Identification of type 2 diabetes- and obesity-associated human β-cells using deep transfer learning

Diabetes affects >10% of adults worldwide and is caused by impaired production or response to insulin, resulting in chronic hyperglycemia. Pancreatic islet β-cells are the sole source of endogenous insulin and our understanding of β-cell dysfunction and death in type 2 diabetes (T2D) is incomplete. Single-cell RNA-seq data supports heterogeneity as an important factor in β-cell function and survival. However, it is difficult to identify which β-cell phenotypes are critical for T2D etiology and progression. Our goal was to prioritize specific disease-related β-cell subpopulations to better understand T2D pathogenesis and identify relevant genes for targeted therapeutics. To address this, we applied a deep transfer learning tool, DEGAS, which maps disease associations onto single-cell RNA-seq data from bulk expression data. Independent runs of DEGAS using T2D or obesity status identified distinct β-cell subpopulations. A singular cluster of T2D-associated β-cells was identified; however, β-cells with high obese-DEGAS scores contained two subpopulations derived largely from either non-diabetic or T2D donors. The obesity-associated non-diabetic cells were enriched for translation and unfolded protein response genes compared to T2D cells. We selected DLK1 for validation by immunostaining in human pancreas sections from healthy and T2D donors. DLK1 was heterogeneously expressed among β-cells and appeared depleted from T2D islets. In conclusion, DEGAS has the potential to advance our holistic understanding of the β-cell transcriptomic phenotypes, including features that distinguish β-cells in obese non-diabetic or lean T2D states. Future work will expand this approach to additional human islet omics datasets to reveal the complex multicellular interactions driving T2D.

Species-specific roles for the MAFA and MAFB transcription factors in regulating islet β cell identity

Type 2 diabetes (T2D) is associated with compromised identity of insulin-producing pancreatic islet β cells, characterized by inappropriate production of other islet cell-enriched hormones. Here, we examined how hormone misexpression was influenced by the MAFA and MAFB transcription factors, closely related proteins that maintain islet cell function. Mice specifically lacking MafA in β cells demonstrated broad, population-wide changes in hormone gene expression with an overall gene signature closely resembling islet gastrin+ (Gast+) cells generated under conditions of chronic hyperglycemia and obesity. A human β cell line deficient in MAFB, but not one lacking MAFA, also produced a GAST+ gene expression pattern. In addition, GAST was detected in human T2D β cells with low levels of MAFB. Moreover, evidence is provided that human MAFB can directly repress GAST gene transcription. These results support a potentially novel, species-specific role for MafA and MAFB in maintaining adult mouse and human β cell identity, respectively. Here, we discuss the possibility that induction of Gast/GAST and other non-β cell hormones, by reduction in the levels of these transcription factors, represents a dysfunctional β cell signature.

Chromatin accessibility differences between alpha, beta, and delta cells identifies common and cell type-specific enhancers

BACKGROUND

High throughput sequencing has enabled the interrogation of the transcriptomic landscape of glucagon-secreting alpha cells, insulin-secreting beta cells, and somatostatin-secreting delta cells. These approaches have furthered our understanding of expression patterns that define healthy or diseased islet cell types and helped explicate some of the intricacies between major islet cell crosstalk and glucose regulation. All three endocrine cell types derive from a common pancreatic progenitor, yet alpha and beta cells have partially opposing functions, and delta cells modulate and control insulin and glucagon release. While gene expression signatures that define and maintain cellular identity have been widely explored, the underlying epigenetic components are incompletely characterized and understood. However, chromatin accessibility and remodeling is a dynamic attribute that plays a critical role to determine and maintain cellular identity.

RESULTS

Here, we compare and contrast the chromatin landscape between mouse alpha, beta, and delta cells using ATAC-Seq to evaluate the significant differences in chromatin accessibility. The similarities and differences in chromatin accessibility between these related islet endocrine cells help define their fate in support of their distinct functional roles. We identify patterns that suggest that both alpha and delta cells are poised, but repressed, from becoming beta-like. We also identify patterns in differentially enriched chromatin that have transcription factor motifs preferentially associated with different regions of the genome. Finally, we not only confirm and visualize previously discovered common endocrine- and cell specific- enhancer regions across differentially enriched chromatin, but identify novel regions as well. We compiled our chromatin accessibility data in a freely accessible database of common endocrine- and cell specific-enhancer regions that can be navigated with minimal bioinformatics expertise.

CONCLUSIONS

Both alpha and delta cells appear poised, but repressed, from becoming beta cells in murine pancreatic islets. These data broadly support earlier findings on the plasticity in identity of non-beta cells under certain circumstances. Furthermore, differential chromatin accessibility shows preferentially enriched distal-intergenic regions in beta cells, when compared to either alpha or delta cells.

A beta cell subset with enhanced insulin secretion and glucose metabolism is reduced in type 2 diabetes

The pancreatic islets are composed of discrete hormone-producing cells that orchestrate systemic glucose homeostasis. Here we identify subsets of beta cells using a single-cell transcriptomic approach. One subset of beta cells marked by high CD63 expression is enriched for the expression of mitochondrial metabolism genes and exhibits higher mitochondrial respiration compared with CD63lo beta cells. Human and murine pseudo-islets derived from CD63hi beta cells demonstrate enhanced glucose-stimulated insulin secretion compared with pseudo-islets from CD63lo beta cells. We show that CD63hi beta cells are diminished in mouse models of and in humans with type 2 diabetes. Finally, transplantation of pseudo-islets generated from CD63hi but not CD63lo beta cells into diabetic mice restores glucose homeostasis. These findings suggest that loss of a specific subset of beta cells may lead to diabetes. Strategies to reconstitute or maintain CD63hi beta cells may represent a potential anti-diabetic therapy.

β-Cell Succinate Dehydrogenase Deficiency Triggers Metabolic Dysfunction and Insulinopenic Diabetes

Mitochondrial dysfunction plays a central role in type 2 diabetes (T2D); however, the pathogenic mechanisms in pancreatic β-cells are incompletely elucidated. Succinate dehydrogenase (SDH) is a key mitochondrial enzyme with dual functions in the tricarboxylic acid cycle and electron transport chain. Using samples from human with diabetes and a mouse model of β-cell-specific SDH ablation (SDHBβKO), we define SDH deficiency as a driver of mitochondrial dysfunction in β-cell failure and insulinopenic diabetes. β-Cell SDH deficiency impairs glucose-induced respiratory oxidative phosphorylation and mitochondrial membrane potential collapse, thereby compromising glucose-stimulated ATP production, insulin secretion, and β-cell growth. Mechanistically, metabolomic and transcriptomic studies reveal that the loss of SDH causes excess succinate accumulation, which inappropriately activates mammalian target of rapamycin (mTOR) complex 1-regulated metabolic anabolism, including increased SREBP-regulated lipid synthesis. These alterations, which mirror diabetes-associated human β-cell dysfunction, are partially reversed by acute mTOR inhibition with rapamycin. We propose SDH deficiency as a contributing mechanism to the progressive β-cell failure of diabetes and identify mTOR complex 1 inhibition as a potential mitigation strategy.

IAPP-induced beta cell stress recapitulates the islet transcriptome in type 2 diabetes

AIMS/HYPOTHESIS

Type 2 diabetes is characterised by islet amyloid and toxic oligomers of islet amyloid polypeptide (IAPP). We posed the questions, (1) does IAPP toxicity induce an islet response comparable to that in humans with type 2 diabetes, and if so, (2) what are the key transcriptional drivers of this response?

METHODS

The islet transcriptome was evaluated in five groups of mice: beta cell specific transgenic for (1) human IAPP, (2) rodent IAPP, (3) human calpastatin, (4) human calpastatin and human IAPP, and (5) wild-type mice. RNA sequencing data was analysed by differential expression analysis and gene co-expression network analysis to establish the islet response to adaptation to an increased beta cell workload of soluble rodent IAPP, the islet response to increased expression of oligomeric human IAPP, and the extent to which the latter was rescued by suppression of calpain hyperactivation by calpastatin. Rank-rank hypergeometric overlap analysis was used to compare the transcriptome of islets from human or rodent IAPP transgenic mice vs humans with prediabetes or type 2 diabetes.

RESULTS

The islet transcriptomes in humans with prediabetes and type 2 diabetes are remarkably similar. Beta cell overexpression of soluble rodent or oligomer-prone human IAPP induced changes in islet transcriptome present in prediabetes and type 2 diabetes, including decreased expression of genes that confer beta cell identity. Increased expression of human IAPP, but not rodent IAPP, induced islet inflammation present in prediabetes and type 2 diabetes in humans. Key mediators of the injury responses in islets transgenic for human IAPP or those from individuals with type 2 diabetes include STAT3, NF-κB, ESR1 and CTNNB1 by transcription factor analysis and COL3A1, NID1 and ZNF800 by gene regulatory network analysis.

CONCLUSIONS/INTERPRETATION

Beta cell injury mediated by IAPP is a plausible mechanism to contribute to islet inflammation and dedifferentiation in type 2 diabetes. Inhibition of IAPP toxicity is a potential therapeutic target in type 2 diabetes.

Navigating the Depths and Avoiding the Shallows of Pancreatic Islet Cell Transcriptomes

Islet gene expression has been widely studied to better understand the transcriptional features that define a healthy β-cell. Transcriptomes of FACS-purified α-, β-, and δ-cells using bulk RNA-sequencing have facilitated our understanding of the complex network of cross talk between islet cells and its effects on β-cell function. However, these approaches were by design not intended to resolve heterogeneity between individual cells. Several recent studies used single-cell RNA sequencing (scRNA-Seq) to report considerable heterogeneity within mouse and human β-cells. In this Perspective, we assess how this newfound ability to assess gene expression at single-cell resolution has enhanced our understanding of β-cell heterogeneity. We conduct a comprehensive assessment of several single human β-cell transcriptome data sets and ask if the heterogeneity reported by these studies showed overlap and concurred with previously known examples of β-cell heterogeneity. We also illustrate the impact of the inevitable limitations of working at or below the limit of detection of gene expression at single cell resolution and their consequences for the quality of single-islet cell transcriptome data. Finally, we offer some guidance on when to opt for scRNA-Seq and when bulk sequencing approaches may be better suited.

Artemether Does Not Turn α Cells into β Cells

Pancreatic α cells retain considerable plasticity and can, under the right circumstances, transdifferentiate into functionally mature β cells. In search of a targetable mechanistic basis, a recent paper suggested that the widely used anti-malaria drug artemether suppresses the α cell transcription factor Arx to promote transdifferentiation into β cells. However, key initial experiments in this paper were carried out in islet cell lines, and most subsequent validation experiments implied transdifferentiation without direct demonstration of α to β cell conversion. Indeed, we find no evidence that artemether promotes transdifferentiation of primary α cells into β cells. Moreover, artemether reduces Ins2 expression in primary β cells >100-fold, suppresses glucose uptake, and abrogates β cell calcium responses and insulin secretion in response to glucose. Our observations suggest that artemether induces general islet endocrine cell dedifferentiation and call into question the utility of artemisinins to promote α to β cell transdifferentiation in treating diabetes.

Virgin Beta Cells Persist throughout Life at a Neogenic Niche within Pancreatic Islets

Postnatal maintenance or regeneration of pancreatic beta cells is considered to occur exclusively via the replication of existing beta cells, but clinically meaningful restoration of human beta cell mass by proliferation has never been achieved. We discovered a population of immature beta cells that is present throughout life and forms from non-beta precursors at a specialized micro-environment or "neogenic niche" at the islet periphery. These cells express insulin, but lack other key beta cell markers, and are transcriptionally immature, incapable of sensing glucose, and unable to support calcium influx. They constitute an intermediate stage in the transdifferentiation of alpha cells to cells that are functionally indistinguishable from conventional beta cells. We thus identified a lifelong source of new beta cells at a specialized site within healthy islets. By comparing co-existing immature and mature beta cells within healthy islets, we stand to learn how to mature insulin-expressing cells into functional beta cells.

Spectrum of DNA variants for non-syndromic deafness in a large cohort from multiple continents

Hearing loss is the most common sensory deficit in humans with causative variants in over 140 genes. With few exceptions, however, the population-specific distribution for many of the identified variants/genes is unclear. Until recently, the extensive genetic and clinical heterogeneity of deafness precluded comprehensive genetic analysis. Here, using a custom capture panel (MiamiOtoGenes), we undertook a targeted sequencing of 180 genes in a multi-ethnic cohort of 342 GJB2 mutation-negative deaf probands from South Africa, Nigeria, Tunisia, Turkey, Iran, India, Guatemala, and the United States (South Florida). We detected causative DNA variants in 25 % of multiplex and 7 % of simplex families. The detection rate varied between 0 and 57 % based on ethnicity, with Guatemala and Iran at the lower and higher end of the spectrum, respectively. We detected causative variants within 27 genes without predominant recurring pathogenic variants. The most commonly implicated genes include MYO15A, SLC26A4, USH2A, MYO7A, MYO6, and TRIOBP. Overall, our study highlights the importance of family history and generation of databases for multiple ethnically discrete populations to improve our ability to detect and accurately interpret genetic variants for pathogenicity.