and Class Switch Recombination in Msh6−/−Ung−/− Double-Knockout Mice

This information is current as Hong Ming Shen, Atsushi Tanaka, Grazyna Bozek, Dan of October 2, 2021. Nicolae and Ursula Storb J Immunol 2006; 177:5386-5392; ; doi: 10.4049/jimmunol.177.8.5386 http://www.jimmunol.org/content/177/8/5386 Downloaded from

References This article cites 44 articles, 20 of which you can access for free at: http://www.jimmunol.org/content/177/8/5386.full#ref-list-1 http://www.jimmunol.org/ Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on October 2, 2021

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2006 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Somatic Hypermutation and Class Switch Recombination in Msh6؊/؊Ung؊/؊ Double-Knockout Mice1

Hong Ming Shen,* Atsushi Tanaka,† Grazyna Bozek,* Dan Nicolae,‡ and Ursula Storb2*†

Somatic hypermutation (SHM) and class switch recombination (CSR) are initiated by activation-induced deaminase (AID). The , and potentially neighboring bases, are processed by error-prone base excision repair and mismatch repair. Deficiencies in Ung, Msh2, or Msh6 affect SHM and CSR. To determine whether Msh2/Msh6 complexes which recognize single- base mismatches and loops were the only mismatch-recognition complexes required for SHM and CSR, we analyzed these ;processes in Msh6؊/؊Ung؊/؊ mice. SHM and CSR were affected in the same degree and fashion as in Msh2؊/؊Ung؊/؊ mice were mostly C,G transitions and CSR was greatly reduced, making Msh2/Msh3 contributions unlikely. Inactivating Ung alone reduced mutations from A and T, suggesting that, depending on the DNA sequence, varying proportions of A,T mutations ؊/؊ ؊/؊ ؅ ؅ arise by error-prone long-patch base excision repair. Further, in Msh6 Ung mice the 5 end and the 3 region of Ig genes Downloaded from was spared from mutations as in wild-type mice, confirming that AID does not act in these regions. Finally, because in the absence of both Ung and Msh6, mutations from C and G likely are “footprints” of AID, the data show that the activity of AID is restricted drastically in vivo compared with AID in cell-free assays. The Journal of Immunology, 2006, 177: 5386–5392.

racil created by activation-induced cytosine deaminase (22). These findings showed that both BER and MMR are involved (AID)3 during somatic hypermutation (SHM) or class in SHM and CSR, and further, suggested that the two mechanisms http://www.jimmunol.org/ U switch recombination (CSR) can be eliminated in vari- may interact in the processing of the AID-created (see ous ways, by direct copying during DNA replication, resulting in Discussion). C/G to T/A transitions, by base excision repair (BER) using Ung Msh2/Ung double-knockout mice produce essentially only C as the major uracil glycosylase, or by mismatch repair (MMR) and G transitions during SHM and have severely reduced CSR (reviewed in Refs. 1–4). In SHM, either the uracil alone and/or (23). We initiated crosses between Msh6Ϫ/Ϫ and UngϪ/Ϫ mice to neighboring bases can be altered, depending on whether BER or test whether MMR and BER are the major mechanisms of creating MMR are involved, and on the number of nucleotides excised and post-AID mutations from A and T, and transversions from C and the types of error-prone polymerases engaged. Inactivation of Ung G. We used Msh6 rather than Msh2 because the former, together results in an almost complete elimination of transversion mutations with Msh2 in a MutS␣ complex, is only involved in recognition of by guest on October 2, 2021 from C and G, while the frequency of mutations from C and G single base-pair mismatches and single base loops, whereas Msh2, overall is not altered and mutations from A and T are still rela- in combination with Msh3 (MutS␤), is also involved in recogniz- Ϫ/Ϫ tively frequent (3, 4). Ung mice also are defective in CSR (5). ing larger loops (24). AID creates single base U/G mismatches; Inactivation of the MMR recognition proteins Msh2 or Msh6 re- larger mismatches and unpaired loops are not direct consequences duces overall mutations and greatly reduces mutations from A or of AID , although loops may arise during CSR in R- Ϫ/Ϫ Ϫ/Ϫ T (6–11). CSR is also diminished in Msh2 and Msh6 mice looped switch regions, and potentially during SHM if AID deami- (2, 6, 7, 12–14). Genes that encode proteins involved in postmis- nates two or more consecutive Cs or if error-prone repair causes a match recognition functions of MMR have also been shown to be mismatch of more than one nucleotide. A potentially different role involved in SHM and CSR. Mice deficient in the MutL homologs of the mismatch repair complex MutS␤ (Msh2/Msh3) compared Mlh1 and Pms2 have altered SHM and CSR patterns (13–20). with MutS␣ (Msh2/Msh6) was suggested, because the pattern of Exonuclease 1-mutant mice have altered SHM and reduced CSR switch joints appeared to differ in Msh2Ϫ/Ϫ and Msh6Ϫ/Ϫ mice (21). Interestingly, mice deficient in Mlh3 have increased SHM, (12). Recently, it was suggested that Msh6 may serve as a scaffold suggesting that Mlh3 normally reduces the SHM load in SHM (25). The Msh6Ϫ/ϪUngϪ/Ϫ mice were also important to further study whether the sparing of the very 5Ј end and the C region of Ig genes *Department of Molecular Genetic and Cell Biology, †Committee on Immunology, during SHM was indeed due to lack of AID access/activity in these ‡ and Department of Statistics, University of Chicago, Chicago, IL 60637 regions. We found previously that the lack of mutations in the very Received for publication May 23, 2006. Accepted for publication July 20, 2006. 5Ј and 3Ј regions of Ig genes was as severe in UngϪ/Ϫ mice as in The costs of publication of this article were defrayed in part by the payment of page wild-type mice and concluded that AID did not function in these charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. regions (26). There remained the possibility that AID-created ura- Ј Ј Ϫ/Ϫ 1 This work was supported by National Institutes of Health Grants AI47380 and cils were present in the 5 and 3 regions of Ung mice, but that AI053130. U/Gs in these flanking regions were repaired error-free by MMR 2 Address correspondence and reprint requests to Dr. Ursula Storb, Department of that proceeded without mistakes, as MMR would elsewhere in the Molecular and Cell Biology, University of Chicago, 920 East 58th Street, . In that case, uracils in the 5Ј and 3Ј ends could have been Chicago, IL 60637. E-mail address: [email protected] faithfully repaired to by MMR. Elimination of both sin- 3 Abbreviations used in this paper: AID, activation-induced cytosine deaminase; SHM, somatic hypermutation; CSR, class switch recombination; BER, base excision gle-mismatch MMR and Ung likely reveals whether error-free repair; MMR, mismatch repair; SRBC, sheep RBC; PNA, peanut lectin agglutinin. copying of uracils occurs in Ig gene SHM.

Copyright © 2006 by The American Association of Immunologists, Inc. 0022-1767/06/$02.00 The Journal of Immunology 5387

Materials and Methods ates). Briefly, 5 ␮g of capture Ab was added to each well in an ELISA Mice plate, and the plate was incubated overnight at 4°C. After rinsing the plate three times with PBS containing 0.05% Tween 20 to remove the extra UngϪ/Ϫ mice were a gift of D. Barnes and T. Lindahl (Imperial Cancer capture Ab, the wells were blocked with 1% BSA. The diluted serum Research Fund, Clare Hall Laboratories, South Mimms, U.K.), and were samples containing Igs of interest were then added to the wells, and the further bred in our mouse facility on a C57BL/6 background. A male ELISA plate was incubated overnight at 4°C. Unbound serum proteins Msh6ϩ/Ϫ mouse was obtained from the Mouse Repository at the National were rinsed out with PBS containing 0.05% Tween 20 before various Abs Cancer Institute (Rockville, MD); the latter mice were originally donated by anti-Ig C regions conjugated with alkaline phosphatase were added. The W. Edelmann (Albert Einstein College of Medicine, Bronx, NY). Both plate was incubated at room temperature on a shaker for 1 h. Unbound Abs UngϪ/Ϫ and Msh6ϩ/Ϫ mice had a C57BL/6 ϫ 129 background. The mice were removed by rinsing the plate five times with PBS containing 0.05% were maintained according to National Institutes of Health instructions. Tween 20. The substrate ( p-nitrophenyl phosphate) was added to the wells ϩ/Ϫ ϩ/Ϫ Ϫ/Ϫ (Ung Msh6 )F1 mice were acquired by crossing Ung mice with the and the OD was read at 405 nm in the Synergy HT reader (Bio-Tek In- Msh6ϩ/Ϫ mouse. Offspring were further crossed twice to generate UngϪ/Ϫ struments) at various time periods. Each serum sample was assayed three Msh6ϩ/ϩ, Ungϩ/ϪMsh6Ϫ/Ϫ, and UngϪ/ϪMsh6Ϫ/Ϫ mice. The experiments times in duplicate. Fig. 2 shows the average of all assays. with mice have been reviewed and approved by the University of Chicago Institutional Animal Care and Use Committee. Statistical analysis Cells and DNAs Pairwise comparisons of Ig levels in different transgenic mice were per- formed using a rank-based nonparametric testing procedure. For each ex- ϩ Ϫ Ϫ Ϫ Mice (all mice were 2 mo old, except for the Ung / Msh6 / mice which periment, the Ig levels of the mice were ranked, and the sum of ranks for were 3 mo old; see Table I) were immunized i.p. with 108 sheep RBC one of the strains was calculated across experiments. An empirical distri- (SRBC; MP Biomedicals) on day 1, boosted with the same amount of bution of this statistic under the null hypothesis of no difference in Ig levels SRBC on day 8, and sacrificed on day 11. Sera were saved for ELISA was constructed using simulations, and two-sided p values were calculated analysis, while spleens were removed and minced in RPMI 1640. RBC in from this distribution. We performed 10,000 simulations, and each simu- Downloaded from the spleen were lysed and the remaining spleen cells were stained with lated rank was drawn by assuming that the Ig levels of the two strains of fluorescein-peanut lectin agglutinin (PNA) and PE-anti-B220. PNAhigh mice are independent identically distributed variables within each experi- ϩ B220 cells were purified using a Mo-Flo sorter (BD Biosciences) at the ment. The statistically significant differences are indicated in Fig. 2. Immunology Core Facility (University of Chicago). The purified cells were lysed in lysis buffer containing proteinase K (500 ␮g/ml), EDTA (50 mM, Results pH 8.0), SDS (1%), and Tris (50 mM, pH8.0) overnight at 37°C. DNAs Ϫ/Ϫ were extracted using phenol/chloroform. The mutation frequencies are not changed in Ung and Ϫ/Ϫ Ϫ/Ϫ Ϫ/Ϫ Ung Msh6 mice but reduced in Msh6 mice with http://www.jimmunol.org/ PCR, cloning and mutation analysis wild-type Ung The PCR method for the H chain gene was described elsewhere (26, 27). We analyzed mutations in the H chain locus starting from the first Briefly, PFU turbo DNA polymerase (Invitrogen Life Technologies) was used during a touchdown PCR in the presence of a pair of primers which nucleotide of the JC intron of VDJH4-rearranged genes. The uracil- ␮ DNA glycosylase-deficient mice underwent efficient somatic hy- specifically annealed to the rearranged VJH558 region and the JH4-C intron (27). Gel electrophoresis was performed to purify a 1.2-kb DNA permutation indicated by the mutation frequencies in mutated fragment containing VJH558 rearranged to JH4 using the QIAquick Gel DNA sequences (3.3 and 2.3/1000 bp, compared with 2.6 and 2.5/ Extraction kit (Qiagen). The purified DNA was inserted into a TOPO II 1000 bp in wild-type mice) (Table I). However, the overall muta- (blunt) vector (Invitrogen Life Technologies). The vector-transformed bac- Ϫ/Ϫ teria were grown overnight at 37°C in a kanamycin plate. Colonies were tion frequencies in mice with the Msh6 background and wild- by guest on October 2, 2021 picked and further amplified in Luria-Bertani medium in the presence of type Ung were reduced (1.4, 1.5, 1.3, and 1.5/1000 bp in mutated kanamycin. The vector DNA was purified using the QIAprep Spin Mini- DNA sequences) (Table I) and mutated sequences accumulated prep kit (Qiagen). Alternatively, colonies were picked and DNA was au- only few mutations (Fig. 1, pie charts). The unchanged mutation tomatically prepared at the sequencing center (University of Chicago). Ϫ/Ϫ DNA was sequenced using automated sequencing (3730XL; Applied Bio- frequencies in Ung mice and the reduced mutations in the sec- Ϫ/Ϫ systems), and sequence analysis was performed using Sequencher 4.1. ondary response of Msh6 mice agrees with previous findings in Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Only mutations in the 1.1-kb DNA sequence starting at the first nucleotide Ung / mice (5, 26), Msh6 / mice (7, 9), and Msh2 / mice of the intron (ϳ0.7 kb from the ) were counted to avoid any (10), respectively. Thus, production of Msh6, as well as Msh2, Ј ambiguity caused by VDJ recombination. For the sequencing of the 5 end which together are involved in mismatch repair of single base mis- of the ␭ 1 gene, the procedure described in Ref. 26 was used. matches or loops, appears to increase the mutation frequency. Al- ELISA ternatively, Msh2/Msh6 may aid the survival of mutating B cells Ϫ/Ϫ Ϫ/Ϫ ELISA was performed using the SBA clonotyping System/AP kit accord- (10) (see Discussion). Interestingly, in Msh6 Ung double- ing to the manufacturer’s instructions (Southern Biotechnology Associ- knockout mice the mutation frequencies in mutated sequences

Table I. Mutation frequencies

Age in Mutated Mutations/1,000 bp Mutations/1,000 bp in Mouse Type Monthsa Titerb Mutations Clones Basesc in Mutated Clones Sequenced Clones

Wild 1 2 1:256 57 20 58,850 2.6 1.0 Wild 2 2 1:384 95 34 65,450 2.5 1.5 UngϪ/ϪMsh6ϩ/ϩ 1 2 1:256 120 33 65,450 3.3 1.8 UngϪ/ϪMsh6ϩ/ϩ 2 2 1:512 84 33 61,600 2.3 1.4 Ungϩ/ϪMsh6Ϫ/Ϫ 1 3 1:16 15 10 92,400 1.4 0.2 Ungϩ/ϪMsh6Ϫ/Ϫ 2 3 1:128 56 33 101,200 1.5 0.6 Ungϩ/ϩMsh6Ϫ/Ϫ 1 2 1:256 21 15 54,450 1.3 0.4 Ungϩ/ϩMsh6Ϫ/Ϫ 2 2 1:128 10 6 30,250 1.5 0.3 UngϪ/ϪMsh6Ϫ/Ϫ 1 2 1:128 81 27 67,650 2.7 1.2 UngϪ/ϪMsh6Ϫ/Ϫ 2 2 1:96 14 6 63,800 2.1 0.2 UngϪ/ϪMsh6Ϫ/Ϫ 3 2 1:256 40 16 41,250 2.3 1.0 UngϪ/ϪMsh6Ϫ/Ϫ 4 2 1:256 41 13 51,150 2.9 0.8

a Mice were immunized when they were 2 or 3 mo old and sacrificed 11 days later. b Anti-sheep RBC Abs. c A total of 1100 bp starting from the first nucleotide of the JC intron were analyzed. Identical sequences in the same PCR were counted only once. 5388 SHM AND CSR IN Msh6Ϫ/ϪUngϪ/Ϫ DOUBLE-KNOCKOUT MICE

FIGURE 1. Mutation patterns in wild-type and dif- ferent knockout mice. The data were combined from two to four mice of each genotype (see Table I). The third box, Ungϩ, Msh6Ϫ/Ϫ, includes two Ungϩ/Ϫ Msh6Ϫ/Ϫ and two Ungϩ/ϩMsh6Ϫ/Ϫ mice (see Table I). The fourth box, UngϪ/ϪMsh6Ϫ/Ϫ, includes 176 muta- tions in the H chain locus (the same region as compiled in the other three boxes) and 54 mutations in VJ␭1 (see Fig. 3). The last column in each box is the percentage adjusted for nucleotide composition in the 1.1-kb region Downloaded from Ј Ϫ/Ϫ Ϫ/Ϫ 3 of JH4 (last column of Msh6 Ung is adjusted for both the heavy and ␭ genes). Identical sequences in the same PCR were counted only once. In the pie charts, the numbers outside the large circles are point muta- tions, those in the central circles are the total clones sequenced. http://www.jimmunol.org/ by guest on October 2, 2021

were as in wild-type mice (2.7, 2.1, 2.3, and 2.9/1000 bp of mu- 34%, compared with 51% in the wild-type mice (Table II, no. 6), tated sequences) (Table I) and many mutated sequences accumu- a reduction in UngϪ/Ϫ by 33%. This is a similar reduction of A,T Ϫ/Ϫ lated large numbers of mutations (Fig. 1, pie charts) (see mutations as previously seen near JH4 in the same Ung mice Discussion). (5, 32) which had 43% A,T mutations, compared with 57% in

Ϫ Ϫ wild-type mice (a reduction by 25%) (Table II, no. 4). Our findings Ung / mice can mutate A and T, but at a reduced frequency Ј here are also similar to the A,T mutations in the 1145 nt 3 of JH4 In UngϪ/Ϫ mice, as described before (5, 26), most of the mutations in another study (26). There, wild-type mice had 47% A,T muta- from C and G were transitions. The UngϪ/Ϫ and wild-type mice tions, and UngϪ/Ϫ mice only 36% (Table II, no. 3). The least showed a similar ratio of mutations at A over those at T, however, reduction of A,T mutations by only 11% was seen in the most the A,T mutation frequency was reduced in the UngϪ/Ϫ mice (Fig. highly mutated region of a ␬ transgene (Table II, no. 1) (26). The Ј 1). The percentage of A,T mutations in the 1100 nt 3 of JH4 was highest reduction of A,T mutations (by 52%) was found toward the

Table II. Ung is responsible for various proportions of mutations at A and T depending on the target sequence

% Mutations from A and T Mutation Frequency Hot Spots/100 nt

Sequencea Wt UngϪ/Ϫ Ung/Wt UngϪ/Ϫ/Wt WRC ϩ GYW (WA ϩ TW) Ref.

1. 917-bp center k-TG 53 47 0.89 0.9 11 (26) 26 2. 864-bp 5Ј k-TG 52 43 0.83 0.4 10.7 (24)b 26 ␮ 3. 1145-bp JH4C- int 47 36 0.77 1.2 10.2 (32) 26 Ј 4. VH 3 flank 57 43 0.75 0.85 8.0 (28) 5, 32 5. 438-bp 5Ј VJ ␭1c 45 30 0.67 0.4 12.6 (20)b 26 ␮ 6. 1100-bp JH4C- int 51 34 0.67 1.2 10.4 (33) This paper 7. 1321-bp JC-int k-TG 65 31 0.48 1 9.4 (42) 26

a k-TG, kappa transgene; int, intron. b Only the mutable region starting at ϩ100 nt is included. c The only gene in this list whose expression depended on B cell selection. The Journal of Immunology 5389

3Ј end of the JC intron of the same ␬ transgene (Table II, no. 7) cept for IgA ( p Ͻ 0.0002). This differs from previous data that (26). Because in this study (26), all regions were sequenced from showed that both IgG1 and IgG3 in Msh6Ϫ/Ϫ mouse sera were PNAhigh B cells of the same mice (Table II, nos. 1, 2, 3, 5, 7), but significantly reduced (7). The mice in that study were 3–6 mo old, the proportion of A,T mutations in UngϪ/Ϫ vs wild-type mice vary our Msh6Ϫ/ϪUngϩ/Ϫ mice were 3 mo old (Table I). Because at from 0.48 to 0.89, the differences are unlikely due to immuniza- these ages residual maternal Abs are unlikely, perhaps unknown tion, but rather may depend on the primary sequence. Taken to- background genes are involved in the different findings. The higher gether, these findings suggest that Ung, and thus presumably BER, IgM titers in UngϪ/ϪMsh6ϩ/ϩ, Ungϩ/ϪMsh6Ϫ/Ϫ, and UngϪ/Ϫ plays a role in A,T mutations (see Discussion). Msh6Ϫ/Ϫ mice were expected (Fig. 2) to compensate for the re- duced or absent IgGs and IgA by overexpressing IgM. Interest- Ϫ/Ϫ Ϫ/Ϫ Ϫ/Ϫ Most mutations in Msh6 and Ung Msh6 mice are ingly, although CSR in the UngϪ/ϪMsh6ϩ/ϩ mice was signifi- transitions from C and G cantly reduced compared with wild-type mice, it was higher than Ϫ Ϫ Ϫ Ϫ A previous study showed that C,G mutations were mainly transi- in Ung / Msh6 / mice ( p Ͻ 0.0002) (Fig. 2), indicating that tions and A,T mutations were greatly diminished in UngϪ/Ϫ both BER and MMR are responsible for CSR and that their roles Msh2Ϫ/Ϫ double-knockout mice (23). To further determine in CSR may be somewhat overlapping. whether the Msh2/6 complex is essential and whether there is any Ј contribution by the Msh2/Msh3 complex, or Msh2 alone, to gen- The 5 region of Ig genes is not accessible to AID erate transversions from C,G or mutations from A,T, we analyzed In a previous study, we showed that, as in wild-type mice, muta- UngϪ/ϪMsh6Ϫ/Ϫ mice along with Ungϩ/ϪMsh6Ϫ/Ϫ mice as con- tions in UngϪ/Ϫ mice start 100–200 bp downstream of the Ig trols (Fig. 1). Only 2% of the mutations in the UngϪ/ϪMsh6Ϫ/Ϫ promoter (26). This finding suggested that AID did not gain access Downloaded from mice were at A or T, the rest were C,G mutations, all of them being to the very 5Ј end of Ig genes, because, if it did, in the absence of CϾT and GϾA transitions. Thus, the findings in the UngϪ/Ϫ Ung one would have expected an AID “footprint” of C:G to T:A Msh6Ϫ/Ϫ are very similar to the mutation profile in the UngϪ/Ϫ transitions (26). However, the possibility remained that C,G tran- Msh2Ϫ/Ϫ mice, indicating that the Msh2/6 complex, but not the sition mutations were not seen because mismatch repair recog- Msh2/Msh3 complex, is required for generating high levels of A,T nized potential AID-created uracils and resulted in faithful, error- mutations. A less severe reduction of A,T mutations was seen in free restitutions of cytosines. To address this question, we Msh6Ϫ/Ϫ mice with wild-type Ung supporting the notion that some analyzed mutations in the 5Ј end of Ig ␭ genes in the UngϪ/Ϫ http://www.jimmunol.org/ mutations at A and T can arise by Ung-dependent BER without the Msh6Ϫ/Ϫ mice. Of 57 mutations in the Ig ␭ L chain gene from 61 help of MMR (but see Discussion) (Fig. 1). clones, no mutations occurred in the first 97 bp and full mutation activity started only ϳ200 bp downstream of the promoter (Fig. 3). Ϫ/Ϫ Ϫ/Ϫ CSR is diminished in Ung Msh6 mice This distribution of mutations is the same as in wild-type or Ϫ Ϫ To ascertain whether CSR was impaired in the UngϪ/ϪMsh6Ϫ/Ϫ Ung / mice (26). Because BER and MMR, the major apparent mice, we conducted ELISA analyses of the sera of SRBC-immu- mechanisms responsible for posturacil SHM events, were absent in nized mice. Data in Fig. 2 are based on the pool of two sets of these mice, we conclude that AID, indeed, is unable to access the Ј ELISA data from two groups of mice (each group has a wild-type, 5 region of Ig genes in vivo. by guest on October 2, 2021 an UngϪ/ϪMsh6ϩ/ϩ,anUngϩ/ϪMsh6Ϫ/Ϫ, and an UngϪ/Ϫ Ј Msh6Ϫ/Ϫ mouse). Compared with the wild-type mice, UngϪ/Ϫ AID activity declines significantly within a 1.1-kb sequence 3 ϩ ϩ Msh6 / mice had significantly lower concentrations of switched of JH4 Igs (IgG1, IgG2a, IgG3 and IgA; p Ͻ 0.0002), except for IgG2b Mutations are also absent in the 3Ј end of Ig genes, both in wild- ( p Ͻ 0.1376), indicating that CSR in the UngϪ/ϪMsh6ϩ/ϩ mice type and UngϪ/Ϫ mice, suggesting that AID does not act beyond was significantly impaired (Fig. 2). In UngϪ/ϪMsh6Ϫ/Ϫ mice se- ϳ1.5–2 kb from the promoter (26). To assess the possibility that rum IgG and IgA were essentially absent. Compared with wild- high fidelity mismatch repair may eliminate uracils in this region, type mice, in the UngϪ/ϪMsh6Ϫ/Ϫ double-knockout mice the we compared the mutation distributions in the 1.1-kb sequence 3Ј IgG1, 2a, 2b, 3, and IgA levels were 12–120 times lower ( p Ͻ 0.0002) (Fig. 2), indicating that CSR is greatly diminished. CSR in the Ungϩ/ϪMsh6Ϫ/Ϫ mice, however, is not severely impaired, ex-

FIGURE 2. CSR activity in wild-type and different knockout mice. The FIGURE 3. Distribution of mutations in the 5Ј region of the ␭1 L chain data were combined from two mice. p, Wild type; s, UngϪ/ϪMsh6ϩ/ϩ; z, gene in UngϪ/ϪMsh6Ϫ/Ϫ mice. The bent arrow indicates the first tran- Ungϩ/ϪMsh6Ϫ/Ϫ; Ⅺ, UngϪ/ϪMsh6Ϫ/Ϫ. The statistical significance is in- scribed nucleotide. Numbers on the x-axis show nucleotide positions up- dicated directly above a column (when less than wild-type mice) or higher stream (Ϫ) and downstream of the start of transcription. The y-axis shows Value of p Յ the total numbers of mutations in 61 clones, of which 14 had mutations. All ,ء .(up above a columns (when more than wild-type mice 0.0002 compared with wild-type. #, Value of p Յ 0.0002 compared with mutations were transitions from C and G, except a deletion of C285 in UngϪ/Ϫ (except for third bar in IgG1 where p ϭ Յ0.003); ϩ, value of p clone 4A; the mutation at Ϫ348 was also in clone 4A. At the bottom is Յ0.0002 compared with Msh6Ϫ/Ϫ. shown a corresponding map of the V␭1 gene. 5390 SHM AND CSR IN Msh6Ϫ/ϪUngϪ/Ϫ DOUBLE-KNOCKOUT MICE

Ϫ/Ϫ Ϫ/Ϫ ␣ of JH4. In both the Ung Msh6 and wild-type mice, the high- mismatch by the MutS- complex, Msh2/Msh6, that is responsible est density of mutations was at the 5Ј end of the JC intron; more for recognizing single-base mismatches and loops, and not the 3Ј the density of mutations was gradually reduced so that muta- MutS-␤ complex, Msh2/Msh3, that recognizes larger mismatches Ј Ј tions at the 3 end of the 1.1-kb region 3 of JH4 were scarce, both and loops. The fact that almost all mutations at C and G are tran- Ϫ Ϫ Ϫ Ϫ in Msh6 / Ung / and wild-type mice (Fig. 4, C and D). Two sitions suggests that they arise by direct replication of the AID factors might contribute to the mutation distribution in this 1.1-kb created uracils and that, besides BER and MMR, no other repair DNA sequence: the distribution of cytosines and the occurrence of mechanism may recognize and process the uracils. It had been the WRC/GYW hotspots. Comparing the 5Ј end to the 3Ј end of shown, by us and others, that nucleotide excision repair deficiency the 1.1-kb DNA fragment, the frequency of cytosines is similar does not alter SHM (28–31). The findings here confirm this. Some (Fig. 4A), while the frequency of WRC/GYW hotspots is actually mutations at A and T still arise in the double-knockout mice, ϳ2% slightly higher 3Ј (Fig. 4B), indicating that the distribution of mu- in the current study (Fig. 1). This is unlikely due to MutS-␤ (Msh2/ tations is independent of either the distribution of cytosines or the Msh3) activity, because a similar proportion is seen in Msh2/Ung occurrence of AID hotspots. Thus, increasing scarcity of AID- double-knockout mice (23). Thus, perhaps other uracil glycosy- induced appears responsible for the waning mu- lases can get involved (32) and, in rare occasions, their action tagenesis toward the 3Ј end, suggesting that, indeed, AID does not during SHM may attract error-prone polymerases. Ј access or does not act at the 3 end of Ig genes. The mutation frequencies in mutated DNA clones are the same in Msh6Ϫ/ϪUngϪ/Ϫ mice as in wild-type mice (Table I). In con- Discussion trast, in Msh6Ϫ/Ϫ mice with wild-type Ung, the mutation frequen- Compared with Msh2Ϫ/ϪUngϪ/Ϫ mice, SHM and CSR are af-

cies are reduced (bold in Table I). Reduction in mutation frequen- Downloaded from Ϫ/Ϫ Ϫ/Ϫ fected in the same way in Msh6 Ung mice. This suggests cies has also been observed in mice with Msh2 deficiency (10, 11) that MMR in these processes is induced by recognition of the U/G but not in mice with Msh2/Ung double deficiency (23). Rada et al. (23) suggested that, in the presence of Ung, CSR was initiated but when MutS-␣ was absent, CSR was not completed and thus the cells may have died, while in the absence of Ung, CSR would not

be initiated. There may also be an interesting alternative explana- http://www.jimmunol.org/ tion, because CSR is not severely compromised in Msh6Ϫ/Ϫ mice. Perhaps, in the presence of Ung but absence of mismatch repair, many of the uracils created by AID may be repaired to cytosine by the normal base excision repair with polymerase ␤, thus reducing mutations. CSR is greatly curtailed in the double-knockout mice, however, it is not eliminated (Fig. 2). There clearly are switched Igs for IgG1, IgG2b, and IgA. Perhaps IgG2a and IgG3 were also present at levels not detectable, because these Ig isotypes are also rare in by guest on October 2, 2021 wild-type mice. Presumably, switch recombination in these cases results from spontaneous single-strand breaks near each other, one on each DNA strand, in S␮ and a downstream S region. If this were the case, it would suggest that in normal CSR the recruitment of nonhomologous end joining complexes and the synapsing of S␮ and another S region would not require participation of AID. In fact, in the absence of AID some IgG1 and IgG2a was observed in AIDϪ/Ϫ mice (33). UngϪ/Ϫ mice had been shown before to retain a high frequency of mutations at A and T (5, 26). This was interpreted to indicate that Ung and consequently base excision repair are not responsible for A,T mutations. Inspecting A,T mutations in the different stud- ies, however, shows that mutations at A and T are consistently lower in UngϪ/Ϫ mice than in wild-type mice, being reduced by 11–52% of the frequencies in wild-type littermates (Table II). This suggests that base excision repair likely does play a role in A,T mutations. Based on the data in Table II, potentially, up to 52% of A,T mutations arise by an error-prone long patch excision during BER initiated by Ung. Outside of SHM, such long patch BER is relatively error-free due to copying of the G-containing strand by pol ␤ or a replicative polymerase, ␦ or ␧ (24). In SHM, error-prone Ϫ Ϫ FIGURE 4. Distribution of mutations in wild-type and Msh6 / polymerases apparently become involved. The proportions of A,T Ϫ Ϫ / Ј Ϫ/Ϫ Ung mice in the 1.1-kb JC intron region 3 of JH4. A, The distribution mutations in Ung vs wild-type mice vary greatly in the differ- of cytosines. Bars above the horizontal axis represent Cs on both DNA ent DNA sequences analyzed (Table II). This seems unrelated to strands. B, The occurrence of WRC/GYW hotspots. Bars above and below the proportion of AID hotspots as would be expected if it were due the horizontal axis are hotspots in the top and bottom DNA strands, re- spectively. C, The mutation distribution in the wild-type mice. D, The to factors acting after AID. The increasing reductions of A,T mu- mutation distribution in the UngϪ/ϪMsh6Ϫ/Ϫ mice. E, Map of H chain tations are slightly parallel with increasing proportions of known gene. Bent arrow, start of transcription; VDJ, V region; C, C region; dotted hotspots of the error-prone polymerase ␩ (WA, and its inverse lines, the sequence of interest. The map is not to scale. TW) (34). This may suggest that pol ␩ is not the major polymerase The Journal of Immunology 5391 interacting with MMR in SHM in the absence of Ung, although it 3. Storb, U., and J. Stavnezer. 2002. Immunoglobulin genes: generating diversity has been reported that pol ␩ can interact with Msh2/Msh6 (35). with AID and UNG. Curr. Biol. 12: R725–R727. 4. Longerich, S., U. Basu, F. Alt, and U. Storb. 2006. AID in somatic hypermutation In the absence of Msh6 but the presence of wild-type Ung, there and class switch recombination. Curr. Opin. Immunol. 18: 164–174. are very few mutations from A and T, although more than when 5. Rada, C., G. Williams, H. Nilsen, D. Barnes, T. Lindahl, and M. Neuberger. 2002. Immunoglobulin isotype switching is inhibited and somatic hypermutation both Ung and Msh6 are inactivated (Fig. 1; (7, 9). This scarcity of perturbed in UNG-deficient mice. Curr. Biol. 12: 1748–1755. A,T mutations in mice deficient for Msh6 alone disagrees with the 6. Martin, A., Z. Li, D. P. Lin, P. D. Bardwell, M. D. Iglesias-Ussel, W. Edelmann, conclusion that Ung is responsible for a proportion of up to 52% and M. D. Scharff. 2003. Msh2 ATPase activity is essential for somatic hyper- mutation at a-T base pairs and for efficient class switch recombination. J. Exp. of mutations from A and T. However, this finding hints at a col- Med. 198: 1171–1178. laboration between Ung and MMR (26). Ung is more active on 7. Martomo, S., W. W. Yang, and P. J. Gearhart. 2004. A role for Msh6 but not ssDNA than dsDNA (36, 37). Presumably, in normal SHM, Ung Msh3 in somatic hypermutation and class switch recombination. J. Exp. Med. 200: 61–68. acts efficiently on ssDNA created by Msh2/Msh6 induced excision 8. Phung, Q., D. Winter, A. Cranston, R. Tarone, W. Bohr, R. Fishel, and of either the U or G containing strand. However, because Ung is P. Gearhart. 1998. Increased hypermutation at G and C nucleotides in immuno- less active in removing the U in dsDNA, long patch base excision globulin variable genes from mice deficient for the MSH2 mismatch repair pro- tein. J. Exp. Med. 187: 17451–17755. Ϫ/Ϫ repair is reduced in Msh6 mice. 9. Wiesendanger, M., B. Kneitz, W. Edelmann, and M. D. Scharff. 2000. Somatic SHM is restricted to transcribed Ig genes between ϩ100 and hypermutation in MutS homologue (MSH)3-, MSH6-, and MSH3/MSH6-defi- ϳ Ј cient mice reveals a role for the MSH2-MSH6 heterodimer in modulating the 200 from the start of transcription to 2kb3 (26, 38–43). This base substitution pattern. J. Exp. Med. 191: 579–584. Ϫ/Ϫ distribution is unaltered in Ung mice (26), suggesting that AID 10. Frey, S., B. Bertocci, F. Delbos, L. Quint, J. Weill, and C. Reynaud. 1998. did not access or not operate in the spared 5Ј and 3Ј ends of Ig Mismatch repair deficiency interferes with the accumulation of mutations in chronically stimulated B cells and not with the hypermutation process. Immunity genes. It was, however, possible that AID-induced uracils were 9: 127–134. Downloaded from produced in the 5Ј and 3Ј ends, but that the U/G mismatches were 11. Rada, C., M. Ehrenstein, M. Neuberger, and C. Milstein. 1998. Hot spot focusing recognized by Msh2/Msh6 and repaired without error. The present of somatic hypermutation in MSH2-deficient mice suggests two stages of muta- Ϫ/Ϫ Ϫ/Ϫ tional targeting. Immunity 9: 135–141. study shows that in Msh6 , Ung mice the 5Ј and 3Ј regions 12. Li, Z., S. Scherer, R. D., M. Iglesias-Ussel, J. Peled, P. Bardwell, M. Zhuang, again have few or no mutations (Figs. 3 and 4). It is thus very K. Lee, A. Martin, W. Edelmann, and M. Scharff. 2004. Examination of Msh6- unlikely that AID targets these regions. This is strikingly different and Msh3-deficient mice in class switching reveals overlapping distinct roles of MutS homologues in antibody diversification. J. Exp. Med. 200: 47–59. from the result of the cell-free study that showed that AID could 13. Schrader, C. E., J. Vardo, and J. Stavnezer. 2002. Role for mismatch repair access cytidines immediately downstream of the promoter in na- proteins Msh2, Mlh1, and Pms2 in immunoglobulin class switching shown by http://www.jimmunol.org/ sequence analysis of recombination junctions. J. Exp. Med. 195: 367–373. ked DNAs (44). Therefore, the interesting questions that remain 14. Schrader, C. E., W. Edelmann, R. Kucherlapati, and J. Stavnezer. 1999. Reduced are whether in vivo AID may be associated with transcription com- isotype switching in splenic B cells from mice deficient in mismatch repair en- plexes only in the ϳϩ100 nt to ϳϩ2 kb region of Ig genes, zymes. J. Exp. Med. 190: 323–330. 15. Schrader, C. E., J. Vardo, and J. Stavnezer. 2003. Mlh1 can function in antibody whether it can only access this region even without association class switch recombination independently of Msh2. J. Exp. Med. 197: with the transcription complex, or whether it is present in com- 1377–1383. plexes with DNA in the 5Ј and 3Ј regions but cannot act. These 16. Ehrenstein, M. R., C. Rada, A. M. Jones, C. Milstein, and M. S. Neuberger. 2001. Switch junction sequences in PMS2-deficient mice reveal a microhomology-me- questions are under further study. diated mechanism of Ig class switch recombination. Proc. Natl. Acad. Sci. USA AID activity in vitro is much higher than the mutation frequen- 98: 14553–14558. by guest on October 2, 2021 cies observed in SHM in vivo. This may be due partially to lower 17. Kim, N., G. Bozek, J. C. Lo, and U. Storb. 1999. Different mismatch repair deficiencies all have the same effects on somatic hypermutation: intact primary concentrations of AID in vivo. However, SHM is tested several mechanism accompanied by secondary modifications. J. Exp. Med. 190: 21–30. days after AID induction in vivo, whereas in vitro, very high mu- 18. Kong, Q., and N. Maizels. 1999. PMS2-deficiency diminishes hypermutation of a ␭1 transgene in young but not older mice. Mol. Immunol. 36: 83–91. tation frequencies are seen within a few minutes (e.g., Ref. 45). 19. Phung, Q. H., D. B. Winter, R. Alrefai, and P. J. Gearhart. 1999. Hypermutation Because BER and MMR appear to be the major mechanisms in- in Ig V genes from mice deficient in the MLH1 mismatch repair protein. J. Im- volved in the processing of AID induced uracils in vivo, the tran- munol. 162: 3121–3124. 20. Cascalho, M., J. Wong, C. Steinberg, and M. Wabl. 1998. Mismatch repair co- sition mutations from C and G in the absence of BER and MMR opted by hypermutation. Science 279: 1207–1210. likely are a “footprint” of AID, namely represent the uracils cre- 21. Bardwell, P. D., C. J. Woo, K. Wei, Z. Li, A. Martin, S. Z. Sack, T. Parris, ated by AID. Thus, it appears that AID action might be curtailed W. Edelmann, and M. D. Scharff. 2004. Altered somatic hypermutation and re- duced class-switch recombination in exonuclease 1-mutant mice. Nat. Immunol. in vivo. An untested alternative is that another uracil glycosylase 5:224229. (such as Smug (32)) is involved in removing a high proportion of 22. Li, Z., J. U. Peled, C. Zhao, A. Svetlanov, D. Ronai, P. E. Cohen, and AID-created uracils during SHM in vivo and that this glycosylase M. D. Scharff. 2006. A role for Mlh3 in somatic hypermutation. DNA Repair 5: 675–682. prevents or does not foster the access of error-prone polymerases. 23. Rada, C., J. Di Noia, and M. Neuberger. 2004. Mismatch recognition and uracil- excision provide complementary paths to both immunoglobulin switching and the second (dA:dT-focussed) phase of somatic mutation. Mol. Cell. 16: 163–171. Acknowledgments 24. Friedberg, E., G. Walker, W. Siede, R. Wood, R. Schultz, and T. Ellenberger. We are grateful to D. Barnes and T. Lindahl for the UngϪ/Ϫ mice and 2006. DNA Repair and Mutagenesis. ASM Press, Washington, D.C. Dr. C. Rada for sequences of UngϪ/Ϫ mice (Table II). We especially thank 25. Li, Z., C. Zhao, M. D. Iglesias-Ussel, Z. Polonskaya, M. Zhuang, G. Yang, Z. Luo, W. Edelmann, and M. D. Scharff. 2006. The mismatch repair protein S. Longerich for insightful comments on this work. We also thank T. E. Msh6 influences the in vivo AID targeting to the Ig locus. Immunity 24: 393–403. Martin for critical reading of the paper. We gratefully acknowledge support 26. Longerich, S., A. Tanaka, G. Bozek, and U. Storb. 2005. The very 5Ј end and the by the University of Chicago Cancer Research Center for the DNA se- constant region of Ig genes are spared from somatic mutation because AID does quencing and Flow Cytometry Facilities. not access these regions. J. Exp. Med. 202: 1443–1454. 27. Jolly, C., N. Klix, and M. Neuberger. 1997. Rapid methods for the analysis of Ig gene hypermutation:application to transgenic and gene targeted mice. Nucleic Disclosures Acids Res. 25: 1913–1919. 28. Kim, N., K. Kage, F. Matsuda, M.-P. Lefranc, and U. Storb. 1997. B lymphocytes The authors have no financial conflict of interest. of xeroderma pigmentosum or Cockayne syndrome patients with inherited de- fects in nucleotide excision repair are fully capable of somatic hypermutation of immunoglobulin genes. J. Exp. Med. 186: 413–419. References 29. Shen, H. M., D. L. Cheo, E. Friedberg, and U. Storb. 1997. The inactivation of 1. Diaz, M., and C. Lawrence. 2005. An update on the role of translesion synthesis the XP-C gene does not affect somatic hypermutation or class switch recombi- DNA polymerases in Ig hypermutation. Trends Immunol. 26: 215–220. nation of immunoglobulin genes. Mol. Immunol. 34: 527–533. 2. Stavnezer, J., and C. E. Schrader. 2006. Mismatch repair converts AID-instigated 30. Wagner, S., J. Elvin, P. Norris, J. McGregor, and M. Neuberger. 1996. Somatic nicks to double-strand breaks for antibody class-switch recombination. Trends hypermutation of Ig genes in patients with xeroderma pigmentosum (XP-D). Int. Genet. 22: 23–28. Immunol. 8: 701–705. 5392 SHM AND CSR IN Msh6Ϫ/ϪUngϪ/Ϫ DOUBLE-KNOCKOUT MICE

31. Jacobs, H., Y. Fukita, G. T. van der Horst, J. de Boer, G. Weeda, J. Essers, 38. Rogerson, B. 1994. Mapping the upstream boundary of somatic mutations in N. de Wind, B. P. Engelward, L. Samson, S. Verbeek, et al. 1998. Hypermutation rearranged immunoglobulin transgenes and endogenous genes. Mol. Immunol. of immunoglobulin genes in B cells of DNA repair-deficient mice. 31: 83–98. J. Exp. Med. 187: 1735–1743. 39. Rada, C., A. Gonzalez-Fernandez, J. M. Jarvis, and C. Milstein. 1994. The 5Ј 32. Di Noia, J. M., C. Rada, and M. S. Neuberger. 2006. SMUG1 is able to excise boundary of somatic hypermutation in a Vk gene is in the leader intron. Eur. uracil from immunoglobulin genes: insight into mutation versus repair. EMBO J. J. Immunol. 24: 1453–1457. 25: 585–595. 33. Muramatsu, M., K. Kinoshita, S. Fagarasan, S. Yamada, Y. Shinkai, and 40. Rada, C., and C. Milstein. 2001. The intrinsic hypermutability of antibody heavy T. Honjo. 2000. Class switch recombination and somatic hypermutation require and light chain genes decays exponentially. EMBO J. 20: 4570–4576. activation-induced cytidine deaminase (AID), a potential RNA editing enzyme. 41. Wu, P., and L. Claflin. 1998. Promoter-associated displacement of hypermuta- Cell 102: 553–563. tions. Int. Immunol. 10: 1131–1138. 34. Rogozin, I., Y. Pavlov, K. Bebenek, T. Matsuda, and T. Kunkel. 2001. Somatic mutation hotspots correlate with DNA polymerase ␩ error spectrum. Nat. Immu- 42. Motoyama, N., T. Miwa, Y. Suzuki, H. Okada, and T. Azuma. 1994. Comparison nol. 2: 530–536. of somatic mutation frequency among immunoglobulin genes. J. Exp. Med. 179: 35. Wilson, T. M., A. Vaisman, S. A. Martomo, P. Sullivan, L. Lan, F. Hanaoka, 395–403. A. Yasui, R. Woodgate, and P. J. Gearhart. 2005. MSH2-MSH6 stimulates DNA 43. Lebecque, S., and P. Gearhart. 1990. Boundaries of somatic mutation in rear- polymerase ␩, suggesting a role for A:T mutations in antibody genes. J. Exp. ranged immunoglobulin genes: 5Ј boundary is near the promoter, and 3Ј bound- Med. 201: 6337–6645. ary is ϳ1 kb from V(D)J gene. J. Exp. Med. 172: 1717–1727. 36. Kavli, B., S. Andersen, M. Otterlei, N. Liabakk, K. Imai, A. Fischer, 44. Shen, H., S. Ratnam, and U. Storm. 2005. Targeting of the activation-induced A. Diurandy, H. Krokan, and G. Slupphaug. 2005. B cells from hyper-IgM pa- cytosine deaminase is strongly influenced by the sequence and structure of the tients carrying UNG mutations lack ability to remove uracil from ssDNA and have elevated genomic uracil. J. Exp. Med. 201: 2011. targeted DNA. Mol. Cell Biol. 25: 10815–10821. 37. Eftedal, I., P. H. Guddal, G. Slupphaug, G. Volden, and H. E. Krokan. 1993. 45. Bransteitter, R., P. Pham, P. Calabrese, and M. F. Goodman. 2004. Biochemical Consensus sequences for good and poor removal of uracil from double stranded analysis of hypermutational targeting by wild type and mutant activation-induced DNA by uracil-DNA glycosylase. Nucleic Acids Res. 21: 2095–2101. cytidine deaminase. J. Biol. Chem. 279: 51612–51621. Downloaded from http://www.jimmunol.org/ by guest on October 2, 2021