Immunosubunits Protect against the Development of CD8 T Cell-Mediated Autoimmune Diseases

This information is current as Dietmar M. W. Zaiss, Cornelis P. J. Bekker, Andrea Gröne, of September 29, 2021. Benedicte A. Lie and Alice J. A. M. Sijts J Immunol published online 29 July 2011 http://www.jimmunol.org/content/early/2011/07/29/jimmun ol.1101003 Downloaded from

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision http://www.jimmunol.org/

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: by guest on September 29, 2021 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 © 2011 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published July 29, 2011, doi:10.4049/jimmunol.1101003 The Journal of Immunology

Proteasome Immunosubunits Protect against the Development of CD8 T Cell-Mediated Autoimmune Diseases

Dietmar M. W. Zaiss,* Cornelis P. J. Bekker,* Andrea Gro¨ne,† Benedicte A. Lie,‡ and Alice J. A. M. Sijts*

Exposure of cells to inflammatory induces the expression of three proteasome immunosubunits, two of which are encoded in the MHC class II region. The induced subunits replace their constitutive homologs in newly formed “so-called” immunoproteasomes. Immunosubunit incorporation enhances the proteasome’s proteolytic activity and modifies the proteasome’s cleavage-site preferences, which improves the generation of many MHC class I-presented peptides and shapes the fine specificity of pathogen-specific CD8 T cell responses. In this article, we report on a second effect of formation on CD8 T cell responses. We show that mice deficient for the immunosubunits b5i/low molecular mass polypeptide (LMP7) and b2i/

multicatalytic endopeptidase complex-like–1 develop early-stage multiorgan autoimmunity following irradiation and bone mar- Downloaded from row transplantation. Disease symptoms are caused by CD8 T cells and are transferable into immunosubunit-deficient, RAG1- deficient mice. Moreover, using the human Type 1 Diabetes Genetics Consortium MHC dataset, we identified two single nucleotide polymorphisms within the b5i/LMP7-encoding sequences, which were in strong linkage disequilibrium, as independent genetic risk factors for type 1 diabetes development in humans. Strikingly, these single nucleotide polymorphisms significantly enhanced the risk conferred by HLA haplotypes that were previously shown to predispose for type 1 diabetes. These data

suggested that inflammation-induced immunosubunit expression in peripheral tissues constitutes a mechanism that prevents http://www.jimmunol.org/ the development of CD8 T cell-mediated autoimmune diseases. The Journal of Immunology, 2011, 187: 000–000.

D8 T cells play an important role in immune protection b2i/ multicatalytic endopeptidase complex-like–1 (MECL-1), and against intracellular pathogens and tumor growth; how- b5i/LMP7 (2, 3), resulting in the formation of immunoprotea- C ever, they also can be mediators of . somes. CD8 T cells recognize short peptide epitopes that are presented on The three immunosubunits have coevolved with the adaptive the cell surface by MHC class I (MHC-I) molecules. These peptides immune system, and two of these, b1i/LMP2 and b5i/LMP7, are usually are generated upon degradation of intracellular by encoded in the MHC class II (MHC-II) region, suggesting an by guest on September 29, 2021 an abundant protease complex, the proteasome (1). The catalytic important role in immune responses. Indeed, the cleavage site activity of is exerted by three subunits, b1, b2, and preferences of immunoproteasomes differ from those of consti- b5, which are constitutively expressed in all cells. In hemato- tutive proteasomes; consequently, the repertoire of peptides pro- poietic cells and cells exposed to inflammatory cytokines, such as duced by immunoproteasomes differs from that produced by TNF-a and IFN-g, these subunits are exchanged for three - constitutive proteasome complexes (4, 5). These qualitative dif- inducible homologs, b1i/low molecular mass polypeptide (LMP)2, ferences in peptide generation are reflected in the immunodomi- nance hierarchy of pathogen-specific CD8 T cell responses, thus several epitopes that are dominant targets of the pathogen- *Division of Immunology, Faculty of Veterinary Medicine, University of Utrecht, specific CD8 T cell response in infected wild type (wt) mice are † 3584CL Utrecht, The Netherlands; Division of Pathology, Faculty of Veterinary not targeted by CD8 T cells in immunosubunit-deficient mice due Medicine, University of Utrecht, 3584CL Utrecht, The Netherlands; and ‡Institute of Immunology, Oslo University Hospital Rikshospitalet, N-0424 Oslo, Norway to inefficient generation (6, 7). However, CD8 T cell responses Received for publication April 6, 2011. Accepted for publication June 21, 2011. to immunoproteasome-generated epitopes are not essential for clearance of most pathogens (5, 8). Recent studies indicated that, This work was supported by National Institutes of Health Grant AI064576 (to A.J.A.M.S.) and the MEERVOUD program of The Netherlands Organisation for under conditions of IFN-induced oxidative stress, immunopro- Scientific Research. This work used resources provided by the Type 1 Diabetes teasomes could play a more general role in the maintenance of Genetics Consortium, a collaborative clinical study sponsored by the National In- stitute of Diabetes and Digestive Kidney Diseases, National Allergy and Infectious homeostasis (9). Whereas IFN-exposed immunoproteasome- Diseases, National Research Institute, National Institute of Child deficient cells were found to accumulate oxidant-damaged, poly- Health and Human Development, and the Juvenile Diabetes Research Foundation ubiquinated unfolded nascent proteins in aggresome-like induced International and supported by National Institutes of Health Grant U01 DK062418. structures, formation of such structures was only transient in cells Address correspondence and reprint requests to Dr. A. Sijts, Department of Infectious Diseases and Immunology, Faculty of Veterinary Medicine, University of Utrecht, with immunoproteasomes, as a result of the enhanced . Yalelaan 1, 3584CL Utrecht, The Netherlands. E-mail address: [email protected] Recently, it has become more appreciated that, in addition to Abbreviations used in this article: BM, bone marrow; CI, confidence interval; IDDM, CD4 T cells, the CD8 T cell subset plays an important role in tissue insulin-dependent diabetes mellitus; IGRP, islet-specific glucose-6-phosphatase cat- destruction during the progression of autoimmune disease (10–13). alytic subunit related protein; LD, linkage disequilibrium; LMP, low molecular mass polypeptide; MECL-1, multicatalytic endopeptidase complex-like–1; MHC-I, MHC In this article, we show that mice, gene-deficient for b2i/MECL-1 class I; MHC-II, MHC class II; o/n, overnight; OR, odds ratio; SNP, single nucleotide and b5i/LMP7, develop CD8 T cell-mediated, early-stage, mul- polymorphism; T1D, type 1 diabetes; T1DGC, Type-1 Diabetes Genetics Consor- titissue autoimmune syndrome after irradiation and bone marrow tium; UTR, untranslated region; wt, wild type. (BM) reconstitution. These data indicate that one important Copyright Ó 2011 by The American Association of Immunologists, Inc. 0022-1767/11/$16.00 function of the immunosubunits, as well as their inducible

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1101003 2 IMMUNOPROTEASOMES PROTECT AGAINST AUTOIMMUNE DISEASES expression in inflamed tissues, is the prevention of CD8 T cell- coding of DRB1, DQA1, and DQB1 haplotypes, was described previously mediated autoimmune reactions. (15). IFN-g–ELISPOT Materials and Methods Ninety-six–well ELISPOT plates (Milipore, Billerica, MA) were coated with 2 mg/ml AN18 (anti-mouse IFN-g) in 100 ml PBS for $2 h at room Mice temperature. Wells were then washed and blocked twice with RPMI 1640 3 6 C57BL/6 (B6), B6 RAG1-, and b5i/LMP7 and b2i/MECL-1 gene-deficient medium. Splenocyte dilutions, starting with 1 10 cells/well, were in- mice (14), backcrossed .10 times onto the B6 background, were maintained cubated for $4 h in the presence or absence of 2 mg/ml synthetic peptide by in-house breeding under specific pathogen-free conditions. RAG12/2 epitopes in 100 ml IMDM medium. Thereafter, the plates were washed mice were crossed with b5i/LMP7 and b2i/MECL-1 gene-deficient mice to with PBS plus 0.01% Tween 20, and IFN-g was detected by incuba- generate b5i/LMP7 and RAG1 gene-deficient mice. PSMB8 deficiency was tion with 2 mg/ml biotinylated XMG1.2, followed by 1 mg/ml alkaline detected by PCR, using the oligonucleotides 59-GGACCAGGACTTTAC- phosphatase-conjugated streptavidin (Jackson Immunobiology, Bar Har- TACGTAGATG-39 on PSMB8 or 59-CCGACGGCGAGGATCTCGTCG- bor, ME) in PBS plus 0.01% Tween 20 supplemented with 2% BSA. The TGA-39 on the neomycin cassette as forward primer and the PSMB8- assay was developed with the Vector AP substrate kit (Vector Laborato- specific oligonucleotide 59-CTTGTACAGCAGGTCACTGACATCG-39 as ries). The plates were then washed, dried, and analyzed using the A.EL. reverse primer. RAG1 deficiency was verified by flow cytometry, using VIS EliSpot Reader (A.EL.VIS, Hannover, Germany). an anti-TCRb and anti-CD19 Ab. To generate BM chimeric mice, BM MHC-I stability cells flushed from the femurs of donor mice were depleted of mature T lymphocytes by incubation with anti-CD4 (clone GK1.5 and RL174.2) RMA-S cells, lacking the TAP transporter, were incubated with or without and anti-CD8 mAb (clone 3–155) and guinea pig complement. Depletion 60 mM synthetic peptide overnight (o/n), at 37˚C, in protein-free hybrid- efficiency was determined by TCRb staining. Recipient mice were irradiated oma medium (Invitrogen, Carlsbad, CA). Cells were then harvested, Downloaded from with 10 Gy as a single dose from an x-ray irradiator and were injected i.v. washed three times with PBS, and chased at 37˚C in the absence of pep- 7 with 10 BM cells. Efficiency of BM replacement was .95% in all tide. Samples were taken after 0, 1, 2, and 4 h; washed with ice-cold PBS b experiments. Water consumption was followed per cage containing at least with 1% BSA and 0.02% NaN3 (PBS buffer); and stained for H2-K class I two mice and, in most experiments, four mice. Blood glucose levels were expression with biotin-conjugated mouse mAbs (AF6-88.5) and PE- determined in a drop of blood from the lateral tail vein using AccuCheck conjugated streptavidin (eBioscience, San Diego, CA). Cells were ana- (Roche). All animal experiments were performed with age-matched mice lyzed on a FACSCalibur, using Cellquest software. and were approved by the Committee on Animal Experiments of the Uni- versity of Utrecht. Statistical analyses of differences between experimental http://www.jimmunol.org/ groups were tested using a two-tailed Mann–Whitney U test. A p value Results ,0.05 was considered statistically significant. Immunosubunit-deficient mice develop signs of autoimmune disease following irradiation and BM reconstitution Glucose challenge C57BL/6 (B6) b5i/LMP7 and b2i/MECL-1 double gene-deficient After measurement of blood glucose levels, mice were deprived of food for mice have a normal, healthy appearance (7, 14). However, fol- $7 h. Fasting blood glucose levels were determined, and mice were b injected i.p. with 1.5 mg D-glucose (Sigma) per gram body weight. Blood lowing irradiation and BM transplantation, we found that 5i/ glucose levels were monitored. LMP7 and b2i/MECL-1–deficient recipients developed different signs of potential autoimmune disease. The most striking symptoms Vasopressin treatment and measurement of urine osmolality were inflammation of the skin (dermatitis) (Fig. 1A) and dramati- by guest on September 29, 2021 Mice were weighed and placed in metabolic cages with free access to water cally increased water consumption (Fig. 1C), both of which were for 2 h. Access to water was removed; 1 h later, the mice were injected s.c. not observed in nonirradiated controls or wt recipients. Both b5i/ with 1 ng/g body weight vasopressin (DDAVP [Sigma]) dissolved in 200 ml LMP7 and b2i/MECL-1–deficient recipients of wt and of b5i/LMP7 PBS or were left untreated. Urine was collected at regular time intervals, and osmolality was determined by freeze-point reduction. and b2i/MECL-1–deficient BM consumed an enhanced amount of water (Fig. 1B), and water consumption progressively increased T cell depletion and transfer of splenocytes over time, until the mice drank ∼10-fold more water per day than Mice were injected twice i.p. with 200 mg purified CD4-depleting (GK1.5) did untreated mice, which constituted about twice their body or CD8-depleting (3–155) Abs. Efficiency of T cell depletion was checked weight (Fig. 1C). Such an increase in water consumption far in the spleen of one mouse, 2 d after the last treatment, by TCRb staining exceeds the regular variation between individual mice or increases and flow cytometry. For splenocyte transfers, b5i/LMP7 and b2i/MECL-1 over the lifespan of a mouse. Taken together, these findings sug- gene-deficient mice were either irradiated and reconstituted with BM or left untreated. Six weeks after irradiation, when water consumption had gested that b5i/LMP7 and b2i/MECL-1–deficient BM recipients increased, mice were sacrificed. Splenocytes were isolated, depleted of had developed different autoimmune diseases. CD4 T cells using anti-CD4 mAbs and complement, and transferred into RAG12/2 or b5i/LMP72/2 and RAG12/2 mice at a ratio of 1.5:1 donor/ Development of central diabetes insipidus contributes to recipient. increased water consumption by immunosubunit-deficient BM-recipient mice Conditional type 1 diabetes risk analysis Dramatic increases in water consumption often are caused either by Sixteen single nucleotide polymorphisms (SNPs) (rs4713598, rs3763365, diabetes insipidus, which results from a lack of function of anti- rs3763349, rs9357155, rs9276810, rs2071543, rs2071541, rs3198005, rs1057373, rs4711312, rs4713600, rs991760, rs241419, rs17587, rs2071476, diuretic hormone (vasopressin) and often has an autoimmune rs241417) covering the PSMB8 and PSMB9 (including 2 kb up- etiology, or by insulin-dependent diabetes mellitus (IDDM), during stream and downstream) were extracted for the Type-1 Diabetes Genetics which autoreactive T cells destroy the insulin-producing b cells of Consortium (T1DGC) MHC fine-mapping dataset (consisting mostly of the islets of Langerhans in the pancreas. To test the potential type 1 diabetes [T1D] families of white origin) and analyzed in combination development of diabetes insipidus, irradiated BM-transplanted with T1D-susceptibility determinants identified in previous fine-mapping studies (15), (i.e., HLA-B, HLA-C, HLA-DPB1, rs4122198, rs1619379, mice were deprived access to water, and urine production was rs1611133, rs2394186, rs6926530, rs3130695, rs4713468, rs2246626, measured. Although urine excretion by wt recipients rapidly de- rs3132959, rs2076522, rs2187818, rs660895, rs6457617, rs3116999, creased, b5i/LMP7 and b2i/MECL-1–deficient recipients contin- rs9368757, rs421446, rs439121, as well as HLA-DRB1-DQA1-DQB1). ued to excrete urine (Fig. 2A), and the experiments had to be Conditional-association analyses were performed by the main effects . test and by homozygous parent transmission disequilibrium test in terminated when the mice had lost 20% of their body weight. UNPHASED 3.1 and 2.403. Linkage disequilibrium (LD) was calculated Osmolality of the urine produced by b5i/LMP7 and b2i/MECL-1– by HaploView. Preparation of the T1DGC MHC dataset, including re- deficient recipient mice did not significantly increase during water The Journal of Immunology 3

(i.e., 1 ng/g body weight). b5i/LMP7 and b2i/MECL-1–deficient recipient mice injected s.c. showed a gradual decline in urine excretion, accompanied by increasing urine osmolality (Fig. 2), indicating that their kidneys had remained sensitive to vasopres- sin. Taken together, these data indicated that b5i/LMP7 and b2i/ MECL-1–deficient recipient mice suffered from central diabetes insipidus, which is caused by a defect in vasopressin production, possibly due to autoimmune attack of vasopressin-producing cells in the hypothalamus. Immunosubunit-deficient BM recipients develop a latent form of IDDM IDDM is well known to lead to enhanced water consumption and is further characterized by defective glucose uptake from the blood. To test for IDDM development, we assessed the blood glucose levels in irradiated BM-recipient mice. Individual b5i/LMP7 and b2i/MECL-1–deficient BM recipient mice displayed highly var- iable glucose levels; however, on average, they were not signif-

icantly increased over those in wt recipient mice (Fig. 3A). In Downloaded from contrast, following o/n fasting, the blood glucose levels in irra- diated b5i/LMP7 and b2i/MECL-1–deficient recipient mice were FIGURE 1. Irradiated b5i/LMP7 and b2i/MECL-1-deficient mice de- significantly higher than those in irradiated wt recipients (Fig. velop signs of autoimmune disease. Irradiated B6 (wt) and b5i/LMP7 and 3B). In addition, glucose clearance from the blood after challenge b2i/MECL-1–deficient (ko) recipients were reconstituted with T cell-de- was significantly delayed in irradiated b5i/LMP7 and b2i/MECL- pleted BM cells of b5i/LMP7 and b2i/MECL-1–deficient donor mice (A, 1–deficient recipients (Fig. 3C). Analysis of the pancreas showed http://www.jimmunol.org/ C) or with BM of either wt or b5i/LMP7 and b2i/MECL-1–deficient (ko) substantially enlarged pancreatic lymph nodes in b5i/LMP7 and B b b donor mice ( ). 5i/LMP7 and 2i/MECL-1–deficient (ko), but not wt, b2i/MECL-1–deficient recipient mice, a loss of Langerhans islet recipients developed dermatitis (A) and consumed increased amounts of fine structure, and infiltration of mononucleated cells (Fig. 3D), water (B, C). A, Formalin-fixed and paraffin-embedded skin samples were + stained with H&E (original magnification 340). Water consumption per of which some were CD3 (data not shown). Overall, no signif- cage was measured 6 wk after reconstitution (B) or over time following icant differences were observed in the size or frequency of islets BM transplantation (C). Experiments were performed at least seven times, of Langerhans in pancreata of b5i/LMP7 and b2i/MECL-1– with two to four mice per group per cage. deficient and wt recipients. Further histological analysis of b5i/LMP7 and b2i/MECL-1–deficient BM recipients revealed deprivation (Fig. 2B), indicating that they were unable to retain enlarged salivary glands, with substantially enlarged draining by guest on September 29, 2021 and concentrate their urine. lymph nodes (data not shown). Taken together, we inferred that To determine whether the continuing urine production in the b5i/LMP7 and b2i/MECL-1–deficient recipients suffer from absence of water intake resulted from a lack of antidiuretic hor- multiorgan disease. mone (vasopressin) or from vasopressin hyposensitivity, mice were CD8 T cell responses to islet -associated Ags in injected s.c. with physiological amounts of synthetic vasopressin immunosubunit-deficient BM recipients Because the immunosubunits are components of the MHC-I Ag- processing pathway and determine which peptides are displayed on the cell surface, we determined whether CD8 T cells play a role in the autoimmune disorders observed in BM-transplanted b5i/ LMP7 and b2i/MECL-1–deficient mice. The Ags targeted dur- ing IDDM have been well studied, and several epitopes targeted by CD8 T cells during experimental diabetes in H-2b mouse models have been described (16). Therefore, we quantified CD8 T cell responses to these epitopes, as well as an IGRP-derived peptide selected for the presence of an H-2Kb–binding motif (see later discussion), in the spleens of the BM-recipient mice ex vivo, using an IFN-g–ELISPOT assay (Fig. 4A,4B). Significant responses were detected in the spleens of b5i/LMP7 and b2i/ MECL-1–deficient mice but not in wt recipient mice or untreated b5i/LMP7 and b2i/MECL-1–deficient mice (data not shown) against each of the epitopes, including the newly selected IGRP- derived H-2Kb binder (Fig. 4B), although there was some variation in responses between individual mice. Remarkably, this epitope FIGURE 2. Development of diabetes insipidus in b5i/LMP7 and b2i/ (IGRP 59untranslated region [UTR] aa 7–14: RWISFIGV; Fig. 4C) MECL-1–deficient recipient mice. B6 (wt) and b5i/LMP7 and b2i/MECL- 1–deficient (ko) recipients of ko BM were deprived of water for the in- is encoded by a 32-aa-long polypeptide that is translated from the dicated time intervals. At 60 min, one group of b5i/LMP7 and b2i/MECL- first ATG codon of the IGRP-encoding mRNA sequence. The full- 1–deficient recipients received 1 ng/g body weight vasopressin s.c. length IGRP protein is translated from the fifth ATG codon only. (arrows). Urine production (A) and osmolality of the urine (B)were Restimulation with IGRP 59UTR7–14 also expanded CD8 T cells determined. Experiments were performed twice, with six mice per group. in primary spleen cell cultures of immunosubunit-deficient, but 4 IMMUNOPROTEASOMES PROTECT AGAINST AUTOIMMUNE DISEASES

stabilized the H2-Kb molecules on RMA-S, but to a lower extent than did TRP2180–188 or OVA257–264. During a subsequent chase in b the absence of peptide, H2-K /IGRP 59UTR7–14 complexes were lost from the RMA-S cell surface with a t1/2 of ∼30 min (Fig. 4E). b b In comparison, H2-K /OVA257–264 and H2-K /TRP2180–188 dis- appeared with a t1/2 of 4 and 2 h, respectively. These data in- dicated that IGRP 59UTR7–14, similar to most diabetogenic CD8 T cell epitopes identified in B6 models, binds its presenting MHC- I molecule (H2-Kb) with low affinity. Development of disease in immunosubunit-deficient BM-recipient mice is CD8 T cell dependent and transferable into RAG1-deficient mice that lack immunosubunit expression Because the immunosubunits play a central role in MHC-I Ag processing, and autoreactive CD8 T cells were detected in diseased b5i/LMP7 and b2i/MECL-1-deficient mice, but not in wt recipient mice, we assumed that CD8 T cells played a causative role in the development of disease in BM-transplanted b5i/LMP7 and b2i/

MECL-1-deficient mice. To test this hypothesis, immunosubunit- Downloaded from deficient BM recipients were injected with CD8 or CD4 T cell- depleting Abs 3 wk after irradiation, prior to the onset of enhanced water intake. Whereas CD8 T cell depletion prevented increases in water consumption, the depletion of CD4 T cells had no effect (Fig. 5A). Moreover, CD8 T cell depletion also abolished the delay in

glucose clearance (Fig. 5B) found in nondepleted b5i/LMP7 and http://www.jimmunol.org/ b2i/MECL-1–deficient recipient mice following challenge (Figs. 2C,5B). Thus, CD8 T cells play a causative role in the disorders FIGURE 3. b5i/LMP7 and b2i/MECL-1–deficient recipient mice de- observed in b5i/LMP7 and b2i/MECL-1–deficient recipient mice. velop a latent form of IDDM. Blood glucose levels of B6 (wt) and b5i/ The notion that only the b5i/LMP7 and b2i/MECL-1–deficient LMP7 and b2i/MECL-1–deficient (ko) BM recipients were determined 6 wks after reconstitution with ko BM. A, Steady-state levels measured in recipient mice, reconstituted with either wt or b5i/LMP7 and b2i/ the afternoon. B, Levels after o/n fasting. C, Levels 30 min after glu- MECL-1–deficient BM, developed autoimmune disease (Fig. 1A), cose challenge. *p , 0.05 considered significant. D, Eight weeks after in contrast to the wt recipients, led us to infer that not the T cell reconstitution with ko BM, B6 (wt) and b5i/LMP7 and b2i/MECL-1– repertoire, but the absence of immunosubunit expression in the deficient (ko) recipients were sacrificed. Formalin-fixed and paraffin- targeted tissues was a causative factor in the observed disorders. by guest on September 29, 2021 embedded pancreata samples were stained with H&E (original magnifi- To test whether the absence of b5i/LMP7 and b2i/MECL-1 in the cation 3200). Experiments in A–C were performed five times, with three peripheral tissues of BM chimeras supports the development of to five mice per group. Experiment in D was performed twice, with six autoimmune disease, RAG-1–deficient mice were backcrossed mice per group. onto an b5i/LMP7-deficient background. CD4 T cell-depleted spleen cells of diabetic b5i/LMP7 and b2i/MECL-1–deficient BM recipients were adoptively transferred into these or control not wt, recipient mice, and analysis of the congenic markers RAG-1–deficient mice. At day 28 after transfer (Fig. 5C), water CD45.1 (present on cells derived from the BM donor mice) and consumption by recipient b5i/LMP7- and RAG-1–deficient mice CD45.2 (present on cells of the immunosubunit-deficient recipient had significantly increased, up to ∼10 ml/mouse/d, compared mice) revealed that the IGRP 59UTR7–14-reactive CD8 T cells with untreated, age-matched mice that consumed ∼5 ml/mouse/d. derived from the BM donors (data not shown). Water consumption by recipient control RAG-1–deficient mice Taken together, we showed the presence of islet Ag-reactive was only slightly elevated, to ∼6 ml/mouse/d, and transfer of CD4 CD8 T cell responses and identified a novel IGRP-derived epi- T cell-depleted spleen cells of untreated b5i/LMP7 and b2i/ tope that is targeted in diseased BM-transplanted b5i/LMP7 and MECL-1–deficient controls did not enhance the water consump- b2i/MECL-1–deficient B6 mice. The presence of islet Ag-specific tion of b5i/LMP7 and RAG1–deficient mice (Fig. 5C). CD8 T cells in the spleens of immunosubunit-deficient, but not Taken together, these data indicated that CD8 T cells are the of wt, recipients further supports our conclusion that the former main mediators of disease in the b5i/LMP7 and b2i/MECL-1– mice develop a CD8 T cell-mediated latent form of IDDM fol- deficient BM recipients and that a lack of immunosubunit ex- lowing irradiation and BM reconstitution. pression in the target tissue plays a causative role in the de- velopment of autoimmune disease. The newly identified IGRP 59UTR7–14 epitope binds MHC-I H2-Kb molecules with low affinity Haplotype-dependent association between b5i/LMP7 allelic Autoreactive CD8 T cells often recognize self-peptides that bind SNPs and T1D development in a human population their presenting MHC-I molecule with relatively low affinity (16). The data obtained so far indicated that immunosubunit expression To determine the relative affinity of IGRP 59UTR7–14 for MHC-I, provides protection against the development of autoimmune dis- we cultured the T cell line RMA-S, which lacks the TAP and, ease in mice. In humans, the T1DGC dataset, which contains data therefore, expresses “empty” MHC-I molecules on its cell surface, on 2957 SNPs across the MHC region for 2321 T1D families, has o/n with synthetic IGRP 59UTR7–14 or with the control peptides allowed the identification of multiple new genetic risk factors for OVA257–264 or TRP2180–188 (a self-peptide derived from an me- T1D development, including several MHC-I alleles (15). Thus, to lanoma-associated Ag). As shown in Fig. 4D, IGRP 59UTR7–14 test whether immunosubunits play a role in the development of The Journal of Immunology 5

0.99 and r2 = +0.86, likely pointing toward the same causal var- iant. In contrast, low LD with SNPs previously identified to be associated with T1D were observed (r2 , 0.23; with the exception of rs3763365 and rs3132959, r2 = +0.54); hence, LD is unlikely to explain the associations seen for the PSMB8 SNPs. In previous analyses, the HLA-B alleles HLA-B8, -B18, and -B39 showed a strong association with T1D (10, 15). Importantly, ex- ploring the T1D association of the PSMB8 SNPs on extended haplotypes, we found these associations to be particularly pro- nounced in combinations with the HLA-B8 and -B18 alleles (Table I). Thus, on the DRB1*03-DQA1*0501-DQB1*0201 haplotype, the PSMB8 SNPs rs3763365 and rs9276810 increased the disease associations conferred by HLA-B8 to odds ratios (ORs) of 3.02 (95% confidence interval [CI], 1.9–4.7) and 3.1 (95% CI 2.1–4.8), respectively, and by HLA-B18 to ORs of 3.6 (95% CI, 1.9–6.9) and 4.2 (95% CI 2.2–8.1), respectively. For rs9276810, the predispos- ing, minor allele was carried by 41% (133/323) of DR3-B8+ sub- jects with T1D and 18% (37/207) of same-haplotype subjects

without T1D, as well as by 78% (146/186) of DR3-B18+ subjects Downloaded from with T1D and 46% (23/50) of same-haplotype subjects without T1D. For rs3763365, the predisposing allele was carried by 35% (114/328) of DR3-B8+ subjects with T1D and 15% (31/209) of the subjects without T1D, as well as by 75% (140/186) of DR3-B18+ subjects with T1D and 45% (21/47) of same-haplotype subjects

without T1D. Thus, the two SNPs occurred significantly more http://www.jimmunol.org/ frequently in persons with T1D than in healthy persons carrying either the HLA-DRB1*03-DQA1*0501-DQB1*0201-B8 or -B18 haplotypes. The effects of HLA-B39 on disease association could not be determined because of the low frequencies of this class I allele among the T1DGC families. In conclusion, our analyses identified two PSMB8 SNPs, rs3763365, located 1048 bp downstream, and rs9276810, in intron 3, with minor allele frequencies of 41.6% and 44.9% in the mainly FIGURE 4. Induction of CD8 T cell responses against epitopes derived white population selected for the T1DGC MHC fine-mapping data by guest on September 29, 2021 from tissue-associated Ags in b5i/LMP7 and b2i/MECL-1–deficient recip- set, as genetic risk factors for T1D development. ients. IFN-g–ELISPOT analysis of splenocytes of b5i/LMP7 and b2i/ MECL-1–deficient recipients (A) and of splenocytes of B6 (wt) and b5i/ LMP7 and b2i/MECL-1–deficient (ko) recipients (B), 8 wk after re- Discussion constitution with ko BM, in a 4-h assay in the presence or absence of The immunosubunits, first discovered during efforts to obtain synthetic IGRP225–233,IGRP241–249, proInsA2–10 (A)orIGRP59UTR7-14 (B) a complete sequence and gene map of the MHC region, are well peptide. Splenocytes of immunosubunit-deficient recipients (A, B) did not known for being expressed in inflamed tissue, as well as for their secrete IFN-g when incubated in the absence of peptide (no-peptide control), dominant role in shaping the fine specificity of pathogen-specific and no peptide-specific responses were detected in wells containing CD8 T cell responses. To our knowledge, in this article, we splenocytes of wt recipients (A). C, The IGRP-encoding mRNA sequence demonstrate for the first time that immunosubunits also may contains five ATG codons. Translation from the first ATG results in a 32-mer b protect the inflamed tissue against autoimmune CD8 T cell re- polypeptide product, which contains a peptide (underlined) with an H2-K – sponses. Thus, our data showed that proteasome immunosubunit- binding motif. D,H2-Kb expression on RMA-S cells after o/n loading with deficient mice developed CD8 T cell-mediated multiorgan im- peptide. Shown are mean fluorescence levels, corrected for the background levels detected on cells stained with SAV-PE only. E, H2-Kb expression was munity, including a latent form of IDDM, following irradiation determined after chase in the absence of peptide. Depicted are percentages and BM transplantation. These findings, as well as the identifica- of remaining MHC-I complexes (time 0 is 100%). tion of two linked PSMB8–b5i/LMP7 allelic SNPs as haplotype- dependent risk factors for development of human T1D, indicated a direct link between immunoproteasome-mediated proteolysis autoimmune disease in a human population, we decided to analyze or Ag processing and autoimmune disease in mice, as well as this dataset for associations between T1D and SNPs within the in humans. Apparently, these two linked SNPs mark a specific PSMB8 and PSMB9 genes, which encode the immunosubunits b5i/LMP7 allele, which occurs with a minor, but significant, b5i/LMP7 and b1i/LMP2, respectively. Both of these genes are frequency in the human population. In contrast to other b5i/LMP7 located in the MHC-II region. alleles, this b5i/LMP7 allelic form fails to confer disease pro- Disease-association analyses conditional on the main genetic tection, at least when expressed in the context of a predisposing HLA determinant DRB1-DQA1-DQB1 (Table I) showed that 4 HLA haplotype. The SNPs in PSMB8 that were found to be as- SNPs in PSMB8 and 1 in PSMB9, out of 16 extracted SNPs lo- sociated with T1D in this study did not fall in any of the regions cated in or in the vicinity of PSMB8 and PSMB9, were signifi- mapped to harbor T1D-associated variants in previous studies (10, cantly associated with T1D development. After correcting also for 15). This probably results from the fact that the association is most the risk factor HLA-B (Table I), two PSMB8 SNPs (rs3763365 evident in combination with specific HLA-B alleles; thus, the and rs9276810) remained significantly associated with T1D de- conferred predisposition went unnoticed in the overall association velopment. These two PSMB8 SNPs are in strong LD, with D9 = analysis. 6 IMMUNOPROTEASOMES PROTECT AGAINST AUTOIMMUNE DISEASES Downloaded from FIGURE 5. Autoimmune disorders in b5i/LMP7 and b2i/MECL-1–deficient recipient mice are mediated by CD8 T cells and transferable into b5i/LMP7 and RAG1-deficient mice. A and B, b5i/LMP7 and b2i/MECL-1–deficient recipient mice transplanted with ko BM were left untreated (squares) or were injected twice with depleting anti-CD8 (diamonds) or anti-CD4 mAbs (circles) in weeks 3 and 4 after BM reconstitution. Water consumption (A) and blood glucose (B) levels, 20 min after glucose challenge, were determined in B6 recipients (wt control), nondepleted, and T cell-depleted b5i/LMP7 and b2i/ MECL-1–deficient BM recipient mice. Experiments were performed twice, with at least four mice per group. C, CD4 T cell-depleted splenocytes of b5i/ LMP7 and b2i/MECL-1–deficient mice were transferred into b5i/LMP72/2 3 RAG12/2 or control RAG1-deficient mice. The donor mice had either been

irradiated 6 wk prior to transfer with ko BM and, thus, showed the first signs of autoimmune diseases (diabetogenic donor) or had been left untreated and http://www.jimmunol.org/ showed no signs of autoimmune diseases (untreated donor). Mean water consumption 4 wk after transfer and values for individual mice (circles) are depicted. *p , 0.05 considered significant.

Most autoimmune diseases progress through different stages that NOD mouse model (18), showing that CD8 T cells mainly play are accompanied by complex immunological cascades, including a role in the early stages of disease. activation of the innate immune system, chronic inflammation, It is not entirely clear how CD8 T cells, recognizing tissue-specific epitope spreading of MHC-I– and MHC-II–presented epitopes, Ags, are activated in our mouse model. Different studies indicated and recruitment of B cells, with each stage controlled by different that immunosubunit expression in lymphoid cells has diverse effects susceptibility loci (17). We found that IDDM in immunosubunit- and can influence immune responses at different levels (e.g., by by guest on September 29, 2021 deficient recipient mice did not proceed further than a latent form, regulating cytokine production) (19–22). Because both immuno- which was characterized by elevated blood glucose levels fol- subunit-deficient mice reconstituted with wt or with immunosub- lowing o/n fasting and glucose intolerance; this led us to infer that unit-deficient BM developed disease (Fig. 1A), whereas disease was immunoproteasomes may play a critical role, most likely in a very absent in wt recipients, we concluded that immunosubunit expres- early stage of development of autoimmune disease. This conclu- sion in the lymphoid compartments and T cell repertoire play a sion would be in good agreement with previous studies in the minor role at most. Thus, these findings suggest a direct regulatory

Table I. Association analyses of SNPs in the PSMB8 and PSMB9 gene regions conditional on the main genetic HLA determinants in T1D

DRB1-DQA1-DQB1

SNP Gene MAF p Valuea HLA-B B18 B8 rs4713598 PSMB8 0.465 ,0.0006 – – – rs3763365 PSMB8 0.416 ,0.0008 ,0.009 1.2 3 1024 7.7 3 1027 rs3763349 PSMB8 0.398 ,0.03 – – – rs9357155 PSMB8 0.186 – – – – rs9276810 PSMB8 0.449 ,0.0009 ,0.02 1.2 3 1025 5.7 3 1028 rs2071543 PSMB8 0.195 – – – – rs2071541 PSMB8 0.105 – – – – rs3198005 PSMB8 0.031 – – – – rs1057373 PSMB8 0.081 – – – – rs4711312 PSMB9 0.124 – – – – rs4713600 PSMB9 0.441 – – – – rs991760 PSMB9 0.091 – – – – rs241419 PSMB9 0.017 – – – – rs17587 PSMB9 0.286 ,0.002 – – – rs2071476 PSMB9 0.462 – – – – rs241417 PSMB9 0.014 – – – – aThe p values given are the least significant p value of either the main effect test or the homozygous parent transmission disequilibrium test. –, not significant; MAF, minor allele frequency. The Journal of Immunology 7 role of immunoproteasomes in inflamed tissue in deviating CD8 medication for the rest of their lives. So far, it has been assumed T cells from tissue-associated Ags. that such treatments may function mainly to prevent possible We propose that the autoimmune phenotype observed in our graft-versus-host side effects. However, our data suggest that, model develops as a consequence of a wave of apoptotic cell death particularly in predisposed patient groups, the suppression of au- (23, 24), induced by irradiation of the recipient mice. Most likely, toimmune diseases may also be an important aspect encouraging these T cells then expand under the lymphopenic conditions fol- the use of such medication. Such a hypothesis is supported by lowing irradiation, a particularly permissive environment for the a number of cases of de novo development of autoimmune dis- initiation of autoimmunity (25, 26). A potential explanation for orders in patients following BM transplantations (33, 34). the particular susceptibility of immunosubunit-deficient mice to In conclusion, we propose that immunoproteasome formation in whole body irradiation-induced autoimmune CD8 T cell responses inflamed tissues forms a protective mechanism that can prevent the was provided by a recent study by Seifert et al. (9), who showed onset or progression of autoimmune diseases. that immunoproteasome formation in cells exposed to IFN was essential for efficient removal of damaged defective ribosomal Acknowledgments products that accumulated as a consequence of IFN-induced We thank J. Jooles for helpful advice regarding measurements of urine ex- oxidative stress. Immunoproteasome deficiency increased the sen- cretion and osmolality and the use of vasopressin, J. Monaco for providing sitivity of IFN-exposed cells to and enhanced clini- B6 b5i/LMP7 and b2i/MECL-1 gene-deficient mice, and M. Eike for cal scores during experimental autoimmune encephalomyelitis- preparation of the T1DGC MHC dataset. induced inflammation in mice. A potential role for CD8 T cells was not addressed. Disclosures Downloaded from However, altered Ag processing in immunosubunit-deficient The authors have no financial conflicts of interest. tissues, compared with immunoproteasome-sufficient tissues, may provide an alternative or complementary explanation for the References development of CD8 T cell-mediated autoimmune reactions in 1. Yewdell, J. W., E. Reits, and J. Neefjes. 2003. Making sense of mass destruction: recipients lacking immunosubunits. An enhanced sensitivity of quantitating MHC class I antigen presentation. Nat. Rev. Immunol. 3: 952–961. immunosubunit-deficient tissue to autoreactive CD8 T cells is 2. Kloetzel, P. M. 2001. Antigen processing by the proteasome. Nat. Rev. Mol. Cell http://www.jimmunol.org/ suggested by Fig. 5, showing that lymphocytes of diseased mice Biol. 2: 179–187. 3. Sijts, A., D. Zaiss, and P. M. Kloetzel. 2001. The role of the - transfer disease symptoms to b5i/LMP7 and RAG1-deficient mice proteasome pathway in MHC class I antigen processing: implications for vac- but barely to RAG1-deficient mice. The effects of immunosubu- cine design. Curr. Mol. Med. 1: 665–676. nits on Ag processing are well documented (5). In particular, b5i/ 4. Toes, R. E., A. K. Nussbaum, S. Degermann, M. Schirle, N. P. Emmerich, M. Kraft, C. Laplace, A. Zwinderman, T. P. Dick, J. Mu¨ller, et al. 2001. Discrete LMP7-deficient cells display less MHC-I on their cell surface than cleavage motifs of constitutive and immunoproteasomes revealed by quantitative wt cells do (14, 21, 27, 28), with levels restored to those on wt analysis of cleavage products. J. Exp. Med. 194: 1–12. cells when cells are incubated with synthetic MHC-I ligands (data 5. Sijts, E. J., and P. M. Kloetzel. 2011. The role of the proteasome in the gener- ation of MHC class I ligands and immune responses. Cell. Mol. Life Sci. 68: not shown). In other words, in the absence of immunoprotea- 1491–1502. somes, a relatively high percentage of MHC-I molecules associ- 6. Chen, W., C. C. Norbury, Y. Cho, J. W. Yewdell, and J. R. Bennink. 2001. by guest on September 29, 2021 ated with low-affinity binders traffic to the cell surface and dis- Immunoproteasomes shape immunodominance hierarchies of antiviral CD8(+) T cells at the levels of T cell repertoire and presentation of viral antigens. J. Exp. integrate there. Such peptides, mainly exposed in immunosubunit- Med. 193: 1319–1326. deficient tissues, may be the targets of autoreactive CD8 T cell 7. Deol, P., D. M. Zaiss, J. J. Monaco, and A. J. Sijts. 2007. Rates of processing responses. determine the immunogenicity of immunoproteasome-generated epitopes. J. Immunol. 178: 7557–7562. Indeed, different studies suggested that the early autoimmune 8. Nussbaum, A. K., M. P. Rodriguez-Carreno, N. Benning, J. Botten, and response in T1D-prone individuals targets peptides bound to MHC J. L. Whitton. 2005. Immunoproteasome-deficient mice mount largely normal molecules with low affinity (16). Also, in mouse models of ex- CD8+ T cell responses to lymphocytic choriomeningitis virus and DNA vaccination. J. Immunol. 175: 1153–1160. perimental diabetes, except for those in which foreign Ags are 9. Seifert, U., L. P. Bialy, F. Ebstein, D. Bech-Otschir, A. Voigt, F. Schro¨ter, expressed from the rat insulin promoter, most described CD8 T. Prozorovski, N. Lange, J. Steffen, M. Rieger, et al. 2010. Immunoproteasomes T cell targets are peptides that bind MHC-I molecules with low preserve protein homeostasis upon interferon-induced oxidative stress. Cell 142: 613–624. affinity (29, 30). The low affinity of the diabetogenic epitope 10. Nejentsev, S., J. M. Howson, N. M. Walker, J. Szeszko, S. F. Field, H. E. Stevens, IGRP 59UTR7–14, targeted in our model (Fig. 4), further sup- P. Reynolds, M. Hardy, E. King, J. Masters, et al; Wellcome Trust Case Control ports this notion. In a recent study (31), diabetes in RIP-B7 trans- Consortium. 2007. Localization of type 1 diabetes susceptibility to the MHC class I genes HLA-B and HLA-A. Nature 450: 887–892. genic mice (32), induced by immunization with an endoplasmic 11. Friese, M. A., and L. Fugger. 2009. Pathogenic CD8(+) T cells in multiple reticulum-targeted preproinsulin cDNA construct that contained sclerosis. Ann. Neurol. 66: 132–141. a low-affinity H2-Kb–presented diabetogenic CD8 T cell epitope, 12. Friese, M. A., K. B. Jakobsen, L. Friis, R. Etzensperger, M. J. Craner, R. M. McMahon, L. T. Jensen, V. Huygelen, E. Y. Jones, J. I. Bell, and L. Fugger. was prevented by coimmunization with a second cDNA construct 2008. Opposing effects of HLA class I molecules in tuning autoreactive CD8+ encoding a peptide that bound this MHC-I molecule with high T cells in multiple sclerosis. Nat. Med. 14: 1227–1235. affinity. These data demonstrated that outcompetition of diabetes- 13. Pinkse, G. G., O. H. Tysma, C. A. Bergen, M. G. Kester, F. Ossendorp, P. A. van Veelen, B. Keymeulen, D. Pipeleers, J. W. Drijfhout, and B. O. Roep. 2005. associated epitopes by higher-affinity MHC-I binders can prevent Autoreactive CD8 T cells associated with beta cell destruction in type 1 diabetes. autoimmune disease. Conversely, we propose that when the Ag- Proc. Natl. Acad. Sci. USA 102: 18425–18430. processing pathway is disrupted at the level of efficient processing 14. Caudill, C. M., K. Jayarapu, L. Elenich, J. J. Monaco, R. A. Colbert, and T. A. Griffin. 2006. T cells lacking immunoproteasome subunits MECL-1 and of high-affinity MHC-I binders, potential targets of autoimmune LMP7 hyperproliferate in response to polyclonal mitogens. J. Immunol. 176: T cells will not be replaced upon inflammation, and a window of 4075–4082. opportunity for autoreactive CD8 T cells will be created that ul- 15. Eike, M. C., T. Becker, K. Humphreys, M. Olsson, and B. A. Lie. 2009. Con- ditional analyses on the T1DGC MHC dataset: novel associations with type 1 timately may initiate autoimmune diseases. diabetes around HLA-G and confirmation of HLA-B. Genes Immun. 10: 56–67. It is tempting to speculate that a similar situation as we described 16. Di Lorenzo, T. P., M. Peakman, and B. O. Roep. 2007. Translational mini-review in this article for b5i/LMP7 and b2i/MECL-1 gene-deficient mice series on type 1 diabetes: Systematic analysis of T cell epitopes in autoimmune diabetes. Clin. Exp. Immunol. 148: 1–16. after BM transplantation could occur in patients undergoing BM 17. Heap, G. A., and D. A. van Heel. 2009. The genetics of chronic inflammatory transplantation. These patients have to take immunosuppressive diseases. Hum. Mol. Genet. 18(R1): R101–R106. 8 IMMUNOPROTEASOMES PROTECT AGAINST AUTOIMMUNE DISEASES

18. DiLorenzo, T. P., R. T. Graser, T. Ono, G. J. Christianson, H. D. Chapman, 28. de Graaf, N., M. J. van Helden, K. Textoris-Taube, T. Chiba, D. J. Topham, D. C. Roopenian, S. G. Nathenson, and D. V. Serreze. 1998. Major histocompat- P. M. Kloetzel, D. M. Zaiss, and A. J. Sijts. 2011. PA28 and the proteasome ibility complex class I-restricted T cells are required for all but the end stages of immunosubunits play a central and independent role in the production of MHC diabetes development in nonobese diabetic mice and use a prevalent T cell receptor class I-binding peptides in vivo. Eur. J. Immunol. 41: 926–935. alpha chain gene rearrangement. Proc. Natl. Acad. Sci. USA 95: 12538–12543. 29. Han, B., P. Serra, A. Amrani, J. Yamanouchi, A. F. Mare´e, L. Edelstein-Keshet, 19. Hayashi, T., and D. Faustman. 1999. NOD mice are defective in proteasome and P. Santamaria. 2005. Prevention of diabetes by manipulation of anti-IGRP production and activation of NF-kappaB. Mol. Cell. Biol. 19: 8646–8659. autoimmunity: high efficiency of a low-affinity peptide. Nat. Med. 11: 645– 20. Hensley, S. E., D. Zanker, B. P. Dolan, A. David, H. D. Hickman, A. C. Embry, 652. C. N. Skon, K. M. Grebe, T. A. Griffin, W. Chen, et al. 2010. Unexpected role for 30. Karges, W., T. Rajasalu, A. Spyrantis, A. Wieland, B. Boehm, and the immunoproteasome subunit LMP2 in antiviral humoral and innate immune R. Schirmbeck. 2007. The diabetogenic, insulin-specific CD8 T cell response responses. J. Immunol. 184: 4115–4122. primed in the experimental autoimmune diabetes model in RIP-B7.1 mice. Eur. 21. Zaiss, D. M., N. de Graaf, and A. J. Sijts. 2008. The proteasome immunosubunit J. Immunol. 37: 2097–2103. multicatalytic endopeptidase complex-like 1 is a T-cell-intrinsic factor influ- 31. Brosi, H., M. Reiser, T. Rajasalu, A. Spyrantis, F. Oswald, B. O. Boehm, and encing homeostatic expansion. Infect. Immun. 76: 1207–1213. R. Schirmbeck. 2009. Processing in the endoplasmic reticulum generates an 22. Groettrup, M., C. J. Kirk, and M. Basler. 2010. Proteasomes in immune cells: epitope on the insulin A chain that stimulates diabetogenic CD8 T cell responses. more than peptide producers? Nat. Rev. Immunol. 10: 73–78. J. Immunol. 183: 7187–7195. 23. Turley, S., L. Poirot, M. Hattori, C. Benoist, and D. Mathis. 2003. Physiological 32. Harlan, D. M., H. Hengartner, M. L. Huang, Y. H. Kang, R. Abe, beta cell death triggers priming of self-reactive T cells by dendritic cells in a type-1 diabetes model. J. Exp. Med. 198: 1527–1537. R. W. Moreadith, H. Pircher, G. S. Gray, P. S. Ohashi, G. J. Freeman, et al. 1994. 24. Kim, H. S., M. S. Han, K. W. Chung, S. Kim, E. Kim, M. J. Kim, E. Jang, Mice expressing both B7-1 and viral glycoprotein on pancreatic beta cells along H. A. Lee, J. Youn, S. Akira, and M. S. Lee. 2007. Toll-like receptor 2 senses with glycoprotein-specific transgenic T cells develop diabetes due to a break- beta-cell death and contributes to the initiation of autoimmune diabetes. Im- down of T-lymphocyte unresponsiveness. Proc. Natl. Acad. Sci. USA 91: 3137– munity 27: 321–333. 3141. 25. Krupica, T., Jr., T. J. Fry, and C. L. Mackall. 2006. Autoimmunity during lym- 33. Sakamoto, K., T. Imamura, F. Niwa, S. Komori, Y. Ishihara, K. Shiga, M. Ito, phopenia: a two-hit model. Clin. Immunol. 120: 121–128. and H. Hosoi. 2011. Dermatomyositis developed in a recipient of allogeneic 26. Guerau-de-Arellano, M., M. Martinic, C. Benoist, and D. Mathis. 2009. Neonatal BMT; the differentiation of chronic GVHD and autoimmune disease. Bone Downloaded from tolerance revisited: a perinatal window for Aire control of autoimmunity. J. Exp. Marrow Transplant. Med. 206: 1245–1252. 34. Narita, A., H. Muramatsu, Y. Takahashi, H. Sakaguchi, S. Doisaki, N. Nishio, 27. Fehling, H. J., W. Swat, C. Laplace, R. Ku¨hn, K. Rajewsky, U. Mu¨ller, and A. Hama, A. Shimada, M. Ito, and S. Kojima. 2011. Autoimmune-like hepatitis H. von Boehmer. 1994. MHC class I expression in mice lacking the proteasome following unrelated BMT successfully treated with rituximab. Bone Marrow subunit LMP-7. Science 265: 1234–1237. Transplant. http://www.jimmunol.org/ by guest on September 29, 2021