Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

The PU.1 and NF-EM5 binding motifs in the Igκ 3′ enhancer are responsible for directing somatic hypermutations to the intrinsic hotspots in the transgenic Vκ gene

The PU.1 and NF-EM5 binding motifs in the Igκ 3′ enhancer are responsible for directing somatic... Abstract Somatic hypermutation is a key mechanism in generating Ig with higher affinities to antigen, a process known as affinity maturation. Using Igκ transgenes, the 3′ enhancer (κE3′) has been shown to play an important role in introducing hypermutations. In order to identify the cis-acting elements that regulate hypermutagenesis, we have generated transgenic substrates containing mutations/deletions in the κE3′ region. Here, we report that base substitutions in the κE3′, either in the PU.1 or in the NF-EM5 binding motif, not only reduce the mutation rate but also disrupt the directed mutagenesis in the intrinsic hotspots of the Igκ transgene. antibodies, B lymphocytes, generation of diversity, transgenic, knock-out CDR complementarity-determining region, DSB double-strand break, GC germinal center, κE3′ Igκ 3′ enhancer, iEκ Igκ intron enhancer, MAR matrix attachment region, PNA peanut agglutinin Introduction Following antigen stimulation, the primary B cell repertoire is shaped by the second wave of somatic DNA changes, known as somatic hypermutation, necessary to generate the variant types of Ig genes coding for the high-affinity antibodies. Nucleotide substitutions are predominantly introduced into three complementarity-determining regions (CDR) of Ig variable (V) segments, at a rate of 10–3 to 10–4 bp/cell generation (1–3). Recently, several groups analyzed the somatic hypermutations in mice that are deficient for DNA repair functions and found that the disruption of the mouse homologues of yeast mismatch repair genes affects the hypermutations at intrinsic hotspots (4–11). In the study of somatic hypermutation, transgenic mice have been quite useful in identifying cis-acting elements for the high-frequency mutagenesis in the Ig gene loci (12,13). Using rearranged Igκ transgenes, it has been shown that both the 3′-enhancer (κE3′) and the intronic enhancer/ matrix attachment region (iEκ/MAR) are critical for introducing somatic hypermutations to the Vκ segments (14,15). Deletion of either enhancer, κE3′ or iEκ, caused a severe drop in the mutation frequency. It has been reported that the extent of mutations is correlated with the level of transcription of the Igκ transgene (16–18). However, findings that the Ig enhancers can even induce hypermutations in the non-Ig sequences indicate a more directive role (19–22). In order to narrow down and identify the cis-acting elements for hypermutagenesis, we have introduced base substitutions in the κE3′ region and analyzed the effects on the introduction of hypermutations in the Igκ transgene. It was found that base substitutions in the PU.1 or the NF-EM5 binding sequence not only reduce the mutation rate, but also affect the biased mutagenesis in the intrinsic hotspots. Methods Sorting cells The B cell fraction of PNAhigh/B220+ was sorted from small intestinal Peyer's patches of unimmunized, 5- to 12-month old, and C57BL/6J-background mice, irrespective of sex (23,24). Cells were incubated with anti-FcγRII/III (2.4G2; PharMingen, San Diego, CA), FITC-conjugated peanut agglutinin (PNA; Vector, Burlingame, CA) and allophycocyanin-coupled anti-B220 (mAb RA3-6B2; PharMingen) in cold PBS supplemented with 2% newborn goat serum, 0.05% sodium azide and 2 mM EDTA (pH 8.0) for sorting with a FACS Vantage cell sorter. PCR amplification The Vκ21C region was amplified by two successive PCR. Genomic DNA was isolated as described (25). The first PCR was performed with 5–10 ng of DNA, using the Expand High Fidelity PCR system (Roche, Mannheim, Germany). The PCR cycle consists of denaturation (94°C, 15 s), annealing (63°C, 30 s) and extension (72°C, 52 s). The cycle was repeated 18 times according to the manufacturer' s instruction. In the second nested PCR, 1/500 of the first PCR product was used. The extension time was 45 s. The cycle was repeated 24 times for the transgenic DNA and 26 times for the non-transgenic DNA. For the detection of the rearranged Vκ21C–Jκ1, two sets of primers were used: (Vκ21C-301) 5′-AAATTTTCTGCTAACACCAACTTCCTGTGG-3′/(Jκ1–30) 5′-GTGCCAGAATCTTGGTTTCAGAGTAAGATT-3′ and (Vκ21C-303) 5′-TGGTGTCTAATAGTGATGTCCGCAGTATTC-3′/(Jκ1–302)5′-TTAGACATAGAAGCCACAGACATAGACAAC-3′. For the semi-quantitative PCR, primer Vκ41–30 was used: 5′-TAACAGTCAAACATATCCTGTGCCATTGTC-3′. PCR products were separated in agarose gels (1.5%), stained with ethidium bromide and quantified with Densitograph software library, version 4.0 (ATTO, Tokyo, Japan). To estimate PCR errors, bone marrow B cells were purified from 12-week-old mice using Lympholyte-M (Cedarlane, Hornby, Canada), followed by the positive isolation with Dynabeads Mouse panB (B220) (Dynal, Oslo, Norway). Detection of hypermutations PCR-amplified DNA was separated by electrophoresis in a 1.5% agarose gel and subcloned into the pGEM-T vector (Promega, Madison, WI). Nucleotide sequences of the Vκ21C (second exon) were determined with the DNA sequencer Model 4000 (LI-COR, Lincoln, NE). Results Detection of somatic hypermutations in the transgenic substrates Three transgenic constructs, 3′-r, rmPU.1 and rmEM5, were used in the present study (Fig. 1A). The control substrate (3′-r) contains the murine Vκ21C and the Jκ– Cκ segments, followed by the 9-kb downstream region that includes the κE3′ (26). The rmPU.1 and rmEM5 substrates are the variants of the 3′-r, carrying base substitutions in the PU.1 and the NF-EM5 binding motifs respectively in the κE3′ (27). We have studied whether the base substitutions in the PU.1 or the NF-EM5 binding motif would influence the frequency of somatic hypermutation. Germinal center (GC) B cells were isolated from Peyer' s patches and sorted for the PNAhigh/B220+ phenotype. To detect hypermutations, rearranged structures of Vκ21C– Jκ1 were amplified from genomic DNA with two sets of PCR primers (Fig. 1A). We first compared the mutation frequency of the wild-type substrate, 3′-r, to that of the endogenous Vκ21C amplified from the non-transgenic (non-Tg) mice. As summarized in Table 1 and Fig. 2, the mutation frequency in the 3′-r was as high as in the endogenous Vκ21C (non-Tg), although there was founder-to-founder variation, most likely due to the integration site of the transgene. Although our PCR primers were not designed to distinguish the transgenic Vκ21C from the endogenous one, the rearranged Vκ21C– Jκ1 structures were mostly derived from the transgenes. This is because the joining frequency of the transgenic Vκ21C is much higher than that of the endogenous Vκ21C, probably due to the closeness of the V and J segments in the transgenic construct (Fig. 1A). To estimate the recombination frequencies of the transgenes, we performed semi-quantitative PCR with the rearranged Vκ21C– Jκ1 segment as described in Methods (Fig. 1B). We first amplified the endogenous Vκ41– Jκ1 sequences from the genomic DNA (10 ng each) and found that they were amplified exponentially without saturating (Fig. 1B, upper panels). Under the same PCR condition, we then simultaneously amplified the Vκ21C– Jκ1 and Vκ41– Jκ1 structures (Fig. 1B, lower panels). In the non-Tg mice, the ratio of the two PCR products (Vκ21C/Vκ41) was 0.04 after 24 cycles. In the transgenic samples, the ratios were reversed: relative amounts of Vκ21C (Vκ21C/Vκ41) were two orders of magnitude higher (6.53– 12.8) than those for the non-Tg samples. Using the Vκ41 band as an internal control, we estimated the contribution of the endogenous Vκ21C to be 0.61% in the total PCR products of Vκ21C. This value is ~1/10 of the PCR error (Table 1). In our statistical analyses, we used mice whose mutation frequencies were higher than the error frequency of PCR. Therefore, we conclude that the contribution of the endogenous Vκ21C rearrangement is not significant enough to disturb our data. In parallel with the 3′-r substrate, we analyzed the rearranged Vκ21C– Jκ1 sequences from the rmPU.1 and the rmEM5 transgenes. In the wild-type control (3′-r), mutation frequencies and occurrences of multiple base changes were comparable to those in the endogenous gene (non-Tg), although founder-to-founder variations were noticed. In the rmPU.1 and the rmEM5 substrates, overall mutation frequencies were significantly reduced (Table 1 and Fig. 2). We then examined the occurrence of multiple base substitutions in the mutant substrates, as previously described by Goyenechea et al. (15). The proportions of non-mutated clones for the rmPU.1 and rmEM5 transgenes were not so much different from those for the control substrate (3′-r) . However, the proportion of clones with a higher incidence of mutations dropped. Interference of targeted mutagenesis in the rmPU.1 and rmEM5 substrates We further examined the nature of the base substitutions in the mutant substrates, rmPU.1 and rmEM5. In the control substrate (3′-r), the distribution of mutations within the Vκ21C sequence was basically the same as that in the endogenous Vκ21C of the non-transgenic mice (non-Tg). In both non-Tg and 3′-r samples, mutations were concentrated in the CDR residues, particularly in CDR1 (Fig. 3). Mutations were also targeted to the hotspot sequences, RGYW, or its complementary decoded sequence, WRCY (where R = purine, Y = pyrimidine and W = A/T) (Figs 3 and 4). When we analyzed the rmPU.1 and rmEM5 substrates, mutations were reduced and distributed evenly throughout the Vκ21C gene. In contrast to the wild-type controls, mutations in the rmPU.1 and rmEM5 substrates were not concentrated in the CDR nor targeted to the RGYW/WRCY hotspot motifs (Figs 3 and 4). For example, variabilities of residues 56, 103, 115 and 128 (indicated by asterisks in Fig. 3) are >8% in the wild-type controls (non-Tg and 3′-r), but lower than the background level in the mutant substrates (rmPU.1 and rmEM5). The accumulation of hypermutations within the CDR and RGYW target sequences was not due to the selection by antigens: we analyzed the non-functionally rearranged (out-of-frame) clones. In the 3′-r clones, 41.6 and 50.3% of 149 mutations in 31 clones were concentrated within the CDR and RGYW motifs respectively (269 mutations analyzed in 53 clones). It should be noted that the CDR regions and RGYW motifs occupy only 29.0% of the Vκ21C region. In contrast, when the out-of-frame clones of the mutant substrates (rmPU.1 and rmEM5) were analyzed (45 mutants in 86 clones), mutations were neither accumulated in the CDR (26.7%) nor in the RGYW motifs (26.7%). These results indicate that the PU.1 and NF-EM5 binding sequences bear important roles in targeting somatic hypermutations to both CDR and hotspot residues. Figure 5 summarizes the directions of base substitutions for each nucleotide. The nature of hypermutations in the transgenic control, 3′-r, was basically the same as that in the endogenous Vκ21C (non-Tg), where a strong strand bias was seen: higher mutation frequencies at A residues than at Ts on the coding strand. No bias was found between the A/T and G/C pairs in the control substrates (Fig. 5B). When both rmPU.1 and rmEM5 substrates were analyzed, the proportions of mutations in the total mutations were increased at T residues, while those at Gs were decreased. The goodness-of-fit χ2-test was used to compare the preference of nucleotide targeting in Tg mice to that in non-Tg mice. In this test, values of P < 0.05 were considered to be statistically significant. The changes in the targeting preference observed in the rmPU.1 and rmEM5 mice were statistically significant, because their P values were 1.8×10–20 and 7.4×10–15 respectively, while that in the wild-type (3′-r) mice was not (P = 0.16). These changes were not related to the selection by antigen, because the mutations in the non-functionally rearranged (out-of-frame) clones were found to be comparable. Proportions of mutated nucleotides after correcting the values for base composition are as follows: 35.2% (A), 33.0% (G), 19.8% (C) and 12.0% (T) for the wild-type (3′-r) substrate, and 40.0% (A), 7.0% (G), 10.7% (C) and 42.3% (T) for the mutant substrates. As a result, the strand bias at A residues over Ts was not as evident in the mutant substrates (Fig. 5A), and proportions of mutation at A/T pairs were higher than at G/C pairs (Fig. 5B). We further analyzed the mutation frequencies (mutations/1000 bp) at different nucleotides. To our surprise, mutation frequencies at T residues never fell for both rmPU.1 and rmEM5, while those at G, C or A residues dropped significantly (Fig. 6). Thus the drops of overall mutation frequencies in the rmPU.1 and rmEM5 substrates, shown in Table 1 and Fig. 2, were probably caused by the selective suppression of mutagenesis at G, C and A residues, but not at Ts. Discussions Somatic hypermutations in the transgenic substrates In the present study, we have analyzed the effect of base substitutions in the PU.1 and NF-EM5 binding motifs on the somatic hypermutations introduced into the Igκ transgene. It was found that the overall mutation frequency was lowered, due to the absence of highly mutated clones (Table 1 and Fig. 2). However, the proportions of non-mutated clones from the rmPU.1 and rmEM5 transgenes were almost the same or only slightly higher when compared with those from the control substrate (Fig. 2). It appears that the base substitutions in the PU.1 and NF-EM5 motifs somehow affect the accumulation of hypermutations, rather than the commitment to mutagenesis. This is in sharp contrast to the previous observations of the Igκ transgenes with a deleted core or the flanking sequences of κE3′: the number of non-mutated clones was doubled compared to the wild-type control with the intact κE3′ (15). Another intriguing observation in the present study was the altered distribution of hypermutations in the rmPU.1 and rmEM5 substrates. It has been reported that somatic hypermutations are preferentially introduced into the CDR and RGYW target residues (1–3). Our transgenic studies with the wild-type construct confirmed this notion (Figs 3 and 4). In contrast, when the mutant substrates, rmPU.1 and rmEM5, were analyzed, mutation frequencies were decreased and somatic hypermutations were concentrated neither in the CDR nor in the target residues. These results indicate that the PU.1 and NF-EM5 binding motifs may be responsible for targeting hypermutations to the mutation hotspots in the Ig Vκ regions. In the wild-type controls (3′-r and non-Tg), mutations accumulated at G, C and A residues in the hotspots, but not at Ts (Fig. 3). In the rmPU.1 and rmEM5, mutagenesis was blocked at G, C and A, but not at Ts, which showed base-level mutations (Fig. 6). As a result, mutation frequencies at T residues in rmPU.1 and rmEM5 became sufficiently higher than the background. Thus, mutations in the rmPU.1 and rmEM5 substrates are biased towards A/T pairs over G/C pairs (Fig. 5). These shifts in the mutation patterns were considered to be independent from the antigen selection, because similar biases were also observed when the non-functionally rearranged (out of frame) clones were analyzed. Possible roles of κE3′ in the hypermutagenesis of the Igκ gene Besides hypermutagenesis, the κE3′ is known to be involved in the regulation of recombination and transcription. In Igκ recombination, the PU.1 binding motif, but not the NF-EM5 motif, appears to act as a cis-acting element that negatively regulates Vκ– Jκ joining (27). In the transcriptional regulation, the Igκ gene is activated by the complex of PU.1/NF-EM5 proteins that binds to the κE3′ core. In the present study, we have demonstrated that both the PU.1 and NF-EM5 binding motifs are equally crucial for the generation of somatic hypermutation. This may suggest that the complex formation of proteins binding these neighboring motifs is important in the regulation of hypermutagenesis. Although the transgenic studies indicate an active role of the κE3′ in hypermutagenesis, knock-out studies gave different results: knocking out the κE3′ region did not abolish the affinity maturation of Vκ sequences (28,29). One explanation for this discrepancy would be a difference in sources of GC B cells. In the present study, we have analyzed Peyer's patch GC B cells for the response to diverse gut antigens. In contrast, in the knock-out studies, VκOX1 response to a specific hapten was analyzed with splenic GC B cells. In the knock-out mouse, mutation frequencies at the residues responsible for the affinity maturation are comparable to those in the wild-type mouse. However, this may be due to the selection by antigen. Actually, mutation frequencies at other residues are lowered in the knock-out mouse. Thus, it is important to study the mutation frequencies in both types of animals, transgenic and knock-out, with the same system in parallel. Since the deletions in the knock-out mice covered an entire κE3′ core, it would be also interesting to generate the knock-in animal that contains base substitutions in the PU.1 or the NF-EM5 binding motif for the study of the total mutational response. When the core or the flanking sequences of the κE3′ is deleted, the level of transcription is decreased, resulting in the drop of mutation frequencies (15). Using the luciferase assay (30), it has been reported that the base substitutions in the PU.1 or NF-EM5 binding motif lower the enhancer activity to approximately one-third of the normal level. It is possible to argue that the increase of non-mutated clones in the rmPU.1 and the rmEM5 substrates may be due to the decreased transcription. However, this does not seem to explain the shift of mutation patterns in the mutant substrates, rmPU.1 and rmEM5, where the mutations were randomly distributed and biased towards A/T pairs over G/C. This is a sharp contrast to some mismatch-repair-deficient mice, such as Msh2– /– , where the mutations are concentrated in the hotspots and have strong preference for G/C pairs over A/T pairs (6– 11). It has been proposed that hypermutations occur through two distinct processes: one is hotspot-focused mutagenesis (G/C biased) and the other is hotspot unfocused mutagenesis (A/T biased) (9). In the rmPU.1 and rmEM5 substrates, not only were the mutation rates lowered, but the mutation patterns were also changed. We assume that this is because the hotspot mutagenesis was disrupted. In other words, the PU.1 and NF-EM5 binding motifs may be involved in the hotspot mutagenesis, but not in the unfocused mutagenesis. Recently, it has been reported that DNA double-strand breaks (DSB) frequently occur in Ig genes of cells undergoing somatic hypermutation (31,32). Since these DSB are targeted to the RGYW motifs and enhanced by Ig enhancer sequences, it has been proposed that hypermutations are triggered by DSB followed by error-prone synthesis (31). Interference of targeted mutagenesis in the rmPU.1 and rmEM5 substrates, which we observed in this study, may be the impairment of mutagenesis induced by such DSB. One possible mechanism is that the PU.1/NF-EM5 complex recruits the DSB enzyme to the hotspots or makes the DNA region accessible to the mutation machinery. Alternatively, the PU.1/NF-EM5 binding region may be necessary for recruiting error-prone polymerase for the DSB repair. In either case, since even bacterial genes can be targets for hypermutation in the presence of κE3′ together with the intron enhancer, there must be a specific cis-acting element in the κE3′ region that recruits the hypermutation machinery. Our present results support the idea that the PU.1 and the NF-EM5 motifs are the candidates for such cis-acting elements. Table 1. Hypermutations in the Vκ21C transgenes Mouse lines    No. of clones  No. of mutations  Mutations/103bp (for all clones)      Total  Mutated      The numbers of clones, mutations and overall mutation frequencies are compared for the three transgenic constructs: 3′-r, rmPU.1 and rmEM5. To obtain the mutation frequencies, the total number of mutations was divided by the total number of nucleotides from the second exon of Vκ21C. The background PCR error frequency was determined with bone marrow B cells as described in Methods. Founders with very low mutation frequencies were omitted for statistical analyses. The following transgenic mice lines were used in the present study: 3′-r [D6 (6 copies) and D11 (1)], rmPU.1 [#15 (1), #53 (13) and #38 (1)] and rmEM5 [#58 (1) and #61 (1)]. To estimate the contribution of the endogenous Vκ21C sequence in the rearranged Vκ21C transgenes, we performed semi-quantitative PCR using another endogenous Vκ gene, Vκ41, as an internal standard (see Methods).  Non-Tg  #1  30  20  96  10.9    #2  30  24  56  5.7  Wild-type (3′-r)  #D6  25  15  140  18.2    #D11  25  14  76  9.9  rmPU.1  #53  49  23  43  2.9    #15  49  14  40  2.7    #38  57  19  21  1.2  rmEM5  #58  58  21  69  3.9    #61  26  6  13  1.6  PCR error          0.5  Mouse lines    No. of clones  No. of mutations  Mutations/103bp (for all clones)      Total  Mutated      The numbers of clones, mutations and overall mutation frequencies are compared for the three transgenic constructs: 3′-r, rmPU.1 and rmEM5. To obtain the mutation frequencies, the total number of mutations was divided by the total number of nucleotides from the second exon of Vκ21C. The background PCR error frequency was determined with bone marrow B cells as described in Methods. Founders with very low mutation frequencies were omitted for statistical analyses. The following transgenic mice lines were used in the present study: 3′-r [D6 (6 copies) and D11 (1)], rmPU.1 [#15 (1), #53 (13) and #38 (1)] and rmEM5 [#58 (1) and #61 (1)]. To estimate the contribution of the endogenous Vκ21C sequence in the rearranged Vκ21C transgenes, we performed semi-quantitative PCR using another endogenous Vκ gene, Vκ41, as an internal standard (see Methods).  Non-Tg  #1  30  20  96  10.9    #2  30  24  56  5.7  Wild-type (3′-r)  #D6  25  15  140  18.2    #D11  25  14  76  9.9  rmPU.1  #53  49  23  43  2.9    #15  49  14  40  2.7    #38  57  19  21  1.2  rmEM5  #58  58  21  69  3.9    #61  26  6  13  1.6  PCR error          0.5  View Large Fig. 1. View largeDownload slide Somatic hypermutations in the transgenic constructs. (A) Schematic diagrams of the transgenes. The wild-type construct, 3′-r, contains the murine Vκ21C and Jκ–Cκ segments including the κE3′ region (26). Mutant substrates, rmPU.1 and rmEM5, contain base substitutions (in italics) in the PU.1 and NF-EM5 binding motifs respectively in the κE3′ (27). Arrowheads in the Vκ and Jκ regions indicate two sets of PCR primers (see Methods). (B) PCR experiments to detect the rearranged Vκ21C– Jκ1 and/or Vκ41C– Jκ1 sequences. The first PCR was repeated 18 times with 10 ng of DNA. In the second PCR, 1/500 of the first PCR product was used. The numbers of cycles for the second PCR are indicated. The rearranged Vκ41 band was used as an internal standard to estimate the ratio of Vκ21C–Jκ1/Vκ41–Jκ1. Numbers in parentheses are the ratios of the two PCR products (Vκ21C/Vκ41) after 24 cycles of the second PCR. Fig. 1. View largeDownload slide Somatic hypermutations in the transgenic constructs. (A) Schematic diagrams of the transgenes. The wild-type construct, 3′-r, contains the murine Vκ21C and Jκ–Cκ segments including the κE3′ region (26). Mutant substrates, rmPU.1 and rmEM5, contain base substitutions (in italics) in the PU.1 and NF-EM5 binding motifs respectively in the κE3′ (27). Arrowheads in the Vκ and Jκ regions indicate two sets of PCR primers (see Methods). (B) PCR experiments to detect the rearranged Vκ21C– Jκ1 and/or Vκ41C– Jκ1 sequences. The first PCR was repeated 18 times with 10 ng of DNA. In the second PCR, 1/500 of the first PCR product was used. The numbers of cycles for the second PCR are indicated. The rearranged Vκ41 band was used as an internal standard to estimate the ratio of Vκ21C–Jκ1/Vκ41–Jκ1. Numbers in parentheses are the ratios of the two PCR products (Vκ21C/Vκ41) after 24 cycles of the second PCR. Fig. 2. View largeDownload slide Hypermutations detected in the second exon of the Vκ21C. Genomic DNA was isolated from PNAhigh/B220+ cells of Peyer' s patch germinal centers. PCR amplification and detection of mutations are described in Methods. For each pie chart, each sector represents a proportion of clones with the indicated number of mutations. Mutations in the endogenous Vκ21C (non-Tg) and PCR errors (see Methods) are also shown. Fig. 2. View largeDownload slide Hypermutations detected in the second exon of the Vκ21C. Genomic DNA was isolated from PNAhigh/B220+ cells of Peyer' s patch germinal centers. PCR amplification and detection of mutations are described in Methods. For each pie chart, each sector represents a proportion of clones with the indicated number of mutations. Mutations in the endogenous Vκ21C (non-Tg) and PCR errors (see Methods) are also shown. Fig. 3. View largeDownload slide Distribution of hypermutations in the second exon of Vκ21C. Variabilities (mutated clones/total clones) are shown as percentage for each nucleotide position of the Vκ21C transgenes. Endogenous Vκ21C sequences of the non-transgenic (non-Tg) mice were analyzed as controls. Horizontal black bars indicate the RGYW/WRCY target sequences. CDR residues are indicated. Asterisks represent typical mutational hotspots, whose variabilities are >8% in both wild-type controls, non-Tg and 3′-r. Fig. 3. View largeDownload slide Distribution of hypermutations in the second exon of Vκ21C. Variabilities (mutated clones/total clones) are shown as percentage for each nucleotide position of the Vκ21C transgenes. Endogenous Vκ21C sequences of the non-transgenic (non-Tg) mice were analyzed as controls. Horizontal black bars indicate the RGYW/WRCY target sequences. CDR residues are indicated. Asterisks represent typical mutational hotspots, whose variabilities are >8% in both wild-type controls, non-Tg and 3′-r. Fig. 4. View largeDownload slide Targeting of mutations to the hotspot motifs (RGYW/WRCY). The proportions of mutations inside and outside of the hotspot motifs (RGYW/WRCY) were plotted. The expected distribution of random mutations was also plotted. Fig. 4. View largeDownload slide Targeting of mutations to the hotspot motifs (RGYW/WRCY). The proportions of mutations inside and outside of the hotspot motifs (RGYW/WRCY) were plotted. The expected distribution of random mutations was also plotted. Fig. 5. View largeDownload slide Biased mutations at A/T residues in the rmPU.1 and rmEM5 constructs. (A) Directions of base substitutions are compared among the Vκ21C transgene sequences (3′-r, rmPU.1 and rmEM5). The endogenous Vκ21C gene of the non-transgenic (non-Tg) mouse was also analyzed as a control. Total numbers of mutations analyzed in each construct are 152 (non-Tg), 216 (3′-r), 104 (rmPU.1) and 82 (rmEM5). Base preferences for mutations are shown in percentage (total mutations = 100%). Among 307 residues in the Vκ21C coding sequence, there are 77 As, 73 Gs, 80 Cs and 77 Ts. Percentages of base substitutions are corrected for the base composition. (B) Proportions of targeted bases for mutations are compared between the A/T and G/C pairs. Three types of transgenic (3′-r, rmPU.1 and rmEM5) and the endogenous (non-Tg) Vκ21C genes are compared. The result on the endogenous Vκ21C gene in non-transgenic mice is also presented. Fig. 5. View largeDownload slide Biased mutations at A/T residues in the rmPU.1 and rmEM5 constructs. (A) Directions of base substitutions are compared among the Vκ21C transgene sequences (3′-r, rmPU.1 and rmEM5). The endogenous Vκ21C gene of the non-transgenic (non-Tg) mouse was also analyzed as a control. Total numbers of mutations analyzed in each construct are 152 (non-Tg), 216 (3′-r), 104 (rmPU.1) and 82 (rmEM5). Base preferences for mutations are shown in percentage (total mutations = 100%). Among 307 residues in the Vκ21C coding sequence, there are 77 As, 73 Gs, 80 Cs and 77 Ts. Percentages of base substitutions are corrected for the base composition. (B) Proportions of targeted bases for mutations are compared between the A/T and G/C pairs. Three types of transgenic (3′-r, rmPU.1 and rmEM5) and the endogenous (non-Tg) Vκ21C genes are compared. The result on the endogenous Vκ21C gene in non-transgenic mice is also presented. Fig. 6. View largeDownload slide Frequencies of hypermutations at different kinds of nucleotides. Base preferences for the mutation are compared among three transgenic Vκ21C constructs (3′-r, rmPU.1 and rmEM5). Mutation frequencies (substituted bases in 103 nucleotides) for each nucleotide are normalized with those of the endogenous Vκ21C gene (non-Tg). Fig. 6. View largeDownload slide Frequencies of hypermutations at different kinds of nucleotides. Base preferences for the mutation are compared among three transgenic Vκ21C constructs (3′-r, rmPU.1 and rmEM5). Mutation frequencies (substituted bases in 103 nucleotides) for each nucleotide are normalized with those of the endogenous Vκ21C gene (non-Tg). Transmitting editor: T. Taniguchi We thank Dr Akio Tsuboi and Hitomi Sakano for helpful discussion and critical reading of the manuscript. This work was supported by the Special Promotion Research Grant from Monbusho and by grants from Toray Science Foundation, Nissan Science Foundation, Mitsubishi Foundation and Japan Foundation for Applied Enzymology. M. K. and R. H. are Research Fellows of the Japan Society for Promotion of Science. H. N. was supported by research grants from Kurata Foundation, Uehara Memorial Foundation and the Naito Foundation. References 1 Maizels, N. 1995. Somatic hypermutation: how many mechanisms diversify V region sequences? Cell  83: 9. Google Scholar 2 Rajewsky, K. 1996. Clonal selection and learning in the antibody system. Nature  381: 751. Google Scholar 3 Dorner, T., Foster, S. J., Brezinschek, H. P. and Lipsky, P. E. 1998. Analysis of the targeting of the hypermutational machinery and the impact of subsequent selection on the distribution of nucleotide changes in human VHDJH rearrangements. Immunol. Rev.  162: 161. Google Scholar 4 Kim, N. and Storb, U. 1998. The role of DNA repair in somatic hypermutation of immunoglobulin genes. J. Exp. Med.  187: 1729. Google Scholar 5 Reynaud, C. A., Bertocci, B., Frey, S., Delbos, F., Quint, L. and Weill, J. C. 1999. Mismatch repair and immunoglobulin gene hypermutation: did we learn something? Immunol. Today  20: 522. Google Scholar 6 Jacobs, H., Fukita, Y., van der Horst, G. T., de Boer, J., Weeda, G., Essers, J., de Wind, N., Engelward, B. P., Samson, L., Verbeek, S., de Murcia, J. M., de Murcia, G., te Riele, H. and Rajewsky, K. 1998. Hypermutation of immunoglobulin genes in memory B cells of DNA repair-deficient mice. J. Exp. Med.  187: 1735. Google Scholar 7 Phung, Q. H., Winter, D. B., Cranston, A., Tarone, R. E., Bohr, V. A., Fishel, R. and Gearhart, P. J. 1998. Increased hypermutation at G and C nucleotides in immunoglobulin variable genes from mice deficient in the MSH2 mismatch repair protein. J. Exp. Med.  187: 1745. Google Scholar 8 Frey, S., Bertocci, B., Delbos, F., Quint, L., Weill, J. C. and Reynaud, C. A. 1998. Mismatch repair deficiency interferes with the accumulation of mutations in chronically stimulated B cells and not with the hypermutation process. Immunity  9: 127. Google Scholar 9 Rada, C., Ehrenstein, M. R., Neuberger, M. S. and Milstein, C. 1998. Hot spot focusing of somatic hypermutation in MSH2-deficient mice suggests two stages of mutational targeting. Immunity  9: 135. Google Scholar 10 Kim, N., Bozek, G., Lo, J. C. and Storb, U. 1999. Different mismatch repair deficiencies all have the same effects on somatic hypermutation: intact primary mechanism accompanied by secondary modifications. J. Exp. Med.  190: 21. Google Scholar 11 Wiesendanger, M., Kneitz, B., Edelmann, W. and Scharff, M. D. 2000. Somatic hypermutation in MutS homologue (MSH)3-, MSH6-, and MSH3/MSH6-deficient mice reveals a role for the MSH2–MSH6 heterodimer in modulating the base substitution pattern. J. Exp. Med.  191: 579. Google Scholar 12 Storb, U., Peters, A., Klotz, E., Kim, N., Shen, H. M., Hackett, J., Rogerson, B. and Martin, T. E. 1998. Cis-acting sequences that affect somatic hypermutation of Ig genes. Immunol. Rev.  162: 153. Google Scholar 13 Storb, U., Peters, A., Klotz, E., Kim, N., Shen, H. M., Hackett, J., Rogerson, B., O'Brien, R. and Martin, T. E. 1998. Immunoglobulin transgenes as targets for somatic hypermutation. Int. J. Dev. Biol.  42: 977. Google Scholar 14 Betz, A. G., Milstein, C., Gonzalez-Fernandez, A., Pannell, R., Larson, T. and Neuberger, M. S. 1994. Elements regulating somatic hypermutation of an immunoglobulin κ gene: critical role for the intron enhancer/matrix attachment region. Cell  77: 239. Google Scholar 15 Goyenechea, B., Klix, N., Yelamos, J., Williams, G. T., Riddell, A., Neuberger, M. S. and Milstein, C. 1997. Cells strongly expressing Igκ transgenes show clonal recruitment of hypermutation: a role for both MAR and the enhancers. EMBO J.  16: 3987. Google Scholar 16 Peters, A. and Storb, U. 1996. Somatic hypermutation of immunoglobulin genes is linked to transcription initiation. Immunity  4: 57. Google Scholar 17 Fukita, Y., Jacobs, H. and Rajewsky, K. 1998. Somatic hypermutation in the heavy chain locus correlates with transcription. Immunity  9: 105. Google Scholar 18 Storb, U., Klotz, E. L., Hackett, J., Jr, Kage, K., Bozek, G. and Martin, T. E. 1998. A hypermutable insert in an immunoglobulin transgene contains hotspots of somatic mutation and sequences predicting highly stable structures in the RNA transcript. J. Exp. Med.  188: 689. Google Scholar 19 Bachl, J. and Wabl, M. 1996. An immunoglobulin mutator that targets G.C base pairs. Proc. Natl Acad. Sci. USA  93: 851. Google Scholar 20 Yelamos, J., Klix, N., Goyenechea, B., Lozano, F., Chui, Y. L., Gonzalez-Fernandez, A., Pannell, R., Neuberger, M. S. and Milstein, C. 1995. Targeting of non-Ig sequences in place of the V segment by somatic hypermutation. Nature  376: 225. Google Scholar 21 Azuma, T., Motoyama, N., Fields, L. E. and Loh, D. Y. 1993. Mutations of the chloramphenicol acetyl transferase transgene driven by the immunoglobulin promoter and intron enhancer. Int. Immunol.  5: 121. Google Scholar 22 Umar, A., Schweitzer, P. A., Levy, N. S., Gearhart, J. D. and Gearhart, P. J. 1991. Mutation in a reporter gene depends on proximity to and transcription of immunoglobulin variable transgenes. Proc. Natl Acad. Sci. USA  88: 4902. Google Scholar 23 Gonzalez-Fernandez, A. and Milstein, C. 1993. Analysis of somatic hypermutation in mouse Peyer's patches using immunoglobulin κ light-chain transgenes. Proc. Natl Acad. Sci. USA  90: 9862. Google Scholar 24 Gonzalez-Fernandez, A., Gilmore, D. and Milstein, C. 1994. Age-related decrease in the proportion of germinal center B cells from mouse Peyer's patches is accompanied by an accumulation of somatic mutations in their immunoglobulin genes. Eur. J. Immunol.  24: 2918. Google Scholar 25 Blin, N. and Stafford, D. W. 1976. A general method for isolation of high molecular weight DNA from eukaryotes. Nucleic Acids Res.  3: 2303. Google Scholar 26 Hiramatsu, R., Akagi, K., Matsuoka, M., Sakumi, K., Nakamura, H., Kingsbury, L., David, C., Hardy, R. R., Yamamura, K. and Sakano, H. 1995. The 3′ enhancer region determines the B/T specificity and pro-B/pre-B specificity of immunoglobulin Vκ–Jκ joining. Cell  83: 1113. Google Scholar 27 Hayashi, R., Takemori, T., Kodama, M., Suzuki, M., Tsuboi, A., Nagawa, F. and Sakano, H. The PU.1 binding site is a cis-element that regulates pro-B/pre-B specificity of Vκ–Jκ joining. 1997. J. Immunol.  159: 4145. Google Scholar 28 Gorman, J. R., van der Stoep, N., Monroe, R. R. Cogne, R., Davidson, L. and Alt, F. W. 1996. The Igκ 3′ enhancer influences the ratio of Igκ versus Igλ B lymphocytes. Immunity  5: 241. Google Scholar 29 van der Stoep, N., Gorman, J. R. and Alt, F. W. 1998. Reevaluation of 3′Eκ function in stage- and lineage-specific rearrangement and somatic hypermutation. Immunity  8: 743. Google Scholar 30 Liu, X., Prabhu, A. and Ness, B. V. 1999. Developmental regulation of the locus involves both positive and negative sequence elements in the 3′ enhancer that affect synergy with the intron enhancer. J. Biol. Chem.  274: 3285. Google Scholar 31 Papavasiliou, F. N. and Schatz, D. G. 2000. Cell-cycle-regulated DNA double-strand breaks in somatic hypermutation of the immunoglobulin genes. Nature  408: 216. Google Scholar 32 Bross, L., Fukita, Y., McBlane, F., Demolliere, C., Rajewsky, K. and Jacobs, H. 2000. DNA double-strand breaks in immunoglobulin genes undergoing somatic hypermutation. Immunity  13: 589 Google Scholar © 2001 Japanese Society for Immunology http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png International Immunology Oxford University Press

The PU.1 and NF-EM5 binding motifs in the Igκ 3′ enhancer are responsible for directing somatic hypermutations to the intrinsic hotspots in the transgenic Vκ gene

Download PDF
8 pages

Loading next page...
 
/lp/oxford-university-press/the-pu-1-and-nf-em5-binding-motifs-in-the-ig-3-enhancer-are-QTwHq4uJm8

References (31)

Publisher
Oxford University Press
Copyright
© 2001 Japanese Society for Immunology
ISSN
0953-8178
eISSN
1460-2377
DOI
10.1093/intimm/13.11.1415
Publisher site
See Article on Publisher Site

Abstract

Abstract Somatic hypermutation is a key mechanism in generating Ig with higher affinities to antigen, a process known as affinity maturation. Using Igκ transgenes, the 3′ enhancer (κE3′) has been shown to play an important role in introducing hypermutations. In order to identify the cis-acting elements that regulate hypermutagenesis, we have generated transgenic substrates containing mutations/deletions in the κE3′ region. Here, we report that base substitutions in the κE3′, either in the PU.1 or in the NF-EM5 binding motif, not only reduce the mutation rate but also disrupt the directed mutagenesis in the intrinsic hotspots of the Igκ transgene. antibodies, B lymphocytes, generation of diversity, transgenic, knock-out CDR complementarity-determining region, DSB double-strand break, GC germinal center, κE3′ Igκ 3′ enhancer, iEκ Igκ intron enhancer, MAR matrix attachment region, PNA peanut agglutinin Introduction Following antigen stimulation, the primary B cell repertoire is shaped by the second wave of somatic DNA changes, known as somatic hypermutation, necessary to generate the variant types of Ig genes coding for the high-affinity antibodies. Nucleotide substitutions are predominantly introduced into three complementarity-determining regions (CDR) of Ig variable (V) segments, at a rate of 10–3 to 10–4 bp/cell generation (1–3). Recently, several groups analyzed the somatic hypermutations in mice that are deficient for DNA repair functions and found that the disruption of the mouse homologues of yeast mismatch repair genes affects the hypermutations at intrinsic hotspots (4–11). In the study of somatic hypermutation, transgenic mice have been quite useful in identifying cis-acting elements for the high-frequency mutagenesis in the Ig gene loci (12,13). Using rearranged Igκ transgenes, it has been shown that both the 3′-enhancer (κE3′) and the intronic enhancer/ matrix attachment region (iEκ/MAR) are critical for introducing somatic hypermutations to the Vκ segments (14,15). Deletion of either enhancer, κE3′ or iEκ, caused a severe drop in the mutation frequency. It has been reported that the extent of mutations is correlated with the level of transcription of the Igκ transgene (16–18). However, findings that the Ig enhancers can even induce hypermutations in the non-Ig sequences indicate a more directive role (19–22). In order to narrow down and identify the cis-acting elements for hypermutagenesis, we have introduced base substitutions in the κE3′ region and analyzed the effects on the introduction of hypermutations in the Igκ transgene. It was found that base substitutions in the PU.1 or the NF-EM5 binding sequence not only reduce the mutation rate, but also affect the biased mutagenesis in the intrinsic hotspots. Methods Sorting cells The B cell fraction of PNAhigh/B220+ was sorted from small intestinal Peyer's patches of unimmunized, 5- to 12-month old, and C57BL/6J-background mice, irrespective of sex (23,24). Cells were incubated with anti-FcγRII/III (2.4G2; PharMingen, San Diego, CA), FITC-conjugated peanut agglutinin (PNA; Vector, Burlingame, CA) and allophycocyanin-coupled anti-B220 (mAb RA3-6B2; PharMingen) in cold PBS supplemented with 2% newborn goat serum, 0.05% sodium azide and 2 mM EDTA (pH 8.0) for sorting with a FACS Vantage cell sorter. PCR amplification The Vκ21C region was amplified by two successive PCR. Genomic DNA was isolated as described (25). The first PCR was performed with 5–10 ng of DNA, using the Expand High Fidelity PCR system (Roche, Mannheim, Germany). The PCR cycle consists of denaturation (94°C, 15 s), annealing (63°C, 30 s) and extension (72°C, 52 s). The cycle was repeated 18 times according to the manufacturer' s instruction. In the second nested PCR, 1/500 of the first PCR product was used. The extension time was 45 s. The cycle was repeated 24 times for the transgenic DNA and 26 times for the non-transgenic DNA. For the detection of the rearranged Vκ21C–Jκ1, two sets of primers were used: (Vκ21C-301) 5′-AAATTTTCTGCTAACACCAACTTCCTGTGG-3′/(Jκ1–30) 5′-GTGCCAGAATCTTGGTTTCAGAGTAAGATT-3′ and (Vκ21C-303) 5′-TGGTGTCTAATAGTGATGTCCGCAGTATTC-3′/(Jκ1–302)5′-TTAGACATAGAAGCCACAGACATAGACAAC-3′. For the semi-quantitative PCR, primer Vκ41–30 was used: 5′-TAACAGTCAAACATATCCTGTGCCATTGTC-3′. PCR products were separated in agarose gels (1.5%), stained with ethidium bromide and quantified with Densitograph software library, version 4.0 (ATTO, Tokyo, Japan). To estimate PCR errors, bone marrow B cells were purified from 12-week-old mice using Lympholyte-M (Cedarlane, Hornby, Canada), followed by the positive isolation with Dynabeads Mouse panB (B220) (Dynal, Oslo, Norway). Detection of hypermutations PCR-amplified DNA was separated by electrophoresis in a 1.5% agarose gel and subcloned into the pGEM-T vector (Promega, Madison, WI). Nucleotide sequences of the Vκ21C (second exon) were determined with the DNA sequencer Model 4000 (LI-COR, Lincoln, NE). Results Detection of somatic hypermutations in the transgenic substrates Three transgenic constructs, 3′-r, rmPU.1 and rmEM5, were used in the present study (Fig. 1A). The control substrate (3′-r) contains the murine Vκ21C and the Jκ– Cκ segments, followed by the 9-kb downstream region that includes the κE3′ (26). The rmPU.1 and rmEM5 substrates are the variants of the 3′-r, carrying base substitutions in the PU.1 and the NF-EM5 binding motifs respectively in the κE3′ (27). We have studied whether the base substitutions in the PU.1 or the NF-EM5 binding motif would influence the frequency of somatic hypermutation. Germinal center (GC) B cells were isolated from Peyer' s patches and sorted for the PNAhigh/B220+ phenotype. To detect hypermutations, rearranged structures of Vκ21C– Jκ1 were amplified from genomic DNA with two sets of PCR primers (Fig. 1A). We first compared the mutation frequency of the wild-type substrate, 3′-r, to that of the endogenous Vκ21C amplified from the non-transgenic (non-Tg) mice. As summarized in Table 1 and Fig. 2, the mutation frequency in the 3′-r was as high as in the endogenous Vκ21C (non-Tg), although there was founder-to-founder variation, most likely due to the integration site of the transgene. Although our PCR primers were not designed to distinguish the transgenic Vκ21C from the endogenous one, the rearranged Vκ21C– Jκ1 structures were mostly derived from the transgenes. This is because the joining frequency of the transgenic Vκ21C is much higher than that of the endogenous Vκ21C, probably due to the closeness of the V and J segments in the transgenic construct (Fig. 1A). To estimate the recombination frequencies of the transgenes, we performed semi-quantitative PCR with the rearranged Vκ21C– Jκ1 segment as described in Methods (Fig. 1B). We first amplified the endogenous Vκ41– Jκ1 sequences from the genomic DNA (10 ng each) and found that they were amplified exponentially without saturating (Fig. 1B, upper panels). Under the same PCR condition, we then simultaneously amplified the Vκ21C– Jκ1 and Vκ41– Jκ1 structures (Fig. 1B, lower panels). In the non-Tg mice, the ratio of the two PCR products (Vκ21C/Vκ41) was 0.04 after 24 cycles. In the transgenic samples, the ratios were reversed: relative amounts of Vκ21C (Vκ21C/Vκ41) were two orders of magnitude higher (6.53– 12.8) than those for the non-Tg samples. Using the Vκ41 band as an internal control, we estimated the contribution of the endogenous Vκ21C to be 0.61% in the total PCR products of Vκ21C. This value is ~1/10 of the PCR error (Table 1). In our statistical analyses, we used mice whose mutation frequencies were higher than the error frequency of PCR. Therefore, we conclude that the contribution of the endogenous Vκ21C rearrangement is not significant enough to disturb our data. In parallel with the 3′-r substrate, we analyzed the rearranged Vκ21C– Jκ1 sequences from the rmPU.1 and the rmEM5 transgenes. In the wild-type control (3′-r), mutation frequencies and occurrences of multiple base changes were comparable to those in the endogenous gene (non-Tg), although founder-to-founder variations were noticed. In the rmPU.1 and the rmEM5 substrates, overall mutation frequencies were significantly reduced (Table 1 and Fig. 2). We then examined the occurrence of multiple base substitutions in the mutant substrates, as previously described by Goyenechea et al. (15). The proportions of non-mutated clones for the rmPU.1 and rmEM5 transgenes were not so much different from those for the control substrate (3′-r) . However, the proportion of clones with a higher incidence of mutations dropped. Interference of targeted mutagenesis in the rmPU.1 and rmEM5 substrates We further examined the nature of the base substitutions in the mutant substrates, rmPU.1 and rmEM5. In the control substrate (3′-r), the distribution of mutations within the Vκ21C sequence was basically the same as that in the endogenous Vκ21C of the non-transgenic mice (non-Tg). In both non-Tg and 3′-r samples, mutations were concentrated in the CDR residues, particularly in CDR1 (Fig. 3). Mutations were also targeted to the hotspot sequences, RGYW, or its complementary decoded sequence, WRCY (where R = purine, Y = pyrimidine and W = A/T) (Figs 3 and 4). When we analyzed the rmPU.1 and rmEM5 substrates, mutations were reduced and distributed evenly throughout the Vκ21C gene. In contrast to the wild-type controls, mutations in the rmPU.1 and rmEM5 substrates were not concentrated in the CDR nor targeted to the RGYW/WRCY hotspot motifs (Figs 3 and 4). For example, variabilities of residues 56, 103, 115 and 128 (indicated by asterisks in Fig. 3) are >8% in the wild-type controls (non-Tg and 3′-r), but lower than the background level in the mutant substrates (rmPU.1 and rmEM5). The accumulation of hypermutations within the CDR and RGYW target sequences was not due to the selection by antigens: we analyzed the non-functionally rearranged (out-of-frame) clones. In the 3′-r clones, 41.6 and 50.3% of 149 mutations in 31 clones were concentrated within the CDR and RGYW motifs respectively (269 mutations analyzed in 53 clones). It should be noted that the CDR regions and RGYW motifs occupy only 29.0% of the Vκ21C region. In contrast, when the out-of-frame clones of the mutant substrates (rmPU.1 and rmEM5) were analyzed (45 mutants in 86 clones), mutations were neither accumulated in the CDR (26.7%) nor in the RGYW motifs (26.7%). These results indicate that the PU.1 and NF-EM5 binding sequences bear important roles in targeting somatic hypermutations to both CDR and hotspot residues. Figure 5 summarizes the directions of base substitutions for each nucleotide. The nature of hypermutations in the transgenic control, 3′-r, was basically the same as that in the endogenous Vκ21C (non-Tg), where a strong strand bias was seen: higher mutation frequencies at A residues than at Ts on the coding strand. No bias was found between the A/T and G/C pairs in the control substrates (Fig. 5B). When both rmPU.1 and rmEM5 substrates were analyzed, the proportions of mutations in the total mutations were increased at T residues, while those at Gs were decreased. The goodness-of-fit χ2-test was used to compare the preference of nucleotide targeting in Tg mice to that in non-Tg mice. In this test, values of P < 0.05 were considered to be statistically significant. The changes in the targeting preference observed in the rmPU.1 and rmEM5 mice were statistically significant, because their P values were 1.8×10–20 and 7.4×10–15 respectively, while that in the wild-type (3′-r) mice was not (P = 0.16). These changes were not related to the selection by antigen, because the mutations in the non-functionally rearranged (out-of-frame) clones were found to be comparable. Proportions of mutated nucleotides after correcting the values for base composition are as follows: 35.2% (A), 33.0% (G), 19.8% (C) and 12.0% (T) for the wild-type (3′-r) substrate, and 40.0% (A), 7.0% (G), 10.7% (C) and 42.3% (T) for the mutant substrates. As a result, the strand bias at A residues over Ts was not as evident in the mutant substrates (Fig. 5A), and proportions of mutation at A/T pairs were higher than at G/C pairs (Fig. 5B). We further analyzed the mutation frequencies (mutations/1000 bp) at different nucleotides. To our surprise, mutation frequencies at T residues never fell for both rmPU.1 and rmEM5, while those at G, C or A residues dropped significantly (Fig. 6). Thus the drops of overall mutation frequencies in the rmPU.1 and rmEM5 substrates, shown in Table 1 and Fig. 2, were probably caused by the selective suppression of mutagenesis at G, C and A residues, but not at Ts. Discussions Somatic hypermutations in the transgenic substrates In the present study, we have analyzed the effect of base substitutions in the PU.1 and NF-EM5 binding motifs on the somatic hypermutations introduced into the Igκ transgene. It was found that the overall mutation frequency was lowered, due to the absence of highly mutated clones (Table 1 and Fig. 2). However, the proportions of non-mutated clones from the rmPU.1 and rmEM5 transgenes were almost the same or only slightly higher when compared with those from the control substrate (Fig. 2). It appears that the base substitutions in the PU.1 and NF-EM5 motifs somehow affect the accumulation of hypermutations, rather than the commitment to mutagenesis. This is in sharp contrast to the previous observations of the Igκ transgenes with a deleted core or the flanking sequences of κE3′: the number of non-mutated clones was doubled compared to the wild-type control with the intact κE3′ (15). Another intriguing observation in the present study was the altered distribution of hypermutations in the rmPU.1 and rmEM5 substrates. It has been reported that somatic hypermutations are preferentially introduced into the CDR and RGYW target residues (1–3). Our transgenic studies with the wild-type construct confirmed this notion (Figs 3 and 4). In contrast, when the mutant substrates, rmPU.1 and rmEM5, were analyzed, mutation frequencies were decreased and somatic hypermutations were concentrated neither in the CDR nor in the target residues. These results indicate that the PU.1 and NF-EM5 binding motifs may be responsible for targeting hypermutations to the mutation hotspots in the Ig Vκ regions. In the wild-type controls (3′-r and non-Tg), mutations accumulated at G, C and A residues in the hotspots, but not at Ts (Fig. 3). In the rmPU.1 and rmEM5, mutagenesis was blocked at G, C and A, but not at Ts, which showed base-level mutations (Fig. 6). As a result, mutation frequencies at T residues in rmPU.1 and rmEM5 became sufficiently higher than the background. Thus, mutations in the rmPU.1 and rmEM5 substrates are biased towards A/T pairs over G/C pairs (Fig. 5). These shifts in the mutation patterns were considered to be independent from the antigen selection, because similar biases were also observed when the non-functionally rearranged (out of frame) clones were analyzed. Possible roles of κE3′ in the hypermutagenesis of the Igκ gene Besides hypermutagenesis, the κE3′ is known to be involved in the regulation of recombination and transcription. In Igκ recombination, the PU.1 binding motif, but not the NF-EM5 motif, appears to act as a cis-acting element that negatively regulates Vκ– Jκ joining (27). In the transcriptional regulation, the Igκ gene is activated by the complex of PU.1/NF-EM5 proteins that binds to the κE3′ core. In the present study, we have demonstrated that both the PU.1 and NF-EM5 binding motifs are equally crucial for the generation of somatic hypermutation. This may suggest that the complex formation of proteins binding these neighboring motifs is important in the regulation of hypermutagenesis. Although the transgenic studies indicate an active role of the κE3′ in hypermutagenesis, knock-out studies gave different results: knocking out the κE3′ region did not abolish the affinity maturation of Vκ sequences (28,29). One explanation for this discrepancy would be a difference in sources of GC B cells. In the present study, we have analyzed Peyer's patch GC B cells for the response to diverse gut antigens. In contrast, in the knock-out studies, VκOX1 response to a specific hapten was analyzed with splenic GC B cells. In the knock-out mouse, mutation frequencies at the residues responsible for the affinity maturation are comparable to those in the wild-type mouse. However, this may be due to the selection by antigen. Actually, mutation frequencies at other residues are lowered in the knock-out mouse. Thus, it is important to study the mutation frequencies in both types of animals, transgenic and knock-out, with the same system in parallel. Since the deletions in the knock-out mice covered an entire κE3′ core, it would be also interesting to generate the knock-in animal that contains base substitutions in the PU.1 or the NF-EM5 binding motif for the study of the total mutational response. When the core or the flanking sequences of the κE3′ is deleted, the level of transcription is decreased, resulting in the drop of mutation frequencies (15). Using the luciferase assay (30), it has been reported that the base substitutions in the PU.1 or NF-EM5 binding motif lower the enhancer activity to approximately one-third of the normal level. It is possible to argue that the increase of non-mutated clones in the rmPU.1 and the rmEM5 substrates may be due to the decreased transcription. However, this does not seem to explain the shift of mutation patterns in the mutant substrates, rmPU.1 and rmEM5, where the mutations were randomly distributed and biased towards A/T pairs over G/C. This is a sharp contrast to some mismatch-repair-deficient mice, such as Msh2– /– , where the mutations are concentrated in the hotspots and have strong preference for G/C pairs over A/T pairs (6– 11). It has been proposed that hypermutations occur through two distinct processes: one is hotspot-focused mutagenesis (G/C biased) and the other is hotspot unfocused mutagenesis (A/T biased) (9). In the rmPU.1 and rmEM5 substrates, not only were the mutation rates lowered, but the mutation patterns were also changed. We assume that this is because the hotspot mutagenesis was disrupted. In other words, the PU.1 and NF-EM5 binding motifs may be involved in the hotspot mutagenesis, but not in the unfocused mutagenesis. Recently, it has been reported that DNA double-strand breaks (DSB) frequently occur in Ig genes of cells undergoing somatic hypermutation (31,32). Since these DSB are targeted to the RGYW motifs and enhanced by Ig enhancer sequences, it has been proposed that hypermutations are triggered by DSB followed by error-prone synthesis (31). Interference of targeted mutagenesis in the rmPU.1 and rmEM5 substrates, which we observed in this study, may be the impairment of mutagenesis induced by such DSB. One possible mechanism is that the PU.1/NF-EM5 complex recruits the DSB enzyme to the hotspots or makes the DNA region accessible to the mutation machinery. Alternatively, the PU.1/NF-EM5 binding region may be necessary for recruiting error-prone polymerase for the DSB repair. In either case, since even bacterial genes can be targets for hypermutation in the presence of κE3′ together with the intron enhancer, there must be a specific cis-acting element in the κE3′ region that recruits the hypermutation machinery. Our present results support the idea that the PU.1 and the NF-EM5 motifs are the candidates for such cis-acting elements. Table 1. Hypermutations in the Vκ21C transgenes Mouse lines    No. of clones  No. of mutations  Mutations/103bp (for all clones)      Total  Mutated      The numbers of clones, mutations and overall mutation frequencies are compared for the three transgenic constructs: 3′-r, rmPU.1 and rmEM5. To obtain the mutation frequencies, the total number of mutations was divided by the total number of nucleotides from the second exon of Vκ21C. The background PCR error frequency was determined with bone marrow B cells as described in Methods. Founders with very low mutation frequencies were omitted for statistical analyses. The following transgenic mice lines were used in the present study: 3′-r [D6 (6 copies) and D11 (1)], rmPU.1 [#15 (1), #53 (13) and #38 (1)] and rmEM5 [#58 (1) and #61 (1)]. To estimate the contribution of the endogenous Vκ21C sequence in the rearranged Vκ21C transgenes, we performed semi-quantitative PCR using another endogenous Vκ gene, Vκ41, as an internal standard (see Methods).  Non-Tg  #1  30  20  96  10.9    #2  30  24  56  5.7  Wild-type (3′-r)  #D6  25  15  140  18.2    #D11  25  14  76  9.9  rmPU.1  #53  49  23  43  2.9    #15  49  14  40  2.7    #38  57  19  21  1.2  rmEM5  #58  58  21  69  3.9    #61  26  6  13  1.6  PCR error          0.5  Mouse lines    No. of clones  No. of mutations  Mutations/103bp (for all clones)      Total  Mutated      The numbers of clones, mutations and overall mutation frequencies are compared for the three transgenic constructs: 3′-r, rmPU.1 and rmEM5. To obtain the mutation frequencies, the total number of mutations was divided by the total number of nucleotides from the second exon of Vκ21C. The background PCR error frequency was determined with bone marrow B cells as described in Methods. Founders with very low mutation frequencies were omitted for statistical analyses. The following transgenic mice lines were used in the present study: 3′-r [D6 (6 copies) and D11 (1)], rmPU.1 [#15 (1), #53 (13) and #38 (1)] and rmEM5 [#58 (1) and #61 (1)]. To estimate the contribution of the endogenous Vκ21C sequence in the rearranged Vκ21C transgenes, we performed semi-quantitative PCR using another endogenous Vκ gene, Vκ41, as an internal standard (see Methods).  Non-Tg  #1  30  20  96  10.9    #2  30  24  56  5.7  Wild-type (3′-r)  #D6  25  15  140  18.2    #D11  25  14  76  9.9  rmPU.1  #53  49  23  43  2.9    #15  49  14  40  2.7    #38  57  19  21  1.2  rmEM5  #58  58  21  69  3.9    #61  26  6  13  1.6  PCR error          0.5  View Large Fig. 1. View largeDownload slide Somatic hypermutations in the transgenic constructs. (A) Schematic diagrams of the transgenes. The wild-type construct, 3′-r, contains the murine Vκ21C and Jκ–Cκ segments including the κE3′ region (26). Mutant substrates, rmPU.1 and rmEM5, contain base substitutions (in italics) in the PU.1 and NF-EM5 binding motifs respectively in the κE3′ (27). Arrowheads in the Vκ and Jκ regions indicate two sets of PCR primers (see Methods). (B) PCR experiments to detect the rearranged Vκ21C– Jκ1 and/or Vκ41C– Jκ1 sequences. The first PCR was repeated 18 times with 10 ng of DNA. In the second PCR, 1/500 of the first PCR product was used. The numbers of cycles for the second PCR are indicated. The rearranged Vκ41 band was used as an internal standard to estimate the ratio of Vκ21C–Jκ1/Vκ41–Jκ1. Numbers in parentheses are the ratios of the two PCR products (Vκ21C/Vκ41) after 24 cycles of the second PCR. Fig. 1. View largeDownload slide Somatic hypermutations in the transgenic constructs. (A) Schematic diagrams of the transgenes. The wild-type construct, 3′-r, contains the murine Vκ21C and Jκ–Cκ segments including the κE3′ region (26). Mutant substrates, rmPU.1 and rmEM5, contain base substitutions (in italics) in the PU.1 and NF-EM5 binding motifs respectively in the κE3′ (27). Arrowheads in the Vκ and Jκ regions indicate two sets of PCR primers (see Methods). (B) PCR experiments to detect the rearranged Vκ21C– Jκ1 and/or Vκ41C– Jκ1 sequences. The first PCR was repeated 18 times with 10 ng of DNA. In the second PCR, 1/500 of the first PCR product was used. The numbers of cycles for the second PCR are indicated. The rearranged Vκ41 band was used as an internal standard to estimate the ratio of Vκ21C–Jκ1/Vκ41–Jκ1. Numbers in parentheses are the ratios of the two PCR products (Vκ21C/Vκ41) after 24 cycles of the second PCR. Fig. 2. View largeDownload slide Hypermutations detected in the second exon of the Vκ21C. Genomic DNA was isolated from PNAhigh/B220+ cells of Peyer' s patch germinal centers. PCR amplification and detection of mutations are described in Methods. For each pie chart, each sector represents a proportion of clones with the indicated number of mutations. Mutations in the endogenous Vκ21C (non-Tg) and PCR errors (see Methods) are also shown. Fig. 2. View largeDownload slide Hypermutations detected in the second exon of the Vκ21C. Genomic DNA was isolated from PNAhigh/B220+ cells of Peyer' s patch germinal centers. PCR amplification and detection of mutations are described in Methods. For each pie chart, each sector represents a proportion of clones with the indicated number of mutations. Mutations in the endogenous Vκ21C (non-Tg) and PCR errors (see Methods) are also shown. Fig. 3. View largeDownload slide Distribution of hypermutations in the second exon of Vκ21C. Variabilities (mutated clones/total clones) are shown as percentage for each nucleotide position of the Vκ21C transgenes. Endogenous Vκ21C sequences of the non-transgenic (non-Tg) mice were analyzed as controls. Horizontal black bars indicate the RGYW/WRCY target sequences. CDR residues are indicated. Asterisks represent typical mutational hotspots, whose variabilities are >8% in both wild-type controls, non-Tg and 3′-r. Fig. 3. View largeDownload slide Distribution of hypermutations in the second exon of Vκ21C. Variabilities (mutated clones/total clones) are shown as percentage for each nucleotide position of the Vκ21C transgenes. Endogenous Vκ21C sequences of the non-transgenic (non-Tg) mice were analyzed as controls. Horizontal black bars indicate the RGYW/WRCY target sequences. CDR residues are indicated. Asterisks represent typical mutational hotspots, whose variabilities are >8% in both wild-type controls, non-Tg and 3′-r. Fig. 4. View largeDownload slide Targeting of mutations to the hotspot motifs (RGYW/WRCY). The proportions of mutations inside and outside of the hotspot motifs (RGYW/WRCY) were plotted. The expected distribution of random mutations was also plotted. Fig. 4. View largeDownload slide Targeting of mutations to the hotspot motifs (RGYW/WRCY). The proportions of mutations inside and outside of the hotspot motifs (RGYW/WRCY) were plotted. The expected distribution of random mutations was also plotted. Fig. 5. View largeDownload slide Biased mutations at A/T residues in the rmPU.1 and rmEM5 constructs. (A) Directions of base substitutions are compared among the Vκ21C transgene sequences (3′-r, rmPU.1 and rmEM5). The endogenous Vκ21C gene of the non-transgenic (non-Tg) mouse was also analyzed as a control. Total numbers of mutations analyzed in each construct are 152 (non-Tg), 216 (3′-r), 104 (rmPU.1) and 82 (rmEM5). Base preferences for mutations are shown in percentage (total mutations = 100%). Among 307 residues in the Vκ21C coding sequence, there are 77 As, 73 Gs, 80 Cs and 77 Ts. Percentages of base substitutions are corrected for the base composition. (B) Proportions of targeted bases for mutations are compared between the A/T and G/C pairs. Three types of transgenic (3′-r, rmPU.1 and rmEM5) and the endogenous (non-Tg) Vκ21C genes are compared. The result on the endogenous Vκ21C gene in non-transgenic mice is also presented. Fig. 5. View largeDownload slide Biased mutations at A/T residues in the rmPU.1 and rmEM5 constructs. (A) Directions of base substitutions are compared among the Vκ21C transgene sequences (3′-r, rmPU.1 and rmEM5). The endogenous Vκ21C gene of the non-transgenic (non-Tg) mouse was also analyzed as a control. Total numbers of mutations analyzed in each construct are 152 (non-Tg), 216 (3′-r), 104 (rmPU.1) and 82 (rmEM5). Base preferences for mutations are shown in percentage (total mutations = 100%). Among 307 residues in the Vκ21C coding sequence, there are 77 As, 73 Gs, 80 Cs and 77 Ts. Percentages of base substitutions are corrected for the base composition. (B) Proportions of targeted bases for mutations are compared between the A/T and G/C pairs. Three types of transgenic (3′-r, rmPU.1 and rmEM5) and the endogenous (non-Tg) Vκ21C genes are compared. The result on the endogenous Vκ21C gene in non-transgenic mice is also presented. Fig. 6. View largeDownload slide Frequencies of hypermutations at different kinds of nucleotides. Base preferences for the mutation are compared among three transgenic Vκ21C constructs (3′-r, rmPU.1 and rmEM5). Mutation frequencies (substituted bases in 103 nucleotides) for each nucleotide are normalized with those of the endogenous Vκ21C gene (non-Tg). Fig. 6. View largeDownload slide Frequencies of hypermutations at different kinds of nucleotides. Base preferences for the mutation are compared among three transgenic Vκ21C constructs (3′-r, rmPU.1 and rmEM5). Mutation frequencies (substituted bases in 103 nucleotides) for each nucleotide are normalized with those of the endogenous Vκ21C gene (non-Tg). Transmitting editor: T. Taniguchi We thank Dr Akio Tsuboi and Hitomi Sakano for helpful discussion and critical reading of the manuscript. This work was supported by the Special Promotion Research Grant from Monbusho and by grants from Toray Science Foundation, Nissan Science Foundation, Mitsubishi Foundation and Japan Foundation for Applied Enzymology. M. K. and R. H. are Research Fellows of the Japan Society for Promotion of Science. H. N. was supported by research grants from Kurata Foundation, Uehara Memorial Foundation and the Naito Foundation. References 1 Maizels, N. 1995. Somatic hypermutation: how many mechanisms diversify V region sequences? Cell  83: 9. Google Scholar 2 Rajewsky, K. 1996. Clonal selection and learning in the antibody system. Nature  381: 751. Google Scholar 3 Dorner, T., Foster, S. J., Brezinschek, H. P. and Lipsky, P. E. 1998. Analysis of the targeting of the hypermutational machinery and the impact of subsequent selection on the distribution of nucleotide changes in human VHDJH rearrangements. Immunol. Rev.  162: 161. Google Scholar 4 Kim, N. and Storb, U. 1998. The role of DNA repair in somatic hypermutation of immunoglobulin genes. J. Exp. Med.  187: 1729. Google Scholar 5 Reynaud, C. A., Bertocci, B., Frey, S., Delbos, F., Quint, L. and Weill, J. C. 1999. Mismatch repair and immunoglobulin gene hypermutation: did we learn something? Immunol. Today  20: 522. Google Scholar 6 Jacobs, H., Fukita, Y., van der Horst, G. T., de Boer, J., Weeda, G., Essers, J., de Wind, N., Engelward, B. P., Samson, L., Verbeek, S., de Murcia, J. M., de Murcia, G., te Riele, H. and Rajewsky, K. 1998. Hypermutation of immunoglobulin genes in memory B cells of DNA repair-deficient mice. J. Exp. Med.  187: 1735. Google Scholar 7 Phung, Q. H., Winter, D. B., Cranston, A., Tarone, R. E., Bohr, V. A., Fishel, R. and Gearhart, P. J. 1998. Increased hypermutation at G and C nucleotides in immunoglobulin variable genes from mice deficient in the MSH2 mismatch repair protein. J. Exp. Med.  187: 1745. Google Scholar 8 Frey, S., Bertocci, B., Delbos, F., Quint, L., Weill, J. C. and Reynaud, C. A. 1998. Mismatch repair deficiency interferes with the accumulation of mutations in chronically stimulated B cells and not with the hypermutation process. Immunity  9: 127. Google Scholar 9 Rada, C., Ehrenstein, M. R., Neuberger, M. S. and Milstein, C. 1998. Hot spot focusing of somatic hypermutation in MSH2-deficient mice suggests two stages of mutational targeting. Immunity  9: 135. Google Scholar 10 Kim, N., Bozek, G., Lo, J. C. and Storb, U. 1999. Different mismatch repair deficiencies all have the same effects on somatic hypermutation: intact primary mechanism accompanied by secondary modifications. J. Exp. Med.  190: 21. Google Scholar 11 Wiesendanger, M., Kneitz, B., Edelmann, W. and Scharff, M. D. 2000. Somatic hypermutation in MutS homologue (MSH)3-, MSH6-, and MSH3/MSH6-deficient mice reveals a role for the MSH2–MSH6 heterodimer in modulating the base substitution pattern. J. Exp. Med.  191: 579. Google Scholar 12 Storb, U., Peters, A., Klotz, E., Kim, N., Shen, H. M., Hackett, J., Rogerson, B. and Martin, T. E. 1998. Cis-acting sequences that affect somatic hypermutation of Ig genes. Immunol. Rev.  162: 153. Google Scholar 13 Storb, U., Peters, A., Klotz, E., Kim, N., Shen, H. M., Hackett, J., Rogerson, B., O'Brien, R. and Martin, T. E. 1998. Immunoglobulin transgenes as targets for somatic hypermutation. Int. J. Dev. Biol.  42: 977. Google Scholar 14 Betz, A. G., Milstein, C., Gonzalez-Fernandez, A., Pannell, R., Larson, T. and Neuberger, M. S. 1994. Elements regulating somatic hypermutation of an immunoglobulin κ gene: critical role for the intron enhancer/matrix attachment region. Cell  77: 239. Google Scholar 15 Goyenechea, B., Klix, N., Yelamos, J., Williams, G. T., Riddell, A., Neuberger, M. S. and Milstein, C. 1997. Cells strongly expressing Igκ transgenes show clonal recruitment of hypermutation: a role for both MAR and the enhancers. EMBO J.  16: 3987. Google Scholar 16 Peters, A. and Storb, U. 1996. Somatic hypermutation of immunoglobulin genes is linked to transcription initiation. Immunity  4: 57. Google Scholar 17 Fukita, Y., Jacobs, H. and Rajewsky, K. 1998. Somatic hypermutation in the heavy chain locus correlates with transcription. Immunity  9: 105. Google Scholar 18 Storb, U., Klotz, E. L., Hackett, J., Jr, Kage, K., Bozek, G. and Martin, T. E. 1998. A hypermutable insert in an immunoglobulin transgene contains hotspots of somatic mutation and sequences predicting highly stable structures in the RNA transcript. J. Exp. Med.  188: 689. Google Scholar 19 Bachl, J. and Wabl, M. 1996. An immunoglobulin mutator that targets G.C base pairs. Proc. Natl Acad. Sci. USA  93: 851. Google Scholar 20 Yelamos, J., Klix, N., Goyenechea, B., Lozano, F., Chui, Y. L., Gonzalez-Fernandez, A., Pannell, R., Neuberger, M. S. and Milstein, C. 1995. Targeting of non-Ig sequences in place of the V segment by somatic hypermutation. Nature  376: 225. Google Scholar 21 Azuma, T., Motoyama, N., Fields, L. E. and Loh, D. Y. 1993. Mutations of the chloramphenicol acetyl transferase transgene driven by the immunoglobulin promoter and intron enhancer. Int. Immunol.  5: 121. Google Scholar 22 Umar, A., Schweitzer, P. A., Levy, N. S., Gearhart, J. D. and Gearhart, P. J. 1991. Mutation in a reporter gene depends on proximity to and transcription of immunoglobulin variable transgenes. Proc. Natl Acad. Sci. USA  88: 4902. Google Scholar 23 Gonzalez-Fernandez, A. and Milstein, C. 1993. Analysis of somatic hypermutation in mouse Peyer's patches using immunoglobulin κ light-chain transgenes. Proc. Natl Acad. Sci. USA  90: 9862. Google Scholar 24 Gonzalez-Fernandez, A., Gilmore, D. and Milstein, C. 1994. Age-related decrease in the proportion of germinal center B cells from mouse Peyer's patches is accompanied by an accumulation of somatic mutations in their immunoglobulin genes. Eur. J. Immunol.  24: 2918. Google Scholar 25 Blin, N. and Stafford, D. W. 1976. A general method for isolation of high molecular weight DNA from eukaryotes. Nucleic Acids Res.  3: 2303. Google Scholar 26 Hiramatsu, R., Akagi, K., Matsuoka, M., Sakumi, K., Nakamura, H., Kingsbury, L., David, C., Hardy, R. R., Yamamura, K. and Sakano, H. 1995. The 3′ enhancer region determines the B/T specificity and pro-B/pre-B specificity of immunoglobulin Vκ–Jκ joining. Cell  83: 1113. Google Scholar 27 Hayashi, R., Takemori, T., Kodama, M., Suzuki, M., Tsuboi, A., Nagawa, F. and Sakano, H. The PU.1 binding site is a cis-element that regulates pro-B/pre-B specificity of Vκ–Jκ joining. 1997. J. Immunol.  159: 4145. Google Scholar 28 Gorman, J. R., van der Stoep, N., Monroe, R. R. Cogne, R., Davidson, L. and Alt, F. W. 1996. The Igκ 3′ enhancer influences the ratio of Igκ versus Igλ B lymphocytes. Immunity  5: 241. Google Scholar 29 van der Stoep, N., Gorman, J. R. and Alt, F. W. 1998. Reevaluation of 3′Eκ function in stage- and lineage-specific rearrangement and somatic hypermutation. Immunity  8: 743. Google Scholar 30 Liu, X., Prabhu, A. and Ness, B. V. 1999. Developmental regulation of the locus involves both positive and negative sequence elements in the 3′ enhancer that affect synergy with the intron enhancer. J. Biol. Chem.  274: 3285. Google Scholar 31 Papavasiliou, F. N. and Schatz, D. G. 2000. Cell-cycle-regulated DNA double-strand breaks in somatic hypermutation of the immunoglobulin genes. Nature  408: 216. Google Scholar 32 Bross, L., Fukita, Y., McBlane, F., Demolliere, C., Rajewsky, K. and Jacobs, H. 2000. DNA double-strand breaks in immunoglobulin genes undergoing somatic hypermutation. Immunity  13: 589 Google Scholar © 2001 Japanese Society for Immunology

Journal

International Immunology – Oxford University Press

Published: Nov 1, 2001

There are no references for this article.