Interactions between Idd5.1/Ctla4 and Other Type 1 Diabetes Kara Hunter, Dan Rainbow, Vincent Plagnol, John A. Todd, Laurence B. Peterson and Linda S. Wicker This information is current as of September 23, 2021. J Immunol 2007; 179:8341-8349; ; doi: 10.4049/jimmunol.179.12.8341 http://www.jimmunol.org/content/179/12/8341 Downloaded from

References This article cites 32 articles, 15 of which you can access for free at: http://www.jimmunol.org/content/179/12/8341.full#ref-list-1

Why The JI? Submit online. http://www.jimmunol.org/

• 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 September 23, 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 © 2007 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Interactions between Idd5.1/Ctla4 and Other Type 1 Diabetes Genes1

Kara Hunter,* Dan Rainbow,* Vincent Plagnol,* John A. Todd,* Laurence B. Peterson,† and Linda S. Wicker2*

Two loci, Idd5.1 and Idd5.2, that determine susceptibility to type 1 diabetes (T1D) in the NOD mouse are on 1. Idd5.1 is likely accounted for by a synonymous single nucleotide polymorphism in exon 2 of Ctla4: the B10-derived T1D-resistant allele increases the expression of the ligand-independent isoform of CTLA-4 (liCTLA-4), a molecule that mediates negative signaling in T cells. Idd5.2 is probably Nramp1 (Slc11a1), which encodes a phagosomal membrane that is a metal efflux pump and is important for host defense and Ag presentation. In this study, two additional loci, Idd5.3 and Idd5.4, have been defined to 3.553 and 78 Mb regions, respectively, on linked regions of chromosome 1. The most striking findings, however, concern the evidence we have obtained for strong interactions between these four disease loci that help explain the association of human CTLA4 with Downloaded from T1D. In the presence of a susceptibility allele at Idd5.4, the CTLA-4 resistance allele causes an 80% reduction in T1D, whereas in the presence of a protective allele at Idd5.4, the effects of the resistance allele at Ctla4 are modest or, as in the case in which resistance alleles at Idd5.2 and Idd5.3 are present, completely masked. This masking of CTLA-4 alleles by different genetic backgrounds provides an explanation for our observation that the human CTLA-4 is only associated with T1D in the subgroup of human T1D patients with anti-thyroid autoimmunity. The Journal of Immunology, 2007, 179: 8341–8349. http://www.jimmunol.org/ enes termed insulin-dependent diabetes (Idd)3 control the de- (liCTLA-4) than does the NOD allele (6). The molecular basis for the velopment of type 1 diabetes (T1D) in NOD mice. The Idd5 splicing difference has been mapped to a single nucleotide polymorphism G region from diabetes-resistant C57BL/10 (B10) or C57BL/6 (SNP) in Ctla4 exon 2 that alters splicing in a similar manner to that (B6) mice provides protection from T1D when introgressed onto the mediated by a SNP in the human CD45 gene (7–9). The liCTLA-4 mol- NOD background (1–2). Idd5 is located on mouse chromosome 1 and ecule mediates negative signaling in T cells thereby predicting that its has been shown by congenic strain analysis to consist of at least two loci, higher expression in mice with the diabetes-protective B10 allele leads to Idd5.1 and Idd5.2, positioned at the proximal and distal ends, respec- reduced T cell activation and/or expansion (10). The Idd5.2 region was tively, of an ϳ15 Mb interval (3). Idd5.1 was defined as a 2.0 Mb B10- localized to a 1.5 Mb interval (3) in which Nramp1 is the most compel- by guest on September 23, 2021 derived resistance interval containing four genes including the candidate ling candidate gene because there is a known functional missense poly- genes Ctla4 and Icos (3, 4), a remarkable finding since human T1D is morphism (Gly169) Ͼ (Asp169) distinguishing the NOD and B10 associated with CTLA4 (5, 6). In addition to the human T1D association Nramp1 alleles. The NOD NRAMP1 protein is wild type and mediates with CTLA4, functional studies support the candidacy of Ctla4 as the protection from certain infectious diseases by contributing to the rapid diabetes gene in the Idd5.1 interval because the B10 allele of Ctla4 pro- acidification of the lysosome whereas the diabetes-resistant B10 duces more of the “ligand-independent” splice form of CTLA-4 NRAMP1 allotype is not functional (11). The likelihood of Nramp1 be- ing Idd5.2 is very high because a knockdown of the gene mimics the biological effect of the natural knockout (12). *Juvenile Diabetes Research Foundation/Wellcome Trust Diabetes and Inflammation Laboratory, Department of Medical Genetics, Cambridge Institute for Medical Re- The current study was initiated to test the hypothesis that Idd5.1/Ctla4 search, University of Cambridge, Cambridge, United Kingdom; and †Department of and Idd5.2/Nramp1 alone are sufficient to account for the diabetes pro- Pharmacology, Merck Research Laboratories, Rahway, NJ 07065 tection originally defined by the larger Idd5 locus containing both Idd5.1/ Received for publication September 11, 2006. Accepted for publication August 23, 2007. Ctla4 and Idd5.2. The unexpectedly high frequency of diabetes observed when resistance alleles at Idd5.1/Ctla4 and Idd5.2/Nramp1 were com- 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 bined revealed the existence of a third locus, Idd5.3, located between with 18 U.S.C. Section 1734 solely to indicate this fact. Idd5.1/Ctla4 and Idd5.2/Nramp1. This locus has been verified by the 1 This study was supported by National Institutes of Health Grant NIH P01 AI039671. development of Idd5.3 congenic strains. In addition, the previously re- The availability of NOD congenic mice through the Taconic Farms Emerging Models Program was supported by grants from the Merck Genome Research Institute, Na- ported strong interaction of the Idd5 and Idd3 protective alleles causing tional Institute of Allergy and Infectious Diseases, and the Juvenile Diabetes Research nearly complete protection from diabetes and insulitis (1, 13) was shown Foundation. L.S.W. and J.A.T. were supported by grants from the Juvenile Diabetes to require resistance alleles at Idd5.2 and Idd5.3, because Idd3 and Idd5.1/ Research Foundation and the Wellcome Trust, and L.S.W. was a Juvenile Diabetes Research Foundation/Wellcome Trust Principal Research Fellow. Ctla4 did not recapitulate the protection observed in mice having resis- 2 Address correspondence and reprint requests to Dr. Linda S. Wicker, Juvenile Di- tance alleles at both Idd3 and Idd5. abetes Research Foundation/Wellcome Trust Diabetes and Inflammation Laboratory, Another novel bicongenic strain consisting of Idd5.2/Nramp1 Department of Medical Genetics, Cambridge Institute for Medical Research, Well- come Trust/Medical Research Council Building, Addenbrooke’s Hospital, Cam- and Idd5.3 had an unexpectedly low frequency of diabetes, leading bridge, U.K. E-mail address: [email protected] to the discovery of a fourth Idd region on chromosome 1, Idd5.4. 3 Abbreviations used in this paper: Idd, insulin dependent diabetes; T1D, type 1 di- In the case of Idd5.4, it is the B10 allele, rather than the NOD abetes; liCTLA-4, “ligand-independent” splice form of CTLA-4; SNP, single nucle- allele, that confers susceptibility to T1D. We demonstrate that an otide polymorphism; AITD, autoimmune thyroid disease. interaction between the B10 Idd5.1/Ctla4 resistance allele and the Copyright © 2007 by The American Association of Immunologists, Inc. 0022-1767/07/$2.00 B10 Idd5.4 susceptibility allele exists and that the B10 Idd5.1/ www.jimmunol.org 8342 GENE INTERACTIONS IN TYPE 1 DIABETES

FIGURE 1. Genetic intervals pres- ent in the Idd5 congenic strains referred to in this study. Filled regions are B10- Downloaded from derived or B6-derived segments of DNA for 1 and 3, respec- tively, defined by the most centromeric and telomeric non-NOD allelic mark- ers. Open regions represent the region between the last non-NOD allelic marker and the first NOD allelic marker http://www.jimmunol.org/ at each boundary. Lines represent NOD-derived DNA. Vertical arrows designate the Idd5.1, Idd5.2, Idd5.3, and Idd5.4 regions. The Idd5.4 interval contains a segment that is NOD-derived due to a double recombination event (initially undetected) that occurred dur- ing the development of the R8 strain.

The diagram is to scale. by guest on September 23, 2021

Ctla4 allele can, when present in different genetic contexts, provide ferred to are summarized in Fig. 1. An asterisk in Fig. 1 indicates the new potent, moderate, or undetectable protection from diabetes. Our study NOD.B10 chromosome 1 congenic lines developed specifically for the also has implications for the ongoing search and characterization of current study. The origins of all of the Idd5 strains in Fig. 1 can be traced to lines R8 or R2. R8 and R2 were used to develop the R974 and R444 lines human T1D susceptibility genes, one of which is CTLA4 (5, 6). and the R2s line, respectively, by backcrossing to the NOD parental strain and identifying recombination events in progeny that were subsequently Materials and Methods selectively bred and selected to be homozygous for the congenic interval. Congenic mouse strains and assessment of diabetes, insulitis, All congenic strains have been backcrossed to NOD 10 to 20 times. The and insulin autoantibodies centromeric segment of R2s was originally produced as a congenic strain from R8 to localize Idd5.1 and it was shown to have a frequency identical The breeding and genotyping strategies for the development of the con- to the NOD strain (L. Wicker, L. Peterson, unpublished observations). genic mouse strains protected from T1D have been reviewed (14). The To perform a direct comparison with R444, the distal segment of R2s was current study extends the fine mapping of the Idd5 region that has been developed from the R2 strain and combined with the centromeric segment previously published (1, 3) and, for clarity, all of the strains used or re- described above and is shown in Fig. 1. Three lines were subsequently The Journal of Immunology 8343

Table I. Novel markers developed for this study

Restriction Enzyme Marker 5Ј Primer 3Ј Primer or Allele Size

Bq460g02 CTGACCTCAAGGCCCTAACA CTGCCTTGTCTGGCTTCAGT Mnl1 RS30280887 CAATAGGTTCACCCTTCACTT TGCCCTTTCACACTTACATTT Tsp45I RS30321818 AGGGCTGCCATAACAAGATG GCAGTATGGCTTTGCTGACA HpyCH4iv RS30471359 GGCTGTGACTGTGCTGAGAA TGCTGACCAAACACACACTG BFA1 Bq411m14-T7 GGTGATTGATGCTGGAGGAC CTGATGGAGAGGCCATGAAG 200/196

developed from R2s: lines 3700, 6359, and 6360. The remaining Idd5 on agarose gels. Primer sequences for novel markers used in this study are congenic strains in Fig. 1 were developed from R444. The NOD.B10 shown in Table I. Idd5R426ϩR52 (R426ϩR52) bi-congenic strain was developed by inter- crossing the R426 and R52 strains, backcrossing the resulting F1 mice to Gene identification the NOD/MrkTac (Taconic Farms) parental strain and selecting for mice heterozygous at both regions, indicating that a recombination event had Inspection of mouse EnsEMBL and the orthologous region in human Idd5.3 occurred between the two Idd intervals. Bicongenic heterozygous mice EnsEMBL was used to determine the gene content of . were intercrossed and homozygous mice selected as founders of the new strain. Results

To confirm the purity of the genetic background of the congenic strains, Discovery of a novel Idd locus on chromosome 1, Idd5.3 Downloaded from DNA from lines R2, R974, R444 and R2s (R8 DNA was not available), was tested by genotyping using a 5K mouse SNP chip and was performed From the analysis of congenic strains of mice, the Idd5 interval on by ParAllele Biosciences. For all of the strains tested, no non-NOD SNPs chromosome 1 consists of two loci, Idd5.1 and Idd5.2 (1, 3). As sum- were identified outside of the defined congenic regions. marized in Fig. 1, the centromeric boundary of Idd5.1 is defined by Three congenic strains having disease resistant Idd alleles on two chro- the R974 strain and is between the markers D1Mit249 and mosomes, Idd3 on chromosome 3, and different portions of Idd5 on chro- mosome 1, were used in the current study. The previously published line Cd28distal3. The distal Idd5.1 boundary is between AL671560TATG 1591 has the R444 and Idd3 genetic segments, line 1573 has the R467 Idd5 and AL671560GA, which is defined by the R46 strain (3). The gene http://www.jimmunol.org/ segment together with Idd3, and line 2402 has the R46 segment together for CTLA-4 is encoded within the 2.0 Mb Idd5.1 region at 61.3 Mb; with Idd3. The Idd3 congenic strain used to make lines 1591, 1573, and it has a sequence variation between the NOD and B10 alleles that 2402 is NOD.B6 Idd3 R450 (15). The NOD.B6 Idd3 R450 was also shown to be free of non-NOD SNPs outside of the congenic region using the 5K alters production of the liCTLA-4 splice form and, consequently, the mouse SNP chip. The R974, R46, R193, R2, R2s, R444, R467ϩIdd3, and level of negative signaling in T cells (3, 6, 10). Thus, Ctla4 is the Idd3 congenic strains are available from Taconic Farms through the prime candidate gene for Idd5.1. The Idd5.2 region has its centro- Emerging Models Program (Lines 974, 2193, 2574, 1092, 1595, 1094, meric and distal boundaries at ϳ75 Mb on chromosome 1 defined by 1573, and 1098, respectively). Line 1591 is available as line 6109. Lines the R193 and R444s strains, respectively (Fig. 1). Nramp1, which is 6360 and 6359 are also available. within the 1.5 Mb Idd5.2 region, is functionally polymorphic between Elevated urinary glucose was detected using Diastix (Miles). Animals were by guest on September 23, 2021 considered diabetic when urinary glucose was over 500 mg/dl. Diabetic mice NOD and B10 mice, and knockdown experiments strongly support also exhibited polydipsia, polyuria, and weight loss. The assessment of insu- the hypothesis that Nramp1 is Idd5.2 (12). litis has been described previously (13). In brief, each pancreas received a To test the hypothesis that the Idd5.1/Ctla4 and Idd5.2/Nramp1 single score for the degree of insulitis observed: 0 no mononuclear cell infil- trates in the islets, 1 mild insulitis, Ͻ20% of the islets have infiltrates; 2 mod- regions as defined above account for the protection conferred by erate insulitis, 20–60% of the islets have infiltrates; 3 severe insulitis, most strains such as R444 or R444s, in which the region between islets (Ͼ60%) are infiltrated; and 4 extensive insulitis, nearly all islets are Idd5.1/Ctla4 and Idd5.2/Nramp1 is also derived from B10, the R52 either completely infiltrated or appear as a residual islet. All procedures were (Idd5.2/Nramp1 from B10) and R426ϩR52 (Idd5.1/Ctla4 com- conducted according to approved protocols of the Institutional Animal Care bined with Idd5.2/Nramp1 from B10) congenic strains were de- and Use Committee of Merck Research Laboratories. veloped (Fig. 1). The frequency of T1D in these two novel strains Statistical analyses was compared with the NOD parental strain as well as the R426 Diabetes frequencies were analyzed with survival curves generated with (Idd5.1/Ctla4 from B10) and R444s (Idd5.1/Ctla4 and Idd5.2/ the GraphPad Prism 4 (GraphPad Software) software package. Survival Nramp1 and the region between from B10) congenic strains (Figs. curves were created using the product limit method of Kaplan and Meier and strain comparisons were performed using the log rank test. The dif- ϫ ference in the degree of insulitis between strains was analyzed by 2 3 100 contingency tables created using GraphPad Prism 4. Comparisons between strains were performed using ␹-square test. The two-tailed unpaired t test 90 5.1 (GraphPad Prism 4) was used to compare “mean age of diabetes onset” 80 Idd5.1/5.3/5.2 R52 R426 R426+R52 NOD between strains. R444s 70 Idd5.2 2E -02 Identification of new microsatellite markers and genotyping 60 5.3 Idd5.1/5.2 NS 50 Idd5.1 NS Mice were initially genotyped by PCR using primers to the previously -09 40 5.2 1.7E

published microsatellite markers. To map recombination points more pre- diabetic % not 30 4.6E -07 cisely, additional microsatellites and SNPs were identified and character- NOD ized as described previously (3). PCR were optimized using B10 and NOD 20 DNA templates. Markers that were polymorphic between these two strains 10 N = 53 67 66 82 100 = P valu es were used in further genotyping. All PCR were performed using Amplitaq 0 Gold with buffer II (Applied Biosystems). When PCR product sizes were 0 50 100 150 200 250 not distinguishable between B10 and NOD samples by 4% agarose gel Time in days electrophoresis, genotyping was performed using an ABI Prism 3100 Ge- netic analyzer (Applied Biosystems). All primers were ordered from Sigma- FIGURE 2. Discovery of Idd5.3. Frequency of diabetes in female Idd5 Genosys. Forward primers were labeled at the 5Ј end with the fluorescent congenic mice reveals the existence of a third Idd5 subregion, Idd5.3. The dye HEX for analysis on the ABI genetic analyzer. SNPs were analyzed p values were obtained by comparing the survival curves of the two indi- following restriction enzyme digestion of the PCR product and visualized cated strains using the log-rank test. NS, Not significant. 8344 GENE INTERACTIONS IN TYPE 1 DIABETES

Ϫ 100 (R426 vs NOD, p ϭ 4.6 ϫ10 7). A similar level of protection was 90 observed when the Idd5.2/Nramp1 region alone was derived from 6360 3700 6359 NOD Ϫ9 80 B10, as in the R52 strain (R52 vs NOD, p ϭ 1.7 ϫ 10 ). Sur- 70 prisingly, when the Idd5.1/Ctla4 and Idd5.2/Nramp1 regions were ϩ 60 5.1 combined in the bicongenic strain, R426 R52, no additional pro- ϭ 50 tection was observed above that with either region alone ( p 0.47 ϭ 40 5.3 vs R52, p 0.24 vs R426). The frequency of diabetes observed in

% not diabetic % not the R426ϩR52 strain was also higher than that in the R444s strain 30 5.2 NS ( p ϭ 0.02). Furthermore, the delay of diabetes observed in R444s 20 1.43E-04 NS 10 as compared with NOD (mean days of onset for R444s and NOD N = 75 55 60 62 =P values are 179 Ϯ 6 and 130 Ϯ 3, respectively, p ϭ 1.4 ϫ 10Ϫ8, t test) was 0 0 25 50 75 100 125 150 175 200 225 also present when R444s was compared with R426 (Idd5.1/Ctla4), Time in days R52 (Idd5.2/Nramp1), and R426ϩR52 (mean day of onset for FIGURE 3. Fine mapping of Idd5.3. Frequency of diabetes in females R426, R52, and R426ϩR52, are 155 Ϯ 5, 143 Ϯ 7, and 146 Ϯ 6, from lines 3700, 6360, and 6359 congenic mice are compared with NOD respectively, p values vs R444s are all Յ8.0 ϫ10Ϫ4, t test). We, females. Line 6359 is not significantly protected from diabetes compared therefore, hypothesize that a B10-derived resistance allele is lo- with the NOD strain. Lines 3700 and 6360 have indistinguishable diabetes cated in a region designated in this study as Idd5.3, which is de- frequencies and are protected from diabetes compared with NOD females Ϫ Ϫ fined by the difference between the distal boundary of the R426 (p values 1.08 ϫ 10 2 and 1.2 ϫ 10 3, respectively) and are also signif- Downloaded from icantly protected from diabetes compared with line 6359 (p values 2.54 ϫ congenic strain and the proximal boundary of the R52 strain, a 10Ϫ2 and 4.1 ϫ 10Ϫ3, respectively). A 1.43 ϫ 10Ϫ4 p value is obtained 7.27 Mb interval. when a meta analysis is done after combining the frequency data of the To verify the existence of Idd5.3 and to further map the region, two strains having resistant alleles at Idd5.3 (lines 3700 and 6360) and the we developed additional congenic strains: lines 6359, 6360, and two strains having susceptible alleles at Idd5.3 (line 6359 and the NOD 3700 (Fig. 1). The frequency of diabetes was assessed in female parental strain). The p values were obtained by comparing the survival mice from lines 3700, 6359, and 6360 and compared with that of http://www.jimmunol.org/ curves of the indicated strains using the log-rank test. NOD mice, which revealed that lines 3700 and 6360 had a significant protection from diabetes ( p ϭ 1.08 ϫ 10Ϫ2 and p ϭ 1.2 ϫ 10Ϫ3 1 and 2). As reported previously (1, 3), the presence of the B10 respectively). Line 6359 had a diabetes frequency indistinguishable resistance allele at Idd5.1/Ctla4 alone confers modest but highly from NOD ( p ϭ 5.57 ϫ 10Ϫ1) (Fig. 3). These new congenic lines significant protection from T1D as compared with NOD mice now define Idd5.3 to a 3.553 Mb region, with the proximal boundary

Mb Human Mouse Chromosome 1 Marker Mb Line Line Line

3700 6360 6359 by guest on September 23, 2021 207.5 KLF7 63.5 208 CREB1 FAM119A D1Mit5 D1Mit300 CCNYL CBO31 FZD5 Klf7 64 208.5 Q7Z4Q0 NP_001073944

CRYGD CRYGC CRYGB Creb1 2310038h17Rik Bq411m14-T7 64.5 Fzd5 9430067k14Rik 209 CRYGA LOC389073 IDH1 PIP5K3 PTHR2 Rpl10a 4921521f21Rik Cryge Crygd Crygc Crygb 65 Cryga D630023f18Rik 209.5 Idh1 Pip5k3 Pthr2 FIGURE 4. Genes in the Idd5.3 re- 65.5 gion. From the T1D frequencies ob- 210 MAP2 Crygf served in Fig. 3, the proximal and dis- 66 tal boundaries of the Idd5.3 region are 210.5 C2ORF21 Q96SS0 Mtap2 RPE NP_689732 defined by strain 6359 and 6360, re- C030018g12Rik RS30280887 ACADL Q9Y348 66.5 BC042720 Rpe spectively. Genes present in the or- 211 MYL1 LANCL1 RS30321818 1110028c15Rik CPS1 thologous human genomic region are Acadl Myl1 Lancl1 67 also indicated. Open regions represent 211.5 Cps1 the region between the last non-NOD allelic marker and the first NOD al- 67.5 212 lelic marker at each boundary. Lines represent NOD-derived DNA. 68 Idd5.3 212.5 ERBB4 = 3.553 Mb

Erbb4 68.5 213

69 213.5 IKZF2 Bq460g02

Ikzf2 69.5 214

RS30471359 SPAG16 70 214.5 Spag16

215 70.5 The Journal of Immunology 8345 defined by line 6359, and the distal boundary defined by lines 3700 A and 6360 (Fig. 4). 100.0 NS 1591 1573 2402 1098 Gene content of Idd5.3 97.5 5.1 As detailed in Fig. 4, there are 11 genes annotated within the 3.553 95.0 5.3 Mb region of Idd5.3, including the functional candidate Ikzf2, Chr -03 2.6E-05 1 3.1E which encodes the Ikaros family zinc finger 2 protein, a T cell 92.5 5.2 restricted transcription factor (16). By virtue of its differential ex- pression in CD4 T cells following activation (17), Acadl, which % not diabetic 90.0 P encodes long chain acyl CoA dehydrogenase (18), is also a can- = values 87.5 N = 176105 91 64 NS didate gene for Idd5.3. Chr 3 85.0 3 Idd5.3 is essential for the nearly complete protection from T1D 0 50 100 150 200 250 caused by resistance alleles at Idd5 and Idd3 Time in days B The combination of protective alleles at Idd5 and Idd3 provides 90 1591 N=40 -05 80 -11 1573 N=40 1.3E nearly complete protection from diabetes, insulitis, and the occur- 1.6E - 2402 N=47 1.2E 03 y rence of insulin autoantibodies (1, 13). To assess the individual 70 r o NOD N=10 g e contributions of the Idd5 subregions to the interaction of Idd3 and t 60 a P c = values Downloaded from s Idd5 (19), two congenic strains combining Idd3 with various por- i h

50 t n tions of the Idd5 region were developed, lines 2402 and 1573 (Fig. i e

40 c i % of mice % of

5A). Line 2402 has its Idd5 interval from R46 (B10 alleles at m

30 D

Idd5.1/Ctla4) and line 1573 has its Idd5 interval from R467 (B10 O N

20 o

alleles at Idd5.1/Ctla4 and Idd5.3). Both lines 2402 and 1573 were N =

Idd3 10 o selected to have resistance alleles at , which were derived r e Z from line 1098. The diabetes frequencies of these two novel strains 0 http://www.jimmunol.org/ none mild insulitis >mild were determined along with those of the parental NOD strain (data Degree of insulitis not shown for clarity in the figure), line 1098 (Idd3 resistance alleles only), and line 1591 (13), a Idd3/5 congenic strain having FIGURE 5. Fine mapping of Idd5.3 by virtue of its ability to increase protection from diabetes in conjunction with protective alleles at Idd3. A, its Idd5 interval derived from the R444 strain and, therefore, hav- The frequency of diabetes was monitored in strains having resistance al- ing resistance alleles at Idd5.1/Ctla4, Idd5.2/Nramp1, and Idd5.3. leles at Idd3 and various subregions of Idd5. p values were obtained by ϭ ϫ Ϫ1 The results (Fig. 5A) showed equivalent ( p 5.9 10 , 1591 vs comparing the survival curves of the two indicated strains using the log- 1573), nearly complete, protection from T1D in lines 1591 ( p ϭ rank test. Note that the y-axis is truncated due to the low frequency of T1D Ϫ5 Ϫ3 2.6 ϫ 10 vs 1098) and 1573 ( p ϭ 3.1 ϫ 10 vs 1098); how- in the strains examined. B, Degree of insulitis in strains having combina- ever, line 2402 was not protected from T1D as completely as lines tions of Idd5 and Idd3 protective alleles. N ϭ Number of pancreata ex- by guest on September 23, 2021 1591 and 1573 because it had a disease frequency equivalent to amined. Each pancreas was examined using two noncontiguous longitudi- line 1098 ( p ϭ 0.76, 2402 vs 1098). These results not only confirm nal sections. Data from strains 1591, 1573, and 2402 are from 7-mo-old the existence of Idd5.3 (line 1573 is more protected from T1D than females. As a positive insulitis control, 3-mo-old NOD males were in- ϫ line 2402) but also show that a resistance allele at Idd5.3 is es- cluded. The p values were obtained using a 2 3 contingency table com- paring the indicated strains. sential for the nearly complete protection from diabetes observed when resistance alleles at Idd5 and Idd3 are both present. It is also notable that resistance alleles at Idd5.1/Ctla4 (R46) and Idd3 to- Idd5.2/Nramp1 resistance alleles were not required for the inter- gether did not increase protection from diabetes as compared with action with Idd3 when diabetes was the phenotype assessed (Fig. Idd3 resistance alleles alone. In contrast to Idd5.3, resistance al- 5A) but are required to provide the genetic interaction causing a leles at Idd5.2/Nramp1 are not required for Idd3/5-mediated dis- marked reduction in insulitis. ease protection because lines 1573 and 1591 were equally pro- tected from T1D. Although resistance alleles at Idd5.1/Ctla4 are B10-derived susceptibility allele distal to Idd5.2/Nramp1 on not sufficient for the interaction with resistance alleles at Idd3, chromosome 1 further analysis of the isolated Idd5.3 congenic strain, line 6360, in Previously, we had observed that the R2 strain, which has B10 conjunction with Idd3 will be needed to determine whether the alleles at Idd5.3 and Idd5.2/Nramp1 (Fig. 1), was not protected Idd5.3 locus is sufficient on its own to mediate the interaction. It is from T1D (1). This is at odds with the protection observed in this possible that resistance alleles at both Idd5.1/Ctla4 and Idd5.3 are study for the R52 strain, which has a resistance allele at Idd5.2/ essential, in combination, for the genetic interaction with resis- Nramp1 alone (Figs. 1 and 2). Because the R52 interval is com- tance alleles at Idd3 to occur. pletely contained within the boundaries of the R2 strain, this result Similar to the diabetes phenotype, we have previously shown suggested that a B10-derived susceptibility allele present in the R2 that as compared with mice having resistance alleles at Idd3 or strain masks the effects of the protective Idd5.2 allele defined by Idd5 alone, the frequency of insulitis is significantly reduced when the R52 strain. This hypothesis was substantiated by the analysis resistance alleles at Idd3 and Idd5 are both present (1, 13). There- (Fig. 6) of a newly derived strain, R2s, which has resistant alleles fore, we assessed the occurrence and severity of insulitis at 7 mo at Idd5.2 and Idd5.3, but like R2, has a NOD allele at Idd5.1/Ctla4 of age, or following the development of diabetes, in females from (Fig. 1). As found for the R52 congenic strain, the R2s strain is lines 1591, 1573, and 2402. In contrast to diabetes, there was a protected from T1D (Fig. 6, p ϭ 2 ϫ 10Ϫ11, R2 vs R2s). Hence, hierarchy in the occurrence and severity of insulitis with line 2402 we proposed the existence of a fourth T1D locus, distal to Idd5.2, having more mice with insulitis than 1573 ( p ϭ 0.05, Fisher’s designated Idd5.4. exact test) and 1573 having more insulitis than 1591 ( p ϭ 0.01) We sought to confirm the existence of the Idd5.4 B10-derived (Fig. 5B). This latter finding is of particular note because the T1D susceptibility allele by hypothesizing that if one dose of 8346 GENE INTERACTIONS IN TYPE 1 DIABETES

100 A 90 100 Idd5.1/5.2/5.3 80 90 5.1 70 4.5E-02 80 5.1 Idd5.3/5.2 60 5.3 70 50 5.3 R2 60 R2s R444s NOD 40 -11 50 R46 5.2 R2s versus R2 2E NOD % not diabetic % not 5.2 30 40

20 5.4 NOD diabetic not % 30 -04 3E 10 N = 72 77 66 67 = P valu es Idd5.3/5.2/5.4 20 5.4 0 10 N = 73 100 = P value 0 50 100 150 200 250 Time in days 0 0 50 100 150 200 250 FIGURE 6. Discovery of Idd5.4. The frequency of diabetes in the R2 Time in days and R2s strains reveals a B10-derived susceptibility allele at the Idd5.4 B locus. The p values were obtained by comparing the survival curves of the 100 two indicated strains using the log-rank test.

90 Downloaded from 80 5.1 the B10 susceptibility allele was detectable in the context of one 70 Idd5.1/Ctla4, NS dose each of B10-derived resistance alleles at 60 5.3 Idd5.2/Nramp1, and Idd5.3, that the frequency of diabetes in R2s R444s Idd5.1/Idd5.2/Idd5.3/Idd5.4 heterozygotes should be greater 50 NOD 40 5.2

than in Idd5.1/Idd5.2/Idd5.3 heterozygotes even though the http://www.jimmunol.org/ Idd5.1/Idd5.2/Idd5.3/Idd5.4 and Idd5.1/Idd5.2/Idd5.3 homozy- diabetic not % 30 gous strains, R974 and R444, respectively, are protected equiv- 20 alently from T1D (3). Thus, Idd5.1/Idd5.2/Idd5.3/Idd5.4 (R974 ϫ 5.4 10 N = 53 51 100 = P value NOD)F and Idd5.1/Idd5.2/Idd5.3 (R444 ϫ NOD)F heterozygous 1 1 0 mice and the three parental strains were tested for their frequencies 0 50 100 150 200 250 of T1D. As seen in Fig. 7A, the frequency of T1D was marginally Time in days ϫ ϫ higher in the (R974 NOD)F1 mice than in the (R444 NOD)F1 C 100

90 by guest on September 23, 2021 A 80 5.1 100 Idd5.1/5.3/5.2/5.4 Homo 90 70 5.3 80 60 -05 R974xNOD NOD 70 R974 R444 R444xNOD 1.8E

Idd5.1/5.3/5.2 Homo R2

50 NOD 60 5.1 5.2 Idd5.1/5.3/5.2 Het 40 50 5.3 40 diabetic not % 30 -

% not diabetic % not 02 30 5.2 4E 20 N = 50 30 100 5.4 20 Idd5.1/5.3/5.2/5.4 Het N= 50 30 70 69 67 10 = P value 10 5.4 =P valu e NOD 0 0 0 50 100 150 200 250 0 50 100 150 200 250 Time in days Time in days B FIGURE 8. Idd5.1/Ctla4 is sensitive to genetic background. The level 100 Idd5.1/5.3/5.2/5.4 Homo of protection conferred by B10 alleles at Idd5.1/Ctla4 is dependent on 90 which alleles are present at other Idd5 subregions. Examples of moderate 80 (A), small or nonexistent (B), and large effects of a protective B10 allele (C) R974xNOD NOD 70 R974 R46xR974 R46 R46xNOD at Idd5.1/Ctla4 are shown. The p values were obtained by comparing the 60 Idd5.1 Homo/ survival curves of the two indicated strains using the log-rank test. NS ϭ 5.1 5.3/5.2/5.4 Het 50 -08 not significant. 5.3 Idd5.1 Homo 8.8E 40 R46xR974 versus % not diabetic % not 30 5.2 Idd5.1 Het R974xNOD 20 Idd5.1/5.3/5.2/5.4 Het N= 50 78 73 66 69 67 5.4 NOD animals (Ⅺ vs E, p ϭ 0.04) even though the R974 and R444 10 =P value 0 parental strains have identical disease frequencies. These data are 0 50 100 150 200 250 Time in days consistent with the existence of a B10-derived susceptibility and that the effect of one dose of this allele is only partially silenced by FIGURE 7. Confirmation of a B10 susceptibility allele at Idd5.4. A, The ϫ ϫ one dose of a B10-derived resistance allele at Idd5.1/Ctla4. higher diabetes frequency in R974 NOD F1 vs R444 NOD F1 mice confirms the influence of Idd5.4. B, Homozygosity for the resistance allele at Gene dosage effects at Idd5.1/Ctla4 Ctla4/Idd5.1 counters the T1D-susceptibility mediated by the B10 allele at Idd5.4. The p values were obtained by comparing the survival curves of the The finding that one dose of a protective allele at Idd5.1/Ctla4 was two indicated strains using the log-rank test. insufficient to provide a high degree of protection from T1D in the The Journal of Immunology 8347 presence of one dose of Idd5.4 led us to test the hypothesis that two is almost completely masked when B10 alleles are present at doses of an Idd5.1/Ctla4 resistance allele could more effectively Idd5.3 and Idd5.2/Nramp1 but not Idd5.4 (Fig. 8B). counter a single dose of a susceptibility allele at Idd5.4. We, there- ϫ ϫ Interactions between Idd5.1/Ctla4 and Idd5.4 fore, compared T1D frequencies in (R974 NOD)F1 and (R46 R974)F1 mice, the latter strain having two doses of the B10 resis- Because disease resistance by the protective B10 allele at Idd5.1/ tance allele at Idd5.1/Ctla4. As seen in Fig. 7B, significantly ( p ϭ Ctla4 is mediated by increased production of the negative-signal- ϫ Ϫ8 ϫ छ 8.8 10 ) fewer (R46 R974) F1 mice ( ) developed diabetes ing molecule liCTLA-4 (10), a logical hypothesis is that the B10- ϫ Ⅺ than (R974 NOD)F1 mice ( ). These results strongly confirm derived susceptibility allele at Idd5.4 causes an event that can be the existence of Idd5.4 and support the hypothesis that resistance reversed or prevented by more negative signaling via liCTLA-4 in alleles at Idd5.1/Ctla4 can, in a dose-dependent manner, mask the one or more cell types. The activity of Idd5.1/Ctla4 is dose-de- increase of diabetes that is mediated by a B10 susceptibility allele pendent when there is one copy of the susceptibility allele at Idd5.4 at Idd5.4. (Fig. 7). This suggests that a homeostatic, quantitative interaction exists between the molecular and/or cellular events caused by Genetic control by the Idd5.1/Ctla4 region is altered in different Idd5.1/Ctla4 and Idd5.4. Thus far, we have only studied the ac- genetic contexts tivity of a B10 allele at Idd5.4 in the context of protective alleles The discovery of Idd5.4 has caused us to the reconsider the genetic at Idd5.3 and Idd5.2/Nramp1 on chromosome 1. It is possible that control mediated by the extended Idd5 region that now contains at if NOD.B10 Idd5.4 congenic mice were developed, they would least four loci: Idd5.1/Ctla4, Idd5.2/Nramp1, Idd5.3, and Idd5.4. have a higher frequency of diabetes than that of the NOD parental As shown in Fig. 8, depending on the genetic context, the protec- strain or that T1D would occur at a younger age. However, it is Downloaded from tive B10 alleles at Idd5.1/Ctla4 provide vastly different levels of also possible that Idd5.4-mediated disease susceptibility is depen- protection from T1D. When present on the NOD background dent on molecular or cellular events caused by resistance alleles at alone, the B10 allele at Idd5.1/Ctla4 provides a small, but signif- Idd5.3 and/or Idd5.2/Nramp1. It would be informative to deter- icant ( p ϭ 3 ϫ 10Ϫ4), level of protection from T1D (Fig. 8A, see mine whether susceptibility at Idd5.4 overcomes the protection also Fig. 2). In contrast, in the presence of B10 protective alleles mediated by other resistance alleles such as Idd3 or Idd9.3 (20). at Idd5.2/Nramp1 and Idd5.3 (Figs. 6 and 8B) there is a small (Fig. Especially as the identity of more Idd loci becomes known, the http://www.jimmunol.org/ 6, p ϭ 4.5 ϫ10Ϫ2, R2s vs R444s) or no (Fig. 8B, p ϭ 0.33, R2s consequences of combining particular resistance and susceptibility vs R444s) effect of the B10 resistant alleles at the Idd5.1/Ctla4 alleles should contribute to an increased understanding of the region. Similarly, the combination of resistance alleles at Idd5.1/ pleiotropic effects of the loci on T1D pathogenesis. Ctla4 and Idd5.2/Nramp1 (Fig. 2) or Idd5.1/Ctla4 and Idd5.3 (1, 3) Combining protective alleles does not always lead to increased does not reduce the frequency of diabetes below that seen with protection from T1D resistance alleles at Idd5.1/Ctla4 alone. Only when B10-derived alleles are present in the Idd5.4, Idd5.3 and Idd5.2/Nramp1 regions Another unexpected result of this study was that no, or only a is highly significant protection provided by the addition of Idd5.1/ slight, increase in diabetes resistance was observed when protec- by guest on September 23, 2021 Ctla4 resistance alleles, as seen in the comparison of the R974 and tive alleles at particular loci are combined: Idd5.1/Ctla4 and R2 strains (4) (Fig. 8C, p ϭ 1.8 ϫ 10Ϫ5 for R974 vs R2). Idd5.2/Nramp1 (Fig. 2) or Idd5.1/Ctla4 with Idd5.2/Nramp1 and Idd5.3 (Figs. 6 and 8B). We also now realize that when protective alleles at Idd5.1/Ctla4 and Idd5.3 are both present, as they un- Discussion knowingly were in several strains studied in a previous report (3), Gene interactions in type 1 diabetes no additional protection was observed above that seen with T1D- This study was initiated to determine whether the Idd5.1/Ctla4 and resistance alleles at Idd5.1/Ctla4 alone (R46, R426, R467, and Idd5.2/Nramp1 T1D resistance alleles derived from the B10 strain R193 all had the same T1D frequencies in Ref. 3). These results at the centromeric and distal regions, respectively, of the Idd5 imply that the mechanisms of protection by Idd5.1/Ctla4 are either region on chromosome 1 are sufficient to explain the disease re- redundant or masked when protective alleles are also present at sistance produced when the intervening region is also derived from either Idd5.2/Nramp1 or Idd5.3 or that the distinct mechanisms the B10 parent, such as in the R444 strain (Fig. 1) (3). Unexpect- provided by Idd5.1/Ctla4 and Idd5.2/Nramp1 or Idd5.1/Ctla4 and edly, this was found not to be the case, and because of these results Idd5.3 do not result in the increased protection observed in, for we hypothesized that additional T1D genes exist in the Idd5 re- example, the combination of T1D-resistant alleles at Idd3 and gion. We designed a series of experiments to test this hypothesis Idd5.1/Idd5.3 (Fig. 5). and defined two new Idd loci on chromosome 1, Idd5.3, which is in the intervening segment between Idd5.1/Ctla4 and Idd5.2/ Genetic control of Idd3/Idd5-mediated protection from T1D Nramp1, and Idd5.4, which is distal of Idd5.2/Nramp1. The most The discovery of Idd5.3 is particularly important when considering unexpected finding in this study was the evidence for strong and the nearly complete T1D protection provided by a combination of complex interactions between the four disease loci: protective al- resistance alleles at Idd3 and the loci on chromosome 1 (18). Re- leles at Idd5.1/Ctla4 could completely mask potent susceptibility sistance alleles at Idd5.3 and Idd5.1/Ctla4 together with Idd3 pro- alleles at Idd5.4; in the absence of protective alleles at Idd5.1/ vided as much protection from T1D as all three Idd5 loci and Idd3 Ctla4, susceptibility alleles at Idd5.4 can, in turn, completely mask (Fig. 5A), although the loss of the protective allele at Idd5.2/ the effect of protective alleles at Idd5.3 and Idd5.2/Nramp1 as seen Nramp1 was observed in the insulitis score (Fig. 5B). In contrast, in the T1D susceptible R2 strain (Fig. 6); disease protection me- protective alleles at Idd5.1/Ctla4 only in conjunction with resis- diated by B10-derived, T1D-resistant alleles at Idd5.1/Ctla4 varied tance alleles at Idd3 were not sufficient to provide this nearly com- depending on the genetic background in which they function. In plete protection. Because of these findings, we have developed an the gene combinations examined in the current study, Idd5.1/Ctla4 Idd5.3 congenic strain and are in the process of developing an is a very potent disease susceptibility gene only when the Idd5.3, Idd5.3/Idd3 double congenic line. It is also important to consider Idd5.2/Nramp1, and Idd5.4 alleles are all derived from the B10 that NOD mice with T1D-resistance alleles at Idd3 (but not T1D- genome (Figs. 7B and 8C). Indeed the effect of variation at Ctla4 resistance alleles at Idd5.1/Ctla4) have increased expression of 8348 GENE INTERACTIONS IN TYPE 1 DIABETES

CTLA-4 on the surface of activated CD4ϩ and CD8ϩ T cells as this strain more or less resistant to an immune modulation protocol compared with activated NOD T cells (20). It is possible that the that is effective in the NOD strain, a possibility that can be tested. increased negative signaling provided by higher levels of It is therefore possible that multiple strains of diabetes-susceptible liCTLA-4 in mice with T1D-resistant alleles at Idd5.1/Ctla4 (10) NOD-related mice could be developed for the purpose of mimick- is somewhat equivalent to the increase in negative signaling that ing a portion of the genetic variation present in humans, especially would be mediated by higher levels of full-length CTLA-4 caused in light of criticisms that rodent models such as the NOD are only by protective alleles at Idd3. This potential overlap in the mech- representative of one case report of human diabetes (24). Such a anism of protection provided by T1D-resistant alleles at Idd3 and congenic strain panel should be more effective than the NOD Idd5.1/Ctla4 could provide an explanation for our observation that model alone for evaluating and prioritizing new therapeutic agents protective alleles at Idd5.1/Ctla4 in conjunction with resistance in the future. alleles at Idd3 did not provide more protection from T1D than protective alleles at the Idd3 region alone (Fig. 5A). Human gene discovery and disease subclassification Implications from this study for mapping and understanding the is sensitive to genetic background Idd5.1/Ctla4 effects of genetic variants on human autoimmune disease are sig- Another lesson learned from the current study is that the mapping nificant. The human CTLA-4 gene is associated with a number of of Idd5.1/Ctla4 may not have been accomplished without the ex- autoimmune diseases; in particular, noncoding SNPs near the istence of the Idd5.4 region, even though we did not know of the CTLA-4 structural gene are strongly associated with autoimmune existence and effect of Idd5.4 at the time (1). Localization of thyroid disease (AITD) (6). These same SNPs are also associated

Idd5.1/Ctla4 via sequential truncations of the proximal end of a with T1D but much more weakly than in AITD (6). Recently, we Downloaded from strain such as R444 would have failed to reveal the Idd5.1/Ctla4 (5) and others (25) have discovered that the CTLA4 SNPs are region because there is very little or no difference in the frequency strongly associated with T1D but in only those T1D cases with of diabetes between strains having T1D resistance alleles at Idd5.1/ AITD autoantibodies or diagnosed AITD. Our interpretation of the Ctla4, Idd5.3, and Idd5.2/Nramp1 and those having protective al- results from these human studies on CTLA4 is based on our current leles at Idd5.3 and Idd5.2/Nramp1. Thus, in the absence of the and previous observations in the NOD mouse model: the strength

B10-derived susceptibility allele at Idd5.4, we most likely would of the effect of allelic variation of Ctla4 is very dependent on http://www.jimmunol.org/ have missed the differential effects of resistant and susceptible al- different combinations of other susceptibility loci, including com- leles at Idd5.1/Ctla4. plete masking of the effect (that is no association with disease), and different combinations of Idd loci give rise to quite different au- Candidate genes toimmune phenotypes within the general NOD genetic back- The development of Idd5.3 congenic strains has allowed the fine ground, including AITD and an autoimmune liver disease (26, 27). mapping of the Idd5.3 region to an interval of 3.553 Mb which It is likely that the genetic basis of human T1D and related diseases contains 11 genes. It is not practical to reduce this region by further that occur more frequently together in T1D cases and families than congenic strain development and, consequently, we are now taking expected, such as AITD and rheumatoid arthritis, is similar: dif- a candidate gene approach. Because we have observed that Acadl ferent sets of alleles from multiple loci give rise to different but by guest on September 23, 2021 is differentially expressed in CD4 T cells obtained from NOD and related diseases. In these sets, there will be susceptibility loci in NOD.Idd3/5 mice (17), it is the primary candidate gene for Idd5.3. common (e.g., HLA class II alleles and haplotypes, and PTPN22) Future studies will address the expression of other polymorphic (28–33), and other loci specific to certain diseases or phenotypes genes within the Idd5.3 interval as well as functional consequences (e.g., the gene encoding insulin for T1D) (34). For the sets of of the differential expression of Acadl. susceptibility genes in humans that can give rise to isolated T1D, The identification of the gene or genes accounting for the Idd5.4 not complicated with AITD disease or related autoantibodies, the genetic effect will require the development of additional strains of CTLA4 SNPs are not associated, or only very weakly, with this congenic mice. Even though global differential expression analy- subclass of T1D, and we proposed that this is due to the presence ses have highlighted Daf1 as a compelling candidate gene (17), the and function of human genes similar to Idd5.3 and Idd5.2/Nramp1 Idd5.4 region is currently quite large (78 Mb) and we have dem- that can mask the effect of variation in the CTLA-4 gene. In the onstrated previously how compelling functional candidate genes minor subgroup of T1D that is complicated with AITD or AITD for Idd regions can be eliminated by refined genetic analyses (22, autoantibodies, ϳ10–15% of T1D cases, CTLA-4 genetic varia- 23). Therefore, to test the candidacy of Daf1, a NOD.B10 Daf1 tion has a strong effect, presumably via its role in regulation of congenic strain is currently being developed and will be tested for peripheral tolerance, in which the disease-associated CTLA4 hap- its ability to alter the frequency of type 1 diabetes. lotype is predisposing to a failure in tolerance to multiple organs or tissues (5). Our current results highlight the complexities of study- Subclassification of autoimmune diabetes ing autoimmunity genes even in a well-controlled experimental The T1D frequency of the R2 strain having B10 alleles at Idd5.3, model, and promote further caution in the design, execution, and Idd5.2/Nramp1, and Idd5.4 is equivalent to that observed in the interpretation of susceptibility gene mapping in humans. NOD parental strain (Fig. 6). However, it is likely that some of the cellular pathways causing autoimmune diabetes are altered by Disclosures the introgression of the B10-derived region on chromosome 1 that The authors have no financial conflict of interest. contains at least two B10-derived protective alleles and one B10- derived susceptibility allele. The resulting disease processes in the References NOD vs R2 strains could be described as two subtypes of type 1 1. Hill, N. J., P. A. Lyons, N. Armitage, J. A. Todd, L. S. Wicker, and diabetes, a situation that we suggest is analogous to the one in L. B. Peterson. 2000. The NOD Idd5 locus controls insulitis and diabetes and overlaps the orthologous CTLA4/IDDM12 and NRAMP1 loci in humans. Dia- human families in which different segregating combinations of betes 49: 1744–1747. susceptibility and resistance alleles cause the development of what 2. Lamhamedi-Cherradi, S. E., O. Boulard, C. Gonzalez, N. Kassis, D. Damotte, L. Eloy, G. Fluteau, M. Levi-Strauss, and H. J. Garchon. 2001. Further mapping is normally assumed to be clinically identical type 1 diabetes. For of the Idd5.1 locus for autoimmune diabetes in NOD mice. Diabetes 50: example, the gene combination present in R2 mice could render 2874–2878. The Journal of Immunology 8349

3. Wicker, L. S., G. Chamberlain, K. Hunter, D. Rainbow, S. Howlett, P. Tiffen, 20. Cannons, J. L., G. Chamberlain, J. Howson, L. J. Smink, J. A. Todd, J. Clark, A. Gonzalez-Munoz, A. M. Cumiskey, R. L. Rosa, et al. 2004. Fine L. B. Peterson, L. S. Wicker, and T. H. Watts. 2005. Genetic and functional mapping, gene content, comparative sequencing, and expression analyses support association of the immune signaling molecule 4-1BB (CD137/TNFRSF9) with Ctla4 and Nramp1 as candidates for Idd5.1 and Idd5.2 in the nonobese diabetic type 1 diabetes. J. Autoimmun. 25: 13–20. mouse. J. Immunol. 173: 164–173. 21. Lundholm, M., V. Motta, A. Lofgren-Burstrom, N. Duarte, M. L. Bergman, 4. Greve, B., L. Vijayakrishnan, A. Kubal, R. A. Sobel, L. B. Peterson, S. Mayans, and D. Holmberg. 2006. Defective induction of CTLA-4 in the NOD L. S. Wicker, and V. K. Kuchroo. 2004. The diabetes susceptibility locus Idd5.1 mouse is controlled by the NOD allele of Idd3/IL-2 and a novel locus (Ctex) on mouse chromosome 1 regulates ICOS expression and modulates murine ex- telomeric on chromosome 1. Diabetes 55: 538–544. perimental autoimmune encephalomyelitis. J. Immunol. 173: 157–163. 22. Prins, J. B., J. A. Todd, N. R. Rodrigues, S. Ghosh, L. S. Wicker, E. Gaffney, 5. Howson, J. M., D. B. Dunger, S. Nutland, H. Stevens, L. S. Wicker, and P. L. Podolin, P. A. Fischer, A. Sirotina, and L. B. Peterson. 1993. Linkage on J. A. Todd. 2007. A type 1 diabetes subgroup with a female bias is characterised chromosome 3 of autoimmune diabetes and defective Fc receptor for IgG in NOD by failure in tolerance to thyroid peroxidase at an early age and a strong asso- mice. Science 260: 695–698. ciation with the cytotoxic T-lymphocyte-associated antigen-4 gene. Diabetologia 23. Podolin, P. L., P. Denny, C. J. Lord, N. J. Hill, J. A. Todd, L. B. Peterson, 50: 741–746. L. S. Wicker, and P. A. Lyons. 1997. Congenic mapping of the insulin-dependent 6. Ueda, H., J. M. Howson, L. Esposito, J. Heward, H. Snook, G. Chamberlain, diabetes (Idd) gene: Idd10, localizes two genes mediating the Idd10 effect and D. B. Rainbow, K. M. Hunter, A. N. Smith, G. Di Genova, et al. 2003. Associ- eliminates the candidate Fcgr1. J. Immunol. 159: 1835–1843. ation of the T-cell regulatory gene CTLA4 with susceptibility to autoimmune 24. Roep, B. O., M. Atkinson, and M. von Herrath. 2004. Satisfaction (not) guaran- disease. Nature 423: 506–511. teed: re-evaluating the use of animal models of type 1 diabetes. Nat. Rev. Im- 7. Lynch, K. W., and A. Weiss. 2001. A CD45 polymorphism associated with mul- munol. 4: 989–997. tiple sclerosis disrupts an exonic splicing silencer. J. Biol. Chem. 276: 25. Ikegami, H., T. Awata, E. Kawasaki, T. Kobayashi, T. Maruyama, K. Nakanishi, 24341–24347. A. Shimada, S. Amemiya, Y. Kawabata, S. Kurihara, et al. 2006. The association 8. Rothrock, C., B. Cannon, B. Hahm, and K. W. Lynch. 2003. A conserved signal- of CTLA4 polymorphism with type 1 diabetes is concentrated in patients com- responsive sequence mediates activation-induced alternative splicing of CD45. plicated with autoimmune thyroid disease: a multicenter collaborative study in Mol. Cell 12: 1317–1324. Japan. J. Clin. Endocrinol. Metab. 91: 1087–1092. 9. Tong, A., J. Nguyen, and K. W. Lynch. 2005. Differential expression of CD45 26. Koarada, S., Y. Wu, N. Fertig, D. A. Sass, M. Nalesnik, J. A. Todd, P. A. Lyons, Downloaded from isoforms is controlled by the combined activity of basal and inducible splicing- J. Fenyk-Melody, D. B. Rainbow, L. S. Wicker, et al. 2004. Genetic control of regulatory elements in each of the variable exons. J. Biol. Chem. 280: autoimmunity: protection from diabetes, but spontaneous autoimmune biliary dis- 38297–38304. ease in a nonobese diabetic congenic strain. J. Immunol. 173: 2315–2323. 27. Yu, S., P. K. Maiti, M. Dyson, R. Jain, and H. Braley-Mullen. 2006. B cell- 10. Vijayakrishnan, L., J. M. Slavik, Z. Illes, R. J. Greenwald, D. Rainbow, B. Greve, ϩ ϩ L. B. Peterson, D. A. Hafler, G. J. Freeman, A. H. Sharpe, et al. 2004. An deficient NOD.H-2h4 mice have CD4 CD25 T regulatory cells that inhibit the autoimmune disease-associated CTLA-4 splice variant lacking the B7 binding development of spontaneous autoimmune thyroiditis. J. Exp. Med. 203: 349–358. domain signals negatively in T cells. Immunity 20: 563–575. 28. Bottini, N., L. Musumeci, A. Alonso, S. Rahmouni, K. Nika, M. Rostamkhani, J. MacMurray, G. F. Meloni, P. Lucarelli, M. Pellecchia, et al. 2004. A functional 11. Fortier, A., G. Min-Oo, J. Forbes, S. Lam-Yuk-Tseung, and P. Gros. 2005. Single

variant of lymphoid tyrosine phosphatase is associated with type I diabetes. Nat. http://www.jimmunol.org/ gene effects in mouse models of host: pathogen interactions. J. Leukocyte Biol. Genet. 36: 337–338. 77: 868–877. 29. Smyth, D., J. D. Cooper, J. E. Collins, J. M. Heward, J. A. Franklyn, 12. Kissler, S., P. Stern, K. Takahashi, K. Hunter, L. B. Peterson, and L. S. Wicker. J. M. Howson, A. Vella, S. Nutland, H. E. Rance, L. Maier, et al. 2004. Repli- 2006. In vivo RNA interference demonstrates a role for Nramp1 in modifying cation of an association between the lymphoid tyrosine phosphatase locus (LYP/ susceptibility to type 1 diabetes. Nat. Genet. 38: 479–483. PTPN22) with type 1 diabetes, and evidence for its role as a general autoimmu- 13. Robles, D. T., G. S. Eisenbarth, N. J. Dailey, L. B. Peterson, and L. S. Wicker. nity locus. Diabetes 53: 3020–3023. 2003. Insulin autoantibodies are associated with islet inflammation but not always 30. Criswell, L. A., K. A. Pfeiffer, R. F. Lum, B. Gonzales, J. Novitzke, M. Kern, related to diabetes progression in NOD congenic mice. Diabetes 52: 882–886. K. L. Moser, A. B. Begovich, V. E. Carlton, W. Li, et al. 2005. Analysis of 14. Lyons, P. A., and L. S. Wicker. 1999. Localising quantitative trait loci in the families in the multiple autoimmune disease genetics consortium (MADGC) col- NOD mouse model of type 1 diabetes. In Genes and Genetics of Autoimmunity, lection: the PTPN22 620W allele associates with multiple autoimmune pheno- Vol. 1. A. N. Theofilopoulos, ed. Karger, Basel, pp. 208–225. types. Am. J. Hum. Genet. 76: 561–571.

15. Lyons, P. A., N. Armitage, F. Argentina, P. Denny, N. J. Hill, C. J. Lord, 31. Carlton, V. E., X. Hu, A. P. Chokkalingam, S. J. Schrodi, R. Brandon, by guest on September 23, 2021 M. B. Wilusz, L. B. Peterson, L. S. Wicker, and J. A. Todd. 2000. Congenic H. C. Alexander, M. Chang, J. J. Catanese, D. U. Leong, K. G. Ardlie, et al. 2005. mapping of the type 1 diabetes locus to a 780 kb region of mouse chromosome PTPN22 genetic variation: evidence for multiple variants associated with rheu- 3: identification of a candidate segment of ancestral DNA by haplotype mapping. matoid arthritis. Am. J. Hum. Genet. 77: 567–581. Genome Res. 446–453. 32. Canton, I., S. Akhtar, N. G. Gavalas, D. J. Gawkrodger, A. Blomhoff, 16. Dovat, S., E. Montecino-Rodriguez, V. Schuman, M. A. Teitell, K. Dorshkind, P. F. Watson, A. P. Weetman, and E. H. Kemp. 2005. A single-nucleotide poly- and S. T. Smale. 2005. Transgenic expression of helios in B lineage cells alters morphism in the gene encoding lymphoid protein tyrosine phosphatase (PTPN22) B cell properties and promotes lymphomagenesis. J. Immunol. 175: 3508–3515. confers susceptibility to generalised vitiligo. Genes Immun. 6: 584–587. 17. Irie, J., B. Reck, Y. Wu, L. S. Wicker, S. Howlett, D. Rainbow, E. Feingold, and 33. Wu, H., R. M. Cantor, D. S. Graham, C. M. Lingren, L. Farwell, P. L. Jager, ϩ W. M. Ridgway. 2006. Genome wide microarray expression analysis of CD4 T N. Bottini, J. M. Grossman, D. J. Wallace, B. H. Hahn, et al. 2005. Association cells from NOD congenic mice identifies Cd55 (Daf1) and Acadl as candidate analysis of the R620W polymorphism of protein tyrosine phosphatase PTPN22 in genes for type 1 diabetes. J. Immunol. In press. systemic lupus erythematosus families: increased T allele frequency in systemic 18. Guerra, C., R. A. Koza, K. Walsh, D. M. Kurtz, P. A. Wood, and L. P. Kozak. lupus erythematosus patients with autoimmune thyroid disease. Arthritis Rheum. 1998. Abnormal nonshivering thermogenesis in mice with inherited defects of 52: 2396–2402. fatty acid oxidation. J. Clin. Invest. 102: 1724–1731. 34. Tait, K. F., J. E. Collins, J. M. Heward, I. Eaves, H. Snook, J. A. Franklyn, 19. Cordell, H., J. Todd, N. Hill, C. Lord, P. Lyons, L. Peterson, L. Wicker, and A. H. Barnett, J. A. Todd, M. Maranian, A. Compston, et al. 2004. Evidence for D. Clayton. 2001. Statistical modeling of interlocus interactions in a complex a Type 1 diabetes-specific mechanism for the insulin gene-associated IDDM2 disease: rejection of the multiplicative model of epistasis in type 1 diabetes. locus rather than a general influence on autoimmunity. Diabetes Med. 21: Genetics 158: 357–367. 267–270.