HLA-A*0201-Restricted T Cells from Humanized NOD Mice Recognize Autoantigens of Potential Clinical Relevance to Type 1 Diabetes

Transcription

1 This information is current as of January 7, References Subscription Permissions Alerts HLA-A*0201-Restricted T Cells from Humanized NOD Mice Recognize Autoantigens of Potential Clinical Relevance to Type 1 Diabetes Toshiyuki Takaki, Michele P. Marron, Clayton E. Mathews, Stephen T. Guttmann, Rita Bottino, Massimo Trucco, Teresa P. DiLorenzo and David V. Serreze J Immunol 2006; 176: ; ; doi: /jimmunol This article cites 53 articles, 30 of which you can access for free at: 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 Information about subscribing to The Journal of Immunology is online at: Submit copyright permission requests at: Receive free -alerts when new articles cite this article. Sign up at: Downloaded from by guest on January 7, 2019 The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD Copyright 2006 by The American Association of Immunologists All rights reserved. Print ISSN: Online ISSN:

2 The Journal of Immunology HLA-A*0201-Restricted T Cells from Humanized NOD Mice Recognize Autoantigens of Potential Clinical Relevance to Type 1 Diabetes 1 Toshiyuki Takaki, 2 * Michele P. Marron, 2,3 Clayton E. Mathews, Stephen T. Guttmann,* Rita Bottino, Massimo Trucco, Teresa P. DiLorenzo, 4 * and David V. Serreze 4 In both humans and NOD mice, particular MHC genes are primary contributors to development of the autoreactive CD4 and CD8 T cell responses against pancreatic cells that cause type 1 diabetes (T1D). Association studies have suggested, but not proved, that the HLA-A*0201 MHC class I variant is an important contributor to T1D in humans. In this study, we show that transgenic expression in NOD mice of HLA-A*0201, in the absence of murine class I MHC molecules, is sufficient to mediate autoreactive CD8 T cell responses contributing to T1D development. CD8 T cells from the transgenic mice are cytotoxic to murine and human HLA-A*0201-positive islet cells. Hence, the murine and human islets must present one or more peptides in common. Islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP) is one of several important T1D autoantigens in standard NOD mice. Three IGRP-derived peptides were identified as targets of diabetogenic HLA-A*0201-restricted T cells in our NOD transgenic stock. Collectively, these results indicate the utility of humanized HLA-A*0201-expressing NOD mice in the identification of T cells and autoantigens of potential relevance to human T1D. In particular, the identified antigenic peptides represent promising tools to explore the potential importance of IGRP in the development of human T1D. The Journal of Immunology, 2006, 176: Type 1 diabetes (T1D) 5 in both humans and NOD mice is a multifactorial disease resulting from T cell-mediated autoimmune destruction of insulin-secreting pancreatic cells. Although T1D is a polygenic disease, the primary susceptibility component in both organisms is particular combinations of MHC genes (reviewed in Refs. 1 and 2). Within the MHC, specific HLA-DQ and HLA-DR class II molecules provide a large component of T1D susceptibility in humans by mediating cell-autoreactive CD4 T cell responses. Similarly, the H-2A g7 class II *Department of Microbiology and Immunology and Department of Medicine (Division of Endocrinology), Albert Einstein College of Medicine, Bronx, NY 10461; The Jackson Laboratory, Bar Harbor, ME 04609; and Department of Pediatrics and Diabetes Institute, Children s Hospital of Pittsburgh, University of Pittsburgh School of Medicine, Pittsburgh, PA Received for publication November 8, Accepted for publication December 22, The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported by National Institutes of Health Grants DK69280 (to M.P.M.), DK64315 (to T.P.D.), DK52956 (to T.P.D.), DK51090 (to D.V.S.), DK46266 (to D.V.S.), AI56374 (to C.E.M.), Cancer Center Support Grant CA34196 (The Jackson Laboratory), and DK20541 (Albert Einstein College of Medicine s Diabetes Research and Training Center), and by grants from the Juvenile Diabetes Research Foundation (to T.P.D., D.V.S., C.E.M., M.T.), and the Alexandrine and Alexander Sinsheimer Foundation (to T.P.D.). The flow cytometry facility at Albert Einstein College of Medicine is supported by National Institutes of Health Cancer Center Grant CA T.T. and M.P.M. contributed equally to this work. 3 Current address: Department of Pediatrics, Division of Genetic and Translational Medicine, University of Alabama at Birmingham, Birmingham, AL Address correspondence and reprint requests to Dr. David V. Serreze, The Jackson Laboratory, 600 Main Street, Bar Harbor, ME 04609; address: dvs@jax.org or Dr. Teresa P. DiLorenzo, Department of Microbiology and Immunology, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY 10461; address: dilorenz@aecom.yu.edu 5 Abbreviations used in this paper: T1D, type 1 diabetes; 2m, 2 -microglobulin; IGRP, islet-specific glucose-6-phosphatase catalytic subunit-related protein; m, murine; h, human; DMK, dystrophia myotonica kinase; MFI, mean fluorescence intensity. variant provides an important component of T1D susceptibility in NOD mice. It is now clear that in addition to the pathogenic effects mediated by CD4 T cells, T1D development also requires contributions from autoreactive MHC class I-restricted CD8 T cells (3 8). Interestingly, the H-2D b and H-2K d class I molecules encoded within the H2 g7 MHC haplotype of NOD mice are common variants also characterizing many strains lacking autoimmune proclivity. However, efficient T1D development in NOD mice is dependent on expression of the class I genes characterizing the H2 g7 haplotype (9 12). The ability of the H2 g7 -encoded class I variants to mediate the development of diabetogenic CD8 T cell responses when expressed in NOD is due to interactive contributions from other susceptibility genes characterizing this strain (13, 14). Association studies suggest certain MHC class I molecules, including the common variant HLA-A*0201 (hereafter designated HLA-A2.1), likely confer an additional risk factor for T1D development in humans when expressed in conjunction with particular class II susceptibility alleles, and perhaps other genes (15 21). Because it is difficult to directly assess the role of class I MHC molecules in T1D development in patients, we have used a humanized NOD mouse model to do so. Previously, we showed that T1D development is accelerated in NOD mice transgenically expressing the human HLA-A2.1 H chain (designated NOD.HLA- A2.1) (22). Furthermore, HLA-A2.1-restricted cell-autoreactive CD8 T cells can be isolated from islets of young transgenic NOD mice. However, NOD.HLA-A2.1 mice also express the murine class I MHC molecules H-2K d and H-2D b, making it impossible to precisely define the ability of HLA-A2.1 alone to mediate autoreactive CD8 T cell responses contributing to T1D development. To address this important issue, we used the chimeric monochain transgene construct designated HHD (23), which encodes human 2 -microglobulin ( 2m) covalently linked to the 1 and 2 domains of human HLA-A2.1, and the 3, transmembrane, and cytoplasmic domains of murine H-2D b. We transgenically introduced Copyright 2006 by The American Association of Immunologists, Inc /06/$02.00

3 3258 HLA-A*0201 MEDIATES DIABETES DEVELOPMENT the HHD construct into NOD mice and then crossed the transgenic mice to the NOD. 2m null strain to eliminate expression of murine class I MHC molecules. In HLA-transgenic models, the frequency of CD8 T cells restricted to human class I MHC molecules is often low due to the poor interaction between murine CD8 and the 3 domain of human class I MHC molecules (24). The presence of a murine 3 domain in HHD molecules is designed to overcome this difficulty. In this study, we show that NOD. 2m null.hhd mice are T1Dsusceptible. This demonstrates that when expressed in the proper context, HLA-A2.1 alone can mediate sufficient cell-autoreactive CD8 T cell responses to elicit T1D. Studies were also conducted to determine whether NOD. 2m null.hhd mice allow identification of candidate cell autoantigens that could be tested in future clinical studies as possible targets of pathogenic CD8 T cells in HLA-A2.1-expressing T1D patients. CD8 T cells isolated from NOD. 2m null.hhd mice were found to lyse human HLA- A2.1-positive islet cells. A subset of these T cells recognize the cell Ag islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP), previously identified as an important target of CD8 T cells in standard NOD mice (25, 26). We have mapped three novel epitopes of IGRP that are recognized by HLA-A2.1- restricted T cells in NOD. 2m null.hhd mice. Importantly, one of these antigenic peptides is conserved between mouse and human IGRP, and the other two are similar. Therefore, humanized NOD. 2m null.hhd mice can be used to identify HLA-A2.1-restricted T cells and cell autoantigens potentially relevant to human T1D. Materials and Methods Mice NOD/LtDvs mice are maintained at The Jackson Laboratory by brothersister mating. Currently, diabetes develops in 90% of female and 63% of male NOD/LtDvs mice by 30 wk of age. NOD. 2m null (3) and NOD.Rag1 null mice (27) have been previously described. The HHD transgene, provided by F. Lemonnier (Institut Pasteur, Paris, France), has been described (23). The transgene was injected directly into NOD zygotes, and founders carrying the HHD transgene were identified by PCR using forward primer 5 -CTTCATCGCAGTGGGCTAC-3 and reverse primer 5 - CGGTGAGTCTGTGAGTGGG-3. Cell surface expression of transgenic HLA-A2.1 molecules in NOD.HHD mice was confirmed by flow cytometry using FITC-conjugated mab CR (28), provided by V. Engelhard (University of Virginia, Charlottesville, VA). To eliminate expression of murine class I MHC molecules, the HHD transgene was transferred to the previously described NOD. 2m null background (3). The 2m null mutation was fixed to homozygosity by backcrossing F 1 progeny to NOD. 2m null mice, then intercrossing to establish the NOD. 2m null.hhd line. The HHD transgene also was transferred to the NOD-scid.Emv30 null background (29) and fixed to homozygosity (designated NOD-scid.HHD mice). All mice were housed under specific pathogen-free conditions. All experiments involving mice were performed in compliance with federal laws and institutional guidelines and have been approved by an institutional animal care and use committee. Assessment of T1D and insulitis development T1D development was defined by glycosuric values of 3 as assessed with Ames Diastix (supplied by Miles Diagnostics). Rates of diabetes development in the indicated experimental groups were assessed for statistically significant differences by Kaplan-Meier life table analysis using Statview 4.5 (Abacus Concepts). Insulitis levels were assessed in nondiabetic 8-wk-old female NOD and NOD. 2m null.hhd mice. Pancreata from these mice were fixed in Bouin s solution, and sectioned at three nonoverlapping levels. Granulated cells were stained with aldehyde fuchsin and leukocytes with an H&E counterstain. Islets (at least 20 per mouse) were individually scored as follows: 0, no lesions; 1, peri-insular leukocytic aggregates, usually periductal infiltrates; 2, 25% islet destruction; 3, 25% islet destruction; and 4, complete islet destruction. An insulitis score for each mouse was obtained by dividing the total score for each pancreas by the number of islets examined. Data are presented as mean insulitis scores SEM for each strain. Flow cytometric analysis of splenic leukocyte populations Single-cell suspensions of splenocytes were analyzed by multicolor flow cytometric analysis. Total B lymphocytes were detected by staining with FITC-conjugated polyclonal anti-mouse Ig (Southern Biotechnology Associates). Total T lymphocytes were detected by staining with FITC-conjugated CD3 -specific mab 145-2C11. Total T cells were then further characterized for CD4 expression using the mab GK1.5 conjugated to the red fluorescent tag Cy3.18-OSu (Cy3; Biological Detection Systems), or for CD8 expression with the mab conjugated to PE whose red fluorescence intensity can easily be distinguished from that of Cy3. CD8 T cell populations were also analyzed by staining with PE-conjugated TCR -chain-specific mab H and FITC-conjugated CD8-specific mab Expression of murine MHC class I molecules was assessed by staining with FITC-conjugated H-2K d -specific mab SF1-1.1, FITC-conjugated H-2D b -specific mab KH95, and FITC-conjugated pan murine class I MHC mab M1/42. Expression of transgenic HLA-A2.1 molecules was assessed by staining with FITC-conjugated mab BB7.2. Cytotoxicity assays using intact islets as targets Human pancreata that were not allocated for whole organ transplantation were obtained from organ procurement organizations (Center for Organ Recovery and Education, Pittsburgh, PA, and National Disease Research Interchange, Philadelphia, PA). Pancreata were harvested using standard multiorgan recovery techniques. Pancreatic islets were isolated using the semiautomated method described previously (30) with minor modifications. Pancreas dissociation was performed using multiple lots of Liberase (Liberase-HI; Roche). Enzymes were reconstituted and dissolved in cold (4 C) HBSS. Before being digested in a modified continuous digestionfiltration device, pancreata were intraductally injected with enzyme solution in a recirculation system designed by Rajotte and colleagues (31). Islets were purified with a COBE 2991 cell separator using discontinuous Euro-Ficoll gradients; the purity (islets/whole tissue) was assessed by dithizone staining, as previously described (32). Preliminary HLA typing was performed by the laboratory of the local donor hospital (HLA-A*02/ A*68 sample) or the typing laboratory of the Allegheny General Hospital (Pittsburgh, PA) (HLA-A*11/A*13 sample). High-resolution HLA typing was performed using the PEL-FREEZ SSP UniTray PCR-based method for specific HLA class I alleles as per the manufacturer s directions (Dynal Biotech). Mouse islets were isolated from female NOD-scid.HHD and NOD.Rag1 null animals using a published collagenase inflation method (33). Islets were hand-picked for purity in HBSS under a dissecting microscope after isolation on a Histopaque 1119 (Sigma-Aldrich) gradient. Cytotoxicity assays using intact islets as targets were performed as described (13, 34). Briefly, human, NOD-scid.HHD, and NOD.Rag1 null pancreatic islets (10 islets/well) were allowed to adhere in 96-well plates during a 10-day incubation at 37 C in low-glucose DMEM. Adherent, monolayered islets were then labeled with 5 Ci/well of 51 Crfor3hat 37 C. Islets were washed and overlaid with 100 l of medium containing various numbers of cultured islet-infiltrating T cells from NOD. 2m null.hhd mice. For establishing E:T ratios, each islet was assumed to contain 750 cells. A minimum of three wells were established for each E:T ratio. Spontaneous release controls consisted of nine wells of labeled islets from each donor cultured in the absence of T cells. Following a 20-h incubation at 37 C, the radioactivity in two fractions from each well was measured. The first fraction was the culture supernatant, and the second was obtained by solubilizing the remaining islets in 200 l of 2% SDS. The percentage of 51 Cr release for each well was calculated by the formula ((supernatant cpm)/(supernatant cpm SDS lysate cpm)) 100%. This allowed us to normalize for the fact that due to variation in the sizes of individual islets, the total levels of 51 Cr incorporation in each well could differ greatly. In turn, the percent-specific cytotoxicity was calculated by subtracting the percent 51 Cr release from islets cultured in medium alone (i.e., spontaneous release) from the release by each well of islets cultured with a given number of T cells. Percent-specific cytotoxicities against the different islet types were assessed for statistically significant differences by Bonferroni/Dunn analysis using SuperANOVA (Abacus Concepts). IGRP peptide library and individual synthetic peptides A peptide library containing all of the 8-, 9-, 10-, and 11-mer peptides that can be derived from murine IGRP was synthesized by Mimotopes using their proprietary Truncated PepSet technology. Each peptide mixture in the library contained four peptides with a common C terminus, but having a length of 8, 9, 10, or 11 residues. The four peptides in each mixture were

4 The Journal of Immunology 3259 present in approximately equimolar amounts. A library of 348 peptide mixtures was synthesized to cover the 355 aa of the IGRP protein. Concentrated peptide stocks (2.75 mm) were prepared in 50% acetonitrile/h 2 O, and 40 M (i.e., 10 M for each peptide in the mixture) working stocks were obtained by serial dilution in PBS (ph 6.5). Murine (m) IGRP (VYLKTNVFL), migrp (RLFGIDLLWSV), migrp (LF GIDLLWSV), migrp (FGIDLLWSV), migrp (GIDLL WSV), migrp (VLFGLGFAI), migrp (ALIPYCVHM), human (h) IGRP (LNIDLLWSV), higrp (AFIPYSVHM), and Flu-MP (GILGFVFTL) peptides were synthesized by standard solidphase methods using Fmoc chemistry in an automated peptide synthesizer (model 433A; Applied Biosystems), and their identities were confirmed by mass spectrometry. Concentrated stocks (10 mm) were prepared in DMSO, and 10 M working stocks were obtained by dilution in PBS. Propagation of islet-infiltrating T cells Islet isolation by collagenase perfusion of the common bile duct was modified from a previously described protocol (35). Briefly, the bile duct was cannulated and the pancreas perfused with collagenase P (Roche). The inflated pancreas was removed and incubated at 37 C to digest exocrine tissue. Following dispersion of digested tissue and three washes with HBSS, islets were resuspended in HBSS containing DNase I (Worthington Biochemical) and handpicked using a siliconized micropipet under a dissecting microscope. Isolated islets were washed with 2% FBS in HBSS, resuspended in RPMI 1640 medium supplemented with 10% FBS (Hy- Clone) and 50 U/ml recombinant human IL-2 (PeproTech), and cultured in 24-well tissue culture plates ( 50 islets/well) at 37 C, 5% CO 2 for 7 days. For the experiment depicted in Fig. 3C, where purified CD8 T cells were used, cultured islet-infiltrating T cells were collected from four 11-wk-old female NOD. 2m null.hhd mice and incubated for 30 min with 40 l of FITC-conjugated anti-cd8 Ab (53-6.7; BD Pharmingen). Then, T cells were washed three times and sorted using a MoFlo Fluorescence Activated Cell Sorter (DakoCytomation). After sorting, 97% of cells were CD8 T cells. IFN- ELISPOT ELISPOT plates (MAHA S45 10; Millipore) were precoated with antimurine IFN- mab (R4-6A2; BD Biosciences/BD Pharmingen) and blocked with 1% BSA (Fraction V; Sigma-Aldrich) in PBS. APC (mitomycin C-treated T2 cells; American Type Culture Collection) were added at cells/well and pulsed with 1 M peptide. Cultured islet-infiltrating T cells were added at cells/well, and plates were incubated at 37 C for 40 h. IFN- secretion was detected with a second, biotinylated anti-murine IFN- mab (XMG1.2; BD Pharmingen). Spots were developed using streptavidin-alkaline phosphatase (Zymed Laboratories) and 5-bromo-4-chloro-3-indolyl-phosphate/NBT chloride substrate (Sigma-Aldrich) and counted using an automated ELISPOT reader system (Autoimmun Diagnostika). Transient transfection COS-7 cells were transfected using a DEAE-dextran protocol as described (36). cdna expression constructs (1 g/ml) for class I MHC molecules (HHD/pcDNA3.1, H-2K d /prsv.5.neo, or H-2D b /pcdnai) were used along with varying concentrations of a murine IGRP cdna expression construct (migrp/pcdna3.1) or pcdna3.1 alone as indicated in the legend to Fig. 3A. migrp/pcdna3.1 was provided by P. Santamaria (University of Calgary, Calgary, Alberta, Canada), H-2D b /pcdnai by N. Shastri (University of California, Berkeley, CA), and H-2K d /prsv.5.neo by T. Hansen (Washington University Medical School, St. Louis, MO). Following coculture with cultured islet-infiltrating T cells, T cell response was measured as IFN- release by ELISA using capture (R4-6A2) and detecting (biotinylated XMG1.2) anti-murine IFN- mabs purchased from BD Biosciences/BD Pharmingen. Plates were developed with streptavidin-conjugated HRP (Southern Biotechnology Associates) and tetramethyl benzidine (Pierce Biotechnology). Peptide-binding assay Peptide binding to the HLA-A2.1 molecule was determined as described (37). Briefly, T2 cells were incubated for 18 h at 26 C and washed in serum-free culture medium. Cells ( )in60 l of serum-free medium were added to U-bottom 96-well culture plates with 20 l of peptide solution or PBS and 20 l of human 2m in PBS (final concentration 15 g/ml; Sigma-Aldrich). The cells were incubated for 20 h at 37 C, 5% CO 2 in humidified air, then transferred into V-bottom 96-well culture plates and incubated in a water bath at 50 C for 3 min. The cells were washed twice in cold (4 C) PBS, stained with FITC-conjugated anti-hla-a2 mab BB7.2 (BD Biosciences/BD Pharmingen), and analyzed by flow cytometry. The up-regulation of HLA-A2.1 is reported as the fluorescence index, defined as the mean fluorescence intensity (MFI) of the experimental sample divided by the MFI in the absence of peptide. Results Accelerated T1D in the presence of transgenic HLA-A2.1 molecules The HHD construct (23) was injected directly into NOD zygotes and a transgenic line was established. As previously observed in NOD.HLA-A2.1 mice (22), expression of the HHD transgene resulted in significant acceleration of T1D onset through 30 wk of age in NOD.HHD females compared with standard NOD controls ( p 0.02; Fig. 1). HLA-A2.1-restricted autoimmune CD8 T cell responses in the absence of murine MHC class I molecules As in our previous study (22), acceleration of T1D in NOD.HHD mice could most simply be attributed to addition of HLA-A2.1- restricted CD8 T cell responses to those elicited by murine class I MHC molecules. Thus, we tested whether HLA-A2.1 molecules could elicit CD8 T cell responses sufficient for T1D development in the absence of murine MHC class I molecules. By crossing the HHD transgene into the NOD. 2m null stock (3), we eliminated expression of murine MHC class I molecules. Flow cytometric analysis of splenocytes confirmed NOD. 2m null.hhd mice express only human HLA-A2.1, and not murine MHC class I molecules (Fig. 2A, Table I). Although low-level surface expression of the H-2D b H chain in the absence of 2m has been reported to occur in cells derived from certain mouse strains (38, 39), we saw no evidence for this in NOD. 2m null.hhd mice (Fig. 2A, Table I). Development of CD8 T cells in NOD. 2m null.hhd mice was compared with NOD and NOD. 2m null controls. As previously reported (3 6), NOD. 2m null mice completely lacked CD8 T cells (data not shown). Introduction of the HHD transgene onto this background restored CD8 T cell development, albeit at considerably reduced numbers compared with standard NOD mice expressing murine MHC class I molecules (Fig. 2B, Table I). The 90% reduction in CD8 T cells is similar to that reported for C57BL/6 background mice expressing the HHD transgene in the absence of murine MHC class I molecules (40). Although female nontransgenic NOD. 2m null mice remained completely free of disease as previously reported (3 6), 55% of HHD transgenic females developed T1D by 30 wk of age ( p 0.006; Fig. 2C). Pancreata from 8-wk-old NOD. 2m null.hhd and NOD female mice were examined histologically for insulitis levels (scale 0 4). This analysis indicated similar levels of leukocytic infiltration in NOD. 2m null.hhd (mean insulitis score, ; n 5) and NOD islets ( ; n 4). FIGURE 1. Transgenically expressed chimeric HLA-A2.1 class I MHC molecules contribute to development of T1D in NOD mice. Female mice were monitored weekly for development of T1D. NOD.HHD mice (E; n 19) had significantly accelerated onset of disease (p 0.02) compared with nontransgenic littermates (f; n 27).

5 3260 HLA-A*0201 MEDIATES DIABETES DEVELOPMENT FIGURE 2. HLA-A2.1 expression and development of HLA-A2.1-restricted CD8 T cells and T1D in NOD. 2m null.hhd mice. Splenocytes from NOD (A and B, left panels) and NOD. 2m null.hhd mice (A and B, right panels) were analyzed by flow cytometry. A, NOD splenocytes express murine MHC class I molecules as indicated. In NOD. 2m null.hhd mice, the level of staining with the murine MHC class I-specific mabs M1/42 (pan murine MHC class I) and KH95 (H-2D b -specific) is reduced to background levels, but transgenically expressed HLA-A2.1 molecules are readily detected using the mab BB7.2. Broken line, isotype control; solid line, specific Ab. B, Splenocytes were stained with anti-tcr and anti-cd8 Abs. Although reduced compared with NOD controls, NOD. 2m null.hhd mice expressing only HLA-A2.1 are able to generate CD8 T cells. Numbers given are mean percentage SEM (see Table I). C, NOD. 2m null.hhd mice (E; n 20) transgenically expressing HLA-A2.1 molecules were susceptible to T1D development while nontransgenic NOD. 2m null littermates (f; n 10), which lack class I MHC molecules, were completely resistant (p 0.006). NOD. 2m null.hhd mice generate T cells that can lyse HLA-A2.1-positive human pancreatic islets T cells were propagated from the islets of NOD. 2m null.hhd mice as described in Materials and Methods. Comparable percentages of CD8 T cells were observed in cultures derived from NOD. 2m null.hhd ( %; n 12) and NOD islets ( %; n 6). All the CD8 T cells propagated from the islets of NOD. 2m null.hhd mice must be HLA-A2.1-restricted, because murine MHC class I molecules are not expressed in this strain. Thus, we asked whether NOD. 2m null.hhd T cell cultures would exhibit cytotoxicity against HLA-A2.1-positive human pancreatic islets. When dissociated islet cells are used as targets in 51 Cr-release cytotoxicity assays, they exhibit a high degree of spontaneous death that can complicate data interpretation. To overcome this problem, we recently developed an alternative assay in which intact islets are allowed to adhere and monolayer in a 96-well plate and are then used as targets (13, 34). Maintenance of cell-cell contact greatly enhances cell survival and leads to low spontaneous 51 Cr-release values. Allowing the islets to adhere and monolayer increases the number of cells that are accessible to effector T cells. Using this assay, we measured the cytotoxic activity of NOD. 2m null.hhd T cell cultures against HLA-A2.1-positive or negative human pancreatic islets (Table II). As further controls, cytotoxic activity was also assessed against NOD-scid.HHD (H- 2D b, H-2K d, and HLA-A2.1) and NOD.Rag1 null (H-2D b, H-2K d ) islets. HLA-A2.1-restricted cytotoxicity against both human and mouse islets was observed. Although these results do not demonstrate the extent of overlap, they indicate that at least some subset of the antigenic peptides targeted by diabetogenic HLA-A2.1-restricted T cells are common to both mouse and human cells. Although cultured islet-infiltrating T cells from NOD. 2m null.hhd mice can recognize islet Ags, they were not able to recognize Con A blasts generated from HLA-A2-positive or HLA-A2-negative humans (data not shown). These collective results indicate that the cytotoxicity observed against HLA-A2.1-positive human islets is both HLA-A2.1- restricted and islet-specific. IGRP peptides are targeted by HLA-A2.1-restricted T cells from NOD. 2m null.hhd mice Residues of IGRP (IGRP ) represent the Ag recognized by the NOD-derived diabetogenic CD8 T cell clone 8.3 (25). The pathogenicity of IGRP reactive T cells is wellestablished (41 45), and 20 30% of CD8 T cells in NOD pancreatic islets recognize this peptide (25). Previously, using a peptide-mhc tetramer reagent, we tested reactivity to IGRP among islet-infiltrating T cells at various prodromal stages of T1D development (5, 7, 11, and 15 wk of age) in individual NOD mice (46). We detected an IGRP response in % of mice in all age groups tested. This response was more frequent than that to the other autoantigens examined, i.e., insulin and dystrophia myotonica kinase (DMK). Importantly, the human IGRP open reading frame is identical in length to murine IGRP and is highly homologous (85% identity; 93% similarity), though a T cell response against IGRP in human T1D patients has not yet been described. For these reasons, we used NOD. 2m null.hhd mice to investigate IGRP as a possible source of HLA-A2.1-binding peptides targeted by diabetogenic CD8 T cells. To determine whether CD8 T cells from NOD. 2m null.hhd mice recognize IGRP peptides, we transiently transfected COS-7 cells with varying concentrations of a full-length murine IGRP cdna expression construct in combination with an HHD, H-2D b, or H-2K d cdna expression construct. Transfected COS-7 cells were assessed for their ability to stimulate T cells propagated from the islets of NOD. 2m null.hhd mice as evidenced by IFN- secretion (Fig. 3A). T cell responses were only elicited in a dosedependent manner when target cells were transfected with both the IGRP and HHD constructs. These results demonstrate that at least some portion of the HLA-A2.1-restricted islet-infiltrating T cells in NOD. 2m null.hhd mice recognize an IGRP-derived peptide(s). In addition, the inability of islet-infiltrating T cells from NOD. 2m null.hhd mice to recognize kidney-derived COS-7 cells transfected with the HHD construct alone further demonstrates that

6 The Journal of Immunology 3261 Table I. Splenocyte composition of NOD. 2m null.hhd mice a % B Cells % CD4 Cells % CD8 Cells H-2D b MFI H-2K d MFI NOD NOD. 2m null.hhd Number of splenocytes ( 10 6 ) No. of B cells No. of CD4 cells No. of CD8 cells Total no. of splenocytes NOD NOD. 2m null.hhd a Splenocytes from 8-wk-old female NOD (n 7) and NOD. 2m null.hhd mice (n 5) were analyzed by flow cytometry for proportions and numbers of lymphocyte populations. Lymphocyte subsets were characterized by staining with specific mabs. Numbers given are percentage of splenocytes SEM or number of splenocytes SEM for each population. Splenocyte expression levels of murine MHC class I molecules were assessed in two 6-wk-old female mice of each strain and are indicated by MFI SD. the response of these effectors against human HLA-A2.1-positive islets shown in Table II is cell type-specific. To identify the IGRP epitopes recognized by CD8 T cells from NOD. 2m null.hhd mice, we screened a peptide library containing all possible 8- to 11-mer sequences that can be derived from murine IGRP. The library consisted of peptide mixtures each containing four sequences with a common C terminus, but having a length of 8, 9, 10, or 11 residues. T cells were cultured from NOD. 2m null.hhd islets and their responsiveness was measured by IFN- ELISPOT assay using HLA-A2.1-expressing human T2 cells as APC. Responses were observed to peptide mixtures 229 and (Fig. 3B). Islet-infiltrating T cell cultures contained, on average, 70% CD8 T cells. However, to ensure we were not detecting CD4 T cell responses, the peptide library was screened again using purified CD8 T cells cultured from the islets of an independent group of NOD. 2m null.hhd mice. These CD8 T cells responded to mixtures 229 and 266 (Fig. 3C), while purified islet-derived CD8 T cells from a third group of mice responded to mixtures 229, 266, and 338 (data not shown). Next, we screened the IGRP peptide library using T cells propagated from NOD.HHD mice that express endogenous H2 g7 -encoded class I MHC molecules (H-2D b and H-2K d ) in addition to the transgenic HLA-A2.1 variant (Fig. 3D). As expected, we detected a strong response to peptide mixtures , corresponding to the H-2K d -binding IGRP epitope. In contrast, T cells from NOD. 2m null.hhd mice do not respond to IGRP (Fig. 3, B and C), consistent with the absence of H-2K d expression in this strain. In addition to mixtures , T cells from NOD.HHD mice also released IFN- in response to peptide mixture 229. In separate screens using T cells from other NOD.HHD mice, responses to mixtures 266 and 338 were also observed (data not shown). These results support our hypothesis that acceleration of T1D in NOD.HHD mice can be attributed, at least in part, to addition of HLA-A2.1-restricted CD8 T cell responses to those elicited by murine MHC class I molecules. Screening of the peptide library using islet-infiltrating T cells from NOD. 2m null.hhd mice (Fig. 3, B and C) detected multiple positive peptide mixtures (229, 266, and ). As mentioned, each peptide mixture consists of four sequences having a common C terminus and varying in length from 8 to 11 residues (Fig. 4A). In mixtures 266 and 338, the 9-mer peptides agreed most closely with the peptide-binding motif described for HLA-A2.1 (i.e., most commonly 9-mers having L or M at position 2 and V or L at position 9) (47). Thus, we predicted that the 9-mer peptides mi- GRP and migrp were responsible for the activity observed with mixtures 266 and 338, respectively. In the case of peptide mixture 229, the 9-mer peptide did not have a preferred residue at position 2. To determine the antigenic target in this mixture, we synthesized all four of the peptides contained within it. T cell reactivity to the 9-mer peptides from mixtures 266 and 338, and to all four peptides from mixture 229, was then assayed (Fig. 4B). Each graph shows a separate experiment using isletinfiltrating T cells cultured from individual 12-wk-old female NOD. 2m null.hhd mice. In these experiments, one mouse showed a response to peptide mixtures 229 and 266, while the other showed a response to mixtures 266 and 338. The results clearly indicated the 9-mer peptides from mixtures 266 and 338 were antigenic. In the case of mixture 229, the 9-mer peptide was the minimal epitope. Recognition of the 10- and 11-mer peptides from this mixture suggests a proteolytic event generated the 9 mer during the assay period. Similarly, the response observed to mixtures 339 and 340 in the original screening likely reflects generation of the migrp mer peptide from longer sequences in each mixture. Table II. Cultured islet-infiltrating T cells from NOD. 2m null.hhd mice show cytotoxic activity against HLA-A2.1-positive human pancreatic islets a E:T Ratio HLA-A*0201/A*68 human islets % Specific Cytotoxicity (% SEM) HLA-A*11/A*13 human islets NOD-scid.HHD islets NOD.Rag1 null islets 50:1 b c d :1 e f g a Data represent the percentage of specific cytotoxicity exerted against the indicated target islets by cultured islet-infiltrating T cells from NOD. 2m null.hhd mice. T cells from five 12- to 14-wk-old nondiabetic mice were pooled and used as effectors. 51 Cr release from islets cultured in medium alone (i.e., spontaneous release) for each target was measured in nine independent wells to calculate specific cytotoxicity as described in Materials and Methods. Spontaneous release values were as follows: HLA-A*0201/A*68 human islets, 9 0.4%; HLA-A*11/A*13 human islets, %; NOD-scid.HHD islets, %; NOD.Rag1 null islets, %. b Six independent wells were set up to calculate average and SEM. c Differed significantly from cytotoxicity against HLA-A*11/A*13 islets tested at the same E:T ratio ( p ). d Differed significantly from cytotoxicity against NOD.Rag1 null islets tested at the same E:T ratio ( p ). e Three independent wells were set up to calculate average and SEM. f Differed significantly from cytotoxicity against HLA-A*11/A*13 islets tested at the same E:T ratio ( p ). g Differed significantly from cytotoxicity against NOD.Rag1 null islets tested at the same E:T ratio ( p ).

7 3262 HLA-A*0201 MEDIATES DIABETES DEVELOPMENT FIGURE 3. Cultured islet-infiltrating T cells from NOD. 2m null.hhd mice recognize murine IGRP peptides. A, COS-7 cells were transiently transfected with varying concentrations of a murine IGRP cdna expression construct and 1 g/ml of an HHD (F), H-2D b ( ), or H-2K d ( ) cdna expression construct. Cultured islet-infiltrating T cells from 12-wk-old female NOD. 2m null.hhd mice were added, and T cell response was measured as IFN- release by ELISA. T cells did not respond to COS-7 cells cotransfected with the murine IGRP expression construct and pcdna3.1, nor did they respond to COS-7 cells cotransfected with the HHD expression construct and pcdna3.1 (data not shown). B, Islets from four 10-wk-old female NOD. 2m null.hhd mice; C, four 11-wk-old female NOD. 2m null.hhd mice; or D, five 12-wk-old female NOD.HHD mice were pooled to obtain sufficient numbers of T cells to screen the murine IGRP peptide library. Islets were cultured in the presence of IL-2, and islet-infiltrating T cells were harvested after 7 days. For the experiment depicted in C, CD8 T cells were purified from the culture. Human T2 cells were used as APC and T cell response was measured by IFN- ELISPOT assay. To confirm the three newly identified IGRP peptides were indeed recognized in the context of HLA-A2.1, we assayed them by ELISPOT using T2 cells (HLA-A2.1) or RMA-S/K d cells (H-2D b, H-2K b, and H-2K d ) as APC (Fig. 4C). APC were washed twice after incubation with the peptides to minimize the quantity of free peptide available for binding to HLA-A2.1 molecules on the responding T cells and subsequent presentation independent of the APC. Clearly, T cell responses were much better when T2 cells were used as APC, though some response likely due to presentation by the T cells themselves was noted for migrp and migrp These results confirmed that recognition of the IGRP peptides occurs in an HLA-A2.1-restricted manner. Individual NOD. 2m null.hhd mice exhibit distinct patterns of CD8 T cell reactivity to IGRP peptides As shown in Fig. 3, B and C, and Fig. 4B, islet-infiltrating T cells exhibited different patterns of recognition to the three HLA-A2.1- binding IGRP peptides. To investigate this in more detail, we independently propagated T cells from the islets of 16 nondiabetic 12- to 13-wk-old female NOD. 2m null.hhd mice. Responses to the three peptides were measured by IFN- ELISPOT assay (Fig. 5A). As expected, the response to each IGRP epitope varied among mice; however, all exhibited a T cell response to at least one of the peptides. When the data are summarized as in Fig. 5B, it is clear that peptide is the immunodominant HLA-A2.1-binding epitope of migrp. Characterization of the human counterparts of HLA-A2.1- binding murine IGRP peptides Up to nearly one-tenth of all CD8 T cells propagated from the islets of NOD. 2m null.hhd mice are IGRP-reactive. This robust response to IGRP suggested recognition of IGRP peptides may have contributed to the cytotoxicity against HLA-A2.1-positive human islet cells demonstrated in Table II. To investigate this possibility, we compared the sequences of the antigenic migrp peptides with the corresponding regions of the human protein (Fig. 6A). IGRP was identical in both the murine and human molecules. The sequences of IGRP and IGRP differed in two amino acids each. Next, we compared the ability of the peptides to bind to HLA-A2.1 using an MHC stabilization assay (Fig. 6B). All three murine IGRP peptides demonstrated HLA-A2.1 binding, as expected. higrp also bound well to HLA-A2.1. However, we were unable to detect any binding of higrp , perhaps due to the change of the anchor residue at position 2 from L to F (47). We then tested the ability of islet-infiltrating T cell cultures containing cells that recognized migrp or migrp to also recognize the corresponding higrp peptides. Probably due to its inability to bind HLA-A2.1, higrp was not recognized by T cells responding to the murine sequence (data not shown). In contrast, three of the eight T cell cultures derived from 12-wk-old NOD. 2m null.hhd mice that exhibited reactivity to migrp also clearly contained effectors capable of recognizing the human version of this peptide (Fig. 6C). There were no examples of cultures having reactivity to higrp , but not its murine counterpart. These collective results suggest that a subset of T cells in certain cultures that recognize migrp are cross-reactive against the corresponding human peptide. higrp and higrp (conserved between human and mouse) bind HLA-A2.1. Bulk T cell populations from NOD. 2m null.hhd mice recognize both human HLA-A2.1-positive islets and these peptides. Taken together, these results suggest higrp and higrp may be presented by human HLA-A2.1-positive islets. Recognition of these IGRP peptides may account, at least in part, for the cytotoxic activity of CD8 T cells from NOD. 2m null.hhd mice against human HLA-A2.1- positive islets. Formal testing of this hypothesis will require the future derivation of clonal T cell populations specific for each of the IGRP peptides.

8 The Journal of Immunology 3263 FIGURE 4. Identification of three antigenic HLA-A2.1-binding peptides from murine IGRP. A, Schematic of three positive peptide mixtures (229, 266, and 338) from the murine IGRP peptide library. Bold letters indicate potential anchor residues based on the HLA-A2.1 peptide-binding motif (47). B, Individual peptides were synthesized, and their T cell stimulation ability ( ) was compared with the original peptide mixtures (f). Each graph shows a separate experiment using islet-infiltrating T cells cultured from individual 12-wk-old female NOD. 2m null.hhd mice and T2 cells as APC. C, T2 cells (HLA-A2.1) or RMA-S/K d cells (H-2D b, H-2K b, and H-2K d ) were used as APC to confirm the HLA-A2.1 restriction of the IGRP-reactive T cells. FIGURE 5. Individual NOD. 2m null.hhd mice exhibit distinct patterns of CD8 T cell reactivity to three HLA-A2.1-binding murine IGRP peptides. A, Islets were isolated from to 13-wk-old female NOD. m null.hhd mice and cultured individually in the presence of IL-2 for 7 days. Response of the resulting T cells to the three peptides was measured by IFN- ELISPOT assay using T2 cells as APC. Although some selection may occur during the culture period, we find that the rank order of prevalence of different T cell populations remains stable from days 0 to 7 of culture (our unpublished observations). Furthermore, islet preparations that are split in half on day 0 and analyzed independently on day 7 reveal identical rank orders of prevalence (our unpublished observations). B, Summary of quantification of ELISPOT assays performed on cultured islet-infiltrating T cells from the sixteen 12- to 13-wkold female NOD. 2m null.hhd mice depicted in A. Discussion T1D is clearly associated with specific class II MHC alleles (1, 2). However, association studies have revealed that particular class I MHC alleles, including HLA-A2.1, likely represent additional risk factors for T1D development (15 21). The frequency of HLA-A2.1 in Finnish children carrying high-risk class II MHC alleles who develop T1D is higher than in those remaining disease-free (15). A U.S.-based study indicated the HLA-A1/HLA-A2 genotype was associated with increased risk for developing islet autoimmunity (17). Also, among an independent group of U.S. children having high-risk class II alleles, HLA-A2 was found to be associated with development of autoantibodies and progression to T1D before 5 years of age (17). Thus, HLA-A2 may confer an increased risk for the development of anti-islet autoimmunity in early life. Although these epidemiological studies suggested HLA-A2.1-restricted CD8 T cells might contribute to T1D pathogenesis, disease development by NOD. 2m null.hhd mice (Fig. 2C) provides critical experimental evidence that this is indeed the case. Individual NOD mice exhibit unique patterns of CD8 T cell reactivity to the three islet Ags IGRP, insulin, and DMK (46). Even within age-matched groups, individuals exhibit variability in the relative proportions of the three Ag-specific T cell populations in islet infiltrates. In this study, we observed similar diverse responses to different epitopes within the IGRP protein (Fig. 5), though peptide was immunodominant overall. These results further confirm that individual NOD mice exhibit unique patterns of CD8 T cell reactivity to cell Ags. As previously proposed (46), these data indicate several antigenic specificities contribute to cell destruction, and in any individual mouse the autoimmune response is dominated by T cells recognizing a unique subset of antigenic epitopes. IGRP is an important target of pathogenic CD8 T cells in NOD mice (25, 26, 46). An examination of islet infiltrates in individual NOD mice revealed T cells responding to IGRP /H-2K d are present in nearly all mice examined, including those as young as 5 wk of age (46). Recently, additional IGRP epitopes targeted by CD8 T cells in standard NOD mice were reported (26). We now show that IGRP antigenicity is not limited to peptides recognized by T cells restricted to murine MHC class I molecules, but extends also to those seen in the context of the human MHC class I molecule HLA-A2.1 (Fig. 3, B and C). Importantly, HLA-A2.1-restricted T cell responses to IGRP peptides can be detected in NOD. 2m null.hhd mice as young as 5 wk of age (data not shown). Our results provide candidate peptides that could allow the potential importance of IGRP in human T1D to be tested in future clinical studies. Of course, it is possible such future investigations could indicate that the candidate IGRP peptides identified in the current study are not targeted as frequently by pathogenic CD8 T cells in HLA-A2.1-expressing T1D patients as is the case in NOD. 2m null.hhd mice. Thus, we have also begun to explore HLA-A2.1-restricted T cell responses to other cell Ags in

9 3264 HLA-A*0201 MEDIATES DIABETES DEVELOPMENT FIGURE 6. Characterization of the human counterparts of the HLA- A2.1-binding murine IGRP peptides. A, Schematic comparison of three HLA-A2.1-binding murine IGRP peptides with the corresponding peptides from human IGRP. Bold and underlined letters indicate amino acid differences between murine and human IGRP peptides. B, HLA-A2.1 stabilization assay for the three HLA-A2.1-binding murine IGRP peptides and their human counterparts. T2 cells were pulsed with the indicated peptides, stained with FITC-conjugated anti-hla-a2 mab BB7.2, and analyzed by flow cytometry. Each bar represents fluorescence index calculated as described in Materials and Methods. Flu-MP is an HLA-A2.1-binding influenza matrix protein (MP) peptide well characterized as a high-affinity HLA-A2.1 binder (37) and used as a positive control. C, migrp and higrp were compared regarding their ability to stimulate cultured islet-infiltrating T cells from NOD. 2m null.hhd mice using IFN- ELISPOT assay. Islets were isolated from eight 12-wk-old female NOD. 2m null.hhd mice and cultured individually in the presence of IL-2 for 7 days. NOD. 2m null.hhd mice. For example, our preliminary results indicate that multiple preproinsulin peptides are recognized by HLA- A2.1-restricted islet-infiltrating T cells in NOD. 2m null.hhd mice (T. Yamada, T. Takaki, M. Marron, D. Serreze, and T. DiLorenzo unpublished observations). These findings are consistent with recent reports suggesting the importance of insulin as a target autoantigen in the development of T1D in both NOD mice (48) and humans (49). Knowledge of the target Ags of CD8 T cells in human T1D patients is extremely limited. A peptide derived from islet amyloid polypeptide precursor protein (50) and several peptides derived from proinsulin (51) are the only ones reported to elicit a response by CD8 T cells from patients without prior in vitro stimulation. Continued characterization of the Ags targeted by MHC class I- restricted T cells in human T1D patients is clearly necessary. Recently, van Endert and coworkers (52) identified peptides from human proinsulin that are immunogenic in HHD-expressing C57BL/6 mice, and these will undoubtedly be examined in the future for their relevance to human T1D. A limitation of human studies is that islet-infiltrating T cells are difficult to obtain. In contrast, application of the HHD transgenic strategy to the autoimmune-prone NOD mouse allows ready access to diabetogenic HLA-A2.1-restricted T cells and here has permitted identification of some of the cell peptides they recognize. We believe this to be the first case in which HHD-expressing mice were used to identify antigenic peptides bound to human MHC class I molecules that are targeted in a natural autoimmune disease process. Importantly, T cells isolated from NOD. 2m null.hhd mice can recognize human HLA-A2.1-positive islets (Table II). This indicates that murine and human HLA-A2.1-positive islets present common disease-relevant peptides. Multiple pathogen and tumor Ags recognized by HLA-A2.1-restricted T cells in HHD mice have been shown to be targets for HLA-A2.1-restricted T cells in patients (23, 53 55). Thus, human counterparts of the antigenic cell peptides identified through the use of NOD. 2m null.hhd mice are prime candidates for exploration in future clinical studies as targets of pathogenic T cells in T1D patients. Acknowledgments We thank the staff of The Jackson Laboratory Scientific Services Microchemistry and Microinjection core facilities for assistance in preparation of the HHD construct and microinjection of NOD zygotes, and the staff of the Flow Cytometry core facility. Disclosures T. P. DiLorenzo has equity and other significant financial interests in and is a proposed consultant to a company that may have a commercial interest in the results of this research and technology. The authors have no other conflicting financial interests. References 1. Serreze, D. V., and E. H. Leiter Genes and cellular requirements for autoimmune diabetes susceptibility in nonobese diabetic mice. Curr. Dir. Autoimmun. 4: Todd, J. A., and L. S. Wicker Genetic protection from the inflammatory disease type 1 diabetes in humans and animal models. Immunity 15: Serreze, D. V., E. H. Leiter, G. J. Christianson, D. Greiner, and D. C. Roopenian Major histocompatibility complex class I-deficient NOD-B2m null mice are diabetes and insulitis resistant. Diabetes 43: Sumida, T., M. Furukawa, A. Sakamoto, T. Namekawa, T. Maeda, M. Zijlstra, I. Iwamoto, T. Koike, S. Yoshida, H. Tomioka, and M. Taniguchi Prevention of insulitis and diabetes in 2 -microglobulin-deficient non-obese diabetic mice. Int. Immunol. 6: Wicker, L. S., E. H. Leiter, J. A. Todd, R. J. Renjilian, E. Peterson, P. A. Fischer, P. L. Podolin, M. Zijlstra, R. Jaenisch, and L. B. Peterson microglobulin-deficient NOD mice do not develop insulitis or diabetes. Diabetes 43: Katz, J., C. Benoist, and D. Mathis Major histocompatibility complex class I molecules are required for the development of insulitis in non-obese diabetic mice. Eur. J. Immunol. 23: Christianson, S. W., L. D. Shultz, and E. H. Leiter Adoptive transfer of diabetes into immunodeficient NOD-scid/scid mice: relative contributions of CD4 and CD8 T lymphocytes from diabetic versus prediabetic NOD.NON- Thy 1 a donors. Diabetes 42: DiLorenzo, T. P., R. T. Graser, T. Ono, G. J. Christianson, H. D. Chapman, D. C. Roopenian, S. G. Nathenson, and D. V. Serreze Major histocompatibility complex class I-restricted T cells are required for all but the end stages of diabetes development in nonobese diabetic mice and use a prevalent T cell receptor chain gene rearrangement. Proc. Natl. Acad. Sci. USA 95: Hattori, M., E. Yamato, N. Itoh, H. Senpuku, T. Fujisawa, M. Yoshino, M. Fukuda, E. Matsumoto, T. Toyonaga, I. Nakagawa, et al Cutting edge: homologous recombination of the MHC class I K region defines new MHClinked diabetogenic susceptibility gene(s) in nonobese diabetic mice. J. Immunol. 163: Pomerleau, D. P., R. J. Bagley, D. V. Serreze, C. E. Mathews, and E. H. Leiter Major histocompatibility complex-linked diabetes susceptibility in NOD/Lt mice: subcongenic analysis localizes a component of Idd16 at the H2-D end of the diabetogenic H2 g7 complex. Diabetes 54: Inoue, K., H. Ikegami, T. Fujisawa, S. Noso, K. Nojima, N. Babaya, M. Itoi-Babaya, S. Makimo, and T. Ogihara Allelic variation in class I K gene as candidate for a second component of MHC-linked susceptibility to type 1 diabetes in non-obese diabetic mice. Diabetologia 47: