Somatic hypermutation (SHM) of Ig genes depends upon the deamination of C nucleotides in WRCY (W = A/T, R = A/G, Y = C/T) motifs by activation-induced cytidine deaminase (AICDA). Despite this, a large number of mutations occur in WA motifs that can be accounted for by the activity of polymerase η (POL η). To determine whether there are AICDA-independent mutations and to characterize the relationship between AICDA- and POL η-mediated mutations, 1470 H chain and 1313 κ- and λ-chain rearrangements from three AICDA−/− patients were analyzed. The Ig mutation frequency of all VH genes from AICDA−/− patients was 40-fold less than that of normal donors, whereas the mutation frequency of mutated VH sequences from AICDA−/− patients was 6.8-fold less than that of normal donors. AICDA−/− B cells lack mutations in WRCY/RGYW motifs as well as replacement mutations and mutational targeting in complementarity-determining regions. A significantly reduced mutation frequency in WA motifs compared with normal donors and an increased percentage of transitions, which may relate to reduced uracil DNA-glycosylase activity, suggest a role for AICDA in regulating POL η and uracil DNA-glycosylase activity. Similar results were observed in VL rearrangements. The residual mutations were predominantly G:C substitutions, indicating that AICDA-independent cytidine deamination was a likely, yet inefficient, mechanism for mutating Ig genes.

Expression of B cell-specific activation-induced cytidine deaminase (AICDA)3 in germinal center (GC) B cells is required for generating somatic hypermutation (SHM) of Ig genes and affinity maturation of Abs during an immune response (1, 2). AICDA-deficient B cells can undergo a GC reaction but fail to accumulate mutations in the Ig variable regions following immunization (3). The lack of mutation may contribute to shifts in the H and L chain repertoires (4). Although AICDA is critical for SHM, other mechanisms contribute to the overall mutational pattern (5) and some mutations may occur independent of AICDA (1, 3). However, the nature of the mutations in the absence of AICDA and the relationship between AICDA-induced mutations and those owing to the activity of other mechanisms, such as error-prone polymerase (POL) η, are not known.

AICDA preferentially deaminates Cs in the WRCY motif (W = A/T, R = A/G, Y = C/T) in both DNA strands (6, 7, 8, 9) and is directly associated with generating the G/C mutation spectrum. The targeting mechanism of AICDA and certain features of Ig H chain gene CDR, such as the presence of overlapping targeted motifs, are associated with favoring CDR as opposed to framework region (FR) mutations and also replacement (R) substitutions that impact the Ag binding site (8, 10) and enable affinity maturation.

The A/T mutation spectrum is associated with the activity of error-prone POL η that preferentially targets A residues in the WA motif. Linkage between AICDA and POL η activity was suggested by a detailed analysis of mutation patterning showing that mutations of POL η-targeted WA motifs were increased when these motifs overlapped AICDA-targeted RGYW motifs (N. S. Longo, P. L. Lugar, S. Yavuz, W. Zheng, P. H. L. Krijger, D. E. Russ, A. C. Grammer, and P. E. Lipsky, manuscript submitted for publication). This pattern of mutations suggested that the AICDA-induced U:G lesions may foster additional mutation in adjacent residues through the recruitment of a WA/TW motif mutator, such as POL η. Consistent with this mechanism, POL η deficiency results in a loss of mutations in WA motifs in AICDA-targeted WRCY/RGYW motifs (11). However, because POL η deficiency also results in a loss of WA mutations remote from RGYW/WRCY motifs, it is possible that this error-prone POL might also have an AICDA-independent function.

To characterize the mutational pattern that occurs in the absence of AICDA and to explore the relationship between AICDA and POL η in more detail as well to characterize the mutational pattern that occurs in the absence of AICDA, we analyzed the H chain and L chain mutations from three AICDA-deficient patients. We focused on mutations in nonproductive rearrangements that are not influenced by selection to identify mutational mechanisms. There was a significant reduction in the frequency of mutations, but still significantly more than could be anticipated in genes of non-B cells. Many of the mutations could be attributed to non-targeted AICDA-independent cytidine deamination resulting in a loss of targeting of mutations to CDRs. The mutation pattern in AICDA deficiency was biased toward transitions and lacked mutations in POL η-targeted WA motifs, indicating that AICDA is necessary for the recruitment of POL η and also uracil-DNA glycosylase (UNG) during SHM.

Three AICDA-deficient patients donated peripheral blood samples after signing informed consent forms in accordance with institutional review board-approved protocols. The control samples (12, 13, 14) and aicda mutations and Ig levels for each patient (P4 age 11, P5 age 7, and P6 age 21) have been described previously (3). Blood mononuclear cells were isolated on Ficoll gradients and stained with mouse IgG1 anti-human CD19 PE (BD Biosciences), mouse IgG2 anti-human IgD FITC (BD Pharmingen), or goat anti-human IgD FITC (Caltag Laboratories) and anti-CD27-PE and isotype controls (BD Pharmingen). Patient mononuclear cells were sorted into CD19+ or CD19+CD27 and CD19+CD27+ subsets using a BD Biosciences FACSVantage SE DiVa or a MoFlo Cell Sorter (DakoCytomation). Sorted populations were 94–98% pure.

Aliquots of approximately one cell were diluted in buffer (10 mM NaCl, 5 mM Tris-HCl (pH 8) at 25°C, and 0.1% Triton X-100) containing 0.4 mg/ml proteinase K (Sigma-Aldrich), and genomic DNA from individual cell lysates was linearly amplified using random 15-mer (Qiagen). Aliquots of the preamplification product were used as template for internal and external nested amplification of the transcribed strand using VH3, VH4, and VH1 H chain-specific upstream primers and downstream JH-specific primers as described previously (14). The AICDA−/− L chains were analyzed with Vκ1, Vκ2, and Vκ3, and Vλ1-, Vλ2-, and Vλ3-specific primers as described previously (15, 16).

PCR products were separated by electrophoresis on 1.2% agarose gels, purified (Edge Biosystems), and directly sequenced using the Applied Biosystems Prism Big Dye Terminator Cycle Sequencing Kit. PCR products were directly sequenced using on an automated capillary sequencer (Applied Biosystems Prism 3100 Genetic Analyzer).

The analysis included 105 nonproductive and 636 productive rearrangements from normal donors that included CD19+IgM+ B cells from accession numbers X87006–X87082, CD19+IgM+CD5+ and CD19+IgM+CD5 B cells from accession numbers Z80363–Z80770 (12, 13, 14), IgD+CD27+ B cells from accession numbers EF542547EF542687, and IgDCD27+ B cells from accession numbers EF542688–EF542796. The AICDA-deficient analysis included 320 nonproductive and 1150 productive CD19+CD27 and CD19+CD27+ VHDJH rearrangements with accession numbers (EU237493EU238970), 294 nonproductive and 482 productive κ rearrangements κ (EU788043-516) (EU788791-919) (EU788982-9157), and 142 nonproductive and 395 productive λ rearrangements λ (EU788517-790) (EU788920-981) (EU789158-362).

Nonproductive and productive VHDJH rearrangements were determined by using the JOINSOLVER sequence and mutation analysis program (17). Enumeration of nucleotides used an algorithm that takes into consideration the finding that some bp positions in the germline genes reside within more than one motif and/or within ambiguous motifs (i.e., AGCT is a RGYW and WRCY motif; TA is a WA and TW motif). This approach avoided a subjective bias in motif nucleotide designation, but also was the source of noninteger values.

Seven new polymorphisms within codons 25–94 in six VH genes were excluded from the mutation analysis (N. S. Longo, P. L. Lugar, S. Yavuz, W. Zheng, P. H. L. Krijger, D. E. Russ, A. C. Grammer, and P. E. Lipsky, manuscript submitted for publication). The following 12 substitutions in L chain genes (ImMuno GeneTics (IMGT) numbering for codons in brackets) appeared to be new polymorphisms and were excluded from the mutation analysis: IGKV1-16*01 [75], IGKV1D-17*01 [26], IGKV1D-17*01 [52], IGKV1-5*03 [57], IGKV1-9*01 [86], IGKV1-9*01 [90], IGKV1-8*01 [77], IGKV1-27*01 [94], IGKV1-12*01 [24], IGLV 13-21*01 [108], and IGLV 6-57*01 [49].

The χ2 test and the p < 0.05 limit were used to determine statistically significant differences. The ratio of nucleotide mutational targeting inside the motif vs mutation of the nucleotide outside the motif was used to compare differences between AICDA and normal donors (see Table III and Fig. 2A).

The PCR error of this technique was initially demonstrated to be 1 × 10−4 by amplifying an unmutated germline Ig sequence 96 times using the same amplification and sequencing conditions used here (19). The low PCR error rate was confirmed by amplifying c-kit genes from B cells (0.8 × 10−4, 6 of 72,261) (20). Finally, unrearranged VH family gene segments were amplified and sequenced from CD19 peripheral blood cells. Analysis of 96 VH4 segments from CD19-negative cells demonstrated a PCR error rate of 0.76 × 10−4 (2 of 25,986).

B cell phenotyping and mutation analysis confirmed previous reports (3) that AICDA-deficient patients have a normal percentage of CD19+CD27+ B cells but unlike normal donors in which CD27+ expression correlates with a mutated memory B cell population, the AICDA−/− CD27+ B cells are infrequently mutated. In this study, 20–28% of the peripheral blood B cells from AICDA−/− patients were CD27+ (data not shown). Both subsets of CD19+CD27+ and CD19+CD27 B cells contained mutated sequences. However, there was no significant difference in the percentage of mutated sequences or the mutation frequency between the CD27 and CD27+ subsets. Since CD27 expression did not correlate with mutation in AICDA−/− B cells and since CD27 human memory cells have been identified (21), the subsets were combined for the mutation analysis. The percentages of mutated VHDJH sequences in AICDA−/− nonproductive and productive repertoires (12 and 9%, respectively) were significantly (p < 0.0001) less than those of normal donors (71–75%) (Table I). The overall mutation frequency was 40-fold less than that of normal controls, whereas the mutation frequency of mutated sequences was only 6.8-fold less than normal controls. Most of the mutated nonproductive AICDA−/− sequences had a unique, single-nucleotide substitution, except four sequences that contained two mutations. In the mutated productive repertoire, 100 (96%) of 104 rearrangements had one to two mutations, three sequences had three mutations, and one sequence had four mutations. These mutations clearly were not polymorphisms and occurred at approximately a 10-fold higher frequency than the PCR error for this method (6 mutations of 75,182 bp; 8 × 10−5) (20) or 2 mutations of 25,986 (0.76 × 10−4) determined for VH4 sequences in non-B cells.

Table I.

Mutation frequency of VH segment of AICDA-deficient B cellsa

AICDA−/−Normal Donors
NPPNPP
Total no. of rearrangements 320 1150 105 636 
Total no. of mutations 43 124 627 4864 
Percentage of sequences mutated 39/320b (12%) 104/1,150b (9%) 76/105 (71%) 479/636 (75%) 
Mutation range 1–2 1–4 1–33 1–38 
Mutation frequency overall 43/64,081b (0.07%) 124/230,514b (0.05%) 627/22,693 (2.8%) 4864/152,977 (3.2%) 
Mutation frequency of mutated rearrangements 43/7,799b (0.6%) 124/19,938b (0.6%) 627/15,149 (4.1%) 4864/113,434 (4.3%) 
AICDA−/−Normal Donors
NPPNPP
Total no. of rearrangements 320 1150 105 636 
Total no. of mutations 43 124 627 4864 
Percentage of sequences mutated 39/320b (12%) 104/1,150b (9%) 76/105 (71%) 479/636 (75%) 
Mutation range 1–2 1–4 1–33 1–38 
Mutation frequency overall 43/64,081b (0.07%) 124/230,514b (0.05%) 627/22,693 (2.8%) 4864/152,977 (3.2%) 
Mutation frequency of mutated rearrangements 43/7,799b (0.6%) 124/19,938b (0.6%) 627/15,149 (4.1%) 4864/113,434 (4.3%) 
a

NP, Nonproductive rearrangements; P, productive rearrangements. The AICDA−/− data include CD19+CD27 and CD19+CD27+ sequences from three patients. The control data come from previous reports of four individuals and three different experiments that collected IgM+, IgM+CD5+, IgM+CD5, IgD+CD27, and IgDCD27+ B cells which represent normal unmutated and mutated populations in peripheral blood. The mutation frequencies were calculated from the number of mutations/total number of bases analyzed or from the total number of bases analyzed from mutated sequences only.

b

Significant (p < 0.0001) difference between AICDA−/− and normal donors.

Because mutations do occur in Ig genes in the absence of AICDA, we next characterized the frequency of mutation in Ig-hypermutable motifs, particularly those in the nonproductive rearrangements that are not influenced by positive and negative selection. An increased focus of mutations in any of the four bases constituting WRCY (W = A/T, R = A/G, Y = C/T) and the inverse complementary motif RGYW are the hallmarks of AICDA-dependent SHM (22, 23). In the nonproductive repertoire, the percentage of mutations that occurred in WRCY motifs was significantly (p = 0.004) reduced compared with normal controls (3% vs 16%; Table II). Similarly, the frequency of mutations in RGYW (10%) was no greater than that expected from random chance (12.5%) and significantly less than in normal controls (24%, p = 0.03). The frequency of mutations in either motif (13%) was also significantly (p = 0.01) less than in normal B cells (40%). These motifs become hot spots for hypermutation primarily because of AICDA targeting C in WRCY on both strands of DNA (N. S. Longo, P. L. Lugar, S. Yavuz, W. Zheng, P. H. L. Krijger, D. E. Russ, A. C. Grammer, and P. E. Lipsky, manuscript submitted for publication and Refs. 7 , 9 , and 24). In the AICDA-deficient B cells, the hypermutable nucleotides (C in WRCY and G in RGYW, which represents an AICDA attack on the opposite strand) were not targeted for mutation more frequently than mutation of C or G residues anywhere else in the H chain variable segment (Table III), consistent with a complete loss of AICDA activity in these subjects. By comparison, there was nearly 4-fold more targeting of these specific nucleotides inside WRCY/RGYW motifs in normal B cells. A similar loss of G/C mutational targeting was noted in the AICDA−/− L chain rearrangements (Fig. 2A). These results indicate that the preferential targeting of the C in WRCY (and G in RGYW) is completely dependent on AICDA activity and provides the opportunity to examine the AICDA dependence of other known components of the SHM machinery.

Table II.

Percentage of mutations in RGYW/WRCY and WA/TW motifsa

AICDA−/−Normal Donors
NPPNPP
WRCY 1.5/43b (3%) 13.8/124c (11%) 98.1/627c (16%) 756.7/4864c (16%) 
RGYW 4.5/43b (10%) 30.3/124 (24%) 152.4/627 (24%) 1285.9/4864 (26%) 
WA 4.5/43b (10%) 25.5/124d (21%) 140.4/627d (22%) 888.5/4864d (18%) 
TW 1.5/43 (4%) 9/124 (7%) 73.7/627 (12%) 490.5/4864 (10%) 
AICDA−/−Normal Donors
NPPNPP
WRCY 1.5/43b (3%) 13.8/124c (11%) 98.1/627c (16%) 756.7/4864c (16%) 
RGYW 4.5/43b (10%) 30.3/124 (24%) 152.4/627 (24%) 1285.9/4864 (26%) 
WA 4.5/43b (10%) 25.5/124d (21%) 140.4/627d (22%) 888.5/4864d (18%) 
TW 1.5/43 (4%) 9/124 (7%) 73.7/627 (12%) 490.5/4864 (10%) 
a

The enumeration of nucleotides was adjusted for nucleotides that occupy a position in more than one motif or ambiguous motifs. This calculation, therefore, can assign a fraction of a mutation for a given motif. NP, Nonproductive; P, productive.

b

Significant (p ≤ 0.05), difference between AICDA−/− and ND.

c

Significant (p < 0.008) difference between WRCY and RGYW.

d

Significant (p < 0.009) difference between WA and TW.

Table III.

Target nucleotide mutation frequency (%) in hypermutable motifs in nonproductive VH rearrangementsa

Mutated BaseMotifAICDA−/−Normal Donor
Inside RGYW 2/2,374b (0.08%) 94/937c (10.0%) 
 All other G 18/13,016b (0.14%) 99/6,093 (1.6%) 
Inside WRC0/2,140b (0%) 62/832c (7.5%) 
 All other C 7/13,596b (0.05%) 75/5,685 (1.3%) 
Inside WA 4.5/8,273b (0.05%) 115.9/2,473c (4.7%) 
 All other A 6.5/8,564b (0.08%) 79.1/3,615 (2.2%) 
Inside TW 1.5/5,180b (0.03%) 56.5/1,550c (3.6%) 
 All other T 3.5/8,288b (0.04%) 45.5/3,714 (1.2%) 
Mutated BaseMotifAICDA−/−Normal Donor
Inside RGYW 2/2,374b (0.08%) 94/937c (10.0%) 
 All other G 18/13,016b (0.14%) 99/6,093 (1.6%) 
Inside WRC0/2,140b (0%) 62/832c (7.5%) 
 All other C 7/13,596b (0.05%) 75/5,685 (1.3%) 
Inside WA 4.5/8,273b (0.05%) 115.9/2,473c (4.7%) 
 All other A 6.5/8,564b (0.08%) 79.1/3,615 (2.2%) 
Inside TW 1.5/5,180b (0.03%) 56.5/1,550c (3.6%) 
 All other T 3.5/8,288b (0.04%) 45.5/3,714 (1.2%) 
a

The targeted nucleotide refers to the base in the underlined position (or two positions in WA, where W = A or T). The mutation frequency of the targeted nucleotide inside a motif was calculated from the number of target nucleotide mutations divided by the total number of the motif (or total number of A in WA motifs). To calculate the mutation frequency outside the motif, the total number of base mutations in the sequence minus the number of targeted nucleotide mutations was divided by the total number of the base in the sequence minus the number of target nucleotides inside the motif. The extent of targeting was determined by calculating the fold difference between the mutation frequency inside a motif vs all other positions.

b

Significant (p < 0.0001) difference between AICDA−/− and normal controls.

c

Significant (p < 0.01) difference between the nucleotide mutation frequency inside a motif vs all other positions.

There is substantial evidence that error-prone POL η plays a significant role in SHM (11, 25, 26). Analysis of Ig gene mutational patterns from individuals deficient in POL η has indicated that this polymerase specifically targets the A in WA motifs, (11, 27) and preferentially targets the TA dinucleotide (N. S. Longo, P. L. Lugar, S. Yavuz, W. Zheng, P. H. L. Krijger, D. E. Russ, A. C. Grammer, and P. E. Lipsky, manuscript submitted for publication). More recently, evidence of WA mutations overlapping RGYW motifs suggested that POL η may be recruited to the site of the AICDA-induced lesion during mismatch repair in normal human B cells. In support of the human data, transgenic mice with SHM substrates acquired the capacity to acquire A/T mutations only when the substrate contained an AICDA-targeted cytosine (28). To examine the role of POL η in the absence of AICDA, we examined mutations in WA motifs in the H chain genes of the AICDA−/− B cells. WA motif mutations (10%) were no more frequent than that expected from random mutation (12.5%) and were significantly less frequent than WA motif mutations found in normal B cells (Table II). Notably, mutations of the hypermutable A in WA motifs were no more frequent than A mutations elsewhere in the variable segment, indicating that POL η activity was greatly reduced (Table III). Similar results were noted with L chain rearrangements (Fig. 2A).

Despite reduced targeting in all hypermutable motifs, mutations in RGYW motifs appeared to be positively selected as indicated by the increased mutation frequency in the VH productive compared with nonproductive repertoires (Table II). A similar trend was observed with the other motifs, but significant differences were not found. Therefore, some form of selection is intact, although this is not evident in normal controls in which a large number of mutations have been analyzed, presumably because large numbers of mutations of Ig genes may disable the Ig molecule and contribute to B cell loss.

Although the known hypermutable motifs were not targeted in AICDA−/− B cells, there was the possibility that another motif was targeted. To address this question, the Ig sequences were divided into oligomers, each of which contained a mutated nucleotide from the AICDA−/− B cells in the context of five nucleotides flanking both sides of the mutation. The sequences were interrogated for shared motifs by using the web-based Genio/logo program (http://www.biogenio.com/logo/logo.cgi). This procedure failed to identify a motif within which residual mutations occurred (data not shown). Other potential patterns of mutational activity were investigated, such as UV-induced dipyrimidine substitutions or oxidative damage generating mutations of guanine dinucleotides, but no pattern with a mutation frequency greater than normal was found (data not shown). We concluded that the major mutation mechanism in AICDA−/− was a random event involving individual Ig gene nucleotides.

In normal nonproductive VH rearrangements, the frequency of mutation in the CDRs in VH genes is 2-fold greater than the frequency of mutations in the FRs after normalization for the number of nucleotides analyzed in each region (Table IV and Ref. 8). By contrast, in the AICDA-deficient VH rearrangements, the preferential mutation targeting of CDRs was lost. These data indicate that AICDA activity is essential for targeting of mutations to CDRs. Enrichment of mutations is normally observed in CDR regions in the productive repertoire when productive and nonproductive VH rearrangements are compared (Table IV). However, in the AICDA−/− rearrangements, an increased frequency of CDR mutations and a decrease in FR mutations was noted in the productive repertoire and, as a result, the ratio of mutations in CDRs vs FRs was normalized, presumably because of positive selection of CDR mutations and negative selection of FR mutations.

Table IV.

The effect of AICDA deficiency on amino acid and nucleotide substitutions in CDRs and FRs of H chain rearrangementsa

AICDA−/−Normal Donors
NPPNPP
CDR R:S 3/1b (3.0) 41/2c (20.5) 170/32b (5.3) 1383/391 (3.5) 
FR R:S 24/15c (1.6) 54/27 (2.0) 329/96b (3.4) 1,940/1,150 (1.7) 
CDR mutation frequency 4/11,393bc (3.5 × 10−443/40,699c (1 × 10−3198/4,826b (4.1 × 10−21741/29,250 (6.0 × 10−2
FR mutation frequency 39/52,688bc (7.4 × 10−481/189,815c (0.4 × 10−3426/20,244b (2.1 × 10−23,092/120,907 (2.6 × 10−2
Mutation frequency 3.5:7.4bc 1:0.4 4.1:2.1 6.0:2.6 
CDR:FR 0.5 2.5 2.0 2.3 
AICDA−/−Normal Donors
NPPNPP
CDR R:S 3/1b (3.0) 41/2c (20.5) 170/32b (5.3) 1383/391 (3.5) 
FR R:S 24/15c (1.6) 54/27 (2.0) 329/96b (3.4) 1,940/1,150 (1.7) 
CDR mutation frequency 4/11,393bc (3.5 × 10−443/40,699c (1 × 10−3198/4,826b (4.1 × 10−21741/29,250 (6.0 × 10−2
FR mutation frequency 39/52,688bc (7.4 × 10−481/189,815c (0.4 × 10−3426/20,244b (2.1 × 10−23,092/120,907 (2.6 × 10−2
Mutation frequency 3.5:7.4bc 1:0.4 4.1:2.1 6.0:2.6 
CDR:FR 0.5 2.5 2.0 2.3 
a

The FR:CDR mutation frequency ratio was normalized to correct for the number of nucleotides analyzed in FRs and CDRs. The analysis excludes the CDR3 and the first 25 codons of FR1.

b

Significant (p < 0.007) difference between nonproductive (NP) and productive (P) rearrangements.

c

Significant (p = 0.03) difference between AICDA−/− and normal donors.

The ratio of replacement:silent mutations (R:S) is 5.3 in normal donor nonproductive CDRs compared with the 3.4 in FRs (Table IV). By comparison, the R:S ratio anticipated by random chance is 4.5 in CDRs and 3.2 in FRs because of codon bias. In normal donors, the R:S ratios of both the CDRs and the FRs in the nonproductive repertoire were 1.5- to 2-fold greater than those of the productive rearrangements, suggesting that large numbers of R mutations in both regions result in the loss of cells, as previously reported (8). In the absence of AICDA, R substitutions in the nonproductive repertoire were less common in both CDRs and FRs than in normal rearrangements. Notably, R substitutions appeared to be positively selected when occurring in CDRs and not eliminated when occurring in FRs, as evidenced by comparison of R:S ratios in nonproductive and productive rearrangements. Presumably, the very low mutational activity in AICDA−/− B cells introduces fewer deleterious mutations so that elimination of rearrangements with R mutations is not greater than the increase resulting from positive selection.

In the nonproductive AICDA-deficient repertoire, G residues were the most frequently mutated (41%), followed by A (24%), C (17%), and T residues (14%) compared with A>G>C>T in normal donors (Fig. 1). Similar to normal controls, a bias favoring A mutations more than T mutations was present in AICDA−/− B cells. Although POL η targets A residues more than T, the absence of targeting in WA motifs suggested that another polymerase, such as POL ζ, POLθ, or POLι, each of which is expressed in human tonsilar B cells, may be contributing to the residual A:T mutations in AICDA−/− B cells (29, 30).

FIGURE 1.

The nucleotide mutation pattern in productive and nonproductive H chain rearrangements in AICDA−/− B cells. The individual nucleotide substitutions detected in the variable segment of (A) nonproductive VHDJH and (B) productive VHDJH rearrangements in peripheral blood B cells from four normal donors and three AICDA−/− patients are presented. The percentages of all mutations (lower panels in each group) are scored after correction for base composition of the target sequence. Transitions appear in boldface. ∗, Significant (p = 0.03) difference between AICDA−/− and normal donors; †, significant (p < 0.0001) difference between transitions and transversions and significant (p < 0.0008) difference in transitions between AICDA−/− and normal donor.

FIGURE 1.

The nucleotide mutation pattern in productive and nonproductive H chain rearrangements in AICDA−/− B cells. The individual nucleotide substitutions detected in the variable segment of (A) nonproductive VHDJH and (B) productive VHDJH rearrangements in peripheral blood B cells from four normal donors and three AICDA−/− patients are presented. The percentages of all mutations (lower panels in each group) are scored after correction for base composition of the target sequence. Transitions appear in boldface. ∗, Significant (p = 0.03) difference between AICDA−/− and normal donors; †, significant (p < 0.0001) difference between transitions and transversions and significant (p < 0.0008) difference in transitions between AICDA−/− and normal donor.

Close modal

Of the few mutations in AICDA−/− VH genes, the percentage of C mutations was similar to normal (AICDA−/−, 17% vs normal, 21%; p = 0.39), whereas the percentage of G mutations was significantly increased in AICDA−/− compared with normal (AICDA−/−, 41% vs normal, 27%; p = 0.03). This suggested that a major source of mutation in AICDA-deficient cells was a G:C-focused mutational mechanism (Fig. 1). AICDA-independent cytidine deamination is the most common form of DNA damage in non-B cells and could contribute to the mutation pattern in AICDA−/− B cells (31). Presumably, this could target both the transcribed and nontranscribed strands with a bias toward the latter with little or no involvement of POL η. By comparing the mutation frequency of AICDA−/− B cells to our estimated mutation frequency of non-B cells, it appears that AICDA-independent C deamination in B cells is subjected to less of the normal repair mechanisms operative in non-B cells.

The individual nucleotide substitutions in the AICDA−/− nonproductive rearrangements were biased toward transitions (77%), compared with normal donors in which transitions and transversions occur at comparable frequencies (Fig. 1). The most evident reason for this mutation shift was the significant (p = 0.01) increase in G>A transitions and to a lesser extent A>G transitions. Similarly, mutations in nonproductive AICDA-deficient L chain rearrangements were skewed significantly toward transitions (Fig. 2 B). One of the mechanisms known to contribute to a loss of transversions is UNG deficiency (32). Even though UNG expression is increased in tonsillar GC cells, UNG may be less active or not accessible when AICDA is not available. A similar phenomenon was noted when mutations in X-HIgM B cells, which are also characterized by less functional AICDA activity, were analyzed. These data suggest that AICDA plays a role not only in the initiation and targeting of SHM, but also in the use of UNG and POL η that are involved in processing the AICDA-induced lesion.

FIGURE 2.

The nucleotide mutation pattern in nonproductive L chain rearrangements in AICDA−/− B cells. The L chain data were collected from AICDA−/− patients P4 and P5 and consists of 18 mutations from 294 nonproductive κ rearrangements and 16 mutations from 142 nonproductive λ rearrangements. A, Individual nucleotide substitutions in the L chain repertoires are presented in which the percentage of all mutations (lower panels) are scored after correction for base composition of the target sequence. B, The mutation frequency of a specific nucleotide inside vs outside an hypermutable motif was calculated as shown in Table III. ∗, Significant (p = 0.02) difference between AICDA−/− and normal donors; †, significant (p < 0.0001) difference between transitions and transversions and significant (p = 0.0008) difference in transitions between AICDA−/− and normal donor.

FIGURE 2.

The nucleotide mutation pattern in nonproductive L chain rearrangements in AICDA−/− B cells. The L chain data were collected from AICDA−/− patients P4 and P5 and consists of 18 mutations from 294 nonproductive κ rearrangements and 16 mutations from 142 nonproductive λ rearrangements. A, Individual nucleotide substitutions in the L chain repertoires are presented in which the percentage of all mutations (lower panels) are scored after correction for base composition of the target sequence. B, The mutation frequency of a specific nucleotide inside vs outside an hypermutable motif was calculated as shown in Table III. ∗, Significant (p = 0.02) difference between AICDA−/− and normal donors; †, significant (p < 0.0001) difference between transitions and transversions and significant (p = 0.0008) difference in transitions between AICDA−/− and normal donor.

Close modal

The current study demonstrated that a small number of mutations can develop in B cell Ig genes of individuals who totally lack expression of AICDA. However, in the absence of AICDA, the mutational frequency is markedly reduced and there is a loss of RGYW/WRCY mutations as well as of mutations in POL η-targeted WA motifs. In addition, UNG-dependent transversions were markedly reduced. These results suggest that AICDA induces a process that involves POL η and UNG to induce the normal spectrum of SHM. The loss of AICDA results in the absence of a mutational focus on CDRs and a bias toward R mutations in CDRs and, therefore, little ability of SHM to contribute to affinity maturation of the Ab response.

Because the measured mutational frequency was relatively low, it was essential to be certain that PCR error did not confound interpretation of the data. PCR amplification of genomic DNA from individual cells with direct sequencing of the PCR products used here has been documented to have a very low PCR error rate (0.8–1.0 × 10−4). To confirm this, we sequenced unrearranged Ig VH gene segments from CD19 non-B cells and confirmed the very low PCR error rate (0.76 × 10−4). Thus, few of the detected bp changes could be accounted for by PCR error. Moreover, because the majority of the data related to the absence of expected mutational targeting (G in RGYW, C in WRCY, A in WA), it was unlikely that these few bp changes potentially introduced by the PCR amplification affected the interpretation of the results.

Importantly, the Ig mutation frequency of AICDA−/− B cells was greater than that expected from DNA instability in non-B cells. We directly measured the mutational frequency of unrearranged VH4 genes in non-B cells and determined that it was no more than the PCR error rate. Thus, the mutation frequency of AICDA−/− B cells was nearly 10-fold that of Ig genes in non-B cells. Other approaches have been used to estimate the mutation frequency of eukaryotic cells, which is generally reported to be ≤10−10 per bp per cell division (33), compared with 1 × 10−3 per bp per cell division for Ig genes in B cells (34). Assuming a DNA length of 240 bp, equivalent to that of VH gene sequences, and that the cells undergo 25 cell divisions, which would yield a predicted Ig gene mutational frequency of 2.5% (similar to that measured for normal B cells), the AICDA-deficient B cells have as much as a 2.8 × 105-fold increase in mutational frequency compared with that of DNA in non-B cells. This estimate does not take into account the increased mutability of Ig genes within B cells, but it clearly is consistent with the current data that the frequency of mutations of Ig genes is significantly greater in AICDA−/− B cells than that of non-Ig genes in B cells or Ig genes in non-B cells. Whether this reflects an increased likelihood of introducing bp changes or an altered tendency for error-prone repair is currently unknown.

During the GC reaction, expression of AICDA and multiple enzymes involved in DNA repair and translesion synthesis is uniquely altered, resulting in high-frequency mutation with specific complex patterns targeted to Ig genes. A greater understanding of how individual components contribute to SHM in humans has resulted from the analysis of the Ig gene mutational pattern in B cells from individuals with a deficiency in one of the enzymes involved in SHM or the GC reaction itself. In normal B cells, WA mutations occur both when motifs overlap RGYW/WRCY tetrads, as well as when they are distant (11). This pattern has made it difficult to distinguish whether the underlying mutation process targeting WA motifs was related to AICDA (35). The loss of POL η activity in xeroderma pigmentosa variant resulted in a reduction in WA mutations, regardless of their position inside or outside RGYW/WRCY motifs, indicated that WA mutations in RGYW/WRCY were related to POL η targeting and not to the direct activity of AICDA (11, 36). Thus, the loss of POL η targeting in WA motifs in the AICDA−/− Ig genes may reflect the observation given that pol η is constitutively expressed in B cells and not up-regulated upon CD40 ligation to a level of expression that is sustained in GCs that are present and accentuated in size in subjects with AICDA deficiency (3). Based upon the current finding that there was a marked reduction of WA targeting in AICDA−/− B cells, it is likely that AICDA may directly or indirectly recruit POL η into a mutation complex. This conclusion is consistent with recent findings that the induction of A/T mutations in a transgenic mutation substrate required a potentially AICDA-targeted G/C site (28). However, the current data extend this finding to implicate POL η as the error-prone polymerase targeting WA motifs in an AICDA-dependent manner. These findings also support the data from X-HIgM studies suggesting a linkage between POL η targeting WA within the AICDA-induced lesion. Notably, however, the current data indicate that AICDA is also responsible for recruiting POL η activity to sites remote from the AICDA-targeted lesion, possibly through long patch repair (37).

The AICDA-deficient B cells demonstrated a marked skewing toward transitions, whereas normal B cells had equal numbers of transitions and transversions. A similar skewing toward transitions was attributed to the absence of UNG in a chicken cell line (32) and mouse (38) and human B cells (39). Importantly, these results suggest a role for AICDA in recruiting UNG2 to sites of AICDA-dependent uracil generation, perhaps by forming a multiprotein structure with proliferating cell nuclear Ag and replication protein A.

The lack of targeting to WRCY/RGYW motifs and the decrease in CDR mutations as well as R substitutions in the CDRs provide support for the proposal that two intrinsic features of Ig genes may have coevolved with AICDA to focus the mutation mechanism on the Ag binding site. First, WRCY/RGYW motifs are more concentrated in CDRs than in the FRs of the VH genes (8). This may bias the mutational machinery initiated by AICDA to induce mutations preferentially into CDRs, since the current data clearly show that in the absence of AICDA, the bias for replacement mutations to localize into CDRs was markedly reduced. The second feature of Ig genes noted in normal B cells to influence mutational patterning is the biased use of codons that favor replacement substitutions. This was highlighted by the increase in the predicted R:S ratio in CDRs (5.2) compared with FRs (3.4) in the current analysis. It is notable that positive selection of R mutations in the CDRs was apparent with the low mutation frequency of AICDA−/− B cells, but was absent from the normal mutated B cells that have a markedly higher mutation frequency compared with AICDA−/− mutated B cells. This may be attributed to the increased loss of B cells that accumulate large numbers of mutations either because the R mutations in the CDRs introduce autoreactivity or R mutations in the FRs alter the structural integrity of the molecule (8). The very low mutation frequency in AICDA−/− B cells may make it less likely that deleterious mutations are introduced and, therefore, more probable that positive selection of mutations can be observed.

The residual mutations in AICDA−/− Ig genes did not appear to be targeted by a new mutator to an identifiable motif. Moreover, the mutations did not exhibit a pattern of targeting within G–G or Y–Y dinucleotides compatible with oxidative damage to DNA. A likely source of the predominant G/C nucleotide mutations was spontaneous cytidine deamination that is common to all cells but apparently occurs more frequently in Ig genes, possibly because of the general down-regulation of conventional high-fidelity repair during a GC reaction. Spontaneous cytidine deamination occurs without any sequence specificity (31) but is strongly dependent on DNA structure. The rate of deamination is 2 orders of magnitude greater for ssDNA than for dsDNA because of cytosine accessibility during replication or transcription (40). Although it is possible that the high rate of B cell proliferation during GC reactions (34) or the level of Ig gene transcription makes spontaneous cytidine deamination more likely in B cells, there is no evidence that this is the case. Alternatively, B cells may be more likely to use error-prone repair mechanisms to correct mutations and thereby be more likely to accumulate G mutations. The nature of such error-prone polymerase is currently unknown but the targeting to G/C nucleotides indicates that POL η is not involved, leaving POL θ or POL ζ as likely candidates. In mice it appears that all A/T mutations are attributed to POL η recruitment by ΜSΗ2 (41). However, A/T mutations remain in the pol η and msh2 single knockouts when AICDA and UNG are available, suggesting that MSH2 may be capable of recruiting other enzymes if POL η is absent or, in the absence of MSH2, error-prone enzymes other than POL η may leave a mild A/T footprint. Thus, the mechanism that accounts for the other 40% of mutations in AICDA-deficient B cells remains unknown.

Overall, the mutations that occur in the absence of AICDA are likely to have little impact on the conventional Ag binding site or the nonvariable FR regions that superantigens bind (42). Despite the positive selection of mutated B cells that occurs in AICDA−/− subjects, the data suggest that the process may not be sufficiently robust to allow B cells to undergo affinity maturation. The lack of affinity maturation and class switching is reflected in the absence of anti-tetanus IgG Abs in AICDA−/− patients after immunization (3), although detailed affinity measurements of these Abs has not been conducted.

In conclusion, the absence of AICDA results in a loss of targeting to known AICDA- and POL η-hypermutable motifs and decreased mutation (and R substitutions) in the CDRs. The current study provides evidence that AICDA not only initiates SHM but is also involved in the recruitment of POL η and, possibly, UNG. The mutation pattern in AICDA−/− B cells suggests that the G:C mutation spectrum is most likely introduced by AICDA-independent cytidine deamination on both DNA strands and a low-level A:T mutation spectrum that is not associated with POL η. Finally, because of the loss of CDR targeting, somatic mutation in AICDA−/− B cells provides an inefficient mechanism for altering the Ag binding site.

We thank Igor Rogozin (National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Bethesda, MD) for comments and discussions. We are grateful for the cooperation of the patients, their families, and normal donors for their participation in the study and thank Jim Simone (Flow Cytometry Section, Office of Science and Technology, National Institute of Arthritis and Musculoskeletal and Skin Diseases, Bethesda, MD) for cell sorting.

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was supported in part by the National Institutes of Health, National Institute of Arthritis and Musculoskeletal and Skin Diseases, Intramural Research Program, grants from the Institut National de la Santé et de la Recherche Médicale, Association de la Recherche Contre le Cancer, the European Community No. PL 006411 (EUROPOLICY-PID)-6th Programme Cadre de Recherche Développment, the Association Nationale pour la Recherche, Institut National du Cancer, and a grant from the Fondazione C. Golgi and the Associazione Immunodeficienze Primitive, Brescia, Italy.

3

Abbreviations used in this paper: AICDA, activation-induced cytidine deaminase; FR, framework region; MMR, mismatch repair; POL, polymerase; R, replacement; S, silent; SHM, somatic hypermutation; GC, germinal center; UNG, uracil-DNA glycosylase.

1
Muramatsu, M., K. Kinoshita, S. Fagarasan, S. Yamada, Y. Shinkai, T. Honjo.
2000
. Class switch recombination and hypermutation require activation-induced cytidine deaminase (AID), a potential RNA editing enzyme.
Cell
102
:
553
-563.
2
Milstein, C..
1987
. Diversity and the genesis of high affinity antibodies.
Biochem. Soc. Trans.
15
:
779
-787.
3
Revy, P., T. Muto, Y. Levy, F. Geissmann, A. Plebani, O. Sanal, N. Catalan, M. Forveille, R. Dufourcq-Labelouse, A. Gennery, et al
2000
. Activation-induced cytidine deaminase (AID) deficiency causes the autosomal recessive form of the hyper-IgM syndrome (HIGM2).
Cell
102
:
565
-575.
4
Meffre, E., N. Catalan, F. Seltz, A. Fischer, M. C. Nussenzweig, A. Durandy.
2001
. Somatic hypermutation shapes the antibody repertoire of memory B cells in humans.
J. Exp. Med.
194
:
375
-378.
5
Barreto, V. M., A. R. Ramiro, M. C. Nussenzweig.
2005
. Activation-induced deaminase: controversies and open questions.
Trends Immunol.
26
:
90
-96.
6
Yu, K., F. T. Huang, M. R. Lieber.
2004
. DNA substrate length and surrounding sequence affect the activation-induced deaminase activity at cytidine.
J. Biol. Chem.
279
:
6496
-6500.
7
Shen, H. M., U. Storb.
2004
. Activation-induced cytidine deaminase (AID) can target both DNA strands when the DNA is supercoiled.
Proc. Natl. Acad. Sci. USA
101
:
12997
-13002.
8
Dörner, T., H. P. Brezinschek, R. I. Brezinschek, S. J. Foster, R. Domiati-Saad, P. E. Lipsky.
1997
. Analysis of the frequency and pattern of somatic mutations within nonproductively rearranged human variable heavy chain genes.
J. Immunol.
158
:
2779
-2789.
9
Xue, K., C. Rada, M. S. Neuberger.
2006
. The in vivo pattern of AID targeting to immunoglobulin switch regions deduced from mutation spectra in msh2−/−ung−/− mice.
J. Exp. Med.
203
:
2085
-2094.
10
Milstein, C., M. S. Neuberger, R. Staden.
1998
. Both DNA strands of antibody genes are hypermutation targets.
Proc. Natl. Acad. Sci. USA
95
:
8791
-8794.
11
Yavuz, S., A. S. Yavuz, K. H. Kraemer, P. E. Lipsky.
2002
. The role of polymerase η in somatic hypermutation determined by analysis of mutations in a patient with xeroderma pigmentosum variant.
J. Immunol.
169
:
3825
-3830.
12
Brezinschek, H. P., R. I. Brezinschek, P. E. Lipsky.
1995
. Analysis of the heavy chain repertoire of human peripheral B cells using single-cell polymerase chain reaction.
J. Immunol.
155
:
190
-202.
13
Dörner, T., H. P. Brezinschek, S. J. Foster, R. I. Brezinschek, N. L. Farner, P. E. Lipsky.
1998
. Comparable impact of mutational and selective influences in shaping the expressed repertoire of peripheral IgM+CD5 and IgM+CD5+ B cells.
Eur. J. Immunol.
28
:
657
-668.
14
Brezinschek, H. P., S. J. Foster, R. I. Brezinschek, T. Dörner, R. Domiati-Saad, P. E. Lipsky.
1997
. Analysis of the human VH gene repertoire: differential effects of selection and somatic hypermutation on human peripheral CD5+IgM+ and CD5IgM+ B cells.
J. Clin. Invest.
99
:
2488
-2501.
15
Monson, N. L., T. Dorner, P. E. Lipsky.
2000
. Targeting and selection of mutations in human Vλ rearrangements.
Eur. J. Immunol.
30
:
1597
-1605.
16
Foster, S. J., H. P. Brezinschek, R. I. Brezinschek, P. E. Lipsky.
1997
. Molecular mechanisms and selective influences that shape the κ gene repertoire of IgM+ B cells.
J. Clin. Invest.
99
:
1614
-1627.
17
Souto-Carneiro, M. M., N. S. Longo, D. E. Russ, H. W. Sun, P. E. Lipsky.
2004
. Characterization of the human Ig heavy chain antigen binding complementarity determining region 3 using a newly developed software algorithm, JOINSOLVER.
J. Immunol.
172
:
6790
-6802.
18
Farner, N. L., T. Dörner, P. E. Lipsky.
1999
. Molecular mechanisms and selection influence the generation of the human VλJλ repertoire.
J. Immunol.
162
:
2137
-2145.
19
Yavuz, A. S., P. E. Lipsky, S. Yavuz, D. D. Metcalfe, C. Akin.
2002
. Evidence for the involvement of a hematopoietic progenitor cell in systemic mastocytosis from single-cell analysis of mutations in the c-kit gene.
Blood
100
:
661
-665.
20
Wei, C., J. Anolik, A. Cappione, B. Zheng, A. Pugh-Bernard, J. Brooks, E. H. Lee, E. C. Milner, I. Sanz.
2007
. A new population of cells lacking expression of CD27 represents a notable component of the B cell memory compartment in systemic lupus erythematosus.
J. Immunol.
178
:
6624
-6633.
21
Rogozin, I. B., N. A. Kolchanov.
1992
. Somatic hypermutagenesis in immunoglobulin genes: II. Influence of neighbouring base sequences on mutagenesis.
Biochim. Biophys. Acta
1171
:
11
-18.
22
Dörner, T., S. J. Foster, H. P. Brezinschek, P. E. Lipsky.
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
-171.
23
Dörner, T., S. J. Foster, N. L. Farner, P. E. Lipsky.
1998
. Somatic hypermutation of human immunoglobulin heavy chain genes: targeting of RGYW motifs on both DNA strands.
Eur. J. Immunol.
28
:
3384
-3396.
24
Mayorov, V. I., I. B. Rogozin, L. R. Adkison, P. J. Gearhart.
2005
. DNA polymerase η contributes to strand bias of mutations of A versus T in immunoglobulin genes.
J. Immunol.
174
:
7781
-7786.
25
Wilson, T. M., A. Vaisman, S. A. Martomo, P. Sullivan, L. Lan, F. Hanaoka, A. Yasui, R. Woodgate, P. J. Gearhart.
2005
. MSH2-MSH6 stimulates DNA polymerase η, suggesting a role for A:T mutations in antibody genes.
J. Exp. Med.
201
:
637
-645.
26
Rogozin, I. B., Y. I. Pavlov, K. Bebenek, T. Matsuda, T. A. Kunkel.
2001
. Somatic mutation hotspots correlate with DNA polymerase η error spectrum.
Nat. Immunol.
2
:
530
-536.
27
Unniraman, S., D. G. Schatz.
2007
. Strand-biased spreading of mutations during somatic hypermutation.
Science
317
:
1227
-1230.
28
Diaz, M., C. Lawrence.
2005
. An update on the role of translesion synthesis DNA polymerases in Ig hypermutation.
Trends Immunol.
26
:
215
-220.
29
Masuda, K., R. Ouchida, M. Hikida, T. Kurosaki, M. Yokoi, C. Masutani, M. Seki, R. D. Wood, F. Hanaoka, O. W. J. .
2007
. DNA polymerases η and θ function in the same genetic pathway to generate mutations at A/T during somatic hypermutation of Ig genes.
J. Biol. Chem.
282
:
17387
-17394.
30
Lindahl, T..
1993
. Instability and decay of the primary structure of DNA.
Nature
362
:
709
-715.
31
Di Noia, J., M. S. Neuberger.
2002
. Altering the pathway of immunoglobulin hypermutation by inhibiting uracil-DNA glycosylase.
Nature
419
:
43
-48.
32
Kunkel, T. A., K. Bebenek.
2000
. DNA replication fidelity.
Annu. Rev. Biochem.
69
:
497
-529.
33
McKean, D., K. Huppi, M. Bell, L. Staudt, W. Gerhard, M. Weigert.
1984
. Generation of antibody diversity in the immune response of BALB/c mice to influenza virus hemagglutinin.
Proc. Natl. Acad. Sci. USA
81
:
3180
-3184.
34
Dörner, T., P. E. Lipsky.
2001
. Smaller role for pol η?.
Nat. Immunol.
2
:
982
-984.
35
Zeng, X., G. A. Negrete, C. Kasmer, W. W. Yang, P. J. Gearhart.
2004
. Absence of DNA polymerase η reveals targeting of C mutations on the nontranscribed strand in immunoglobulin switch regions.
J. Exp. Med.
199
:
917
-924.
36
Modrich, P., R. Lahue.
1996
. Mismatch repair in replication fidelity, genetic recombination, and cancer biology.
Annu. Rev. Biochem.
65
:
101
-133.
37
Rada, C., G. T. Williams, H. Nilsen, D. E. Barnes, T. Lindahl, M. S. Neuberger.
2002
. Immunoglobulin isotype switching is inhibited and somatic hypermutation perturbed in UNG-deficient mice.
Curr. Biol.
12
:
1748
-1755.
38
Imai, K., G. Slupphaug, W. I. Lee, P. Revy, S. Nonoyama, N. Catalan, L. Yel, M. Forveille, B. Kavli, H. E. Krokan, et al
2003
. Human uracil-DNA glycosylase deficiency associated with profoundly impaired immunoglobulin class-switch recombination.
Nat. Immunol.
4
:
1023
-1028.
39
Frederico, L. A., T. A. Kunkel, B. R. Shaw.
1990
. A sensitive genetic assay for the detection of cytosine deamination: determination of rate constants and the activation energy.
Biochemistry
29
:
2532
-2537.
40
Delbos, F., S. Aoufouchi, A. Faili, J. C. Weill, C. A. Reynaud.
2007
. DNA polymerase η is the sole contributor of A/T modifications during immunoglobulin gene hypermutation in the mouse.
J. Exp. Med.
204
:
17
-23.
41
Silverman, G. J..
1998
. B cell superantigens: possible roles in immunodeficiency and autoimmunity.
Semin. Immunol.
10
:
43
-55.