Presenter
Abhishek Kulkarni (University of Chicago)

Authors
Abhishek Kulkarni, Sarah A. Tersey, Fei Huang, Annie R. Pineros, Farooq Syed, Hongyu Gao, Kara Orr, Yunlong Liu, Maureen Gannon, Marcia McDuffie, Jerry L. Nadler, Margaret A. Morris, Raghavendra G. Mirmira

Purpose
In recent years, type 1 diabetes (T1D) has become viewed as a disease initiated and propagated by islet β cells. 12-lipoxygenase (12-LOX), an enzyme in arachidonic acid metabolism, is expressed in β cells and macrophages, and prior studies have shown that its deletion globally can protect against T1D in the NOD mouse model. In this study, we hypothesized that 12-LOX within the β cell may be an initiator of autoimmunity in T1D through activation of the integrated stress response (ISR).

Methods
We interrogated NOD mice with islet-specific deletion of the gene encoding 12-LOX for activation of the ISR and diabetes outcome, and leveraged tissues from the nPOD repository for interrogation of the ISR in human T1D. In addition, we utilized an inhibitor of the human 12-LOX enzyme (ML355) in studies of human islets and in studies using NOD mice containing targeted replacement of the mouse enzyme with the human enzyme (“humanized mice”).

Summary of Results
We backcrossed both mice with floxed alleles of the gene encoding 12-LOX (Alox15) and mice with the PdxPB-CreERT transgene onto the NOD background at greater than 99.7% of genomic loci. Tamoxifen-induced deletion of islet Alox15 at 6 weeks of age led to preservation of β-cell mass, suppression of insulitis, and near-complete protection against autoimmune diabetes. Single cell RNA-sequencing and mass cytometry analyses revealed that the loss of islet Alox15 led to an increase in a population of β cells expressing Cd274, encoding the immune checkpoint protein PD-L1 [5], and promoted the reprogramming of the immune response. The reprogramming included the expansion of anti-inflammatory macrophages and regulatory T cells and suppression of cytotoxic T cells. The increase in PD-L1 protein was coincident with the suppression of the integrated stress response (ISR) in these mouse islets, as evidenced by reductions in stress granules and levels of phospho-eIF2α. Administration of a PD-L1 blocking antibody led to recovery of the diabetes phenotype in islet Alox15 knockout mice. In humans, increases in islet 12-LOX and the ISR (evidenced by formation of stress granules) was observed in pancreas tissue sections from AAb+ donors. In human islets, inhibition of either 12-LOX using ML355 or the ISR using ISRIB resulted in the upregulation of PD-L1 levels on both β cells and in their liberated exosomes. To assess if inhibition of 12-LOX using the human-specific 12-LOX inhibitor modifies T1D progression, we generated mice in which the gene encoding mouse 12-LOX (Alox15) was replaced by the gene encoding human 12-LOX (ALOX12). These “humanized” mice were subsequently backcrossed onto the NOD background, then treated in the prediabetic phase with the 12-LOX inhibitor ML355. Male and female humanized mice developed T1D at the expected frequency, but those receiving ML-355 showed significant delay of T1D.

Conclusions
Our results support several key, previously unappreciated findings: (a) that cell-autonomous inflammatory signaling via 12-LOX in the β-cell dictates a dialog that propagates innate and adaptive immunity, likely via the suppression of the immune checkpoint PD-L1, (b) 12-LOX activity is linked to the ISR in T1D, and that the ISR may be increasing the susceptibility of β cells to autoimmune attack, and (c) targeting of the 12-LOX pathway and the ISR provides an opportunity for the potential modification of T1D disease progression.

Presenter
Farooq Syed (Indiana University)

Authors
Farooq Syed, Olivia Ballew, Chi0Chun Lee, Jyoti Rana, Angela Castela, Stephane Demine, Maria Ines Alvelos, Kara Orr, Garrick Chang, Staci A. Weaver, Jacqueline Del Carmen Aquino, Kentaro Yamada, Jing Liu, Donalyn Scheuner, Decio L. Eizirik, Carmella Evans-Molina

Purpose
T1D results from a complex interaction that occurs between the invading immune cells, which release a variety of chemokines and cytokines, and immunogenic signals produced by injured or dying β cells. Multiple lines of evidence suggest a prominent role for type 1 interferons, particularly interferon-α (IFNα), in T1D pathogenesis. A type I IFN-inducible transcriptional signature is present in the blood of children with a high genetic risk of T1D, even prior to the development of T1D-associated autoantibodies, and interferon-stimulated gene signatures are present in biopsies of pancreatic islets from individuals with recent-onset T1D. IFNα has been shown to induce endoplasmic reticulum stress, chemokine production, and HLA class I overexpression in human β-cells – three histological hallmarks of islets in human T1D. The effects of IFNα are mediated via its receptor (IFNAR1) and the protein tyrosine kinases JAK1 and TYK2 that phosphorylate and activate the signal transducers and activators of transcription (STAT1 and STAT2) proteins. JAK inhibitors have shown promise in mouse and in vitro models of T1D and are being tested in a clinical trial in adolescents and adults with recent-onset T1D. Notably, polymorphisms that decrease TYK2 activity are protective against T1D, and TYK2 inhibitors (TYKi) are being tested in other autoimmune conditions such as psoriasis. However, whether strategies to inhibit TYK2 have similar efficacy in models of T1D has not been tested.

Methods
We tested two small molecule inhibitors of TYK2 (BMS-986165 and BMS-986202). Both compounds target the pseudokinase Janus homology 2 regulatory domains (JH2) of TYK2, leading to inhibition of receptor-mediated TYK2 activation. Isolated human islets from cadaveric organ donors and EndoC-βH1 cells were pre-incubated with TYKis for 2hrs followed by exposure to IFNα alone or in combination with IL-1β or TNFα for 24hrs or 48hrs. Expression of HLA class I, CXCL10, and MX1 were evaluated by RT-qPCR; STAT phosphorylation was assessed by immunoblot. Apoptosis was evaluated by Hoechst/Propidium Iodide staining. To determine the in vivo efficacy of TYK2 inhibition, we utilized two mouse models: 1) the RIP-LCMV-GP model, which is a transgenic and virally-induced model of T1D, and 2) the NOD mouse model of spontaneous T1D development. RIP-LCMV mice were pre-treated with BMS-986202 for 2 days prior to the LCMV injection (0.5×105 PFU), and NOD mice were treated with BMS-986202 from 6-12 weeks of age. In both models, vehicle and drug-treated mice were monitored for diabetes incidence. Immunostaining for insulitis and β cell mass and single-molecule RNA Fish (smRNA FISH) was performed in pancreatic sections from treated mice. Flow cytometry was used to profile immune cell populations in the blood, pancreatic lymph node (PLN), and spleen.

Summary of Results
Results from the in vitro studies showed that BMS-986165 and BMS-986202 pre-treatment prevent IFNα-induced upregulation of CXCL10, MX1, and HLA-ABC in EndoC- βH1 and human islets at 24hrs and 48hrs (p<0.05). TYK2 inhibition also decreased STAT1/2 phosphorylation and prevented cytokine (i.e. IFNa + IL1-β) -induced apoptosis in dispersed human islets. BMS-986202 reduced diabetes incidence by 80% in RIP-LCMV mice (n=18 vehicle/18 TYK2i; p<0.001). Flow cytometry analysis of blood, spleen, and PLN 3 days after LCMV injection revealed a significant decrease in the percentage of CD11b+F4/80+ macrophages (M1) and CD11b+CD49+ NK cells in the PLN and blood and increased circulating CD11b-CD49+ tolerogenic-NK cells in TYK2i treated RIP-LCMV mice. At day 7 and 14 post LCMV induction, PDI+CD8+T-cells in the blood, spleen, and PLN and PDI+FOXP3+ Treg cells were increased in the spleen of BMS-986202-treated RIP-LCMV mice. Treatment with BMS-986202 resulted in a 44% decrease in diabetes incidence in NOD mice (n=32 vehicle/ 34 TYK2i; p=0.0075) and significantly reduced insulitis (p<0.05). smFISH analysis revealed decreased β cell expression of STAT1 and MX1 expression in TYK2i-treated RIP-LCMV and NOD mice compared to vehicle-treated mice.

Conclusions
Treatment with TYK2 inhibitors protected human islets in vitro, reduced diabetes incidence in two mouse models of T1D, and was associated with early changes in innate immune and late changes in adaptive immune signatures, most notably leading to increased T regulatory cells and increased PDI expressing CD8+ T cells, a phenotype which has been linked T cell exhaustion, Taken together, these data highlight the potential role for TYK2 inhibitors as a novel disease-modifying therapy in T1D, with beneficial effects in both the immune and β cell compartments.

Presenter
Mark Mamula (Yale University)

Authors
Mei-Ling Yang, Cate Speake, Carmella Evans-Molina, Lut Overbergh, Eddi James, Li Wen, Kevan Herold, Mark Mamula

Purpose
Emerging evidence indicates that autoimmunity often arises to post-translational protein modifications (PTMs), leading to chronic inflammation by infiltrating lymphocytes of target organs such as the pancreas in type 1 diabetes (T1D). While the identification and assessment of autoimmunity to novel PTM biomarkers is important to understanding pathogenic processes in T1D, it is similarly critical to determine if PTM proteins alter key processes in beta-cell biology, namely glucose sensing and insulin production, folding and release. This study specifically addresses these biologic functions of PTM proteins associated to T1D.

Methods
Carbonylation is the major protein modification in response to oxidative stress. Citrullination is mediated by calcium-dependent peptidylarginine deiminase (PAD) enzymes, which catalyze deimination, the conversion of arginine into the non-classical amino acid citrulline. Various proteomic analyses, including mass spectrometry, were performed in both murine and human pancreatic islets to identify carbonylated- and citrullinated-islet proteins. Glucose-stimulated proinsulin/insulin secretion was examined in human islets stressed with inflammatory cytokines and physiologic oxidative conditions in the presence or absence of PAD inhibitors. In addition, the effect of citrullination on the enzyme kinetics of glucokinase (Vmax and Km), the first rate-limiting step of glycolysis in the liver and pancreas, was also examined.

Summary of Results
We identify autoantibodies against to beta subunit of prolyl-4-hydroxylase (P4Hb; native and carbonylated form) and glucokinase (GK; native and citrullinated form) in both human T1D and murine models. By mass spectrometry, six carbonyl residues and sixteen citrulline residues were mapped in oxidative P4Hb and PAD-treated GK, respectively. In regards to glucose metabolism, the carbonylated-P4Hb is amplified in stressed human islets coincident with decreased glucose-stimulated insulin secretion and increased proinsulin to insulin ratios. Citrullination alters GK biologic activity (Km) and suppresses glucose-stimulated insulin secretion. Moreover, PAD2/4 inhibitor can partially correct IFNgamma +IL-1beta suppressed glucose stimulated insulin secretion in INS-1E beta cells.

Conclusions
The major function of pancreatic beta cells is to secret insulin in response to glucose uptake in order to maintain the blood glucose level. Glucokinase is expressed in hepatocytes to regulate glycogen synthesis, and in pancreatic beta cells as a glucose sensor to initiate glycolysis and insulin signaling. P4Hb is a critical macromolecule for the accurate folding of insulin. Our studies implicate the crucial enzymes, glucokinase and P4Hb, as the biomarkers, providing new insights into creating autoantigens and define the impact of PTMs on the aberrant beta cell functions of T1D. Targeting glucose metabolism by PTMs, such as PAD inhibitors, may lead to preventing diabetes autoimmunity and restoring beta cell function.

Presenter
Julia Panzer (University of Miami)

Purpose
Julia Panzer, Alejandro Tamayo, Alejandro Caicedo

Methods
Increased glucagon secretion from the pancreatic alpha cell is the first and most important defense against hypoglycemia. In type 1 diabetes this defensive mechanism is lost, increasing the mortality risk. Thus, understanding the pathophysiological mechanisms that make alpha cells unresponsive to a drop in glycemia during disease progression is therefore of utmost is importance to improve the management of diabetes.

Summary of Results
Studying alpha cell physiology in type 1 diabetes has met major technological roadblocks, as methods conventionally used are very difficult to apply to type 1 diabetic donors. We overcome these limitations by using living pancreas slices, which allow functional assessments of damaged and infiltrated islets within their native environment. We used slices from non-diabetic donors and donors with type 1 diabetes to determine alpha cell responses to (a) changes in glycemia, (b) agonists, antagonists, and positive allosteric modulators of glutamate receptors, and (c) reference stimuli such as adrenaline and KCl depolarization using functional recordings. We further performed in vivo studies using mouse models with defective glucose counter regulation to determine whether alpha cell responses to hypoglycemia can be restored.

Conclusions
We found that alpha cells in slices from type 1 diabetic donors had normal glucagon content and responded to KCl depolarization but failed to respond to decreases in glucose concentration. Furthermore, we found severely dimished Ca2+ responses to both lowering in glucose concentration and glutamate receptor stimulation. By reactivating residual glutamate receptor function with the positive allosteric modulators cyclothiazide and aniracetam we could rescue glucagon secretion in response to hypoglycemia in human tissue slices from donors with T1D.

Presenter
Yuval Dor (The Hebrew University of Jerusalem)

Authors
Yuval Dor, Ayelet Peretz, Judith Magenheim, Netanel Loyfer, Agnes Klochendler, Benjamin Glaser, Tommy Kaplan

Purpose
The methylome of cells encodes and memorizes the combined influence of pre-existing genetics, cell differentiation and maintenance programs, and environmental exposures. Previous studies have described extensive plasticity of the beta cell methylome during healthy development and in diabetes. However such studies have typically relied on material from whole islets, where different preparations may have different cellular compositions. In addition, analyses have often examined the methylation status of individual CpG sites, while the actual functional unit of DNA methylation, controlling chromatin organization and protein binding, is a block of multiple adjacent CpGs that are either all methylated or all unmethylated. The purpose of this study was to characterize the methylome of human beta cells as compared to all other major human cell types.

Methods
We have generated a comprehensive cell type-specific human methylome atlas, based on whole genome bisulfite sequencing of individual cell types sorted from dissociated surgical material. The atlas contains the methylomes of 210 samples from 86 cell types, including triplicates of samples from alpha, beta and delta cells from non-diabetic donors; duplicates of alpha, beta and delta cells from donors with type 2 diabetes; and triplicates of samples from acinar cells and duct cells.

Summary of Results
Comparative analysis of the atlas provides interesting insights, including:
1. When examined through the lens of methylation blocks, the methylomes of beta cells from different donors are nearly identical. Out of 3 million methylation blocks in the genome, only 1000 (0.03%) vary in their methylation patterns between beta cells of healthy individuals. This degree of variation is seen also when comparing material from healthy donors and people with type 2 diabetes.
2. The methylome of beta cells is most similar to that of alpha and delta cells, followed by similarity to acinar cells, duct cells and hepatocytes. Strikingly, the methylomes of beta cells and neurons are highly dissimilar, despite the phenotypic similarity. Thus, methylomes reflect the lineage history of beta cells more than their current phenotype.
3. There are hundreds of loci unmethylated uniquely in beta cells. These are typically enhancers that control expression of beta cell genes.
4. Loci that are methylated uniquely in beta cells are typically binding sites to CTCF, suggesting that these are regulators of beta cell-specific chromatin looping.

Conclusions
The methylome of beta cells, examined in the context of the methylomes of other cell types, opens a window into beta cell biology including regulatory circuits, interpretation of GWAS hits and more.

Presenter
Mia Smith (Barbara Davis Center for Diabetes)

Authors
Zachary Stensland, Hali Broncucia, Adam Magera, Peter Gottlieb, Mia Smith

Purpose
In healthy individuals autoreactive B cells that escape central tolerance mechanisms are normally silenced by anergy, a type of B cell tolerance wherein autoreactive cells occupy peripheral lymphoid organs but are antigen unresponsive. Previously we have found that young new-onset T1D subjects exhibit a loss of anergic insulin-binding B cells (IBCs) in their peripheral blood. Given that an increase in B cells in both the blood and pancreas is associated with the aggressive form of T1D that is seen in younger patients, we hypothesize that loss of anergic IBCs in the peripheral blood reflects their activation and trafficking to tissues rich in autoantigen (e.g. pancreas and pancreatic lymph node (pLN)) where they participate in disease pathogenesis.

Methods
Since antigen-specific B cells occur in very low frequency in the periphery, we developed a magnetic nanoparticle-based scheme to enrich for IBCs from blood and tissue. Aided by enrichment, we explored the frequency of various IBC and non-IBC subsets and their activation status in the peripheral blood of T1D subjects and spleen and pLN of nPOD donors using both mass cytometry and spectral flow cytometry.

Summary of Results
Both manual gating and unsupervised clustering algorithms identified ‘anergic’ B cells in the peripheral blood of T1D donors are highly activated (e.g. increased expression of CD86, CD11c, and CXCR3), particularly in insulin-binding B cells compared to non-insulin-binding B cells. Analysis of the spleen and pLN of nPOD donors revealed similar findings with these B cell subsets further enriched in these tissues, especially the pLN. Trajectory and phenotypic analysis suggests these activated ‘anergic’ B cells could be precursors to extrafollicular antibody secreting cells, as well as act as potent antigen-presenting cells to diabetogenic T cells.

Conclusions
Our results suggest that autoreactive B cells that are normally silenced by anergy in healthy individuals, become activated in T1D subjects and traffic to the pLN, where they likely participate in disease pathogenesis through antibody secretion and/or antigen presentation to T cells.

Presenter
Alberto Pugliese, University of Miami

Authors
Helena Reijonen, Isaac Snowhite, Francesco Vendrame, George W. Burke III, Alberto Pugliese

Purpose
Patients with autoimmune, type 1 diabetes (T1D) and end stage renal disease may become recipients of simultaneous pancreas-kidney (SPK) transplants to restore insulin secretion and kidney function. We previously reported that some SPK recipients may develop T1D recurrence (T1DR) on follow-up despite immunosuppression that prevents rejection; in some patients T1DR was confirmed by the demonstration of insulitis and beta cell loss in a pancreas transplant biopsy. Overall, we have observed T1DR in about 5% of SPK recipients on extended follow-up. We previously reported that seroconversion for multiple autoantibodies is a risk factor for T1DR. The aim of this study was to determine whether autoreactive T cells in the circulation of SPK recipients are associated with T1DR and define key phenotypic features of these cells.

Methods
We studied 7 SPK recipients who had developed T1DR and 16 with normal glucose tolerance (NGT), of whom 5 were classified as autoantibody converters at risk for future T1DR (NGT-C) and 11 were autoantibody negative or had stable autoantibody levels from prior to the transplant (stable, NGT-S). We evaluated autoreactive T cells in peripheral blood using a pool of HLA class II tetramers loaded with T1D-associated peptides from multiple autoantigens (GAD65, proinsulin and ZnT8), and a viral antigen (flu) was used as a control. Besides antigen specificity, T cells were analyzed for lineage and phenotype by flow cytometry staining CD4, CD45RA, CD45RO, PD1, CXCR3, CCR7, CCR6 and CCR4. We examined naïve (CD45RA+, CCR7+), effector memory (EM, CD45RO+, CCR7-), central memory (CM, CD45RO+, CCR7+), total memory (EM+CM) and terminally differentially memory (TEMRA, CD45RA+, CCR7-). We measured these phenotypes in the total CD4 T cell compartment and in tetramer positive CD4 T cells. We assessed number of tetramer+ cells/1×106 CD4 T cells and as % of the various subset among the tetramer+ CD4 T cells; Value ranks among patient groups were compared using the non-parametric Mann-Whitney test; two-tailed p values are reported. For some analysis, we used ROC (Receiving Operating Curves).

Summary of Results
T1DR patients had significantly higher numbers of autoreactive CD4 T cells compared to NGT-S patients, and this was observed for naïve, total memory and effector memory compartments. In contrast, T1DR patients had decreased numbers of autoreactive CD4 T cells among TEMRAs compared to NGT-S recipients. For NGT-C patients, numbers of autoreactive CD4 T cells were similar to those of NGT-S patients, except these numbers were higher among naïve T cells.
We then examined the frequency of various subsets among autoreactive CD4 T cells, specifically naive, total memory, effector memory and TEMRA. In T1DR patients there were increased proportions of naïve and total memory autoreactive CD4 T cells compared to NGT-S patients. T1DR patients had very low frequencies of TEMRA autoreactive CD4 T cells compared to NGT-S patients (% mean 1.143 + SEM 0.6335 vs 46.82 + 8.781, respectively, p<0.0026); NGT-C patients also had lower frequency TEMRAs among the autoreactive CD4 T cell (% mean 14.60 + SEM 10.35) compared to NGT-S patients (p= 0.0279). The T1DR and NGT-C patients combined had much lower frequencies of TEMRA autoreactive CD4 T cells compared to NGT-S patients (p<0.00001). Using ROC, the proportions of TEMRA autoreactive CD4 T cells distinguished T1DR and NGT-C from NGT-S patients (AUC= 0.94, p=0.0003, 90% sensitivity, 92% specificity). We did not observe significant differences in the frequencies of naïve and memory subsets among the three patient groups, except that TEMRA CD4 T cells were reduced in T1DR compared to NGT-S patients (% mean 3.286 + SEM 1.358 vs 6.250 + 1.023, respectively, p<0.04).
The frequency autoreactive CD4 T cells expressing PD1, CCR6, or CCR4 did not differ among the three patient groups. However, the frequency of CXCR3-positive CD4 autoreactive T cells was significantly higher in T1DR and NGT-C patients compared to NGT-S patients. The frequency of CXCR3-positive autoreactive CD4 T cells among T1DR and NGT-C patients combined was % mean 25.25 + SEM 6.361 vs % mean 4.455 + SEM 3.431 among NGT-S patients (p= 0.0063). ROC analysis showed that the proportions of CXCR3-positive autoreactive CD4 T cells distinguished T1DR and NGT-C from NGT-S patients (AUC= 0.80, p=0.01, 82% sensitivity, 75% specificity).

Conclusions
Autoreactive CD4 T cells were increased in T1DR compared to NGT-S patients. We show that this applied to both naïve and total memory compartments. NGT-C patients had higher frequencies then NGT-S patients only in the naive compartment. Remarkably, the autoreactive CD4 T cells had very low frequency in the TEMRA compartment of T1DR and NGT-C patients, and this was a distinguishing feature that may be exploited as a biomarker of T1DR. Our results also demonstrate an association of CXCR3-positive autoreactive CD4 T cells with T1DR and autoantibody conversion. The expression of CXCR3 by circulating autoreactive CD4 T cells help identify a stage in which autoreactive T cells may migrate to the pancreas and infiltrate the islets. While our findings are from a relatively small number of patients, given the low frequency of T1DR among SPK recipients, they support CXCR3 as a potential therapeutic target to antagonize recurrent islet autoimmunity.

Presenter
Matti Nykter (Tampere University)

Authors
Jake Lin, Elaheh Moradi, Karoliina Salenius, Suvi Lehtipuro, Tomi Hakkinen, Jutta E. Laiho, Sami Oikarinen , Sofia Randelin, Hemang Parikh, Jeffrey Krischer, Anette-G. Ziegler, Jorma Toppani, Ake Lernmark, Joe Petrosino, Nadim Ajami, Jin-Ziong She, Beena Akolkar, William A. Hagopian, Marian J. Rewers, Richard E. Lloyd, Kirsi Granberg, Heikki Hyoty, Matti Nykter

Purpose
Genetic basis of type 1 diabetes has been characterized extensively and several non-genetic risk-modifying factors have been identified. While several studies have pointed out the role of environmental factors in the pathogenesis, possible contribution of host responses that are induced by these environmental factors is not known. The distinct genetic background of the two type 1 diabetes endotypes, that is children who initially develop autoantibodies against either insulin or GAD65, suggest that different gene-environment interactions may play a role in their pathogenesis. The purpose of this study was to elucidate these interactions in the emergence of the two divergent antibody patterns that predispose to type 1 diabetes.

Methods
We utilized 2376 matched longitudinal transcriptome sequencing data to characterize dynamic host responses during the prospective follow-up of children within the The Environmental Determinants of Diabetes in the Young (TEDDY) birth cohort study using nested case-control (NCC) design. Transcriptomics response profiles were analysed in children who later developed autoimmunity and in matched control children, and correlated them with IAA and GADA autoantibody patterns and with 4536 virome profiles in stool and plasma samples. DESeq2 and temporal filtering were applied for selecting of differentially expressed genes. Comprehensive immune cell type proportions were estimated by regression analysis with elastic net regularization. Conditional logistic regression was used to assess the associated odds of virome exposures, the expression of the selected genes, and cell type proportions to the IA outcome in the NCC setting.

Summary of Results
We identified distinct temporal gene expression patterns and proportions of immune cells in children with the first appearing autoantibody against either insulin or GAD65. Applying statistical testing between the cases and controls (adjusted p-value<0.05, LFC>0.5, 2+ timepoints), we identified 17 genes in GADA first and 3 genes in IAA first seroconversion cohorts. GAD65 linked genes include ZBED6, associated with pancreatic beta cell survival and FABP5, fatty acid gene modulating inflammation. The integration with enterovirus infections diagnosed by metagenomic sequencing data from stool and serum samples showed that the enterovirus-induced host response was weaker in children who later developed islet autoimmunity compared to enterovirus-infected autoantibody negative children exhibiting enriched type 1 interferon production (p-value<0.003), signaling (p-value<3.35e-11) and response (p-value< 0.0001). Case children who developed IAA as the first autoantibody had an elevated monocyte component, associated with chronic inflammation, throughout the time course. We showed that transcriptomic data provides additional independent information on top of genetic and environmental markers that can be used for improved islet autoimmunity prediction.Conclusions
In conclusion, our study showed immune related transcriptomic differences between cases and controls prior to islet autoimmunity. These are presented differently in children with the first appearing autoantibody against either insulin or GAD65. We also found that enterovirus infections lead to a stronger antiviral response in control children than in children who develop autoantibodies. A major strength of our study is the comprehensive integration of transcriptomic profiles, virome and genetics while incorporating immune cell type alterations prior to seroconversion. Taken together, our analysis provides transcriptomic and immunogenic characterization of host responses in the context of type 1 diabetes-related autoantibody patterns and environmental triggers of the disease.

Presenter
Eddie James (Benaroya Research Institute)

Authors
Perrin Guyer, David Arribas-Layton, Cate Speake, Carla Greenbaum, Decio Enzirik, Sally Kent, Roberto Mallone, Eddie James

Purpose
A recent discovery effort investigated tissue specific mRNA splice variants and other novel secretory granule antigens within human islets, demonstrating that unique peptide sequences from these proteins are present within HLA-class I peptidome of human β cells and documenting their recognition by CD8+ T Cells from peripheral blood and human islets. Our goal was to investigate the relevance of CD4+ T cell recognition of epitopes derived from these target antigens.

Methods
In this study, we applied a systematic epitope discovery process to identify novel CD4+ T cell epitopes derived from mRNA splice variants and novel secretory granule antigens. We first predicted potential epitopes spanning unique junctions of mRNA splice variants and within conventional secretory granule antigens contained in the data set. Peptides with DRB1*04:01 motifs were screened for in vitro binding and used to generate HLA class II tetramers. The corresponding tetramers were used to assess peptide immunogenicity, isolate T cell clones, and label and detect CD4+ T cells specific for these putative epitopes in peripheral blood. We further investigated the relevance of these epitopes by examining their characteristics in subjects with established T1D and by investigating their recognition by islet derived T cell lines.

Summary of Results
We observed detectable populations of T cells that recognize three novel epitopes in the peripheral blood of subjects with T1D at frequencies that were similar to an immunodominant proinsulin epitope. T cells that recognized these epitopes were present in peripheral blood at higher frequencies in subjects with T1D than in controls and predominantly exhibited a Th1-like surface phenotype. Among the three novel epitopes, responses to a peptide derived from the CCNI-008 splice variant tended to be the most frequent. T cells with this specificity also exhibited a more differentiated memory phenotype. Furthermore, T cells that respond to these epitopes were present among islet infiltrating T cells.

Conclusions
These results reveal novel epitopes that are recognized by CD4+ T cells in human T1D. This further establishes alternative splicing as a mechanism that contributes to the loss of tolerance in T1D.

Presenter
Yi-Chun Chen (University of British Columbia)

Authors
Yi-Chun Chen, Agnieszka Klimek-Abercrombie, Kathryn Potter, Lindsay Pallo, Galina Soukhatcheva, Dai Lei, Melena Bellin, C. Bruce Verchere

Purpose
Pancreatic islet transplantation has therapeutic potential in T1D and is also an established therapy for patients with chronic pancreatitis; however, long-term transplant outcomes are modest. Identifying indicators and predictors of graft function could aid in the improvement of transplant outcomes and glycemic control.

Methods
We analyzed plasma beta-cell (pro)hormones including proinsulin, C-peptide, amidated islet amyloid polypeptide (IAPP), and proIAPP1-48 levels in a retrospective cohort of autologous total pancreatic islet transplant patients (n = 28), as well as in a mouse model of optimal versus sub-optimal human islet transplantation. We also performed histological analysis of islet grafts retrieved from mice to evaluate islet prohormone processing machinery in situ.

Summary of Results
Proinsulin-to-C-peptide (PI/C) and proIAPP-to-total IAPP (proIAPP/IAPP) ratios measured at 3 months post-autologous islet transplant were significantly higher in patients who remained insulin dependent at 1 year follow-up (PI/C: 8.83 ± 1.03 vs. 5.76 ± 0.93, p < 0.05; proIAPP/IAPP: 0.62 ± 0.10 vs. 0.30 ± 0.02, p < 0.05). In a mouse model of sub-optimal human islet transplantation, we found that mice that later became hyperglycemic displayed significantly higher PI/C ratios than mice that remained normoglycemic (1.08 ± 0.10 vs. 0.25 ± 0.41, n = 8 and 41, respectively; p < 0.05). Histological analysis of islet grafts retrieved from mice showed reduced insulin- and proinsulin-positive area, but elevated glucagon-positive area in grafts that experienced greater secretory demand. Increased prohormone convertase 1/3 (PC1/3) immunoreactivity was detected in glucagon-positive cells, and glucagon-like peptide 1 (GLP-1) immunoreactive area was elevated in grafts from mice that displayed hyperglycemia or elevated plasma PI/C ratios, demonstrating intra-islet incretin production in metabolically challenged human islet grafts.

Conclusions
In failing islet transplants, incomplete beta cell prohormone processing may be an early indicator of graft dysfunction and future insulin dependency. Alpha cell prohormone processing is also likely altered, leading to intra-islet GLP-1 production.

Authors

Adam Curnock, Emma Henderson, Rita Figueiredo, Katherine Wiseman, Ronan O’Dwyer, David Overton, Veronica Gonzalez, Nicola Smit, Lorraine Whaley, Hemza Ghadbane, Shephen Hearty, Tara Mahon, Peter Weber.

Purpose

The success of immunosuppressive therapies to preserve pancreatic islets and treat Type I Diabetes (T1D) has, so far, been limited either due to safety concerns or lack of persistent efficacy. To overcome these issues, we designed a novel targeted immune-suppressive approach, called ImmTAAI, that binds specifically β-cell in pancreatic islets, engages autoreactive T cells and suppresses their effector functions only when bound to target cells. By restricting T cell inhibition to the pancreas, we aim to achieve potent localized immune suppression while avoiding systemic immunosuppression.

Methods

To create a β-cell specific PD-1 agonist ImmTAAI, an affinity-enhanced T cell receptor (TCR) specific for the peptide-HLA-A*02 complex of pre-pro-insulin PPI15-24, a well-characterized autoantigen found on pancreatic β-cells, was fused to an agonistic PD-1 antibody to mimic the activity of PD-L1, the endogenous ligand of the PD-1. To verify that the TCR targeting domain is able to bind its cellular target, β-cell lines and disaggregated β-islets derived from healthy donors were stained with biotinylated PPI15-24 TCR. Furthermore, the subcellular localisation of PPI PD-1 agonist ImmTAAI at the β-cell – T cell interface was studied by confocal microscopy. To evaluate the impact of the β-cell bound PD-1 agonist bispecific on TCR signalling, a β-cell line – Jurkat NFAT cellular reporter assay was used. Finally, inhibition of T cell functions by cell-bound ImmTAAI was measured in cytotoxicity co-culture imaging experiments using CD8+ T cell clones that recognize the PPI6-14 peptide-HLA presented by EndoH-BH2 target cells.

Summary of Results

Microscopy studies showed that the PPI TCR bound specifically to immortalized β-cell lines and primary β-cells from disaggregated pancreatic islets. In co-culture localization experiments we observed that, once bound to β-cells, the PPI PD-1 agonist ImmTAAI engaged the PD-1 receptor on attacking T cells, accumulated at the T cell synapse and inhibited TCR signaling. The engineered bispecific did not compete with PD-L1 binding to PD-1 and was additive with PD-L1 in inhibiting TCR signaling. At picomolar concentrations, the PPI PD-1 agonist ImmTAAI suppressed secretion of inflammatory cytokines and inhibited β-cell killing by autoreactive CD8+ T cell clones. Notably, this PD-1 agonist was unable to inhibit T cells when free in solution and not targeted to β-cells.

Conclusions

We have generated a TCR bispecific inhibitory ImmTAAI molecule that is targeted to β-cells to suppress autoreactive T cells and protect pancreatic islets. These data support the concept that this bispecific inhibitor with the potential to bind β-cells with high specificity, enhance natural immune suppression delivered by PD-L1 and inhibit attacking T cells. Importantly, the molecule is inactive when free in solution and therefore has the potential to deliver localized immune suppression in islets while avoiding systemic immunosuppression. These features make PPI PD-1 agonist ImmTAAI molecules an attractive and novel approach to potentially treat T1D.

Presenter
Michelle Lee (University of Pennsylvania)

Authors
Michelle Lee, Jonathan Schug, Elisabetta Manduchi, Eric Waite, Erin Duffy, Danielle Jaffe, Ali Naji, Mark Atkinson, Alvin Powers, Klaus Kaestner.

Purpose
Deep molecular profiling of pancreatic islet cells via single-cell RNA-sequencing (scRNA-seq) has revealed cell type specific transcriptome profiles. However, scRNA-seq suffers from ‘dropout’, or missing signals, for genes expressed at low levels, such as those encoding transcription factors or signaling molecules. In addition, there is significant of cross-contamination of ambient RNA from lysed cells in each cell, complicating the detection of rare and/or transitional cell types or states. Therefore, additional molecular profiles are required to strengthen our understanding of cell type identity and reveal potential subpopulations of cells. Single nucleus Assay for Transposase-Accessible Chromatin sequencing (snATAC-seq) reveals regulatory element sequence and key transcription factor motifs that function in a cell type specific manner, complementing gene expression. Therefore, we employed snATAC-seq to advance our understanding of cell type identity for key human pancreatic cell types.

Methods
We performed snATAC-seq on pancreatic islets obtained from 33 deceased organ donors that were non-diabetic, auto-antibody positive but normoglycemic, type 1 diabetic, or type 2 diabetic, and performed computational analyses.

Summary of Results
By unsupervised clustering, we found that chromatin accessibility profiles alone can identify the major pancreatic cell types, including five endocrine cell populations. Furthermore, the accessible chromatin regions of the various cell types showed enrichment for specific transcription factor binding motifs. Moreover, we observed a population of cells displaying a transitional phenotype related to but distinct from alpha cells. Cells of this population showed accessibility at the glucagon promoter, the defining marker of alpha cells, but also displayed multiple open chromatin regions enriched for the binding motifs of developmental transcription factors, raising the possibility that even adult human islets contain a subset of transitional and possibly progenitor type cells.

Conclusions
We demonstrate that chromatin accessibility captures important characteristics of cell type identity in the human pancreas. Moreover, snATAC-seq discovered a subpopulation of cells that show a progenitor-like phenotype. Thus, snATAC-seq reveal novel information and putative transitional cell types that complement gene expression analysis by scRNA-seq. Future integration of these data promises to further our understanding of cell type identity and function of human pancreatic cells.

Presenter

Maria Redondo, Baylor College of Medicine

Authors

Maria J. Redondo, Sarah Richardson, Daniel Perry, Charles Minard, Alice Carr, Irina Kusmartseva, Alberto Pugliese, Mark Atkinson

Purpose

The type 2 diabetes (T2D)-linked TCF7L2 variants are associated with features atypical for type 1 diabetes (T1D) suggesting that autoimmunity and beta cell loss are less severe. Hence, we determined the influence of TCF7L2 SNPs on insulin containing islet (ICI)% in T1D pancreata.

Methods

We analyzed 111 nPOD donors (www.jdrfnpod.org) with T1D who had data on the presence or absence of residual beta cells. In those donors with remaining beta cells, formalin fixed paraffin embedded (FFPE) pancreas tissue sections from regions containing beta cells (as determined by the nPOD Pathology Core) were stained for the presence of insulin-positive and glucagon-positive cells. The ICI% was calculated by quantifying the number of islets containing insulin-positive beta cells, and expressing this as a percentage of the total islets analyzed within each donor. The observed distribution of ICI% was examined to determine an interesting cutoff value. The largest difference between consecutive observations under 10% occurred between 2.2 and 6.4. Therefore, high ICI% was defined as ≥5% for this study. Mean age at T1D onset was 12.2 years (sd=7.9) (range=0-36), diabetes duration was 15.2 years (sd=13.7) (range=0-74), BMI was 24.5 (sd=4.6), 53% were male, 80% non-Hispanic white, 13% African American, and 7% Hispanic.

Summary of Results

Mean ICI% was 9.7 (sd=21.5) (range=0-92.2). At the TCF7L2 rs7903146 locus, 45.5% (50/110) of donors carried the T2D-associated T allele (41.8% as TC and 3.6% as TT). Donors with high (≥5) (n=30, 27%) vs low ICI% (<5) were older at onset (15.3±6.9 vs 11.1±7.9 years, p=0.013), had shorter diabetes duration at procurement (7.0±7.4 vs 18.3±14.3 years, p<0.001), and had higher African ancestry score (0.2±0.3 vs 0.1±0.2, p=0.043) and lower European ancestry score (0.7±0.3 vs 0.9±0.3, p=0.023). There were no significant differences in sex, BMI or TCF7L2 SNPs allele distribution but African American race was more common among high ICI (23.3%) than low ICI (8.6%, p=0.038). In multivariable logistic regression predicting high ICI % with adjustmentfor age of onset (p=0.086), diabetes duration (p<0.001), BMI (p=0.796), sex (p=0.417) and African American race (p=0.050), donors with the TCF7L2 rs7903146 T allele (TC or TT), compared to those without it (CC) were 2.91 times (95%CI=1.02-8.3) more likely to have high ICI% (p=0.046). A similar model was also fit including the interaction between African American race and rs7903146 T allele (p=0.021). Among African American donors, the odds of high ICI were 146.6 (95%CI=4.1-5300) times greater among donors with the T allele compared to those without the T allele. Among non-African American donors, the odds ratio was 1.7 (95% CI=0.5-5).

Conclusions

Among nPOD donors with T1D, carriers of the T2D-linked TCF7L2 variant had a higher number of residual ICIs, possiblyrepresenting a disease endotype characterized by T2D-like features and less severe beta cell loss.

Presenter Herbert Gaisano (University of Toronto)

Authors

Tairan Qin, Tao Liang, Li Xie

Purpose

The pancreatic islet is a complex micro-organ that regulates glucose homeostasis from the actions of major hormones secreted from three endocrine cell types (beta, alpha and delta cells) which become severely perturbed in type-1 diabetes (T1D), particularly in their cross-talk. A major complication in T1D is iatrogenic hypoglycemia wherein alpha-cells become insensitive to low glucose induced by insulin treatment. alpha-cell ‘glucose blindness’ has been attributed to delta-cell somatostatin (SST) inhibition of alpha-cell that becomes accentuated when beta-cells are destroyed in T1D, but the precise mechanism to explain these effects is not completely understood.

Methods

Syntaxin-1A is the key SNARE protein that forms the putative membrane-fusion SNARE complex mediating secretory granule exocytosis in neurons and neuroendocrine cells, including the 3 major islet cell types. We here induced delta-cell exocytosis blockade by treating Syntaxin(STX)-1A flox/flox mice with intraperitoneal injection of adeno-associated virus (AAV8)-SST promoter-Cre/RFP.

Summary of Results

Immunocytochemistry demonstrated STX-1A deletion only in the red-colored delta-cells as verified by SST and Cre antibody staining. Delta-cell STX1A-KO, compared to WT mice, showed improved glucose homeostasis with larger biphasic rise in blood insulin levels during IPGTT; and ITT (insulin injection) showed improved insulin sensitivity with larger increases in glucagon release (60 min and 120 min). Whole pancreas perfusion performed on KO vs WT mice showed the following: 1) confirmed the reduction and flattening of stimulated SST secretion; 2) a higher increase in glucose (10 mM)-stimulated insulin secretion that was suppressible by exogenous SST to the same level as WT mouse pancreas; and 3) a much higher hypoglycemic (1 mM glucose) induced glucagon release that was also suppressible by exogenous SST to the same level as WT mouse pancreas. Initial exocytosis imaging with NPY-pHluorin that’s specifically expressed in delta-cells in islets within pancreatic slices showed 10 mM glucose increased abundant exocytosis in WT delta-cells but with almost no exocytosis from STX1A-KO delta-cells.

Conclusions

Future studies. We plan to induce T1D in these WT and delta-cell STX-1A KO mice with streptozotocin (STZ) to assess whether the clinically-observed loss of hypoglycemia-induced glucagon release causing the ‘hypoglycemic blindness’ could be restored by the delta-cell STX1A deletion. We will then assess by exocytosis imaging (NPY-pHluorin) and electrophysiology of the islets within pancreatic slices prepared from these mice the effects of delta-cell exocytotic blockade caused by STX-1A deletion on whole islet alpha-cell (and also beta-cell) secretory responses and the corresponding changes in alpha-cell electrophysiology before and also after the STZ treatment. As well, we will recapitulate some of these findings in human T1D pancreatic slices vs normal pancreatic slices obtained from nPOD. This study when completed will elucidate the mechanistic basis by which delta-cell exocytotic blockade restores the alpha-cell secretory response to iatrogenic insulin-induced hypoglycemia.

Supported by Helmsley Charitable Trust and the Canadian Institute of Health Research

Presenter
Nagesha Guthalu Kondegowda (Arthur Riggs Diabetes and Metabolism Institute)

Authors
Nagesha Guthalu Kondegowda, Joanna Filipowska, Nancy Leon-Rivera, Rollie Hampton, Rosemary Li, Selassie Ogyaadu, Clive Wasserfall, Mark Atkinson, Helna Reijonen, Yuan Yate-Ching, Navneet Dogra, Daniel Roeth, Markus Kalkum, Carol Levy, Susmita Sahoo, Rupangi C. Vasavada

Purpose
The goal of these studies is to examine the role of humoral factors and circulating extracellular vesicles (cEVs) in the pathogenesis of Type 1 diabetes (T1D). T1D is an autoimmune disease in which beta cell loss and dysfunction play a vital role in its pathogenesis. EVs, small membrane bound structures, are secreted into the circulation by almost every cell type. Due to their distinct cargo, EVs act as molecular messengers and inter-organ communicators, and play an important role in both normal and pathophysiological conditions. EV cargo content is based on the environment of the tissue it is being secreted from and can alter the phenotype of recipient cells. Humoral factors in the serum from T1D subjects can be cytotoxic to rodent beta cell lines and islets. Also, serum EVs of long term T1D patients carry distinct miRNA cargo and cause defect in β-cell secretion. Therefore, we hypothesized that cEVs from T1D subjects, at different stages of the disease, are detrimental to islet health, through their differential RNA and protein cargo, with the potential to serve as disease biomarkers.

Methods
Serum and plasma samples were obtained from early (1-5 years since diagnosis) and late (>10 years since diagnosis) stage T1D patients, from autoantibody positive (Aab+) donors, and age, sex and ethnicity matched non-diabetic healthy donors (HD). cEVs and EV-depleted fractions prepared from T1D and HD plasma using ultracentrifugation and exoquick methods, were characterized by Dynamic light scattering (DLS), Nanoparticle Tracking Analysis, Western blotting and transmission electron microscopy. PKH26 labeled cEVs were used for uptake studies in human islet cells. Cytotoxicity assays were performed on rat insulinoma cell line (INS1), mouse and human islet cells cultured for 24h in media in which fetal calf serum (FCS) was substituted for human serum (10%v/v), or in islet media treated with cEVs, from the donors indicated above. Cleaved caspase 3 staining, TUNEL with insulin and glucagon co-staining, were used to assess cell death. Mouse serum from T1D NOD female mice at different stages of the disease was examined similarly. Seahorse analysis was used to assess the effects on mitochondrial function in human islets. RNAseq and proteomic analyses of cEV cargo and the functional contribution of candidate miRNAs towards b-cell cytotoxicity was assessed.

Summary of Results
Our initial findings show that there were no obvious differences in physical characteristics, cell surface markers, size and number, in cEVs from T1D versus HD plasma. cEVs from T1D and HD donors tagged with PKH26 dye showed uptake in INS1 cells and human beta cells. Serum and plasma from T1D donors (both early and late-stage) (n=5-9) significantly increased cell death in INS1 cells, and in human beta cells, comparable to cell death levels induced by pro-inflammatory cytokines, suggesting that humoral cytotoxicity may persist with disease progression. The same findings hold true when serum from early and late stage of the disease from female NOD mice was used. Our initial data indicate that serum from Aab+ subjects (n=8) induce cytotoxicity in human beta cells, similar to that induced by serum from T1D subjects, suggesting that the humoral beta cell cytotoxicity occurs early, before disease onset. Seahorse analysis suggests that T1D serum-treatment reduces the mitochondrial respiratory spare capacity of human islets compared to HD serum (n=4).
We then tested our hypothesis that the humoral cytotoxicity against beta cells seen in the serum and plasma of T1D patients is mediated by cEVs. Indeed, cEVs from T1D donors, but not HD (n=8/group), significantly induced human beta cell death, and this was not observed with the EV-depleted fraction. Relevant to disease pathology, T1D cEV-induced cytotoxic effect was specific to human beta cells but did not induce cell death in human alpha cells in the same islet prep. To investigate the cargo and molecular mechanisms, we performed a pilot RNASeq and proteomic analysis on cEVs (n=5) from T1D and HD subjects. We identified differential miRNA, lncRNA, mRNA and protein cargo in cEVs from T1D vs HD subjects. A functional analysis of candidate miRNAs in T1D cEVs showed specific miRNAs are detrimental to beta cells.

Conclusions
Serum, plasma, and plasma-derived cEVs from human T1D subjects induce human beta cell but not alpha cell death in vitro. Humoral cytotoxicity on beta cells may initiate early in pre-disease at-risk Aab+ individuals, and likely persist in late-stage disease. The differential RNA and protein cargo in T1D cEVs likely mediates the detrimental effects on human beta cells. Our findings suggest that cEVs, at various stages of the disease, play a role in the pathogenesis of T1D as related to the beta cell. Our ongoing studies are investigating the effects of the cEVs on the immune system, their molecular mechanisms of action, as well as the differential RNA and protein cargo in cEVs from different stages of the disease, with the potential to develop new biomarkers and therapeutics for T1D.