Single-Cell Analysis of CD4 T Cells in Type 1 Diabetes: from Mouse to Man, How to Perform Mechanistic Studies

1886 Diabetes Volume 68, October 2019

Single-Cell Analysis of CD4 T Cells in Type 1 Diabetes: From Mouse to Man, How to Perform Mechanistic Studies

Siddhartha Sharma,1 Jeremy Pettus,2 Michael Gottschalk,3 Brian Abe,1 Peter Gottlieb,4 and Luc Teyton1

Diabetes 2019;68:1886–1891 | https://doi.org/10.2337/dbi18-0064

Type 1 diabetes is the prototypical CD4 T cell–mediated acid at position b57, unlike any other HLA-DR, -DQ, or -DP autoimmune disease. Its genetic linkage to a single poly- molecules, HLA-DQ2 (HLA-DQb0201), and HLA-DQ8 (HLA- morphism at position 57 of the HLA class II DQb chain DQb0302) (Fig. 1). This genetic terrain allows multiple makes it unique to study the molecular link between HLA environmental factors to emerge as disease triggers (5). and disease. However, investigating this relationship has At first glance, an association of a CD4 T cell–mediated been limited by a series of anatomical barriers, the small disease with HLA class II gene products, whose function is size and dispersion of the insulin-producing organ, and to present peptides to CD4 T cells, appears easily explain- the scarcity of appropriate techniques and reagents to able. Over the years, the most obvious, nonexclusive fi interrogate antigen-speci c CD4 T cells both in man and theories have been tested: instability and poor peptide rodent models. Over the past few years, single-cell tech- binding of diabetogenic HLA class II molecules (6), unique nologies, paired with new biostatistical methods, have peptide repertoire of the same molecules (7), T cells fo- changed this landscape. Using these tools, we have cused on the recognition of HLA-DQb57 (8), failed thymic identified the first molecular link between MHC class II selection of autoreactive T cells (9), and abnormal T-cell and the onset of type 1 diabetes. The translation of these – observations to man is within reach using similar binding to autoimmune peptide MHC complexes (10). approaches and the lessons learned from rodent models. While all of those might bear truth and give some level of understanding of what the b57 residue might do, none could formally associate the mutation to a molecular Type 1 diabetes is a CD4 T cell–mediated autoimmune mechanism leading to diabetes. The closest one to explain- PERSPECTIVES IN DIABETES disease that results in the destruction of the pancreatic ing the association of the same mutation with a disease b-cells that produce insulin. Type 1 diabetes is also the was in the context of celiac disease, where the same poster child of autoimmune diseases linked to genetic HLA-DQ molecules are strongly predisposing to onset susceptibility. While more than 40 genes have been de- and also promote a frequent association with type 1 di- scribed in this inherited landscape (1), the association that abetes (11). In this instance, it was shown in transgenic stands out is with the HLA class II locus on chromosome HLA-DQ8 mice, and for some human CD4 T-cell clones, 2 6 with a P value of 10 123, suggesting not only influence that gliadin peptides were recognized by T-cell receptors but causality. This notion is reinforced by the fact that the (TCRs) bearing a negatively charged residue in the first fine mapping of this linkage with the HLA class II region segment of the CDR3b loop (12). Interestingly, most identifies a single polymorphism at position 57 of the native gliadin peptides are glutamine rich and neutral HLA-DQb chain as being responsible for most of the unless deamidated by tissue transglutaminase (13) but association (2,3), while non-HLA loci and genes likely play can still be presented by HLA-DQ2 and -DQ8 molecules, aroleininfluencing the progression to disease onset (4). which usually prefer peptides with negatively charged Only two common haplotypes of HLA-DQ carry a nonaspartic amino acids in their C-termini (7). The acidic residues

1Department of Immunology and Microbiology, The Scripps Research Institute, La Received 16 July 2019 and accepted 21 July 2019 Jolla, CA B.A. is currently afﬁliated with Division of Immunology & Rheumatology, Stanford 2 Division of Endocrinology and Metabolism, University of California, San Diego, San University School of Medicine, Stanford, CA. Diego, CA © 2019 by the American Diabetes Association. Readers may use this article as 3University of California, San Diego Medical Center, San Diego, CA long as the work is properly cited, the use is educational and not for proﬁt, and the 4Department of Pediatrics and Department of Immunology & Microbiology, work is not altered. More information is available at http://www.diabetesjournals University of Colorado School of Medicine, and Barbara Davis Center for Diabetes, .org/content/license. Denver, CO Corresponding author: Luc Teyton, [email protected] diabetes.diabetesjournals.org Sharma and Associates 1887

also the making of T cell–detecting reagents such as MHC tetramers. The results from our structural studies were surprising; while we expected the CDR3b negative charge to sit in proximity to the positive patch of the b57 residue and establish a salt bridge, shifting the TCR over the COOH- terminal part of the p-MHC complex, the TCR was found in a normal diagonal position putting the CDR3b far away from b57 (8). However, biophysical studies demonstrated that the complementation of charges between TCR and p-MHC were operating through Coulombic interactions, a phenomenon that allows surfaces of opposite charges to enhance dehydration and increase on-rates of binding interaction. In any case, this deep knowledge of I-Ag7 and HLA-DQ molecules could not establish a direct link between MHC class II, position b57, and type 1 diabetes. In the absence of — Figure 1 Depiction of the three-dimensional structure of HLA-DQ8, a rodent model and MHC tetramers, the celiac disease the prototypical diabetogenic molecule. In this top view of the molecule, the peptide binding groove is horizontal and limited at observation could not be tested further, while the absence the top by the a helix of the a chain (gray) and at the bottom by the a of antigen-specific reagents was impeding our studies in helix of the b chain (purple); the peptide is in yellow with its 9th mice. In addition, studying type 1 diabetes offers addi- residue represented in spheres. Position b57, also represented in spheres, and colored in green, limits the outside of the P9 pocket tional challenges that were insurmountable for decades where the P9 residue of the peptide is sitting. The nature of the both in mouse and man. The two most challenging were relationship b57-P9 residue and its interpretation by TCRs will drive thesizeoftheorganthatproducesinsulin(,1.5 g of tissue anti–b-cell autoimmunity. Image was generated using pdb 1JK6 in a human) and the asynchrony of the lesions across the from the Protein Data Bank (39). ;1 million islets. These numbers and the low efficiency of the autoimmune process, which takes on average 15% of a life span to reach completion (5 years in humans, occupy the P9 pocket of MHC and compensate for the loss 15 weeks in mice), likely translate to a very small number of the aspartic acid at position b57, which is an integral of anti–b-cell–specific CD4 T cells locally and in circulation. part of the outside wall of this MHC pocket (14) (Fig. 1). This situation is very consequential, making the diagnosis The need for side chains anchoring into pockets for of the preclinical phase of disease extremely difficult and peptide binding as we see for MHC class I and HLA-DR mechanistic studies very challenging. molecules has been lost for HLA-DQ molecules (14,15), As often in science, advancements in technology open allowing more diverse and promiscuous binding. As a con- access to the next level of understanding. About a decade sequence of this mode of binding, peptide repertoire is ago, microfluidic systems allowed the isolation of single much broader for HLA-DQ than for HLA-DR molecules, cells in microchambers that can be used as reaction ves- but the affinities of peptide binding are much lower, often sels. Concomitant with the rise of next-generation in the mid to high micromolar range (12). This biophys- sequencing, this new engineering gave birth to the rapidly ical detail is often overlooked, although it informs us of expanding world of single-cell technologies. We can now two important convergent features of autoimmunity: probe with single-cell resolution genetic differences, dif- first, because there is a threshold to activate T cells, ferential gene expression, genome-wide epigenetic mod- low affinity peptides must be abundant to compensate ifications, and large sets of unique proteins. Most for short binding half-lives; second, homozygosity of the importantly, these approaches have allowed us to inter- susceptibility HLA genes, as often observed in autoim- rogate very small numbers of cells as we expect in biopsies munity, is essential to increase cell surface expression of or circulating blood. As it stands today, single-cell tech- the diabetogenic peptide–MHC complexes. In type 1 di- nologies have allowed us to refine our translational studies abetes, this latter issue is further compounded by the fact from mouse to man and approach mechanistic under- that HLA-DQ2/DQ8 heterozygotes are also at a higher standing of disease, and, most importantly, might open risk of disease due to the expression of transdimers in the possibility of an early preclinical diagnosis and the which the b57 position always lacks the normal aspartic monitoring of treatment in man. acid (16). In an effort to understand how neutral peptides bound to HLA-DQ8 could select TCRs with a negatively A Nonexhaustive Review of Single-Cell Technologies charged residue in their CDR3b loop, we extended the Bulk analysis techniques have been the primary method observation to the mouse model and a neutral peptide utilized thus far to understand autoimmunity in type from hen egg lysozyme in order to gain structural in- 1 diabetes. While these have been invaluable in develop- formation (8). Indeed, the very low affinity of gliadin ing our basic understanding of the disease, they are built peptides for HLA-DQ8 precluded structural studies and on the principle that populations are homogenous and 1888 Single-Cell Analysis of CD4 T Cells Diabetes Volume 68, October 2019 static, and therefore readings of a particular marker can MHC Class II Tetramers and Antigen Tetramers be averaged over the population (17). Unlike bulk tech- Our ability to interrogate specificity of T cells with MHC niques, single-cell techniques are powered to account for tetramers dates from 1996 (27). While relatively successful variability. There are excellent reviews that discuss in for MHC class I and the detection of CD8 T cells in patients detail the application of single-cell approaches to address (28,29), the use of tetramers for isolating CD4 T cells has cellular and biological identity (17–20). All of these been difficult and has not moved yet to the clinic. The main techniques rely on large data sets with computational reasons for this situation are inherent to the MHC class II and statistical analysis. While instrument manufacturers molecules associated with type 1 diabetes, HLA-DQ in provide basic analysis packages, thorough analysis and humans and I-A in mice. This class of molecules, unlike the production of informative figures requires the in- MHC class I or the ancestral MHC class II molecules, volvement of computational biologists. This bottleneck HLA-DR and I-E, do not necessarily bind and select pep- seriously limits the access to these technologies and also tides using anchor side chains in binding pockets, but they creates a landscape where no one way is yet accepted as can accommodate many more peptides by using an anchor- the standard. less mode of binding (15). The consequences of this di- Single-cell genomics are the most used single-cell vergence are a promiscuity of binding, the ability to display techniques; they can interrogate both genomic DNA multiple registers of a single peptide, and a much larger and the transcriptome (exome). These approaches have peptide repertoire. These characteristics are advantageous primarily been used to describe cellular diversity and to fight infection, but they inherently impede an efficient representation in tissues for two main applications: deletion of autoreactive cells. embryogenesis/development (21) and cancer. While When it comes to expressing recombinant HLA-DQ/ sequencing techniques are highly accurate, “depth of I-A molecules for functional and/or structural studies, interrogation” and “coverage” of the genome or tran- low-affinity peptide binding results in poor protein sta- scriptome are the two limiting factors that should be bility and often a mix of different registers, while the kept in mind. These two parameters vary dramatically inefficient MHC a and b chain dimerization further limits based on the method used for isolating single cells (micro- expression. If the latter issue can be addressed by leucine fluidic vs. droplet isolation) and for sequencing (full length zippers forced pairing (30), low peptide affinity, and the vs. 39 counting methods) (22). Very recent technologies display of single registers remain without universal using the principle of fluorescent in situ RNA hybridization answers. As a result, very few good reagents have been in a multiplex and repetitive format on tissues have added produced for mouse and human studies of CD4 T cells in spatial information to exome data (23,24). Similarly, epi- autoimmunity; none have been brought to the clinic. genetic information would nicely complement transcrip- Finally, an appreciable number of patients do not carry tomics analysis, but single-cell epigenomics techniques diabetogenic HLA genes and cannot be interrogated with remain difficult and limited (25). MHC tetramers; therefore, detection of anti–b-cell CD4 In all cases, regardless of the depth of the single-cell T cells using non-HLA tetramer–based technologies must genomics interrogation, profiles of RNA expression must be developed. be compared with the levels of proteins made by the cells The isolation of antigen-specific B cells using tetramer- of interest. ized antigens is a more recent approach that has proven For blood cells and cells dissociated from tissues, this useful in following anti-infectious responses (31,32) but validation step at single-cell resolution is dominated by remains confidential in autoimmunity (33). multiparametric flow cytometry analysis and variants of it In both instances, the tetramer isolation of single cells such as mass cytometry. All are currently limited to 20–50 allows the determination of TCR and B-cell receptor chain parameters, a dimensionality that is largely sufficient to sequences, their pairing, and most importantly their reex- separate immune cell populations for which differential pression in surrogate cells and/or as recombinant soluble markers have been very well described. For tissue sections, molecules for functional and mechanistic studies. automated immunofluorescence techniques with succes- sive rounds of hybridization-photobleaching or imaging Translational Immunology via Single-Cell Methods mass cytometry achieve similar levels of resolution. The ThedebateaboutthevalueoftheNODmousemodelfor immune infiltrate in pancreatic islets from healthy donors, human type 1 diabetes is still current with no signs that patients with recent-onset type 1 diabetes, and patients either of the two sides will compromise anytime soon. In with established diabetes was recently published using our opinion, arguments should always be ranked in bi- these approaches (26). ological order. In this respect, the confounding similar- While these cutting-edge technologies allow the de- ities of the genetics of the disease between mouse and scription of phenomenology at its highest granularity man are at the top, especially the common mutation at and might generate interesting hypotheses, mechanistic position 57 of the b chain of MHC class II. The next immunological studies require one more level of techno- arguments on the list are clinical and pathological: spon- logical advance to address the central theme of immune taneous onset, prolonged preclinical period (15% of a life- recognition: antigen specificity. time), allochronicity of the lesions, b-cell exclusivity, and diabetes.diabetesjournals.org Sharma and Associates 1889 similar comorbidity (thyroid, salivary glands). Therefore, its TCR. Using this approach in NOD mice, we have been we would argue that the great divide between our col- able to not only characterize CD4 T cell populations in leagues is more a methodological issue than an issue with thymus, spleen, pancreatic lymph nodes, and islets, but to the model itself. However, to bridge mouse and man, determine the gene expression signature in each organ methodology is difficult and faces obvious challenges: (35). The main conclusion of this examination was that, first, in the two species the work is usually done at based on 96 “T-cell genes,” we could easily identify the different phases of disease—preclinical in the mouse, residence of each T cell. While peripheral T cells (spleen established disease in man. We rarely keep mice under as well as peripheral lymph nodes) expressed nearly no insulinotherapy to study the post-destruction phase of signaling genes, T cells in the pancreatic lymph nodes the disease, whereas in humans, access to preclinical showed signs of cytokine activation but no stigma of samples is limited to the small TrialNet cohort and newly TCR engagement, and intraislet cells displayed the bona diagnosed cases remain difficult to obtain in large num- fide signature of TCR triggering and activation. This first bers. Beyond these fundamental discrepancies, the tis- layer of examination indicated that the antigen-specific sues that we sample are also dramatically different. Blood process was mostly limited to the islet, not the draining is the only readily available steady source of immune cells lymph node. The second layer of information that was in humans but is challenging to survey in mice because gained was about the nature of the early antigen, insulin. of volume; conversely, internal tissues such as the spleen, MHC tetramers displaying insulin peptides showed that at lymph nodes, and infiltrating cells of the islets are easily 6 weeks of age, about two-thirds of all intraislet CD4 T cells accessible in mice but not in humans. To cap this difficult recognized a single epitope of the B9-23 segment of in- situation, informative samples from recently diagnosed sulin, B12-20, and that at 12 weeks of age this epitope patients, accessible through the Network for Pancreatic dominance was gone. This series of experiments firmly Organ Donors with Diabetes, are extremely rare and established that insulin was the main initiator and driver coveted by many laboratories. Being conscious of these antigen of anti–b-cell autoimmunity (35). Aside from this important issues allowed us to redesign our studies to very important piece of information, our single-cell anal- compare similar stages of disease, sample blood from ysis was capable of supporting two more mechanistic single mice, use identical markers and reagents, and use studies. The first one was about the mode of T-cell similar protocols. Single-cell approaches are remarkably recognition that allows breakage of T-cell tolerance and suited to this effort. that we described earlier as the “P9 switch,” in which diabetogenic MHC molecules displaying peptides with Why Are These Techniques the Breakthrough We neutral residues at P9 are recognized by TCRs bearing Needed for Type 1 Diabetes Research? an acidic amino acid in the N-terminal segment of their As mentioned, the top advantage of single-cell techniques CDR3b (8,12). The P9 switch was demonstrated by is to allow the interrogation of very few cells. In the mouse, sequencing and reexpressing TCRs isolated with I-Ag7 access to cell numbers was until now addressed by pooling B12-20 insulin tetramers and by showing their mode of together organs from a large number of animals. Indeed, activation in vitro. Most importantly, while the P9 switch the frequencies of autoreactive T cells in most organs is dominated the anti-insulin response at 6 weeks of age, it always low, from 0.05% to 0.2% for CD4 T cells, numbers was nearly gone by 12 weeks, indicating that its role is that preclude any characterization at the very early pre- essential in the initiation of disease, not its progression. clinical stage of disease when only a few cells are present in Also, this P9 switch–driven anti-insulin response was only a few islets. Now the recovery of 100–200 cells from eliminated from mice in which the b57 mutation was the 200–300 islets from a single mouse is exploitable and corrected to an aspartic acid, and these mice were resistant reaches statistical power to determine the diversity of the to disease. The second mechanistic study that the single- CD4+ T-cell population, the gene expression, and idiotypic cell approach gave access to was the examination of receptor sequences of each cell in it. Similarly, we can whether anti-islet CD4 T cells recirculated and were de- isolate and study 25–50 CD4 anti-insulin T cells that we tectable in peripheral blood. To access this information, we isolate from 0.5–1 mL of blood with MHC tetramers first profiled intraislet CD4 T cells and then examined (frequencies in this tissue vary widely from 0.1% to blood for cells with similar phenotypes and TCRs, tracing 0.01% both in mice and humans). Because we know the once again specificity with insulin I-Ag7 tetramers. Our biology of T cells in such detail, the examination of 96 genes early studies, both in mice and humans, are encouraging by single-cell quantitative PCR allows us, based on gene but still preliminary (Fig. 2). Anti-insulin cells were found module examination, to categorize subsets of CD4 T cells in peripheral blood at a very early stage (3 weeks of age in (Th1, Th2, Th17, Treg) while simultaneously evaluating mice), but it appeared that their state of activation varied their state of activation (Fig. 2) (34). Although this exercise greatly. If activation state correlates with the progression of parsing cellular diversity remains very descriptive, it of autoimmunity, a single-cell profiling of gene expression provides a very deep interrogation of the functional state of anti-insulin CD4 T cells from blood might constitute of each cell and can easily indicate whether the cell is a very early diagnostic test of autoimmunity in which not dormant, activated by cytokines, and/or activated through numbers but activation status is evaluated (Fig. 2). The 1890 Single-Cell Analysis of CD4 T Cells Diabetes Volume 68, October 2019

Figure 2—Isolating and analyzing anti-insulin CD4 T cells from peripheral blood of patients. A: HLA-DQ8 tetramers loaded with one of the main insulin epitopes, B12-20, were used to isolate circulating anti-insulin CD4 T cells from a just-diagnosed patient. Positive cells were sorted by ﬂow cytometry as 1 cell per well in 96-well plates. B: Quantitative PCR (Fluidigm; Biomark) single-cell gene expression analysis of the tetramer-sorted cells. In this heat map, each column is a unique gene, whereas each row is a single cell. Low to high expression goes from black to yellow. In this particular experiment, about 20% of the cells have a higher expression of the activation genes that are tested in our panel (red asterisks).

confrontation of this approach with the detection of anti- our animal studies translate to humans, TCR sequencing of islet antibodies in at-risk patients will answer this very single CD4 T cells could improve the accuracy and timing of important question. diagnosis by showing T cells capable of recognition through a P9 switch; these cells should be the earliest Where We Need to Go toappear.Weshouldalsobeabletoconfirm that the Other groups have made progress in the diagnostic anti-insulin response comes first and drives disease pro- isolation of circulating autoreactive anti–b-cell T cells gression. Deeper RNA sequencing and epigenome mea- and/or T-cell populations from the peripheral blood surements should also shed light on the mechanisms of at-risk or just diagnosed patients. In one study, multi- leading to the breakage of tolerance. This issue is critical parametric flow cytometry with HLA class I tetramers to address if we want to be able to understand environ- was used to isolate anti-ZnT8–specific CD8 T cells, whereas mental factors at the molecular level and, practically in the other flow cytometry was used to isolate CD4 T cells speaking, to parse the at-risk population into “at high and examine those by single-cell RNA sequencing (29,36). risk” and “at low risk” of progression. In this process, it is It is likely that the technological progress for single-cell also highly probable that we might uncover some expla- analysis and the examination of a larger cohort of patients nation for the increasing heterogeneity in disease pre- will deliver a reliable diagnosis within the next 5 years. Of sentation and evolution. course, it is in at-risk populations that these approaches This perspective would not be complete if we were need to demonstrate their value; in this respect, the not addressing some financial issues related to single- TrialNet cohort is of the greatest importance and a re- cell analysis. The lofty goal of diagnosing type 1 diabetes in source that each type 1 diabetes researcher should appre- its preclinical phase and monitoring immune interventions ciate and advocate for. The added interest of assessing aimed at protecting b-cells is hindered by the cost of autoimmunity from peripheral blood is twofold: one is to single-cell technologies and is a major issue. Between establish precedent for other autoimmune diseases and the cost of goods, the cost of processing (cell sorting on generalize the approach, and the other one is to appreciate a cytometer, preparation of DNA libraries, sequencing, in real time or near real time the progression of disease and deconvolution of metadata), and the cost for highly qual- the impact of therapies on its course. Indeed, immuno- ified personnel, each mouse and each patient that are logically based interventions at diagnosis or in at-risk examined represent an investment of $10,000–$12,000. patients are currently monitored on residual b-cell As the cost of DNA sequencing is now near bottom, the mass, not on the driving autoimmune process itself as deployment of single-cell technologies in the clinic will it should be (37,38). require a complete change of the business model used by It is also likely that a single-cell approach will add the private sector that provides equipment and consum- a deeper understanding of mechanisms of disease. If ables. It is understood that the current effort was intended diabetes.diabetesjournals.org Sharma and Associates 1891 for basic research, but the future is in the clinic with 16. Aly TA, Ide A, Jahromi MM, et al. Extreme genetic risk for type 1A diabetes. thousands of patients who will need regular evaluation Proc Natl Acad Sci U S A 2006;103:14074–14079 over a lifetime. 17. Tanay A, Regev A. Scaling single-cell genomics from phenomenology to mechanism. Nature 2017;541:331–338 18. Wagner A, Regev A, Yosef N. Revealing the vectors of cellular identity with Funding. This work is supported by National Institute of Diabetes and Digestive single-cell genomics. Nat Biotechnol 2016;34:1145–1160 and Kidney Diseases grant 1R01DK117138 to J.P., M.G., and L.T. S.S. is a TL1 19. Giladi A, Amit I. Single-cell genomics: a stepping stone for future immunology scholar supported by a National Institutes of Health Clinical and Translational Sci- discoveries. Cell 2018;172:14–21 ence Award issued to the Scripps Translational Science Institute (UL1TR002550, 20. Trapnell C. Defining cell types and states with single-cell genomics. Genome TL1TR002551). B.A. was a KL2 scholar (KL2TR001112) from the same Clinical and Res 2015;25:1491–1498 Translational Science Award. 21. Veres A, Faust AL, Bushnell HL, et al. Charting cellular identity during human Duality of Interest. No potential conflicts of interest relevant to this article in vitro b-cell differentiation. Nature 2019;569:368–373 were reported. 22. Ziegenhain C, Vieth B, Parekh S, et al. Comparative analysis of single-cell RNA sequencing methods. Mol Cell 2017;65:631–643.e4 References 23. Shah S, Takei Y, Zhou W, et al. Dynamics and spatial genomics of the nascent 1. Barrett JC, Clayton DG, Concannon P, et al.; Type 1 Diabetes Genetics transcriptome by intron seqFISH. Cell 2018;174:363–376.e16 Consortium. Genome-wide association study and meta-analysis find that over 24. Eng CL, Lawson M, Zhu Q, et al. Transcriptome-scale super-resolved im- 40 loci affect risk of type 1 diabetes. Nat Genet 2009;41:703–707 aging in tissues by RNA seqFISH. Nature 2019;568:235–239 2. Todd JA, Bell JI, McDevitt HO. HLA-DQ beta gene contributes to susceptibility 25. Adey AC. Integration of single-cell genomics datasets. Cell 2019;177:1677– and resistance to insulin-dependent diabetes mellitus. Nature 1987;329:599–604 1679 3. Hu X, Deutsch AJ, Lenz TL, et al. Additive and interaction effects at three 26. Damond N, Engler S, Zanotelli VRT, et al. A map of human type 1 diabetes amino acid positions in HLA-DQ and HLA-DR molecules drive type 1 diabetes risk. progression by imaging mass cytometry. Cell Metab 2019;29:755–768.e5 Nat Genet 2015;47:898–905 27. McHeyzer-Williams MG, Altman JD, Davis MM. Enumeration and char- 4. Insel RA, Dunne JL, Atkinson MA, et al. Staging presymptomatic type acterization of memory cells in the TH compartment. Immunol Rev 1996;150: 1 diabetes: a scientific statement of JDRF, the Endocrine Society, and the 5–21 American Diabetes Association. Diabetes Care 2015;38:1964–1974 28. Kvistborg P, Philips D, Kelderman S, et al. Anti-CTLA-4 therapy broadens the 5. Todd JA. Etiology of type 1 diabetes. Immunity 2010;32:457–467 melanoma-reactive CD8+ T cell response. Sci Transl Med 2014;6:254ra128 6. Carrasco-Marin E, Shimizu J, Kanagawa O, Unanue ER. The class II MHC 29. Culina S, Lalanne AI, Afonso G, et al. Islet-reactive CD8+ T cell frequencies in I-Ag7 molecules from non-obese diabetic mice are poor peptide binders. J Im- the pancreas, but not in blood, distinguish type 1 diabetic patients from healthy munol 1996;156:450–458 donors. Sci Immunol 2018;20:eaao4013 7. Suri A, Walters JJ, Gross ML, Unanue ER. Natural peptides selected by 30. Scott CA, Garcia KC, Carbone FR, Wilson IA, Teyton L. Role of chain pairing for diabetogenic DQ8 and murine I-Ag7 molecules show common sequence speci- the production of functional soluble IA major histocompatibility complex class II ficity. J Clin Invest 2005;115:2268–2276 molecules. J Exp Med 1996;183:2087–2095 8. Yoshida K, Corper AL, Herro R, Jabri B, Wilson IA, Teyton L. The diabetogenic 31. Franz B, May KF Jr, Dranoff G, Wucherpfennig K. Ex vivo characterization and mouse MHC class II molecule I-Ag7 is endowed with a switch that modulates TCR isolation of rare memory B cells with antigen tetramers. Blood 2011;118:348–357 affinity. J Clin Invest 2010;120:1578–1590 32. Moody MA, Haynes BF. Antigen-specific B cell detection reagents: use and 9. Kishimoto H, Sprent J. A defect in central tolerance in NOD mice. Nat quality control. Cytometry A 2008;73:1086–1092 Immunol 2001;2:1025–1031 33. Hamilton JA, Li J, Wu Q, et al. General approach for tetramer-based 10. Hahn M, Nicholson MJ, Pyrdol J, Wucherpfennig KW. Unconventional to- identification of autoantigen-reactive B cells: characterization of La- and snRNP- pology of self peptide-major histocompatibility complex binding by a human reactive B cells in autoimmune BXD2 mice. J Immunol 2015;194:5022–5034 autoimmune T cell receptor. Nat Immunol 2005;6:490–496 34. Holt M, Costanzo A, Gioia L, Abe B, Su AI, Teyton L. Gene profiling and T cell 11. Festen EA, Goyette P, Green T, et al. A meta-analysis of genome-wide receptor sequencing from antigen-specific CD4 T cells. Methods Mol Biol 2018; association scans identifies IL18RAP, PTPN2, TAGAP, and PUS10 as shared risk 1712:217–238 loci for Crohn’s disease and celiac disease. PLoS Genet 2011;7:e1001283 35. Gioia L, Holt M, Costanzo A, et al. Position b57 of I-Ag7 controls early anti- 12. Hovhannisyan Z, Weiss A, Martin A, et al. The role of HLA-DQ8 b57 insulin responses in NOD mice, linking an MHC susceptibility allele to type 1 polymorphism in the anti-gluten T-cell response in coeliac disease. Nature 2008; diabetes onset. Sci Immunol 2019;4:eaaw6329 456:534–538 36. Heninger AK, Eugster A, Kuehn D, et al. A divergent population of auto- 13. Arentz-Hansen H, Körner R, Molberg O, et al. The intestinal T cell response to antigen-responsive CD4+ T cells in infants prior to b cell autoimmunity. Sci Transl a-gliadin in adult celiac disease is focused on a single deamidated glutamine Med 2017;9:9 targeted by tissue transglutaminase. J Exp Med 2000;191:603–612 37. Herold KC, Hagopian W, Auger JA, et al. Anti-CD3 monoclonal antibody in 14. Corper AL, Stratmann T, Apostolopoulos V, et al. A structural framework for new-onset type 1 diabetes mellitus. N Engl J Med 2002;346:1692–1698 deciphering the link between I-Ag7 and autoimmune diabetes. Science 2000;288: 38. Herold KC, Bundy BN, Long SA, et al. An anti-CD3 antibody, teplizumab, in 505–511 relatives at risk for type 1 diabetes. N Engl J Med 2019 15. Scott CA, Peterson PA, Teyton L, Wilson IA. Crystal structures of two 39. Pettersen EF, Goddard TD, Huang CC, et al. UCSF Chimera–a visualization I-Ad-peptide complexes reveal that high affinity can be achieved without large system for exploratory research and analysis. J Comput Chem 2004;25:1605– anchor residues. Immunity 1998;8:319–329 1612