Analysis of Half a Billion Datapoints Across Ten Machine-Learning Algorithms Identifies Key Elements Associated With Insulin Transcription in Human Pancreatic Islet Cells.

Machine learning (ML)-workflows enable unprejudiced/robust evaluation of complex datasets. Here, we analyzed over 490,000,000 data points to compare 10 different ML-workflows in a large (N=11,652) training dataset of human pancreatic single-cell (sc-)transcriptomes to identify genes associated with the presence or absence of insulin transcript(s). Prediction accuracy/sensitivity of each ML-workflow was tested in a separate validation dataset (N=2,913). Ensemble ML-workflows, in particular Random Forest ML-algorithm delivered high predictive power (AUC=0.83) and sensitivity (0.98), compared to other algorithms. The transcripts identified through these analyses also demonstrated significant correlation with insulin in bulk RNA-seq data from human islets. The top-10 features, (including IAPP, ADCYAP1, LDHA and SST) common to the three Ensemble ML-workflows were significantly dysregulated in scRNA-seq datasets from Ire-1αß-/- mice that demonstrate dedifferentiation of pancreatic ß-cells in a model of type 1 diabetes (T1D) and in pancreatic single cells from individuals with type 2 Diabetes (T2D). Our findings provide direct comparison of ML-workflows in big data analyses, identify key elements associated with insulin transcription and provide workflows for future analyses.

A Pro-Endocrine Pancreatic Islet Transcriptional Program Established During Development Is Retained in Human Gallbladder Epithelial Cells.

BACKGROUND & AIMS: Pancreatic islet ß-cells are factories for insulin production; however, ectopic expression of insulin also is well recognized. The gallbladder is a next-door neighbor to the developing pancreas. Here, we wanted to understand if gallbladders contain functional insulin-producing cells. METHODS: We compared developing and adult mouse as well as human gallbladder epithelial cells and islets using immunohistochemistry, flow cytometry, enzyme-linked immunosorbent assays, RNA sequencing, real-time polymerase chain reaction, chromatin immunoprecipitation, and functional studies. RESULTS: We show that the epithelial lining of developing, as well as adult, mouse and human gallbladders naturally contain interspersed cells that retain the capacity to actively transcribe, translate, package, and release insulin. We show that human gallbladders also contain functional insulin-secreting cells with the potential to naturally respond to glucose in vitro and in situ. Notably, in a non-obese diabetic (NOD) mouse model of type 1 diabetes, we observed that insulin-producing cells in the gallbladder are not targeted by autoimmune cells. Interestingly, in human gallbladders, insulin splice variants are absent, although insulin splice forms are observed in human islets. CONCLUSIONS: In summary, our biochemical, transcriptomic, and functional data in mouse and human gallbladder epithelial cells collectively show the evolutionary and developmental similarities between gallbladder and the pancreas that allow gallbladder epithelial cells to continue insulin production in adult life. Understanding the mechanisms regulating insulin transcription and translation in gallbladder epithelial cells would help guide future studies in type 1 diabetes therapy.

Machine learning workflows identify a microRNA signature of insulin transcription in human tissues.

Dicer knockout mouse models demonstrated a key role for microRNAs in pancreatic ß-cell function. Studies to identify specific microRNA(s) associated with human (pro-)endocrine gene expression are needed. We profiled microRNAs and key pancreatic genes in 353 human tissue samples. Machine learning workflows identified microRNAs associated with (pro-)insulin transcripts in a discovery set of islets (n = 30) and insulin-negative tissues (n = 62). This microRNA signature was validated in remaining 261 tissues that include nine islet samples from individuals with type 2 diabetes. Top eight microRNAs (miR-183-5p, -375-3p, 216b-5p, 183-3p, -7-5p, -217-5p, -7-2-3p, and -429-3p) were confirmed to be associated with and predictive of (pro-)insulin transcript levels. Use of doxycycline-inducible microRNA-overexpressing human pancreatic duct cell lines confirmed the regulatory roles of these microRNAs in (pro-)endocrine gene expression. Knockdown of these microRNAs in human islet cells reduced (pro-)insulin transcript abundance. Our data provide specific microRNAs to further study microRNA-mRNA interactions in regulating insulin transcription.

The intestinal epithelial insulin-like growth factor-1 receptor links glucagon-like peptide-2 action to gut barrier function.

Glucagon-like peptide-2 (GLP-2) is an intestinal growth-promoting hormone used to treat short bowel syndrome. GLP-2 promotes intestinal growth through a mechanism that involves both IGF-1 and the intestinal-epithelial IGF-1 receptor (IE-IGF-1R). GLP-2 also enhances intestinal barrier function, but through an unknown mechanism. We therefore hypothesized that GLP-2-enhanced barrier function requires the IE-IGF-1R and is mediated through alterations in expression and localization of tight junction proteins. Conditional IE-IGF-1R-null and control mice were treated with vehicle or degradation-resistant Gly(2)-GLP-2 for 10 days; some animals also received irinotecan to induce enteritis. Mice were then examined for gastrointestinal permeability to 4-kDa fluorescein isothiocyanate-dextran, jejunal resistance using Ussing chambers, tight junction structure by electron microscopy, and expression and localization of tight junction proteins by immunoblot and immunohistofluorescence, respectively. GLP-2 treatment decreased permeability to 4-kDa fluorescein isothiocyanate-dextran and increased jejunal resistance (P <.05-.01), effects that were lost in IE-IGF-1R-null mice. Electron microscopy did not reveal major structural changes in the tight junctions in any group of animals. However, the tight junctional proteins claudin-3 and -7 were upregulated by GLP-2 in control (P <.05-.01) but not null mice, whereas IE-IGF-1R deletion induced a shift in occludin localization from apical to intracellular domains; no changes were observed in expression or distribution of claudin-15 and zona occludins-1. Finally, in irinotecan-induced enteritis, GLP-2 normalized epithelial barrier function in control (P < .05) but not knockout animals. In conclusion, the effects of GLP-2 on intestinal barrier function are dependent on the IE-IGF-1R and involve modulation of key components of the tight junctional complex.

Ghrelin, the proglucagon-derived peptides and peptide YY in nutrient homeostasis.

Dysregulation of nutrient homeostasis is implicated in the current epidemics of obesity and type 2 diabetes mellitus. The maintenance of homeostasis in the setting of repeated cycles of feeding and fasting occurs through complex interactions between metabolic, hormonal and neural factors. Although pancreatic islets, the liver, muscle, adipocytes and the central nervous system are all key players in this network, the gastrointestinal tract is the first tissue exposed to ingested nutrients and thus has an important role. This Review focuses on several of the endocrine hormones released by the gastrointestinal tract prior to or during nutrient ingestion that have key roles in maintaining energy balance. These hormones include the gastric orexigenic hormone, ghrelin, and the distal L cell anorexigenic and metabolic hormones, glucagon-like peptide (GLP)-1, GLP-2, oxyntomodulin and peptide YY. Each of these hormones exerts a distinct set of biological actions to maintain nutrient homeostasis, the properties of which are currently, or might soon be, exploited in the clinic for the treatment of obesity and type 2 diabetes mellitus.