

Presenter
Esra Karakose, PhD (Icahn School of Medicine at Mount Sinai)
Authors
Esra Karakose, Luca Lambertini, Xuedi Wang, Saul Carcamo, Dan Hasson, Peng Wang, Huan Wang, Carmen Argmann, Andrew Stewart
Purpose
Type 1 diabetes results from insufficient numbers of insulin-producing beta cells. Our previous studies have shown that insulinomas – rare and benign human pancreatic adenomas – hold the transcriptomic and genomic ‘recipe’ for inducing human beta cell replication. They yield suggestions for therapeutic regenerative drugs for human beta cells, such as the DYRK1A inhibitor harmine, TGF beta Inhibitors, KDM6A inhibitors, and GLP-1 receptor agonists. Interestingly, we observed epigenetic misregulation in majority of the insulinomas we surveyed in our study. Therefore, we took a deeper dive into insulinoma epigenetics in order to discover novel therapeutic targets for human beta cell replication.
Methods
We performed ATAC-seq and H3K27me3 ChIP-seq experiments on FACS-sorted human beta cells and compared these to human insulinomas. In downstream bioinformatics analyses, we integrated and correlated the ATAC-seq and ChIP-seq results with RNA-seq dataset on human beta cells and insulinomas.
Summary of Results
We identified more than 19,000 open chromatin regions specific to insulinomas. Further, we also found more than 7,000 open chromatin regions specific to human beta cells. Our H3K27me3 ChIP-seq results yielded ~900 insulinoma- and beta cell-specific enhancers. Integration of ATAC-seq and H3K27Ac ChIP-seq datasets, together with transcription factor motif enrichment analysis, suggested distinct patterns of transcriptional networks in beta cells vs insulinomas. Integration of accessible chromatin regions with the enhancer regions in the genome yielded four separate clusters: 1- Unique insulinoma enhancers with an open chromatin signature; 2- Unique beta cell enhancers with an open chromatin signature; 3- Enhancers with an open chromatin signature shared between insulinomas and beta cells; and, 4- Open chromatin regions where no enhancer was detected. Interestingly, motif enrichment analysis indicated that AP-1 family members were highly significantly enriched in unique beta cell enhancers with open chromatin signature (cluster 2). Our preliminary knock-down and over-expression experiments on AP-1 family members in cadaveric human islets validated that they are important regulators of human beta cell replication and differentiation. These studies are currently ongoing, and the results will be further discussed.
Peptidome and immunopeptidome studies of human islets are limited by the short supply of high-quality human islets available, low MHC expression level, and poor recovery of low-affinity peptides. To understand the causality and relevance of known autoantigens in T1D, uncover new autoantigens (i.e. stress-generated neoantigens), and identify the iPSC-derived autoantigens responsible for reactivation of memory CD4 T cells, we will leverage our access to large quantities of iPSC-islets. Comparison of the peptidome of primary human and iPSC-derived islets revealed very similar percentages (P > 0.05, Ordinary 2-way ANOVA) of insulin, glucagon, somatostatin, pancreatic polypeptide, and ghrelin peptides. Additionally, iPSC-islets and primary human islets shared nearly identical profiles for all insulin peptides. Importantly, these results are reproducible across multiple independent experiments (primary human islets, n = 3 biological replicates; iPSC-islets, n = 4 biological replicates). Finally, we tested the sensitivity and feasibility of our novel peptide binding/isolation assay – which we will use to uncover the MHC-II immunopeptidome of iPSC-islets. All the insulin peptides identified in the presence of 1 µg IAg7 and the peptidome extracted from the equivalent of 1X10E5 MIN6 cells were represented as overlapping peptides in the published MHC-II immunopeptidome of mouse islets.
Conclusions
Collectively, these studies indicate that insulinoma beta cells have altered ATAC-seq and histone ChIP-seq signatures when compared to cadaveric human beta cells which is also reflected in the downstream differences in expression of genes that control cell cycle and beta cell function. Most importantly, these studies identify AP-1 family as an important regulator of human beta cell replication and differentiation. Finally, our results clearly demonstrate that insulinomas continue to serve as a model to identify potential targets for novel therapies to induce human beta cell regeneration.
Presenter
Isabella Altilio, BS (University of Miami)
Authors
Isabella Altilio, Mayur Doke, Silvia Alvarez-Cubela, Dagmar Klein, Ricardo Pastori, Juan Dominguez-Bendala
Purpose
A longitudinal scRNAseq study of long-term cultured human pancreatic slices, previously published by our lab, revealed population shifts with ductal progenitor activation and detection of insulin-producing cells that appear to be transitioning from a ducto-acinar intermediate stage. Using emerging spatial transcriptomics approaches, we will link this groundbreaking dynamic transcriptomic landscape to real anatomical features in sections from T1D donors. In particular, we expect to detect extensive ductal remodeling and sprouting of putative beta cells, directly linking transitional clusters of cells from our dynamic scRNAseq analysis with distinct anatomical features from live pancreatic samples.
Methods
For spatial transcriptomic analysis, formalin-fixed paraffin-embedded (FFPE) human pancreatic tissue sections from selected type 1 diabetes (T1D) donors (nPOD cases #6240, #6263, #6083, #6380), autoantibody-positive donors (nPOD cases #6116, #6158), and non-diabetic donors (nPOD cases #6393, #6416) were subjected to immunofluorescence staining with CK7 (ductal), insulin (β-cells), and DAPI (nuclei). This staining enabled the identification of distinct ductal and endocrine regions, as well as hypothesized transitional β-cells at the duct-islet interface. Immunofluorescence facilitated the selection of regions of interest containing these transitional cells for further transcriptomic analysis. The 10X Genomics Visium HD Spatial Gene Expression platform will be applied, enabling whole transcriptome spatial analysis at single-cell resolution. Each capture area (6.5 mm x 6.5 mm) comprises approximately 11 million barcoded oligonucleotide capture spots (2 μm x 2 μm), allowing precise spatial mapping of gene expression. The spatial data will be processed in collaboration with 10X Genomics and visualized using the Loupe Browser. Custom R and Python tools will be used for downstream analysis, with reference scRNAseq data (GSE223713) providing context for observed expression profiles.
Summary of Results
We conducted a Visium HD spatial transcriptomic analysis using Seurat version 5 on Type 1 Diabetes (T1D), Ab+, and control pancreas samples from nPOD cases #6240, #6263, #6038, #6380, #6116, #6158, #6393, #6416. Our preliminary analysis aimed to assess cellular composition changes, particularly focusing on islet and ductal regions. In control samples, we observed a clear presence of beta cells within islet regions. However, in T1D samples, the islet regions were largely devoid of beta cells, consistent with disease pathology. Interestingly, in T1D samples, we identified a subset of cells near the ducts that expressed TFF1 and SPP1, known markers associated with progenitor-like states. Notably, some of these progenitor marker-expressing cells also showed expression of the insulin (INS) gene.
Conclusions
These initial findings suggest potential progenitor cell activity near ductal structures in T1D-affected pancreas tissue. Deeper ongoing analyses are expected to offer valuable insights into the cellular heterogeneity in T1D and provide a basis for further exploration of progenitor cell contributions to pancreatic remodeling and regeneration in T1D.
Presenter
Leslie Wagner, PhD (Indiana University)
Authors
Leslie Wagner, Olha Melnyk, Charanya Muralidharan, Matthew Arvin, Michelle Martinez-Irizarry, Bryce Duffett, Justin Crowder, Elisabetta Manduchi, Klaus Kaestner, Joseph Brozinick, Amelia Linnemann
Purpose
Type 1 diabetes (T1D) is characterized by the autoimmune destruction of β cells hypothesized to result from a combination of genetic and environmental factors, likely including early childhood viral infection. Viral infections elicit the generation of type 1 interferons, such as interferon alpha (IFN-α), a cytokine that induces markers of early diabetes development including ER stress and HLA class I overexpression. IFN-α is also expressed in islets of donors with early T1D, suggesting a role in T1D pathogenesis. Here, we evaluated the direct impact of IFN-α on β cell physiology, primarily focusing on reactive oxygen species (ROS) generation, as excess ROS generated by numerous cytokines are thought to correlate with negative outcomes during disease pathogenesis.
Methods
To visualize ROS generation in vivo, human islets were transduced with β cell selective adenovirally packaged GRX1-roGFP2, a ratiometric and reversible ROS biosensor (INS-GRX1- roGFP2). Biosensor expressing islets were transplanted under the kidney capsule of NOD scid gamma (NSG) mice. Intravital imaging was performed on anesthetized mice two weeks later to evaluate the acute response to systemic human IFN-α elevation. To investigate the specific ROS being produced, we treated human islets in vitro with human IFN-α then stained them with either dihydroethidium (DHE) or 2′,7′-dichlorofluorescein diacetate (DCFDA) to measure superoxide or hydrogen peroxide, respectively. To determine if the observed cytoplasmic ROS first originates in the mitochondria, we treated EndoC-βH1 cells with human IFN-α and stained with either MitoSOX Green, a mitochondrion-specific superoxide fluorescent probe, or transfected cells with a plasmid expressing mito-roGFP2, a mitochondrion-specific ROS biosensor. To determine the unique molecular signature predisposing this subset of β cells to ROS production, we treated human islets with recombinant human IFN-α in vitro and flowsorted the ROS accumulating cells to perform bulk RNA sequencing, which was evaluated on its own or in comparison to single cell sequencing data collected by the Human Pancreas Atlas Program (HPAP).
Summary of Results
IFN-α exposure leads to acute robust ROS elevation in a subset of human β cells (20%-60%, donor specific), termed ‘responders’, in as little as 15 minutes. Comparison of our data with phenotypic information provided by the Integrated Islet Distribution Program (IIDP) indicated that healthier donors (with lower BMI and HbA1C) had a higher number of responders. In vitro IFN-α treatment of human islets elicited a heterogenous increase in DHE intensity, suggesting that superoxide is the specific ROS being produced under these conditions. This response is significantly blunted in the islets of donors with type 2 diabetes. EndoC-βH1 treated with IFN-α accumulated mitochondrial superoxide that could be reversed by the mitochondrially targeted antioxidant, MitoQ. Lastly, RNA sequencing identified an enrichment of genes involved in inflammatory and immune response in the responders and a reduction in genes involved in apoptosis. Gene set enrichment analysis (GSEA) on HPAP data indicated that genes upregulated in responders are enriched in controls as compared to T1D donors.
Conclusions
IFN-α induces a heterogenous ROS accumulation in human β cells both in vivo and in vitro. The accumulating ROS species is primarily superoxide that is generated in the mitochondria. The number of ROS accumulating cells is strongly negatively correlated with donor BMI and HbA1C, and islets from donors with type 2 diabetes do not accumulate ROS in response to acute IFN-α exposure in vitro. The RNA signature of ROS accumulating cells also negatively correlates with the signature of β cells from donors with type 1 diabetes. Therefore, our data collectively suggests that ROS accumulation in response to IFN-α may be beneficial to the β cell and promote β cell survival.
Presenter
Patricia Velado, PhD (Helmholtz Munich)
Authors
Patricia Velado, Teresa Rodriguez-Calvo
Purpose
Type 1 diabetes (T1D) is a complex and heterogeneous disease characterized by the immunemediated destruction of pancreatic beta cells. This destruction might be linked to beta cell stress, which can lead to abnormal protein processing and accumulation, potentially triggering immune responses. While most of the research has focused on proinsulin processing errors in beta cells, other molecules within the insulin granules, specifically chromogranins, secretogranin (collectively known as granins) and islet amyloid polypeptide (IAPP) could undergo similar processing errors and be targeted by the immune system. Granins are essential for granule assembly, insulin sorting and condensation. In addition, granins like Secretogranin 5 (SCG5) function as a chaperone for the prohormone convertase 2 (PC2) enzyme, while secretogranin 8 (SCG8 or PCSK1N) inhibits the prohormone convertase 1/3 (PC1/3) enzyme. This study aims to determine the expression, cellular localization, and molecular interactions of chromogranins, secretogranins, and IAPP in pancreatic beta cells.
Methods
We analyzed the expression of chromogranins, secretogranins, and IAPP during the progression of T1D by performing immunohistochemistry on pancreatic tissue samples from non-diabetic (ND), double autoantibody-positive (dAAb+), and T1D donors. We characterized the cellular composition of each islet (beta and alpha cells) using whole-slide images and quantified the percentage of beta and alpha cells positive for CHGA, CHGB, SCG2, SCG3, SCG5, SCG7, SCG8, and IAPP. Additionally, we assessed intensity features as a proxy for protein expression, both at the islet level and within individual islet cells.
Summary of Results
We observed a reduction in the proportion of beta and alpha cells expressing CHGA, CHGB, SCG2, SCG3, SCG5 and SCG7 in dAAb+ donors. This indicates that granin expression is altered in both endocrine cell types prior to clinical onset of the disease, potentially indicating impaired biogenesis and assembly of secretory granules. Interestingly, the remaining beta cells in T1D donors showed granin expression levels similar to those of non-diabetic donors. Consistent with previous findings, we observed reduced IAPP in beta cells from T1D donors. Moreover, the fluorescence intensity of CHGA and SCG2 progressively increased with disease progression, with higher levels observed in both beta and alpha cells of dAAb+ and T1D donors compared to non-diabetic individuals.
Conclusions
Our findings suggest that alterations in granin expression, especially in the early stages of T1D, may serve as an important marker of beta cell dysfunction. In the dAAb+ donors, beta cells with disrupted granin expression may be preferentially targeted by the immune system, whereas beta cells that retain functional granin cargo may be among the last to be destroyed. The observed increase in granin intensity, particularly CHGA and SCG2, in beta cells from dAAb+ and T1D donors suggests that the remaining beta cells may compensate for the beta cell loss by upregulating their granin expression. Overall, these results imply that alterations in granin expression occur before disease onset and could serve as early indicators of T1D.
Presenter
Julia Panzer, PhD (City of Hope)
Authors
Julia Panzer, Anna Lang, Pablo Garcia, Alberto Pugliese, Alejandro Caicedo
Purpose
Alpha cells are critical regulators of glucose homeostasis, secreting glucagon to prevent hypoglycemia. Recent studies indicate that alpha cells are not a uniform population but instead exhibit significant functional heterogeneity, with subpopulations differing in their signaling dynamics, hormone secretion, and responsiveness to stimuli. This heterogeneity suggests that alpha cells play diverse and specialized roles in maintaining metabolic balance. Understanding these functional differences is essential for advancing our knowledge of alpha cell physiology and the broader regulatory networks within the islet microenvironment.
Methods
We initially utilized a mouse model in which alpha cells express the genetically encoded calcium indicator GCaMP6, enabling visualization of calcium dynamics across the entire alpha cell population. We now started to extend these studies to human samples using both isolated islets and pancreatic tissue slices from both non-diabetic donors and donors with type 1 diabetes. We assess alpha cell function under three primary conditions: (a) varying glycemic levels, (b) application of agonists and antagonists targeting autocrine and paracrine pathways, and (c) stimulation with reference agents such as adrenaline and KCl. Functional outcomes are measured using dynamic hormone secretion assays and real-time calcium imaging to evaluate cellular activity.
Summary of Results
Our functional studies reveal evidence for significant alpha cell heterogeneity. Alpha cell responses to glucose differ markedly from beta cells, as they exhibit no sustained glucagon secretion. Instead, secretion is transient and shows hysteresis, meaning that alpha cells do not faithfully track glucose concentrations but instead depend on prior glycemic states. Calcium imaging revealed that only about 30% of alpha cells respond to changes in glucose, highlighting functional heterogeneity within the alpha cell population. Calcium imaging also identified subpopulations of alpha cells with distinct responses to other stimuli, including varying reactions to paracrine signals like serotonin and somatostatin. Hormone secretion studies confirmed this variability: blocking serotonin signaling with a 5-HT1F receptor antagonist nearly doubled glucagon secretion in low glucose, while blocking somatostatin signaling with cyclosomatostatin increased glucagon secretion by ~50%.
Conclusions
Our findings provide strong evidence for functional heterogeneity among alpha cells, with distinct subpopulations exhibiting differential responses to regulatory cues and hysteresis. This suggests the hypothesis that alpha cell dysfunction in T1D may be limited to specific subpopulations rather than affecting the entire population uniformly. Studying these subpopulations in T1D is crucial, as it opens the possibility of targeting dysfunctional groups pharmacologically to restore glucagon secretion and improve glucose regulation.
Presenter
Sandra Ferreira, PhD (University of Florida)
Authors
Sandra Ferreira, Austin Stis, Adrienne Widener, Edward Phelps
Purpose
Intracellular calcium [Ca2+]i is a crucial messenger responsible for modulating several functions inside of the cell. In pancreatic β-cells, the amount of insulin secreted is proportional to the transient amount of [Ca2+]i. Ca2+ oscillations, driven by bursts of action potentials, in turn, give rise to the oscillatory characteristics of insulin secretion. GABA (γ-aminobutyric acid) is a neurotransmitter synthesized and secreted by pancreatic β-cells. The high GABA concentration in β-cells, second only to neurons, indicates an important role in islet function. However, while numerous effects of applied exogenous GABA have been studied, there has never been a way to study the endogenous GABA because global knockout is embryonically lethal, and a GABA-synthesizing enzyme β-cell specific knockout has never been produced until now. GABA can signal through a chloride channel, the GABAA Receptor (GABAAR), in β-cells. The ion channel opens upon GABA binding, leading to Cl- movement through the plasma membrane, directly affecting the membrane potential. Our previous results showed that the modulation of GABAAR activity by pharmacological agonists and antagonists affects [Ca2+]i, indicating endogenous GABA may modulate Ca2+ behavior. Here, we aimed to evaluate the Ca2+ oscillations characteristic in islets from knockout mice lacking GABA in pancreatic β-cells
Methods
We isolated islets from conditional knockout mice lacking the GABA-synthesizing enzymes GAD65 (encoded by Gad2) and GAD67 (encoded by Gad1), specifically in pancreatic beta cells (β-Gad1,2-/-) and Ins1-Cre mice. To confirm the knockout, GABA content was measured by HPLC, and GAD65 and GAD67 expression were assessed using Western Blot. Insulin secretion and Ca2+ oscillations were evaluated on different glucose concentrations.
Summary of Results
HPLC and Western blot confirmed that knockout of GAD65 and GAD67 results in islets devoid of GABA. β-Gad1,2-/- islets secreted more insulin in all glucose incubations (8, 11.1, and 16.7 mM) and KCl than Ins1-cre islets, indicating that endogenous GABA can inhibit the β-cell function in an autocrine fashion. To evaluate if GABA modulates insulin secretion through the Ca2+ signal, we recorded [Ca2+]i in different glucose concentrations. We observed that β-Gad1,2-/- islets responded more rapidly to glucose (around 2 min before) and remained maximally activated longer before initiating rhythmic Ca2+ oscillations. Evaluation of the oscillations showed that the β-Gad1,2-/- islets exhibited a significantly longer active phase and shorter silent phase, which reduced the trough-to-peak amplitude. This increase in active phase duration resulted in the Ca2+ levels failing to return to baseline before starting a new Ca2+ wave and, thus, oversecretion of insulin.
Conclusions
We concluded that GABA is crucial to modulating the Ca2+ wave components, maintaining the equilibrium between the active and silent phases, resulting in insulin secretion at a proper level.
Presenter
James Johnson, PhD (University of British Columbia)
Authors
James Johnson, Weston Elison, Hanna Mummey, Patrick MacDonald, Xiao-Qing Dai, Theodore dos Santos, James Lyon, Jasmine Maghera, Vira Kravets, Nirmala Balasenthilkumaran, Francis Lynn, Chieh Min Chu, Natalie Nahirney, Kyana Chen, Abby Gordon, Søs Skovsø, Kyle Gaulton
Purpose
Mechanisms underlying the early stages of type 1 diabetes (T1D) pathogenesis remain to be elucidated. The insulin gene (INS) is the 2nd largest risk driver, explaining ~10% of heritability. INS locus fine mapping identified 3 independent T1D risk signals. It remains unknown how risk variants at the INS locus contribute to disease. Published reports suggest that risk variants may paradoxically increase beta cell insulin production, consistent with lower glucose infants with INS risk alleles. We therefore sought to understand the (patho)physiological and molecular mechanisms by which INS gene variation drives T1D risk.
Methods
Genome-wide association, eQTL mapping, single-cell multiomics, patchSeq, live-cell imaging, proteomics, and functional assays in primary human beta cells and stem-cell derived beta-cells, and NOD mice, were used.
Summary of Results
Genetic risk was associated with increased INS regulatory activity in human beta cells in single cell multiome data. Patch-seq showed that beta cells from non-diabetic donors with the at-risk C/C allele at rs3842735 and significantly higher INS mRNA and the potential for hyper-activity, with larger Ca2+ currents and greater exocytotic capacity. Ca2+ currents and exocytosis were increased in beta-cells isolated from T1D donors, compared with non-diabetic donors. Primary beta cells and stem cell-derived beta-like cells with higher INS gene activity, high protein synthesis rates, and more mature transcriptome and proteome profiles were more susceptible to death across an array of stresses in vitro. In mice, beta cells with higher Ca2+ currents and exocytosis were more likely to be located in alpha cell-rich islet regions characterized by increased T-cell contacts. In female and male NOD mice, partial reduction of Ins2 in a subpopulation of beta cells with high insulin promoter activity resulted in a delay in T1D onset. We hypothesize that relatively higher insulin production, Ca2+ entry, and exocytosis can lead to increased stress in beta-cells, resulting in fragility and susceptibility to cell death, as well as more neo-autoantigens that can drive early autoimmune processes in T1D.
Conclusions
Risk factors for T1D may trigger the initiation of T1D in pancreatic beta cells. Because insulin production is a modifiable factor, this conceptual framework may be leveraged in therapeutic strategies to prevent T1D initiation and/or progression.
Presenter
Alejandra Petrilli, PhD (AdventHealth)
Authors
Alejandra Petrilli, Carley Glass, Hayley Nielson, Charles Rowe, Yury Nunez Lopez, Richard Pratley, Anna Casu
Purpose
The pancreatic stellate cells (PSCs) constitute a heterogenous mesenchymal cell population. They have been implicated in the pathogenesis of pancreatic exocrine diseases. PSCs have been shown to sustain local inflammation, secrete pro-inflammatory cytokines, modulate myeloid cell
activation and contribute to their phenotype determination, participate in the extracellular matrix secretion and reabsorption and, in their activated state, reduce insulin secretion and induce pancreatic islet fibrosis. On the other hand, their role seems beneficial to islet transplant survival.
While all these functions could be important for type 1 diabetes (T1D) pathogenesis, the role of PSCs in T1D has not been studied.
Our central hypothesis is that PSCs activated by the inflammatory/autoimmune process in T1D contribute to β-cell functional decline through factors released in the microenvironment that act on the islets modifying their cell function and/or gene expression.
Methods
PSCs were obtained from pancreas slices (obtained from nPOD) and pancreas pieces (obtain from OurLegacy, the local organ procurement organization) of donors with and without T1D as described by Bachem et al. The isolated cells were positive for PSC markers (αSMA, vimentin, GFAP). Conditioned media from 80% confluent PSC cultures at passage 3 were collected and stored at -80°C. PSCs were harvested for mRNA extraction and RNA sequencing (Novogene Co.). Pancreatic islets from a healthy donor were cultured for 24 hours with PSC conditioned media
from T1D and non-T1D donors, and non-conditioned media. Islet equivalents (IEQ) were counted and glucose-stimulated insulin secretion was assessed using dynamic perifusion in Krebbs Ringer Bicarbonate Buffer (KRBB) + 0.1% albumin. mRNA sequencing was performed on the islets postperifusion. Descriptive statistics are presented. Differences were analyzed using t-tests, ANOVA, or non-parametric tests. Gene expression data were analyzed with systemPipeR in R, using edgeR for differential expression and GSEA for enrichment analysis.
Summary of Results
PSCs were obtained from 10 organs from donors with T1D and 10 organs from normoglycemic control donors of similar age, sex and race, matched by donor source (T1D: 8/10 M, 8/10 White, age 33.8±9.5, BMI 30.0±8.4, diabetes duration 9.0±7.1 Controls: 8/10 M, 6/10 White, age 33.0±11.7, BMI 28.0±5.3). Cultured PSCs from T1D donors maintained transcriptome differences compared to normoglycemic donors, with 37 differentially expressed genes (11 upregulated, 26 downregulated). Key downregulated genes (PDGFRA, TCF21, RGCC, EDNRB) were linked to vascular and organ morphogenesis and mesenchymal cell differentiation. GSEA showed significant downregulation of stellate cell markers in T1D PSCs, suggesting defects in pancreas homeostasis and repair. Upregulated genes included IL20RB and IL26, associated with autoimmune and inflammatory diseases.
Islets exposed to conditioned media from T1D and control PSCs showed no difference in basal insulin secretion (during perifusion with KRBB + 2.8 mM glucose). However, glucose-stimulated insulin secretion (GSIS) standardized per IEQ was lower in T1D (AUC 70.16±2.84 (mU/L)·min/100 IEQ) compared to controls (AUC 81.40±2.94 (mU/L)·min/100 IEQ) (p=0.0132), with a significantly higher stimulated peak in controls (p=0.029). The stimulation index behaved similarly (AUC p=0.00388 and peak p=0.067). No differences were observed when adding IBMX or KCl.
Conclusions
PSCs in T1D exhibit distinctive characteristics. Insulin secretion is reduced by conditioned media of PSCs obtained from donors with T1D. Further studies are needed to confirm these preliminary findings and identify the underlying mechanisms and mediators.
Presenter
Paola Apaolaza, PhD (Helmholtz Munich)
Authors
Paola Apaolaza, Yi-Chun Chen, C. Bruce Verchere, Teresa Rodriguez-Calvo
Purpose
In Type 1 diabetes (T1D), glucose dysregulation impacts various endocrine cells in distinct ways. In alpha cells, proglucagon is processed by PC2 enzyme to produce glucagon (GCG), and in gut L-cells is processed by PC1/3 enzyme (PC1) to synthesize glucagon-like peptide-1 (GLP-1). We have recently demonstrated that PC1-mediated proinsulin processing might be reduced in beta cells and mis-expressed in alpha cells from donors with T1D. Whether PC1 mediates GLP-1 processing in alpha cells of donors with T1D remained controversial. The aim of this study was to address the ongoing debate regarding the presence of PC1 in alpha cells by quantifying the co-expression of GCG, GLP-1, and PC1 at the single-cell level in donors with T1D.
Methods
FFPE-Pancreatic sections from sex, BMI, and age-matched (mean age 20.8 ± 4.5 years) non-diabetic (n=10 ND) and T1D donors with short-duration of the disease (<5 years after diagnosis) (n=10 T1D), were immunostained with amidated GLP-1, PC1, and GCG. All antibodies were validated in KO-mice models and by Western blot. Confocal microscopy images from islets (16.8 ± 4.4 islets/donor) were analyzed via the open-source software QuPath. First, islets were detected, and 6 different alpha cell populations were identified: single-positive for GCG or GLP-1; double-positive for GCG and GLP-1, GCG and PC1, or GLP-1 and PC1; and triple-positive for GCG, GLP-1, and PC1 (tGCG+GLP-1+PC1+). The percentage of alpha cells, as well as the protein relative fluorescence intensity levels (RFU) of GCG, GLP-1, and PC1 on a single cell level were quantified. Colocalization of PC1 with either GCG or GLP-1 was studied by measuring Pearson´s colocalization coefficient (PCC).
Summary of Results
The percentage of single and double positive alpha cells was low and did not differ between ND and T1D donors. The percentage of tGCG+PC1+GLP-1+ was significantly higher in T1D compared to ND donors (70.1 ±19.7% vs. 38.5 ± 21%). While cell mean intensity levels of GCG in tGCG+PC1+GLP-1+ alpha cells remained comparable among both donor groups, GLP-1 intensity was significantly elevated in T1D compared to ND donors (6.5 ± 2.8 vs. 4.1 ± 2.1 RFU). Although not significant, PC1 intensity levels were slightly lower in triple positive cells of T1D, 19.3 ± 6.7 vs. 26.6 ± 6.6 RFU in ND donors. Interestingly, colocalization of PC1 with GCG (0.78 ± 0.1 in T1D vs. 0.59 ± 0.1 in ND) and GLP-1 (0.65 ± 0.1 in T1D vs. 0.48 ± 0.1 in ND) was increased in T1D
Conclusions
In T1D, alpha cells exhibit dysfunctional behaviour, including excessive glucagon secretion. This study aimed to uncover the role of the PC1 enzyme in alpha cells. In our T1D cohort, we observed an increased number of cells positive for GCG, GLP-1, and PC1, though with reduced enzyme production per cell based on intensity measurements. Our findings indicate that PC1 in T1D alpha cells contributes to heightened GLP-1 expression, likely resulting from alternative proglucagon processing into GLP-1 via PC1. The alternative processing of GLP-1 in alpha cells may represent an adaptive response to stimulate insulin secretion and support beta cell survival amid beta cell loss and glucose dysregulation. Expanding the study to include more donors, particularly at-risk ones, would further illuminate the timing and extent of alpha cell changes over T1D progression. Nonetheless, these results could potentially give a rationale for developing new beta cell therapies by targeting alpha cell function.
Presenter
Kristen Wells, PhD (University of Colorado)
Authors
Kristen L. Wells, Thu A. Doan, Jennifer C. Whitesell, Robin S. Lindsay, Alan E. Buser, Sambra D. Redick, Alan G. Derr, Mason W. Tarpley, David M. Harlan, Sally C. Kent, Rachel S. Friedman
Purpose
Type 1 diabetes (T1D) is a chronic autoimmune disease characterized by a T cell mediated loss of pancreatic beta cells, resulting in a disruption in glucose homeostasis. Evidence from mouse models suggest that myeloid cells play a role in modulating T cell pathogenesis. They contribute to beta cell destruction through antigen presentation, stimulation of autoreactive T cells, and production of proinflammatory mediators. Conversely, they can also protect islets by promoting beta cell regeneration, conducting efferocytosis, and suppressing autoreactive T cells. However, little is known about the function of myeloid cells in human islets. Thus, our goal is to determine the function of human islet myeloid cells and their influence on the islet T cell response.
Methods
To understand the role of islet myeloid cell mediated efferocytosis in mice, we used the nonobese diabetic (NOD) mouse model of T1D and inhibited the myeloid-expressed efferocytosis receptor Mertk. We analyzed T cell activation in islets and T1D disease progression. To explore myeloid cell phenotypes in human T1D progression, we acquired single cell RNAsequencing datasets of all islet cells (HPAP) and of islet myeloid cells (collaborators) of nondiabetic (n = 35), autoantibody positive (n = 3), and short (n = 8) and long term (n = 6) T1D donors. The phagocytic nature of myeloid cells complicated the analysis. Passenger transcripts described by Lantz et al, are defined as transcripts that are taken up from the environment rather than transcripts produced by the cell. Because the cellular environment between the individuals with and without T1D was dramatically different, we developed a computational pipeline to predict and remove these passenger transcripts.
Summary of Results
Inhibition of Mertk in the NOD mouse model resulted in enhanced islet T cell activation and increased diabetes incidence, indicating that Mertk signaling prevents T1D progression. Immunofluorescent staining of human pancreas sections obtained from nPOD showed increased number of cells expressing Mertk in T1D islets containing insulin compared to the islets of non-diabetic donors, suggesting that Mertk-expressing myeloid cells may protect beta cells from autoimmune destruction. Transcriptional analysis of the single cell RNA-seq from human islet cells identified acinarspecific genes (ie CEL3A) and islet endocrine cell genes (ie INS and GCG) as top differentially expressed genes. This indicated that separating phagocytosed passenger transcripts from the myeloid cell-expressed transcripts is an important step for accurate identification of differentially expressed genes. Our novel pipeline was able to specifically remove environmentally derived genes (ie. CEL3A, INS, GCG) from the dataset while not impacting expression of myeloid cell genes (ie HLA-DQB1). Implementation of our pipeline in the human islet myeloid cell dataset allowed us to identify an enhancement in efferocytosis and suppression of antigen processing and presentation in islet myeloid cells from T1D donors with a short disease duration. We also identified islet myeloid cell functional heterogeneity including one population of macrophages with high expression of the efferocytosis genes.
Conclusions
Our findings from both the mouse and human data indicate that islet myeloid cells play a protective role against T1D progression in response to efferocytosis. This protection involves a Mertk-dependent mechanism that restrains T-cell mediated autoimmunity. In humans, islet myeloid cells increase expression of efferocytosis pathways and downregulate antigen processing and presentation.
Presenter
Maki Nakayama, MD (University of Colorado)
Authors
Laurie Landry, Amanda Anderson, Kristen Wells, Andrew MacMillan-Ladd, Jessie Barra, Aaron Michels, Holger Russ, Maki Nakayama
Purpose
Islet-specific T cells play an essential role in beta cell destruction in type 1 diabetes (T1D). While T cells in the pancreas and draining lymph nodes are crucial sources for understanding T1D pathogenesis, only a subset of these T cells may be specific to beta cells. Identifying phenotypes and functions of islet-specific T cells residing in pancreatic lymph nodes (PLN) will provide essential insights into understanding islet autoimmunity. This study aimed to establish a high-throughput T cell receptor (TCR) screening system to identify antigen-specific T cells and apply this platform to determine molecular signatures of islet antigen-specific T cells in PLNs of organ donors with T1D.
Methods
We developed a high-throughput TCR screening strategy, named “tZip,” to identify antigenspecific TCRs from thousands of T cells. Using tZip, we determined TCR clonotypes specific to preproinsulin, viral peptides, and stem cell-derived beta-like cells (sBC) from T cells in the pancreas and PLN of T1D organ donors. Additionally, we performed gene expression profiling of the PLN sample to determine phenotypes of beta cell-specific T cells.
Summary of Results
To validate the tZip platform’s efficiency, we analyzed 850 TCRs expressed by CD4 T cells isolated from the pancreas of T1D organ donors for their response to preproinsulin peptides presented by HLA-DQ8 and DQ8-trans. The final tZip vector libraries achieved 96-98% correct TCR alpha and beta pairing and over 99% correct TRAV and TRBV ligations. Over 95% of topscored TCRs enriched by tZip responded to proinsulin peptides, leading to the identification of dozens of novel TCRs. These TCRs, identified from multiple donors, recognize one of four peptide regions, enabling comprehensive identification of major epitopes involved in T1D pathogenesis. Furthermore, TCR clonotypes specific to sBC and viral peptides were successfully enriched from 1,450 CD8 T cells in PLNs of a new-onset T1D donor. Initial phenotypic analysis revealed that both sBC and viral peptides-specific T cells were clustered into either the stem cell memory or the effector memory cell subset. sBC-specific T cells express higher levels of Nur77 (recent antigen-exposure marker) compared to viral-specific T cells, whereas both express similar levels of CD5 (signature of past antigen exposure), which is higher than the rest of T cells.
Conclusions
The tZip platform demonstrates high efficiency and reliability in identifying antigen-specific TCRs from T1D donors, providing a powerful tool for studying T cell responses in autoimmune conditions. Our findings reveal distinct phenotypic signatures between beta cell-specific and viral-specific T cells in PLNs, suggesting ongoing beta cell-specific immune responses in T1D patients. This approach opens new avenues for understanding T cell-mediated autoimmunity.
Presenter
Brandon Dinner, BS
Authors
Brandon Dinner, Denise A. Berti, Anne Costanzo, Lisa Kain, Han Zhu, Christopher C.W. Hughes, Jeffrey R. Millman, Maike Sander, Luc Teyton
Purpose
Type 1 diabetes (T1D) is an autoimmune disease that results in the destruction of insulin-producing pancreatic β cells. Autologous iPSC-derived islet-like organoid (iPSC-islet) grafts are gaining traction as a therapeutic strategy to address clinical complications with insulin therapy and poor outcomes with allogenic islet transplantation. While clinical trials are underway, many basic immunological principles have been overlooked. Firstly, T1D is a CD4 T cell mediated disease; thus, initiating antigen presenting cells (APCs) must present MHC-II. Whether the β cell is directly capable of being this APC is debated, yet the mechanisms of β cell antigen presentation and destruction are likely recapitulated in iPSC-islet grafts. Secondly, β cells produce the autoantigens that drive the pathogeny of T1D. Our access to a large supply of iPSC-islets gives us the unique opportunity to investigate: 1) whether iPSC-derived β cells are a good surrogate to study human autoantigens, 2) the impact of inflammation on post-translational modifications, neo-antigens formation, and antigen spreading, and 3) the nature of iPSC-derived peptides supporting the reactivation of memory CD4+ T cells.
Methods
iPSC-islets were differentiated from iPSC cell lines of three diabetic donors and three non-diabetic donors. Peripheral blood mononuclear cells (PBMCs) from the 6 donors were isolated and frozen. Because genetic risk for T1D is strongly associated with HLA-II, donors were selected for their HLA-DQβ haplotype: HLA-DQ2 (HLA-DQB1*0201) and HLA-DQ8 (HLA-DQB1*0302). MIN6 cells, EndoC-βH1 cells, iPSC-islets, or primary human islets were cultured in vitro for 72 hours with or without 10 ng/mL IFNγ. Cell surface expression of MHC-I and -II was measured by flow cytometry and immunofluorescence. ATAC-seq was performed on ESC-derived and primary human β cells to assess chromatin accessibility of antigen presentation genes. For peptidome studies, we extracted whole peptidomes from iPSC-islets, primary human islets, MIN6 cells and EndoC-βH1 cells using a modified neuropeptide isolation protocol. The consequences of inflammation were evaluated after a 72h exposure of EndoC-βH1 cells to 0.5 ng/mL IL1-β, 10 ng/mL IFNγ, and 1 ng/mL TNFα. Raw LC-MS/MS data was analyzed in PEAKS X, PEAKS PTM, or PEAKS SPIDER against the UniProt human or mouse databases. Peptide filtering in silico was performed as previously described. To move from peptidome to immunopeptidome and overcome some limitations of classical methods, especially the loss of lower affinity peptides, a new binding/isolation assay was created. In this format, whole cell extracted peptides were incubated at 37ºC with recombinant MHC-II at acidic pH, +/-HLA-DM, for various time, and pMHC complexes were isolated by gel filtration (spin columns). Bound peptides were then analyzed by LC-MS/MS as described above
Summary of Results
After 72-hour treatment with IFNγ we did not detect MHC-II expression on the mouse insulinoma cell line MIN6, human EndoC-βH1 cells, primary human β cells, or iPSC-β cells, as measured by flow cytometry and immunofluorescence with the α-IAb antibody (M5/114.15.2) for mouse cells or α-HLA-DR (L243), α-HLA-DQ (SPVL3) and α-HLA-DP (B7/21) antibodies for human cells. Interestingly, upon IFNγ stimulation, both human primary islets and iPSC-derived islets express HLA-II on non-insulin, non-glucagon secreting cells. Otherwise, all cell types responded to IFNγ stimulation by an increase in MHC-I expression. As IFNγ response elements are the same for all MHC promoter regions, an epigenetic mechanism repressing MHC-II gene expression is likely. Following this idea, we performed ATAC-seq on embryonic stem cell-derived and primary human β cells for antigen presentation (AP) genes. In those cells, enhancers and transcription start sites (TSSs) of HLA-DR, DQ, and DP genes were inaccessible. TSSs of HLA-DM and CD74 remained accessible, but their enhancers were inaccessible. Similar trends follow for alpha cells, showing that these cells repress HLA-II transcription as well. Collectively, a non-α, non-β cell type is expressing HLA-II and might be responsible for presenting autoantigens to CD4 T cells in the peri-islet space.
Peptidome and immunopeptidome studies of human islets are limited by the short supply of high-quality human islets available, low MHC expression level, and poor recovery of low-affinity peptides. To understand the causality and relevance of known autoantigens in T1D, uncover new autoantigens (i.e. stress-generated neoantigens), and identify the iPSC-derived autoantigens responsible for reactivation of memory CD4 T cells, we will leverage our access to large quantities of iPSC-islets. Comparison of the peptidome of primary human and iPSC-derived islets revealed very similar percentages (P > 0.05, Ordinary 2-way ANOVA) of insulin, glucagon, somatostatin, pancreatic polypeptide, and ghrelin peptides. Additionally, iPSC-islets and primary human islets shared nearly identical profiles for all insulin peptides. Importantly, these results are reproducible across multiple independent experiments (primary human islets, n = 3 biological replicates; iPSC-islets, n = 4 biological replicates). Finally, we tested the sensitivity and feasibility of our novel peptide binding/isolation assay – which we will use to uncover the MHC-II immunopeptidome of iPSC-islets. All the insulin peptides identified in the presence of 1 µg IAg7 and the peptidome extracted from the equivalent of 1X10E5 MIN6 cells were represented as overlapping peptides in the published MHC-II immunopeptidome of mouse islets.
Conclusions
Our data supports the argument that, in response to pro-inflammatory cytokines, MIN6 cells, human EndoC-βH1 cells, primary human β cells, and iPSC-β cells can upregulate MHC-I expression but are not MHC-II inducible. ATAC-seq performed on embryonic stem cell-derived and primary human β cells for AP genes identified epigenetic repression of MHC-II antigen presentation genes. Using iPSC-islets at various stages of differentiation and ATAC-seq, we will address whether terminal differentiation is responsible for this repression. Furthermore, β cells undergo chronic ER stress from insulin production demands and addition of MHC-II production may trigger death by the unfolded protein response (UPR). We expect insulin knockout or drug attenuation of the UPR should allow β cells to become MHC- II inducible upon IFNγ stimulation. Finally, our work on peptidomics aims to provide the first peptidome and MHC-II immunopeptidome of iPSC-islets. We conclude, that at least on a peptidome level, iPSC-islets and primary human islets are remarkably similar. Thus, we propose that iPSC-islets are an appropriate tool to study human autoantigens. To address the generation of stress-related peptides, iPSC-islets and primary human islets will be exposed to metabolic, chemical, and environmental stressors. Peptidomes extracted from stressed or unstressed cells will then be subjected to peptide binding assays. This series of experiments should identify the most relevant β cell peptides capable of binding to recombinant HLA-DQ2, -Q8, Q2/Q8, -DR4, and -DQ6 (all produced and purified as CLIP-MHC complexes). HLA-DM dependency will also be determined. New peptides identified in whole peptidome binding assays will then be synthesized, and HLA-II tetramers made to interrogate the frequency of CD4+ T cells specific for the identified antigen in PBMCs of our T1D patient donors.
Presenter
Timothy Tree, PhD (King’s College London)
Authors
Timothy Tree, Eleni Christakou, Evangelia Williams, Stephanie Hannah, John Gregory, Colin Dayan, Danijela Tatovic, Vipin Narang
Purpose
Assays that assess the frequency and functional phenotype of T cells involved in beta cell destruction are uniquely positioned to offer valuable insights into the pathogenesis and progression of type 1 diabetes. They can enhance our understanding of the disease at every stage, from the initial loss of immune tolerance to the onset of clinical symptoms and may provide crucial information about the rate of beta cell loss after diagnosis. The T cell biomarkers identified are becoming an essential component of immunotherapy trials in type 1 diabetes, helping to identify and refine intervention targets and offering new insights into the reasons behind treatment success or failure.
Methods
We utilized a suite of highly sensitive and specific assays to investigate the frequency and phenotype of islet-specific T cells in individuals with and without T1D including samples from clinical trials aimed at preventing c-peptide loss in early stage 3 T1D. Assays included an unbiased activation-induced marker assay linked to single-cell transcription profiling incorporating paired TCR A and B sequencing (10x AIM) and a multiparameter cytokine FLUOROSPOT.
Summary of Results
Using the 10x AIM assay we demonstrate that whilst the total frequency of islet-specific T cells in the peripheral blood of individuals with and without new-onset type 1 diabetes is not significantly different, they display very different patterns of gene expression and TCR clonal expansion. Islet-specific T cells from those without T1D are 41haracterized by the expansion of islet-specific T cells with a predominantly regulatory phenotype, including populations of resting and memory FOXP3 Tregs and cytolytic CD4 regulatory T cells. In contrast, islet-specific T cells from individuals with T1D display a distinctive proinflammatory phenotype with a pronounced expansion of cells with a polyfunctional phenotype expressing multiple effector cytokines known to be directly toxic to beta cells (IFN-gamma, IL17A, IL-17F, IL-2 and GM-CSF), follicular helper T cells (Tfh) and activated FOXP3 Tregs. Moreover, profiling islet-specific T cells from individuals from the USTEKID clinical trial confirmed that the expansion of islet-specific T cells with a polyfunctional proinflammatory phenotype is a feature of new-onset T1D and that targeting these cells by IL-12/IL-23 inhibition was associated with a preservation of C-peptidestrengthening the evidence that these cells play a role in islet destruction.
Analysis of TCR use based on paired A and B chain sequencing showed evidence of T cell clonal expansions in islet-specific T cells in those with and without T1D. However, the frequency of clonally expanded T cells was higher in those with T1D and displayed a distinctive pattern. Whereas, in individuals without T1D, TCR sharing was only observed between T cells with similar phenotypes (e.g. sharing between resting and memory FOXP3 Tregs or between memory effector T cells), individuals with T1D showed extensive TCR sharing between cells with opposing phenotypes. In particular, TCR clonal sharing was observed between activated FOXP3 Tregs and polyfunctional proinflammatory effector T cells as a unique feature in the T1D group.
Conclusions
The expansion of islet-specific T cells with a distinctive polyfunctional phenotype in those with T1D and the link between their targeting and beta cell preservation suggests they may play an important role in islet destruction. Furthermore, TCR sharing raises the possibility that this proinflammatory population may arise due to FOXP3 Treg instability or transdifferentiation.
Presenter
Jessica Prendergast, PhD (Colorado University)
Authors
Jessica Prendergast, Catherine Nicholas, Amanda Anderson, Peter Gottlieb, Maki Nakayama, Kristen Wells, Mia Smith
Purpose
Type 1 diabetes (T1D) is a prevalent autoimmune disease affecting millions globally. While historically, T1D has been classified as a T cell mediated disease, B cells have a crucial role in disease development by functioning as antigen presenting cells (APCs) and modulating the cytokine milleu. Additionally, studies in non-obese diabetic (NOD) mice have provided evidence for autoreactive B cell involvement in T1D disease progression. Previous work from our group has identified public (shared) B cell receptors (BCRs) in the peripheral blood of individuals with Stage 1, 2, or 3 T1D but not in non-diabetic donors, some of which are reactive to islet-antigens. This study aims to determine whether there are public clones that are also present in the pancreatic lymph nodes (PLN) of individuals with T1D.
Methods
We performed single cell RNA- and BCR-sequencing on B cells enriched with islet-reactivity isolated from peripheral blood mononuclear cells (PBMCs) from individuals that are nondiabetic first-degree relatives (n=5), autoantibody positive (Aab+) Stage 1 or 2 T1D (n=6), and recent onset Stage 3 T1D (n=5). In addition, we performed single-cell RNA and BCR sequencing on PLN samples obtained from nPOD from 2 Aab+ and 3 recent-onset T1D donors. To understand the B-cell receptor (BCR) characteristics and to identify shared sequences, we searched for shared clones between samples from the PLN and PBMCs. Clonal identity was defined by the presence of identical V and J gene segments and a CDR3 sequence similarity of at least 80%.
Summary of Results
Of the more than 143,000 B cells sequenced, we identified 89 public clones among our PBMC samples, 66 of which were only found in AAb+ and T1D donors. The majority of these clones were reactive with at least one islet antigen, including insulin, IA-2, or GAD65. We identified shared AAb+ and T1D-specific public clones between the PLN and PBMC samples, some of which were islet-reactive.
Conclusions
These results demonstrate that public T1D-specific BCR sequences found in the peripheral blood are also present in the PLN of recent-onset T1D donors. Future studies will analyze the PLN BCR repertoire of non-diabetic control donors to determine whether these sequences are truly restrictive to T1D donors. In addition, we will compare the gene expression patterns of B cells from the PLN in Aab+, T1D, and non-diabetic donors to determine whether unique B cell phenotypes or subsets are present in autoimmune subjects but not in controls.
Presenter
Matthew Brown, BS (University of Florida)
Authors
Matthew Brown, Jessie Barra, Marcus Pina, James Proia, Todd Brusko, Holger Russ
Purpose
Generating pancreatic β-cells by differentiating human pluripotent stem cells (hPSC) provides a near-limitless pool of biocompatible cells for replacement therapies, overcoming the critical challenge of islet donor shortages. However, engineering durable therapies that can withstand
interactions with autoreactive T cells will be crucial for restoring long-term endogenous insulin production for individuals with type 1 diabetes (T1D). Across a cohort of 38 nPOD donors (n=18 HC, n=20 T1D), preliminary data show that laser-microdissected islets from individuals with T1D exhibit altered expression of the gene PVR, encoding the immunomodulatory receptor, CD155, compared to healthy controls. Here, we present a novel strategy for reducing the immunogenicity of human stem-cell-derived pancreatic β-cells (sBCs) by overexpressing a higher-affinity mutant (rs1058402; G>A; Ala67Thr) of CD155. This further promotes the immunosuppression of T cell activation by augmenting co-inhibitory signaling through the T cell checkpoint inhibitor, TIGIT, bolstering β-cell evasion of autoimmunity.
Methods
Using TALEN-mediated gene-editing, we engineered hPSC to overexpress either the mutant (CD155 Mut) or wild-type CD155 (CD155 WT) before differentiation into sBCs. Detailed phenotypical analyses, including dynamic glucose-stimulated insulin secretion assays (GSIS), were used to confirm genome-engineered lines and controls differentiated equally efficiently into glucose-responsive sBCs. To validate differences in CD155 affinity for its ligands, CD226 and TIGIT, fluorescently labeled immunoglobulin (Ig) fusion proteins were used to measure binding across a range of 0-25,000 ng/mL in hPSC and sBC of each genotype. To assess the allogeneic proliferative T cell responses induced by CD155 Mut, CD155 WT, and control sBCs, co-cultures with proliferation-dye labeled naïve CD8+ T cells were performed. Separately, we cultured sBCs with CD8+ T cell avatars specific for the β-cell autoantigen preproinsulin (PPI15- 24, clone 1E6) or a melanoma antigen (MelanA27-35, clone MART-1) as an irrelevant control. Chromium-release assays and supernatant cytokine multiplex assays were performed after 48 hours of co-culture to assess sBC immunogenicity and subsequent T cell cytotoxicity
Summary of Results
Compared to the unmodified parental hPSC line, engineered hPSC lines retained pluripotent stem cell markers and efficiently differentiated into functional sBCs. CD155 Mut sBCs displayed greater binding affinities for TIGIT-Ig and CD226-Ig compared to CD155 WT sBCs, as well as more potent inhibition of allogeneic CD8+ T cell proliferation. Preproinsulin (PPI)- specific 1E6 T cell avatars produced significantly lower levels of the cytolytic molecules, FasL, granzyme B, perforin, and granulysin, when co-cultured with CD155 Mut sBCs, compared to conditions with CD155 WT sBCs. In conjunction with these findings, CD155 Mut sBCs demonstrated augmented protection from T cell-mediated lysis (CML) in the contexts of both bystander killing with MART-1 T cell avatars as well as antigen-specific killing with 1E6 T cell avatars, compared to CD155 WT sBCs. Notably, this protective effect was ablated upon pretreatment of 1E6 T cell avatars with anti-TIGIT blocking mAbs to abrogate co-inhibitory CD155:TIGIT signaling upon co-culture with CD155 Mut sBCs.
Conclusions
Our findings collectively support the continued development of engineered sBCs overexpressing immunomodulatory proteins to suppress T cell activation as a targeted approach to protect pancreatic β-cell replacement therapies from alloimmune and reoccurring autoimmune destruction.
Presenter
Emerson Parks (University of Florida)
Authors
Emerson Parks, Leeana Peters, Matthew Brown, Amanda Posgai, Todd Brusko
Purpose
CD226 has been implicated in the development of numerous autoimmune diseases, including type 1 diabetes (T1D). The CD226 gene contains the risk variant rs763361 (C>T), which has been shown to contribute to dysregulated signaling and pro-inflammatory cytokine production. The role of CD226 and its associated risk variant in CD8+ T cells in T1D has yet to be fully defined. We examined the impact of CD226 expression on CD8+ T cell phenotype and antigenspecific cytotoxicity in peripheral blood samples from healthy donors. Our group has previously shown that treatment with a blocking antibody specific for CD226 (ɑCD226) in mice was therapeutically beneficial. Here, we investigated the effects of ɑCD226 treatment on human CD8+ T cell function. We also evaluated the expression patterns of CD226 and the competing co-inhibitory receptor, TIGIT, in nPOD organ donor derived spleen.
Methods
To assess the role of CD226 expression on human CD8+ phenotype, CD226+TIGIT-, CD226-TIGIT+, and CD226-TIGIT- CD8+ T cells were FACS sorted from whole blood and stimulated with ɑCD3/28 coated beads for 24 hours; then RNA transcripts were evaluated with the Nanostring nCounter system. CRISPR/Cas9 mediated CD226 knockout (KO) CD8+ human T cells were evaluated for phenotypic alterations by flow cytometry. IGRP-reactive CD8+ T cells were generated via lentiviral transduction and cocultured with βLox5 cells in the presence of ɑCD226 or isotype control to assess βLox5 cell death and CD8+ cytokine production via flow cytometry. Lastly, we examined the relevance of these phenotypes to tissue using flow cytometry of nPOD donor spleen.
Summary of Results
Transcriptomic analysis revealed that relative to the stimulated CD226-TIGIT+ subset, CD226+TIGIT- CD8+ T cells demonstrated increased expression of effector molecules such as GZMB (Log2 fold change=2.80), as well as the IL-2 family to JAK-STAT signaling pathway (p=0.009) and cytokine to JAK-STAT signaling pathway (p=0.014). We observed downregulation of negative regulatory molecules, CD96 (Log2 fold change=-2.18), and IKZF2 (Helios) (Log2 fold change= -8.98) in the CD226+TIGIT-population as compared to the CD226-TIGIT+ population. We confirmed this with CD226 KO in CD8 T cells, which expressed reduced CD96 relative to mock-edited CD8 T cells (p=0.048) after polyclonal T cell receptor (TCR) stimulation. Antigenspecific CD8+ T cells treated with ɑCD226 exhibited a significant reduction in target cell killing at the effector:target ratios of 5:1 (p=0.009) and 10:1 (p=0.0037). Moreover, at the 10:1 ratio, we observed a significant reduction in effector cytokine production in the ɑCD226 condition for granulysin (p<0.001), granzyme A (p=0.0017), granzyme B (p<0.001), interferon gamma (p<0.001), and perforin (p<0.001). In an external validation cohort of nPOD donor spleen, we observed reciprocal expression of CD226 and TIGIT between Helios- and Helios+ CD8+ T cell subsets, where Helios+ subsets expressed reduced CD226 and increased TIGIT.
Conclusions
These data support the notion that CD226 signaling is important for effector cytokine production, acts to oppose negative regulatory programs, and contributes to T1D antigenspecific cytotoxicity. Our data supports a growing body of literature that CD226 could be a viable therapeutic target to prevent or slow the progression of T1D.
Presenter
Chaitra Rao, PhD (Indiana University)
Authors
Chaitra Rao, Saptarshi Roy, Carmella Evans-Molina, Jon Piganelli, Decio Eizirik, Raghavendra Mirmira, Emily Sims
Purpose
Intercellular crosstalk between pancreatic β cells and immune cells significantly influences the progression of type 1 diabetes (T1D). Human leukocyte antigens (HLA), especially class I molecules (HLA-A, B, and C), are crucial for antigen presentation and immune recognition in T1D. As a protective mechanism against inflammation, surviving β cells increase their expression of programmed death-ligand 1 (PD-L1). PD-L1 interacts with programmed death-1 (PD-1) on immune cells, potentially limiting β cell destruction.
Extracellular vesicles (EVs) play a significant role in communication between islet cells. However, the specific contributions of different EV subtypes in β cell function and their independent or synergistic effects during T1D development remain unclear. This study aims to investigate how EV-associated PD-L1 and HLA-positive EVs influence immune responses and β cell function in the context of ongoing autoimmunity in T1D. Understanding these mechanisms could provide new insights into T1D pathogenesis and potential therapeutic strategies.
Methods
Human β cell lines and islets were treated with proinflammatory cytokines to model the T1D microenvironment. EVs were isolated and analyzed for HLA-A and PD-L1 content and function. Functional assays were conducted to test the ability of EV PD-L1 to bind PD-1 and inhibit NOD CD8+ T cells. Plasma EV PD-L1 levels were compared among islet autoantibody-positive (Ab+) individuals, recent-onset T1D patients, and non-T1D controls.
Summary of Results
PD-L1 and HLA class I molecules were detected on the surface of β cell EVs. Treatment with IFN-α or IFN-ɣ for 24 hours induced over a two-fold increase in EV PD-L1 and HLA cargo without a corresponding increase in the number of EVs. Only a small percentage (<0.5%) of the EVs showed colocalization of PD-L1 and HLA, while HLA colocalized with another tetraspanin EV marker (commonly associated with EVs) in about 60% of cases. Functional experiments demonstrated specific effects of β cell EV PD-L1 in suppressing proliferation and cytotoxicity of NOD CD8+ T cells. Plasma EV PD-L1 levels were increased in Ab+ individuals, particularly in those with single Ab positivity. Additionally, in individuals with either Ab+ status or T1D, but not in controls, plasma EV PD-L1 positively correlated with circulating C-peptide levels
Conclusions
Our findings demonstrate that β cell-derived EVs upregulate both PD-L1, a pro-tolerogenic protein, and HLA class I molecules, which are pro-immunogenic, in response to IFN exposure. Interestingly, only a small percentage of EVs co-express both molecules simultaneously, suggesting a heterogeneous population of EVs with potentially divergent immunomodulatory functions. β cell EV PD-L1 binds PD1 and inhibits CD8+ T cell proliferation and cytotoxicity. Clinical data show that circulating EV PD-L1 levels are elevated in Ab+ individuals compared to controls. Moreover, EV PD-L1 levels positively correlate with residual C-peptide at various stages of T1D progression, suggesting a possible protective role for β cell function. The dual presence of pro-tolerogenic and pro-immunogenic proteins on β cell EVs may contribute to the complex immunomodulatory landscape in T1D, potentially influencing disease heterogeneity and residual β cell function.
Presenter
Charles Lazimi, MS (University of Florida)
Authors
Charles Lazimi, Austin Stis, Clayton Mathews, Edward Phelps
Purpose
Real-time calcium imaging in human pancreatic tissues has traditionally relied on chemical sensors using strategies that continuously monitor a single islet at a time to capture dynamic calcium activity upon glucose stimulation. Such methodologies are powerful but result in a low-throughput approach. Transduction of live tissues with a Ca2+ biosensor, CaMPARI2, enables the simultaneous analysis of glucose-responsiveness in entire islet populations, minimizing user-introduced selection bias. Upon glucose stimulation and simultaneous exposure to 405nm light, CaMPARI2 fluorescence undergoes irreversible photoconversion from green to red, permanently marking Ca2+ activity over a defined time window. Tissues expressing CaMPARI2 can be fixed to enable multiplexed assays, correlating islet function with protein or RNA expression, offering insights into the mechanisms underlying functional heterogeneity in islets or islet dysfunction. This approach facilitated the functional assessment of up to 60 islets per nPOD human donor case, including those with T1D, T2D, mAab+, and non-diabetic cases.
Methods
To implement CaMPARI in pancreas slices, we obtained the second-generation expression construct CaMPARI2 from Addgene and subcloned it into an Adenoviral vector with CMV promoter (Ad-CMV-CaMPARI2). Human pancreatic tissue slices were received from nPOD and transduced via Ad-CMV-CaMPARI2. Widespread CaMPARI2 expression was confirmed at 24 hours. After the transduction period, slices were stimulated with 3mM glucose, 16.7 mM glucose, or KCl while simultaneously exposed to 405 nm photoconverting light (PC light) via a multi-well LED array. Slices were then fixed and immunostained for insulin and a monoclonal antibody that specifically enhances the signal from only the red photoconverted CAMPARI2 form. Tissues slices were then imaging with a confocal microscope and analysis of photoconversion analyzed with ImageJ.
Summary of Results
Calcium activity is quantified by measuring photoconversion within cells, reported as the CaMPARI2 Red:Green ratio. CaMPARI2 signal is normalized to expression to account for any differences in transduction efficiency. We observed strong β cell photoconversion in only the slices exposed to the combination of high glucose and PC light. We then assessed the function of pancreatic tissue slices from nPOD donors across clinical groups. Dysregulated glucose sensing in the diseased nPOD cases, observed by using CaMPARI2, was validated by insulin secretion (perifusion) and real-time calcium imaging data (by fluorescent microscopy). In contrast, the ND control nPOD case showed normal glucose responsiveness resulting in significant photoconversion.
Conclusions
Our findings demonstrate the utility of CaMPARI2 as a functional biosensor for monitoring calcium activity in human pancreatic tissue slices. This tool can be implemented across disease states and different stages of disease progression, offering a powerful, higher throughput means to assess calcium dynamics and glucose responsiveness in islets of whole pancreatic tissues. In future works, we plan to use CaMPARI2 to correlate functional and structural parameters within the islets of tissue slices. For instance, using this tool we can explore the relationship between calcium responsiveness and islet architecture, such as the size or morphology of the islet, or protein expression to uncover mechanisms of dysfunctional. Furthermore, CaMPARI2 can be combined with immune cell staining techniques to quantify T cell infiltration during insulitis, offering insights into how immune cell presence might impact β-cell function and survival.
Presenter
Nathan Steenbuck, PhD (University of Zurich)
Authors
Nathan Steenbuck, Nicolas Damond, Stefanie Engler, Irina Kusmartseva, Amanda Posgai, Denise Drotar, MacKenzie Williams, Clive Wasserfall, Natalie De Souza, Todd M. Brusko, Mark Atkinson, Bernd Bodenmiller
Purpose
The natural history and pathogenesis of type 1 diabetes (T1D), particularly during the autoantibody positive (AAb+) stages preceding clinical onset, is not completely understood, in part due to the limited availability of human pancreatic samples. We performed a novel, single-cell analysis of immune and metabolic dysregulation within the human pancreas over T1D progression, using highly multiplexed imaging.
Methods
We analyzed pancreas sections from 88 organ donors from the Network for Pancreatic Organ donors with Diabetes (nPOD) using imaging mass cytometry (IMC). The study included single AAb+ (N=28), multiple AAb+ (N=10), recent-onset T1D (N=21), and long-standing T1D donors (N=14), plus age, gender, and BMI-matched non-diabetic controls (N=15). We applied two 45-plex antibody panels and profiled over 12,000 islets and 16 million cells, creating, to our knowledge, the largest human multiplexed T1D dataset to date.
Summary of Results
We first examined β-cell evolution, identifying a gradual loss of islet amyloid polypeptide (IAPP) and upregulation of MHC-I as dysregulated events in the AAb+ stages of the disease.
We annotated over 1 million immune cells and developed an infiltration score that integrates immune cell abundance and proximity to islets. By combining this score with β-cell pseudotime and differential spatial analysis, we identified gradual infiltration of myeloid, B, CD4+ T helper (T-CD4) and CD8+ cytotoxic T cells (T-CD8) into islets. T-CD4 and T-CD8 cells co-infiltrated islets and all four cell types were associated with β-cell MHC-I and insulitic β-cell MHC-II expression, suggesting that they are key drivers of T1D progression.
We further identified a spatial pancreatic myeloid microenvironment, and showed that, along disease progression, activated macrophages and conventional dendritic cells (DC) acquired M1-polarized and mature DC phenotypes respectively and localized near T cells at the islet edge.
Finally, we detected islet infiltration of PD1+ T-CD4 and T-CD8 cells in both AAb+ stages and identified a PD1+ T-CD8 subtype that was strongly enriched in islets and neighbored the proinflammatory myeloid phenotypes.
Conclusions
We provide an unprecedented dataset for in situ mapping of cellular networks during T1D progression. We identified β-cell state, PD1+ T cells, and pro-inflammatory myeloid phenotypes as critical indicators of early disease progression, suggesting that these are potential therapeutic targets.
Presenter
Teifion Luckett, PhD (University of Exeter)
Authors
Teifion Luckett, Kathryn Murrall, Christine Flaxman, Christiana Lekka, Stephanie Hunter, James
Shaw, Noel Morgan, Sarah Richardson
Purpose
We previously reported that Type 1 Diabetes (T1D) is a heterogenous disease linked to age-atonset. Initiation of islet-autoimmunity coincides with a dramatic increase in pancreas size post- birth; however, little is known on how the pancreatic architecture changes during this period. We hypothesised that pancreas maturity at the onset of autoimmunity contributes towards T1D heterogeneity.
Methods
We employed AI-based classifiers to demarcate endocrine and acinar tissue in historically stained tissue sections from archival (EADB, Seattle, HDBR) and contemporary (nPOD, QUOD, EUnPOD, DiViD) pancreas biobanks stained for Chromogranin A/CK19 and Insulin/glucagon. A total of 569,751 endocrine objects (EO) were annotated across 346 cases. EO area was transformed into bins (smallest 0 – 9 biggest). EO object morphology, % positively stained area and spatial distribution was then assessed against a range of clinical variables.
Summary of Results
The % endocrine area and object density significantly decrease in the first 18 years post birth, coinciding with acinar expansion. Larger EO’s take up an increasing proportion of the total endocrine area with advancing age. Strikingly, small Ins+ only EO’s (bins 0-2) with a diameter between 15-45µM account for ~50% of total EOs in healthy controls but were found to be virtually absent in T1D. Absence of beta cells within mid-sized EOs, but persistence in large EO’s, suggests beta cells in mid-sized islets are also susceptible to immune attack, but those within larger EOs appear better protected. Donors with young T1D-onset whose autoimmunity will have initiated earlier in life, had significantly fewer large EO’s, which manifests as reduced large EO density in adulthood compared to age-matched healthy controls.
Conclusions
Collectively, the data presented support the hypothesis that single beta cells or those contained in small-mid-sized clusters are effectively eliminated by the autoimmune process, whereas those in larger EOs are better protected. Our data suggest that initiation of autoimmunity at a young age, when more beta cells are contained in smaller EOs, results in a more efficient destruction of beta cell mass, impacting EO growth and resulting in reduced large EO density in adulthood. These findings highlight the importance of early intervention to prevent young onset of T1D and have important clinical implications for islet replacement therapies, and endogenous beta cell regeneration.
Presenter
Valeriia Muradova, MD (University of Florida)
Authors
Valeriia Muradova, Laura Jacobsen, Irina Kusmartseva, Mark Atkinson, Desmond Schatz
Purpose
The mean weight of pancreata from U.S. organ donors with type 1 diabetes (T1D) is decreased compared to age/BMI-matched individuals without this disease. However, it remains unknown whether this finding is consistent across all races and ethnicities.
Based on observations that African American (AA) individuals tend to experience poorer clinical outcomes in diabetes, including higher A1C levels compared to non-Hispanic Whites (NHW), we hypothesized that AA individuals without diabetes have reduced pancreatic weight compared to NHW individuals without diabetes. Additionally, we investigated whether individuals with AA or Hispanic/Latino background, with T1D or type 2 diabetes (T2D), exhibited a diminished pancreatic weight compared to individuals having a NHW background with T1D or T2D.
Methods
We analyzed 389 pancreata recovered by two organ donor initiatives: the Network for Pancreatic Organ donors with Diabetes (nPOD, n = 302) and the Human Pancreas Analysis Program (HPAP, n = 87). Donors from these programs (aged 6 to 89 years, with median age = 30.1 years; 41.6% females, 58.3% males) were categorized into three groups: no diabetes (n = 188), T1D (n = 139), and T2D (n = 62). Relative pancreatic weight (RPW; pancreatic weight [g] / body weight [kg]) was calculated to account for potential weight-related differences. Subjects were stratified for analyses by age group: 6 to 17 years (ped) and those >18 years (adult). Statistical comparisons were made using a multiple linear regression model.
Summary of Results
The mean RPW for AA adult donors with no diabetes was 0.96 (n = 19, standard deviation [SD] = 0.22), compared to NHW adults with no diabetes, 1.03 (n = 75, SD = 0.29; p-value = 0.649). Mean RPW for AA adults with T1D was 0.62 (n = 12, SD = 0.25), compared to NHW adult donors with T1D, 0.56 (n = 85, SD = 0.18; p-value = 0.46). Mean RPW for AA adult donors with T2D was 0.91 (n = 16, SD = 0.26), compared to NHW adults with T2D, 0.97 (n = 31, SD = 0.25; p-value = 0.39). In addition, there were no differences in mean RPW in children with no diabetes, T1D, or T2D by race. The number of pancreata from Hispanic/Latino donors was limited, so only data for adults were reported, with no differences observed in mean RPW compared to NHW donors with no diabetes, T1D, or T2D.
Conclusions
Our findings suggest that there are no significant racial/ethical differences in RPW between individuals with AA and NHW or between Hispanic and NHW backgrounds, with or without diabetes.
Presenter
Verena van der Heide, PhD (Icahn School of Medicine at Mount Sinai)
Authors
Sarah McArdle, Michael S. Nelson, Karen Cerosaletti, Sacha Gnjatic, Zbigniew Mikulski, Amanda L. Posgai, Irina Kusmartseva, Mark Atkinson, Dirk Homann
Purpose
Type 1 diabetes (T1D) is an autoimmune condition that culminates in the loss of insulinproducing beta cells. Histopathology of the human pancreas provides essential insights into disease initiation and progression. However, an integrated overview of pathogenic processes in situ is lacking in part due to limited sample availability, the dispersed nature of anatomical lesions, and restricted analytical dimensionality. Here, we combined multiplexed immunostaining, high-magnification whole-slide imaging, and customized semi-automated digital pathology strategies to interrogate pancreatic tissue sections across T1D stages at scale.
Methods
Pancreatic head and tail sections from 25 non-diabetic control, autoantibody+ (stage 1/2), short (including at-onset)- and long-duration T1D donors were obtained through the nPOD consortium. Expression of eight endocrine hormones (CHGA, ProINS, INS, IAPP, GCG, ProCGC, SST, PPY) as well as islet-infiltrating/surrounding CD45+ immune cells were visualized in successive staining cycles using the recently published Multiplexed Immunohistochemical Consecutive Staining on Single Slide (MICSSS) platform followed by whole-slide brightfield scanning at high magnification (40x). We next applied machine learning algorithms in with customized scripts in the open-source digital pathology software QuPath to achieve semi-automated image alignment of ~500 whole-slide scans and assessed total islets per tissue section, their architectural features and spatial relations (e.g., area, density, shape descriptors), endocrine hormone contents (including combinatorial metrics such as hormone co-expression) as well as islet-associated immune cell distributions. Finally, conducting nonlinear UMAP dimensionality reduction analysis of >100 islet-derived parameters, we reveal core histopathological signatures and derived clustering at the islet level across T1D disease stages.
Summary of Results
Deconvolution of architectural features, endocrine cell compositions, immune cell burden, and spatial relations of ~25,000 islets confirms central tenets of pancreatic T1D histopathology, including recent findings obtained by 3D pancreas mapping. Moreover, our data uncover multiple novel aspects about the dynamic organization of the human pancreas in health and T1D disease. Notably, we find that islet size may serve as a basic organizing principle for their heterogenous endocrine cell composition and that a fundamental similarity of head and tail islets is skewed by a phenotypically unique islet pool in the uncinate process of the pancreatic head. Our results further identify striking histopathological correlates of the stage 1/2 pancreas (e.g., reorganization of islet UMAP cluster magnitudes and spatial tissue distributions as well as significantly enhanced islet-associated immune cell burden) and indicate that early T1D autoimmune destruction preferentially targets small islets accompanied by a relative loss of IAPP and proinsulin. Finally, we observe that insulin-containing islets without immune cell association nevertheless present with distinct pathological alterations, that T1D progresses in an unexpectedly synchronized fashion throughout the pancreas, and that long-standing T1D is associated not only with exacerbated deterioration of islet architecture but also a loss of alpha and delta cell mass.
Conclusions
Our integrated analysis of the human pancreas in health and T1D disease conceptually organizes a broad range of histopathological alterations permitting the reconstruction of a revised timeline of natural T1D history. Specifically, the striking histopathological features detected in the stage 1/2 pre-diabetic pancreas may account for the seemingly sudden onset of clinical disease. This study may thus provide a foundation for future targeted investigations into T1D autoimmunity and the informed consideration of interventional modalities.
Presenter
Alexandra Rippa, PhD (University of Florida)
Authors
Alexandra Rippa, Amanda Posgai, Maigan Brusko, Seth Currlin, Clive Wasserfall, Irina Kusmartseva, Martha Campbell-Thompson, Mark Atkinson
Purpose
Two-dimensional (2D) analyses of pancreata from non-diabetic (ND) individuals and those with or at increased risk for type 1 diabetes (T1D) demonstrate well-described heterogeneity for islet number and size, and their endocrine cell composition from random cross-sections. Recent 3D studies of healthy human pancreas samples have provided more complete islet information. Here, we characterized islet size and endocrine cell composition via 3D light sheet fluorescence microscopy (LSFM) utilizing pancreases from ND, GAD autoantibody-positive (GADA), and short duration T1D donors with residual insulin (INS).
Methods
Formalin-fixed human pancreas (X=5mm, Y=10mm, Z=3mm) from ND (n=3; age range 14-23yr), GADA (n=3; 18-22yr), and T1D donors (n=6; 11-33yr) were cleared using a modified iDISCO protocol, stained for INS and glucagon (GCG), and imaged using the fully automated Miltenyi UltraMicroscope Blaze™. An average tissue volume of 18.5mm3 (3D surface) was used to characterize INS+ and GCG+ signals (Imaris). Cell clusters with volume >3000µm3 (mean ± standard deviation [SD] in T1D n=509±250; GADA=1918±515 in; ND=2268±1573) were binned by volume size (cell clusters 10³-104, small 104-105, medium 105-106, large ≥106 µm3 islets). Data were analyzed by one- or two-way ANOVA with Tukey’s test for multiple comparisons.
Summary of Results
Interestingly, islet density (islets/mm3) was lower in T1D (27.8±14.3, all p<0.01) relative to ND (130.7±61.5) and GADA (111.3±28.1). T1D pancreata also had a reduced percentage of INS-containing islet (ICI) (T1D mean±SD=13.8±11.9% vs ND=84±1.45%, GADA=83±5.62%; all p<0.0001) as well as reduced percentage of beta cell volume (T1D=0.5±0.36%, ND=2.8±1.0%, GADA=2.3±0.4%; all p<0.005), as expected. The percentage of alpha cell volume was similar across groups (T1D=1.7±0.76%, ND=1.4±0.76%, GADA=1.5±0.57%).
In the T1D pancreas, the INS+GCG- islet fraction was significantly reduced in T1D both as a percentage of total islet count (T1D=1.3±1%, ND=59.2±3.39%, GADA=59.8±15%; all p<0.0001) and as a percentage of total islet volume T1D=1.9±2.89%, ND=10.2±1.12%, GADA=15.4±7.14%; all p<0.04). The INS+GCG+ islet fraction was reduced in T1D as a percentage of total islet count (T1D=12.5±11.88%, ND=25.6+4.18%, GADA=23.2±10.46%; all p=ns) and significantly reduced as a percentage of total islet volume (T1D=44.7±24.74%, ND=88.9±0.75%, GADA=83.6±6.49%; all p<0.004).
T1D pancreata contained significantly reduced percentages of total ICI count in the cell cluster (T1D=3.55±5.33%, ND=26.3±7.09%, GADA=16.6±4.16%; all p<0.02) and small islet size bins (T1D=4.5±3.54%, ND=44±4%, GADA=44.6±9.5%; all p<0.0001), but significantly increased percent of total ICI in the large size bin (T1D=63.3±18.51%, ND=7.3±4.93%, GADA=7.6±2.08%; all p<0.0001). As a percentage of total ICI count, T1D pancreata displayed significantly reduced INS+GCG- islet fractions in the cell cluster (T1D=2.3±3.88%, ND=25±6.55%, GADA=16±4.58%; all p<0.001) and small islet size bins (T1D=4.3±3.83%, ND=37.3±0.57%, GADA=40.3±7.09%; all p<0.001) alongside no difference in medium size INS+GCG+ and increased fractions of large size INS+GCG+ islets (T1D=60±17.61%, ND=7±4.35%, GADA=7.6±2.08%; all p<0.0001).
Conclusions
In a healthy pancreas, INS+GCG- islets make up approximately 60% of the total islet count but only 10-15% of the total islet volume due to their small size. INS+GCG+ islets are larger, comprising approximately 25% of the total islet count but more than 80% of the endocrine volume. T1D pancreas from donors with residual INS had significantly reduced islet density and dramatic loss of INS+GCG- islets. While INS containing cell clusters and small-sized islets were preferentially lost, medium and large islets containing both INS and GCG were preserved.
Presenter
Yu Shen, BS (Johns Hopkins University)
Authors
Yu Shen, Won June Cho, Saurabh Joshi, Swarnagouri Naganathanhalli, Benjamin Wen, Kyu Sang Han, Wen-Chen Chen, Mia Grahn, Bridgette Kim, Andre Forjaz, Casey Grubel, Maria Beery, Kacie Geelhoed, Irina Kusmartseva, Pei-Hsun Wu, Mark Atkinson, Denis Wirtz, Ashley Kiemen
Purpose
Islet-specific T cells play an essential role in beta cell destruction in type 1 diabetes (T1D). While T cells in the pancreas and draining lymph nodes are crucial sources for understanding T1D pathogenesis, only a subset of these T cells may be specific to beta cells. Identifying phenotypes and functions of T1D is an autoimmune condition associated with a loss of the insulin-producing beta cells in the pancreatic islets. Histological analysis plays a pivotal role in understanding the pathophysiology of T1D, particularly in examining the cellular composition of the islets and their microenvironment. Hematoxylin and eosin (H&E) staining provides general morphological insights, while immunohistochemistry (IHC) enables staining molecules of interest, such as insulin and glucagon within the alpha and beta cells of the islets. Both stains are crucial for studying T1D. Leveraging advanced deep learning techniques, we developed a novel approach to transform H&E-stained images of the pancreas into multiplex IHC stains of insulin, glucagon, and CD3+ T cells. To integrate pathological information within the islets and the surrounding pancreatic parenchyma, we further applied a novel tissue mapping platform, CODA, to profile the multi-faceted spatial interactions with the onset of T1D in 3D space.
Methods
We obtained paired H&E and IHC histology from the pancreatic head, body, and tail of 72 pancreatic donors, with an equal number of nondiabetic, autoantibody-positive, and type 1 diabetic samples. The IHC was stained with hematoxylin and counterstained with immunohistochemical labels for insulin (in red), glucagon (in blue), and CD3+ T cells (in brown). We designed an algorithm enabling rapid generation of high-quality training and testing tiles with a size of 256x256x3 pixels centered around the pancreatic islets. We trained five generative models to produce virtual IHC-stained images from H&E-stained human pancreas tissue and compared their performance using several quantitative metrics.
To extend our analysis to 3D, we collected thick slabs of T1D, autoantibody-positive, and nonT1D human pancreas tissue. Samples were formalin-fixed, paraffin-embedded, and serially sectioned. Histological slides were stained with H&E and IHC and digitized. We used CODA to segment eight pancreatic microanatomical components in the images: islets of Langerhans, pancreatic ducts, fat, vasculatures, extracellular matrix, acinus, non-tissue, and nerves. Using hue, saturation, variation (HSV) color masks, we quantified the antibody signal of the IHC images to spatially map twelve additional components: alpha cells, beta cells, delta cells, NK & activated T cells, lymphatic vessels, endothelium, leukocytes, antigen-positive cells, two different T cell subtypes, B cells, and nucleic acids. As a result, we integrated antibody labeling with our segmentation model to map the cellular compositions of individual islets and quantify the structural transformation of hundreds of diabetic islets.
Summary of Results
We successfully trained five distinct virtual staining models. To test their performance, we applied six quality metrics to an independent testing dataset of 699 virtually stained IHC tiles. We then compared islet morphological features between real H&E and virtual IHC tiles, and stain composition similarities between real and virtual IHC tiles. We discovered that diffusion models can generate morphologically accurate islets while maintaining consistent insulin andglucagon stain compositions.
In our 3D analysis, we reconstructed the complete anatomical structures of three cm3-sized samples of human pancreas tissue. Further, we quantified and compared the anatomical changes of the diabetic pancreas to those of the nondiabetic and autoantibody-positive samples. We discovered a total increase in the bulk composition of islets, pancreatic ducts, vasculatures, fat, extracellular matrix, and nerves, and decreases in acinar content in the diabetic sample. The increase in the composition of islets of Langerhans can be caused by obtaining samples from different parts of the pancreas. This discovery emphasizes the need for investigating microarchitecture and molecular variations in pancreatic compositions with the onset of diabetes. Cellular changes associated with immune inflammations were quantified based on the subcellular segmentation of alpha, beta, and delta cell compositions of healthy and diabetic islets, and the inflammation map revealed an elevation in immune content in the diabetic pancreas and a large inter-sample heterogeneity in islet composition and immune aggregation.
Conclusions
The results demonstrate high fidelity in the virtual IHC images, facilitating a more efficient and less invasive means of studying islet pathology in T1D. This approach not only enhances the ability to investigate the cellular and molecular composition of islets but also has the potential to accelerate research by reducing reliance on physical staining processes, thereby contributing to a deeper understanding of T1D pathogenesis and the development of targeted therapies. Integrating various histological staining results using CODA allows collective evaluation of the anatomical environment in T1D. The spatial analysis of diabetic and nondiabetic pancreases at tissue and cellular levels provides a comprehensive understanding of the disease progression in T1D patients. Current results indicate that advances in artificial intelligence systems can facilitate the discovery of autoimmune and pancreatic pathogenesis of T1D.
Presenter
Ery Petropoulou, PhD (Helmholtz Center Munich)
Authors
Ery Petropoulou, Jose Zapardiel-Gonzalo, Nuria Punet Valls, Joy Paramor, Sophia Forsskahl, Jonna Saarimaki-Vire, Diego Balboa, Andrew Pepper, Teresa Rodriguez-Calvo
Purpose
High HLA-I expression is considered one of the hallmarks of Type 1 Diabetes (T1D). Previous studies aiming to decipher its role showed that it is strongly associated with the presence of beta cells in islets from T1D donors, and that interferons are powerful drivers of its expression (in vitro). Here, we sought to assess: 1) the magnitude of HLA-I expression by collecting quantitative data regarding the islets that present high (H), elevated (E) and low (L) HLA-I expression in the pancreas, 2) the timeline of HLA-I expression, infiltration and beta cell loss and 3) the dynamics of HLA-I upregulation and its correlation with beta cell function.
Methods
Pancreatic sections of 41 donors (10 non-diabetic (ND), 7 single auto-antibody positive (sAAb+), 7 double auto-antibody positive (dAAb+), and 17 T1D donors) were obtained through nPOD. We stained sections for Insulin (INS), Glucagon (GCG), HLA-ABC (HLA-I), CD3 and CD8. We categorized the islets into HLA-I high (H), elevated (E) and low (L), based on mean islet intensity. Next, we acquired high-resolution images from 5 islets per category with a confocal microscope to assess INS clustering within the cells. Finally, isolated human and stem-cell (SC) derived islets were treated with IFNα and IFNγ for 0h, 24h, 48h, 7 days and 14 days. HLA-I, INS and GCG expression, as well as glucose-stimulated insulin secretion (GSIS) were performed.
Summary of Results
By analyzing more than 9500 islets we found that high HLA-I expression affects approx. 8% of dAAb+ islets and 15% of T1D islets. Recent-onset T1D donors show hyperexpression on 39% of their islets. Immune infiltration densities are highest in HLA-I high islets of all donor groups; on average, 98% of dAAb-H and 83% of T1D-H islets are infiltrated by at least 1 CD3+ T-cell, compared to 68% and 63% in the respective low islets. The highest beta cell content is found in ND-L, dAAb-H and T1D-H islets. Low islets in T1D are insulin-deficient islets, or seriously depleted of beta cells. Interestingly, INS clusters of E and H islets in T1D donors are comparable to those of ND donors, whereas beta cells of T1D L islets have decreased INS clusters. Isolated human and SC-islets treated with INFs showed increased HLA-I expression at the cell membrane. An analysis of intracellular insulin levels and clustering showed no significant changes between treated and not treated islets, but the GSIS revealed a decreased stimulation index.
Conclusions
This is the first study to provide quantitative data regarding the association of HLA-I expression, immune infiltration, beta cell content and insulin dynamics in T1D. High HLA-I expression and immune infiltration are prominent in dAAb+ and T1D donors, peaking around disease onset. Islet beta cell proportion declines steadily with disease progression, reaching a nadir in long duration T1D donors . The remaining beta cells of E and H islets of T1D donors retain their INS content, whereas the beta cells of L islets show a decrease in INS cluster density and size, pointing to beta cell dysfunction and defective INS synthesis or processing at later stages of the disease. In human and SC-derived islets HLA-I is quickly upregulated after IFN stimulation, it localizes at the cell membrane and remains high after the removal of the stimulus. IFNdependent HLA-I upregulation does not affect the intracellular levels of INS, but it affects its secretion and therefore, beta cell function. Therefore, we hypothesize that strategies aiming at decreasing HLA-I expression may delay or avoid disease progression and preserve beta cell mass and function, alone, or in combination with strategies focusing on eliminating immunemediated beta cell destruction.