cells, expressed the antisense transcript, it remained possible that the phenotype was due to indirect effects, such as diminished T cell function. In addition, SHM was reduced in human B cell lines in which the Rev3 gene encoding the catalytic domain of DNA polymerase ζ was inhibited by antisense oligonucleotides, suggesting a direct role for this polymerase in an in vitro model of hypermutation (21). A conditional knockout mouse model of Rev3 using the CD21 promoter also resulted in a reduced SHM frequency that was difficult to discern from a proliferation defect (22).
To circumvent the problem of embryonic lethality and non-specific effects from Pol ζ deficiency in non-B cells, we generated mice with B-cell specific deletion of Rev3. We also constructed mice with a knocked-in leucine (L) to phenylalanine (F) mutation at residue 2610 in Rev3 (Rev3L2610Fmice). The homologous change in Saccharomyces cerevisiae increases spontaneous and UV light-induced mutagenesis and is associated with a specific error signature (23, 24). We reasoned, that if Pol ζ plays a direct role in SHM, a more mutagenic variant would increase the frequency of mutation at Ig loci and its error signature would be accentuated. Indeed, we show here that Rev3 knock-out mice experienced a dramatic reduction in SHM frequency, while the Rev3L2610Fmice showed a significant increase in SHM frequency and an altered SHM specificity. The results indicate a direct role for DNA polymerase ζ in SHM.
A linearized targeting vector containing loxP sites flanking exon 26 of Rev3 was generated and electroporated into embryonic stem (ES) cells from C57BL6 mice (Supplemental Figure 1A and 1B). Exon 26 of Rev3 encodes the metal binding domain of DNA polymerase ζ and it is essential to its DNA synthesis function (25, 26). Recombinants were electroporated with Cre recombinase to eliminate the loxP site-flanked neo site. Clones still retaining the loxP sites flanking exon 26 of Rev3 (Rev3-floxed mice) were identified by Southern hybridization. Recombinant clones were confirmed by PCR analysis using primers N17delckF; 5’- GTTTGGGGCATTGGTTTACAGGTG - 3’ and N17del3R; 5’- CTCCTTACTGCTGGGGATACTCATGTG -3’. Several clones were selected for blastocyst injection resulting in male chimeras. Chimeras were breed with C57BL/6 Taconic mice and heterozygotes obtained. Rev3-floxed mice were crossed with the cre-recombinase transgenic line C.Cg-Cd19tm1(cre)Cgn (Balb/c) (27) which we will refer to as CD19+) and reporter mice B6.129X1-Gt(Rosa)26Sortm1(EYFP)Cos/J (C57BL/6) (which we will refer to as YFP+), strains purchased from The Jackson Laboratory(Bar Harbor, ME). CD19+ mice were crossed with YFP+ mice to give, CD19+/−, YFP+/− these mice were then backcrossed to the Rev3F/F background (10 generations) to give CD19+/−, YFP+/−, Rev3F/F mice in a C57BL6 background. The resulting mice had B cell specific deletion of Rev3 exon 26 and the recombinant B cells can be collected by FACS by virtue of their YFP expression (Supplemental Figure 1B). Alleles of the targeted floxed exon 26 Rev3 gene were examined by PCR with primers N1753F (5'-TTTCTATCAGCTTGTGCCCTTATCCTTACT-3') plus N17cKR (5'-ACAAGATTTCTTTGTGTAACAGCCCTGG-3') for the 5’ loxp site, give WT and mutant products of 474bp and 624bp, respectively. Genotyping of the whole exon 26 region using N1753F and N17del3R resulted in a WT product of 1330bp and mutant product of 1630bp (Supplemental Figure 1C). Upon cre-mediated deletion of the floxed region, the above primers resulted in a product of 632bp (Supplemental Figure 1C).
To detect the CD19 cre-recombinase transgene primers, CD19 5’ (5′-CTATCTGAAAAATATTTAACAGGTGCCAC-3′), CD19 3’ (5′-CACTATCCTCCACGTTCACTGTCCA-3′) and CD19 cre (5′-GGCAAATTTTGGTGTACGGTCAGTAA-3), resulting in a WT product of 950bp and a transgene product of 850bp were used ((Supplemental Figure 1D). YFP genotyping primers were oIMR0316 (5’-GGAGCGGGAGAAATGGATAT-3’), 0IMR0883 (5’-AAAGTCGCTCTGAGTTGTTAT-3’) and 0IMR4982 (5’AAGACCGCGAAGAGTTTGTC-3’) which amplify a WT fragment of 600bp and a mutant fragment of 320bp (Supplemental Figure 1E).
To generate Rev3L2610F (mutator knock-ins), a second vector was made with a mutation replacing the leucine at position 2610 to phenylalanine in exon 23 (Supplemental Figure 2A). The homologous mutation in yeast reveals a 2–3 fold increase in Rev3-mediated mutagenesis (23–24) Rev3L2610Fcells with the L to F mutation that had undergone homologous recombination were identified by Southern Hybridization and confirmed for the mutation by PCR using primers: N29B5PCRF: 5’- CTTTCATGTTCTCCATCAGCGTTTCC- 3’ and N29B5PCRR: 5’- GGTTAGCTGGGCTACATTCCAATTCATC -3’, followed by sequencing with primer: N29B5PCRR: 5’GGTTAGCTGGGCTACATTCCAATTCATC -3’. (Supplemental figure 2B–C).
For Rev3L2610Fmice, detection of the mutant allele was done by PCR with primers: N29BNeodelF 5’- GTTAAATCAGCTTCCGTTGCAGCACT- 3’ , and N29BNeodelR: 5’- GCTTCCACAAGTGTTTCCTATGAGAGTTG-3’. The presence of the mutation in the mRNA encoding Rev3 in thèse mice was confirmed using cDNA as template and using primers : F5’ATgAgAgCCCCACAgTgTgTT and R5’CAACCCTAgCACCCCATTTCT and the nested primers, 5’gTCTCgTTTCTATAgCAACTCTg and 5’ CTggCTTTgTgAACAATgCTATC. The nested primers were also used for sequencing (Supplemental Figure 2D). All mice were kept at the pathogen specific free animal facility at NIEHS and were maintained in microisolator cages on hardwood bedding, and provided with autoclaved food and reverse osmosis, deionized water.
For detection of WT transcripts, cDNA from Rev3-deleted B cells were amplified using Taq DNA polymerase (Invitrogen) and using Rev3 junct 25/26 F (5’-GCCGTGCATTGAGGTTGGTGATA-3’) and Rev3 exon 27 R (5’-CTTCAGTTTCACTGGCCTAGGATTAGTA-3’) resulting in a 229bp WT fragment. Mutant Rev3 transcripts were detected using Rev3 junct 25/27 F (5’-TGCCGTGCATTGAGTATGTTTGTACTACT-3’) and Rev3 exon 30 R (5’-CAGCGTTTCATACATGTAGCCCACAT-3’) resulting in a 191bp mutant Rev3 transcript (Supplemental Figure 1F). The RT-PCR of mouse GAPDH was performed with GAPDH F primer (5’-ACCACAGTCCATGCCATCAC-3’) and GAPDH R primer (5’-TCCACCACCCTGTTGCTGTA-3’).
8–12-week-old mice were immunized by intraperitoneal injection of 100µl (0.1ml of 1mg/ml) of 4-Hydroxy-3-nitrophenylacetyl hapten, or 2,4,6-Trinitrophenyl hapten, conjugated to chicken γ-globulin (NP-CGG, N-5055-5, and TNP-CGG, T5052-1; Biosearch Technologies, Novato, CA) prepared in an alum-precipitated suspension. Mice were euthanized and their spleens and serum recovered for analysis 14–15 days post-immunization.
Spleen and bone marrow cell suspensions were created as describe in (28). Peyer’s patches were collected from mice at ages of 8, 12 and 26–52 weeks and cell suspensions generated as described previously (29).
Bone marrow progenitor B cells were enriched using allophycocyanin (APC)-labelled rat anti mouse B220/CD45R (clone RA3-6B2). Combined staining with either phycoerythrin (PE)-labelled rat anti mouse CD117/c-kit (clone ack45), CD25/IL-2Rα-chain (clone Pc61) or IgM (cloneR6-60.2) isolated pre-BI, pre-BII and immature B cells, respectively. Splenic B cells were isolated using PE-labelled rat anti mouse-CD19 (1D3) and IgM (cloneR6-60.2), plus APC-labelled rat anti mouse B220/CD45R (clone RA3-6B2) all from PharMingen (San Diego, CA). Activated splenic B cells from NP-immunized mice were stained for APC-labelled rat anti mouse B220/CD45R (clone RA3-6B2) (BD PharMingen, San Diego, CA) and biotinylated rat anti mouse Ig λ1, λ2 & λ3 light chain (clone R26-46) (BD PharMingen, San Diego, CA). Biotinylated antibodies were revealed using streptavidin-phycoerythrin (Southern Biotechnology Assoc, Birmingham, AL), strepavidin-allophycocyanin (BD PharMingen, San Diego, CA), or strepavidin-tri-colour (Caltag laboratories). Peyer’s Patches B cells were enriched by staining with PE-labelled rat anti mouse-B220/CD45R (clone RA3-6B2) and APC-labelled Anti-mouse Ly-77 (clone GL7) from eBioscience (San Diego, CA). Cell sorting was carried out using a Becton Dickinson FACSVantage SE Flow Cytometer or FACSDiVa (Franklin Lakes, NJ). Data was analysed using the FlowJo (Ashland, OR).
RNA from B220+ YFP+ λ+ cells sorted cells was prepared in TRIzol. One microgram of RNA was used as template for cDNA synthesis in the reverse transcriptase reaction by using a SuperScript II First-Strand Synthesis System (RT-PCR; Invitrogen). cDNA (2µl) was amplified using Phusion DNA polymerase (New England BioLab, Ipswich, MA) using the VH186.2 specific primers VH186.2 F (5’-AGCAGCCTGGGGGCTGAGCTT-3’) and IgH Cγ1 R (5’-CAGGGGCCAGTGGATAGACAGA-3’) and the following PCR conditions: 94°C for 2 mins; 35 cycles of 94°C for 1 minute, 60°C for 45 seconds and 72°C for 45 seconds; and 72°C for 5 mins, resulting in an ~400bp product. Sorting for λ+ cells and amplifying the VH186.2 region enriches for Ig sequences that have experienced SHM during the NP response (30). The reverse primer was anchored in the gamma constant domain to ensure amplification of the Ig HC from B cells likely to have participated in the GC reaction. The IgH intronic region downstream of rearranged V genes was amplified by PCR as previously described (28).
GC morphology was examined via biotin-labeled peanut agglutinin (PNA) (Vector Laboratories) as previously described (20). Briefly, spleens from 8–12 week-old mice were frozen in Tissue-Tek OCT (Sakura) and sectioned on a Leica CM 3050 S cryostat (6 µm) and affixed to slides with Rapid Fix solution (Shandon-Lipshaw). Protein blocking was carried out with an avidin-biotin blocking kit (Vector Laboratories). Incubation with PNA was done at a 1/1000 dilution and labeling with Biogenex streptavidin label. The stain was developed with diaminobenzidine chromogen (DakoCytomation) and the slides were counterstained with hematoxylin and visualized with a fluorescence microscope.
For CFSE analysis, splenic B cells from 8–12 week-old C57BL/6J, CD19+(B6), Rev3L2610F and Rev3-deleted mice were CD43-depleted using CD43 MicroBeads and MACS cell separator magnetic columns following the manufacturer’s instructions (Miltenyi Biotech GmbH, Bergisch Gladbach, Germany). Cells were incubated with 5nM CFSE for 15 min at 37°C following the manufacturer’s instructions (Invitrogen, Molecular Probes Carlsbad, CA). Cells were washed twice with PBS and incubated in RPMI 1640 medium. Cells (106 cells/ml) were then stimulated over 4 days using LPS alone (20ug/ml, Sigma), or together with IL-4 (25ng/ml, R & D Systems. All samples were analyzed as duplicates. The absorbance (Ex. 492nm / Em. 517nm) was measured using BD Biosciences FACSVantage SE Flow Cytometer. Activated B cells were determined using flow cytometry using APC-Cy7-labelled rat anti mouse B220/CD45R (clone RA3-6B2) and PE-labelled IgM (cloneR6-60.2) and IgG1 (clone 15H6) all from BD PharMingen (San Diego, CA). There were 105 cells analysed for each sample. A total of 5 mice per group were used.
The difference in the distribution of mutational classes (number of clones with a particular number of mutations) among groups was examined for significance using the Kolmogorv-Smirnov test. The differences in the mutation frequency among genotypes in the spleen and PP were examined using Mann Whitney statistic and Kolmogov-Smirnov . The difference in the proportion of insertions and deletions, of tandem multiples, or of mutations from deoxythymidine among all mutations was tested for significance among groups using Fisher’s exact test or Chi-Square statistic. All probability values were considered significant if less than 0.05.
To test the hypothesis that DNA polymerase ζ directly induces mutations at Ig loci during SHM, we generated mice with B-cell specific deletion of Rev3 exon 26 which encodes the metal binding domain, and mice with a mutation in Rev3 that, in yeast, renders the polymerase more mutagenic (23, 24) (Supplemental Figures 1–3). Rev3 knock-out mice were immunized with the NP hapten and mutations at the Vh186.2 heavy chain (HC) were analyzed using cDNA from splenic B cells. To sort the cells, we used B220+, λ+, and YFP+B cells to make the cDNA and amplified rearranged Vh186.2 regions that had switched to IgG. This regimen enriches for NP-reactive, isotype switched B cells. YFP is a marker for expression of cre-recombinase activity and thus strongly correlates with Rev3 deletion, as YFP+ cells lacked the floxed exon (Supplemental Figures 1A–B, 1F). Vh186.2 is the Ig HC associated with enhanced affinity to NP, and Ig’s bearing Vh186.2 HCs and λ light chains experience a 10-fold increase in affinity when tryptophan at position 33 is mutated to leucine (30). Affinity maturation to a specific antigen such as NP can profoundly influence the mutational pattern and frequency at the relevant Ig loci. Therefore, we also examined the accumulation of mutations in the intronic region downstream of rearranged J558 Ig HC V genes in B cells from the Ileal Peyer’s Patches (PP) of Rev3-altered mice. Because the region analyzed is non-coding, mutations here are a more accurate representation of the intrinsic SHM machinery in the absence of selection.
Impaired Pol ζ function did not impact B cell development in the bone marrow or the spleen compared to CD19-driven cre-recombinase transgenics and C57BL6 controls (Figure 1A–B). Over 90% of B220+ B cells were also YFP+, the marker for Rev3 deletion (Figure 1C–D). The number of GC B cells (GL7+) in B-cell specific Rev3 knock-out mice was similar to C57BL6 controls. While over 90% of all splenic B cells were YFP+, the percentage decreased to 56% among GL7+ B cells (Figure 1B). This was accentuated dramatically in Peyer’s Patches, where only 3% of the GL7+ B cells were YFP+ (Figure 1B). Splenic or Peyer’s Patches B cells from Rev3L2610Fmice displayed no defects in B cell development or in the number of germinal center (GC) cells (Figure 1A; data not shown).
Following immunization with NP-CGG in alum, YFP+, Rev3-deleted B cells from Rev3 conditional knock-outs that had been crossed to CD19 promoter-driven cre-recombinase transgenics were examined for mutations at the rearranged Vh186.2 locus. Over 40% of the clones from these mice had either zero or 1 mutation compared to less than 15% of the clones from either C57BL6 controls or CD19 cre-recombinase transgenics without the floxed Rev3 gene (Figure 2A –BSupplemental Figure 4). This distribution of mutational frequencies was significantly different from controls (Mann-Whitney and Kolmogorov-Smirnov test, p= 0.00041 and p = 0.001, respectively). In addition, among VH186.2 clones with at least one mutation, 46% had the affinity-enhancing mutation W to L at amino acid 33 (Figure 2B). Remarkably, in some clones this was the only mutation. GC’s from these mice were indistinguishable from those of control mice (Figure 1B) in terms of both morphology and abundance, 15 days following immunization. The fact that GC morphology was intact and that the same proportion of clones from these mice acquired the NP-affinity enhancing mutations as in controls suggests that the decrease in the mutation frequency is not due to a defect in proliferation in these cells (see below).
Following immunization, most of the unique clones derived from Rev3L2610Fmice B cells contained slightly fewer mutations than C57BL6 controls (Figure 2C). However, there was an increase in the fraction of clones containing at least 10 mutations, with a few clones harboring more than 15 in the rearranged V region (Figure 2C). While it did not achieve significance at 0.05, there was a trend for an increase in the fraction of Vh186.2 clones with the characteristic W to L change at amino acid 33 that is associated with increased affinity to NP (> 60% of the clones, Figure 2C). These data may be explained by an increased rate of SHM due to an enhanced mutagenic potential or activity of DNA polymerase ζ. For example, B cells from these mice may acquire the affinity-enhancing mutations in VH186.2 more efficiently, and thus terminate SHM early in the GC reaction. The accumulation of mutations in conditions where the impact of selection is minimized (see below) helps clarify these data.
Given that only 3% of GL7+ B cells in Peyer’s Patches and 56% of GL7+ B cells from the spleen were YFP+ (compared to over 90% of all B cells), and to rule out a defect in proliferation from Rev3 deficiency, we measured proliferation in Rev3-deleted B cells. Following activation with LPS, Rev3-deleted B cells consistently exhibited a significant defect in proliferation compared to CD19 transgenic WT controls (Figure 3A), as LPS-only treated Rev3-deleted B cells appeared to have an excess of cells in earlier stages of division. This defect was less obvious in LPS plus IL4 treated cells when compared to CD19 transgenics with normal Rev3 function (Figure 3B). Rev3L2610F mice activated with LPS and IL4 displayed normal proliferation (Figure 3B). As CSR is tied to proliferation, we examined switch to IgG1 following activation with LPS and IL4 and found that Rev3KO cells displayed decreased numbers of IgG1+ cells when compared to CD19 transgenic controls while Rev3L2610F had levels that were similar to C57Bl6 controls (Figure 3C).
Given the profound impact that antigen-driven selection has on the frequency and the pattern of SHM, we examined SHM in Rev3-altered mice in the intronic region downstream of rearranged VDJ segments from Peyer’s Patches B cells. This region has been characterized as a target of the SHM machinery with a reported peak mutation frequency of 0.01–0.15 mutations/base pairs (31, 32). Peyer’s Patches B cells are chronically stimulated and accumulate mutations over time, but by 5 months of age most mice reach their peak mutation frequency (33). All but one of the clones from mice with B-cell specific deletion of Rev3 were not mutated, even at 6 months of age (Figure 4A). In contrast, over 75% of the clones from control mice had at least one mutation by 6 months of age. Since only 3% of GL7+ B cells in Peyer’s Patches were YFP+ in the Rev3 knock-out mice, it is possible that the low mutation frequency of Rev3-deleted B cells puts them at a selective disadvantage in the GC environment of the Peyer’s Patches, specially given the pressure exerted by microbial flora in the small intestine. Alternatively, a proliferation defect may put these cells in a competitive disadvantage.
Rev3L2610Fmice were also examined for mutations at the intronic region downstream of rearranged V genes. There was a highly significant increase in the mutation frequency in GC B cells from Peyer’s Patches of these mice (Figure 4A). The number of mutations per base pair in Rev3 knock-in B cells was around 0.035, which represents a near 2-fold increase over the wild type (WT) controls. Because of the elevated mutation frequency, we divided the 1.2kb region for analysis into proximal and distal regions as defined by the distance from the rearranged V region, with the notion that the distal region may be less saturated and a more accurate assessment of the mutational frequency and pattern will be possible. This analysis revealed that the distal region from Rev3L2610Fmice had experienced a 3-fold increase in mutation frequency (0.014) over the same region in the WT controls (0.004) that was also statistically significant (Figure 4, Supplemental Figure 3D). Strikingly, 40% of the clones derived from Rev3L2610Fmice had accumulated more than 15 independent mutations, with many of those having 30–40 mutations in the 420bps of the proximal region alone (Figure 4B). There were also highly significant changes in the pattern of mutation in B cells derived from Rev3L2610Fmice. Tandem mutations, such as adjacent doublets and triplets, were increased 3–4 fold, and insertions or deletions (mostly +1 or −1 bp) were 5-fold higher than in controls (Figure 4C; Figure 5). Finally, when examining the non saturated sequences (distal region or mutations from 2 month old mice) mutations at thymidine bases were moderately increased in these cells (Figure 6), suggesting strand bias by Pol ζ in SHM at A:T base pairs.
In this study, we tested the hypothesis that DNA polymerase ζ plays a direct role in SHM using mice with altered Rev3 function, the gene encoding the catalytic subunit of DNA polymerase ζ. Mice with a B-cell specific deletion of Rev3 displayed a significant reduction in the accumulation of mutations at the VH186.2 locus during the immune response to the NP hapten, and PP B cells from these mice displayed a dramatic reduction in mutation frequency at the intronic region downstream of rearranged V genes. Indirect effects on SHM, a possibility plaguing previous studies, is minimized here as B-cell specific deletion of Rev3 eliminated the impact of Rev3 deletion in other cells such as in helper T cells or antigen presenting cells. Furthermore, in our model, Rev3-deleted B cells could be specifically recovered because cre-recombinase deletion of Rev3 (under control of the CD19 promoter) coincides with YFP expression (resulting from cre-recombinase-mediated deletion of a floxed stop codon upstream of the YFP gene). Thus, Rev3- deleted cells could be isolated based on YFP expression. This minimized the impact of incomplete Rev3 deletion, particularly in the highly competitive GC environment. Indeed, while over 90% of the splenic B cells were YFP+, only 56% of GL7+ B cells, a GC marker, were YFP+. PP B cells with Rev3 deletion displayed almost no mutation accumulation even at 6 months of age. B cells from Rev3 knock-out mice formed morphologically normal GC’s, yet displayed a proliferation defect following activation with LPS. Among those GC B cells that were mutated, a similar proportion had acquired the affinity-enhancing tryptophan to leucine change at amino acid 33 of VH186.2 following immunization with NP, suggesting that eventual acquisition of high affinity antibodies was not significantly altered in these cells, although it is possible that the affinity maturation rate may have been altered. These results are similar to a previous study using conditional Rev3 knock-out mice in terms of reduced mutation frequency but differ in that smaller germinal centers from Rev3 knock-out mice were reported in the previous study, while GC appeared normal in our study (22). This discrepancy may be due to the different cre-recombinase transgenic strains used in the studies (CD19 vs CD21). For instance, with earlier Rev3 deletion (as the CD19 promoter is active earlier than the CD21 promoter), it is possible Rev3-deleted cells may have partially compensated in terms of proliferative capacity from Rev3 deficiency, since this is not a polymerase that is essential for DNA replication. To circumvent the proliferation issue, and to determine if the SHM frequency reduction observed in both studies is the result of DNA polymerase ζ participating in the SHM process rather than entirely explained by a defect in proliferation, we examined mutation at Ig loci in Rev3 KI mice with a hypermutagenic polymerase ζ and an altered mutational spectra. An alteration in the pattern of mutations consistent with the observed error signature for the Rev3Ki polymerase, is strong evidence for a direct involvement and unlikely to be explained by a defect in proliferation.
In the Rev3Ki mice, the Rev3L2610F mice, a highly significant increase in mutation frequency at the intronic region downstream of rearranged V genes was seen in PP B cells. In yeast, the equivalent leucine to phenylalanine replacement in the Rev3 gene increases the error rate of DNA polymerase ζ about 2–3 fold without significantly altering its catalytic activity (23, 24). It is also associated with an increase in complex mutations (more than one mutation within a short patch presumed to occur in a single event), and with an increase in insertions and deletions. Given the high mutation frequency in the Rev3L2610F clones, it was difficult to ascertain which complex mutations originated from an accumulation of single mutational events or from a single DNA polymerase transaction that yielded complex mutational changes. However, limiting the analysis to tandem doublets or triplets revealed a 3- to 4- fold increase in these events. There was also a 5-fold increase in insertions and deletions, mostly adding or deleting a single nucleotide. These are similar to the mutational pattern observed in yeast (23, 24). Interestingly, an increase in mutations from thymine was also observed, suggesting Pol ζ strand bias during SHM. This is reminiscent of strand bias at A:T base pairs in Ig templates from DNA polymerase η deficient patients, and may imply complementary roles for Pol ζ and Pol η in SHM. Together, the results from both mouse models can best be explained by a direct role for DNA polymerase ζ in SHM.
While there were significant changes in both the frequency and pattern of SHM in mice with altered Rev3 function, there was no obvious difference between mutations at G:C vs A:T base pairs in either strain: all base pairs experienced either a decrease in Rev3-deleted B cells or an increase in Rev3L2610F B cells (with the exception of a modest increase in mutations from T’s but not from A’s in the Rev3L2610Fmice). The mutation frequency for Rev3 knock-out B cells and the lack of G:C bias is consistent with data from previously described models of Rev3-deletion, including mice expressing antisense transcripts to Rev3 (20, 22). These results are different from human B cells deficient in DNA polymerase η, where the loss of mutations was primarily at A:T base pairs (10), and suggests that DNA polymerase ζ participates in both the G:C and A:T phase of SHM.
On the surface, it appears difficult to reconcile the fact that Pol ζ is directly involved in introducing mutations at both G:C and A:T base pairs with the fact that AID triggers SHM by deaminating cytosines in both strands. One possibility, however, is that Pol ζ is involved in the mutagenic bypass of the abasic sites generated by the removal of AID-generated uracils in the DNA of Ig V regions. Continued synthesis pass the abasic site would also result in mutations at A:T base pairs, perhaps in complex with Pol η, and this would generate mutations at all base pairs. Pol ζ is uniquely efficient at extending synthesis past a variety of DNA lesions, including abasic sites, alone or in complex with REV1, a co-factor shown to contribute to mutations at AID-mediated abasic sites in a cell line (34, 35). Furthermore, it has been shown to extend mismatches generated by other TLS polymerases (36, 37). In Pol ζ deficient B cells, AID-generated abasic sites may be repaired in an error-free fashion reducing mutations at all base pairs, not just G:C base pairs. Inefficient recruitment of error-prone DNA polymerases to compensate for Pol ζ ablation may occur explaining why 15% of Rev3 deleted clones derived from splenic B cells had more than 6 mutations . Alternatively, Pol ζ may assemble a targeted mutasome complex to Ig loci which include AID. This would explain an impact on mutation at all base pairs, but it appears less likely since it would predict loss of all mutations, not just a reduction in the mutation frequency, with Rev3 deletion. Also, the results using Rev3L2610Fmice indicate a role for Pol ζ as a DNA polymerase, not just as a targeting complex.
It is interesting that Rev3-deleted B cells appear to be at a disadvantage in the PP environment compared to wild type B cells within the same animal. While close to 90% of B220+ B cells were YFP+ and therefore Rev3-deleted, only 3% of GC B cells in Peyer’s Patches expressed YFP. In the spleen, the reduction was less marked (56% of GL7+ B cells), and is more in agreement with other models of CD19 promoter-driven cre-recombinase deletion (38). This defect may be due to impaired proliferation as seen with splenic cells, or the result of intense competition in the GC environment for antigen binding and increased affinity through SHM. A reduced mutation frequency may impact the rate at which affinity-enhancing mutations are acquired, and this may place Rev3-deleted B cells in a competitive disadvantage in the small intestine Ileum, which is chronically stimulated by microbial flora. Interestingly, B cells from Rev3L2610Fmice in the spleen were more likely to have affinity-enhancing mutations to NP hapten (60% vs. 40% of controls). This raises the possibility that the converse is true: a higher rate of SHM with a mutagenic Pol ζ is associated with an enhanced affinity maturation rate for those cells. Having these novel strains that differ in the ability of their B cells to undergo SHM will be useful in examining the impact of SHM rate in the rate of affinity maturation to certain pathogens, specially those associated with specific antibodies bearing a large number of mutations, as appears to be the case for HIV (39) and influenza (40).
We are grateful to John W. Drake, and Dmitry Gordenin for critical readings of the manuscript. Natasha Clayton, Tiwanda Masinde for assistance with immunohistochemistry, and Carl Bortner, and Maria Sifre for assistance with flow cytometry. We are grateful to Yingbin Ouyang and Leanne Zhu from Xenogen and Manas Ray from NIEHS for chimera generation. We are also, grateful to Scott Jenkins for generation of the Rev3-deleted yeast strain.
¶This work was supported by Project Z01 ES101603 to M.D. and Project Z01 ES065070 to T.A.K. from the Division of Intramural Research of the NIH, NIEHS.