Abstract:Impaired insulin secretion from pancreatic islet β-cells is a major cause of metabolic dysregulation and type 2 diabetes mellitus. Complete transcriptomic characterization of islets in patients with type 2 diabetes mellitus has yet to be completed, and it remains challenging to link insulin secretion dysfunction with precise changes in gene expression. There are several ongoing initiatives aimed at enabling the discovery of regulatory molecules that might contribute to insulin secretion dysfunction and whole-body glucose homeostasis impairment. In one such line of research, Bacos et al.1 obtained evidence suggesting that the gene PAX5 might play an important role in impaired insulin secretion in human islets (Figure 1). Given the established differences between mouse and human islets, type 2 diabetes mellitus candidate genes that were identified in mice might not have the same regulative effects on human islets. There is an unmet need for powerful transcriptomic analyses that can be applied to human islets isolated from individuals with type 2 diabetes mellitus and nondiabetic controls. Due to the difficulties associated with obtaining human islets, most human islet transcriptomic studies thus far have involved small cohorts and have lacked functional validation. Bacos et al.1 generated a ribonucleic acid sequencing (RNA-Seq) resource bank from the large Lund University Diabetes Center pancreatic islet cohort. It is one of the largest existing type 2 diabetes mellitus human islet cohorts in existence, providing an extensive gene expression resource based on 309 islet preparations in total from individuals with type 2 diabetes mellitus and nondiabetic controls. They then identified 395 differentially expressed genes (DEGs) from the Lund University Diabetes Center cohort, and further performed some functional validation of DEGs in vitro. The utility of transcriptomic resources can be affirmed by robust replication of DEGs across studies and databases. To date, few DEG replication studies in type 2 diabetes mellitus human islets have been reported, and the various studies in the literature showing replication have been relatively small, including the work by Bacos et al.1 Unsurprisingly, the variance in demographic and pathophysiological profiles among sample donors affects DEG overlap across study cohorts. There remains a need to analyze human islets from a more diverse donor pool, and larger cohorts to clarify whether islet gene expression differs among individuals with type 2 diabetes mellitus in relation to demographic and pathophysiological variables. Another important factor affecting DEG overlap is the particular screen technology applied. RNA-Seq, microarray analysis and other screening technologies have been used to identify genes with altered expression in type 2 diabetes mellitus human islets, generating transcriptional resources with differing sensitivities and reliabilities of characteristics. State-of-the-art RNA-Seq technology continues to improve our understanding of the transcriptional atlas of type 2 diabetes mellitus human islets, as shown in the paper by Bacos et al.1 Notably, previous work provides clear evidence that type 2 diabetes mellitus alters the cellular composition of human islets, relative to that of nondiabetic islets. Single-cell RNA-Seq has the potential to circumvent challenges in the clarification of type 2 diabetes mellitus-associated DEGs by enabling cells to be classified according to cell type, thus enabling a cell-type resolved analysis, such that transcriptional variations can be examined at a single-cell resolution level. In a study that included pancreata from 34 human donors with and without diabetes, Camunas-Soler et al.2 showed that single-cell RNA-Seq could be combined with electrophysiological measurements of exocytosis and channel activity to link endocrine physiology to transcriptomic data on a single-cell level. Furthermore, a recently developed method called spatial transcriptome sequencing allows information on the spatial location and gene expression of islet cells to be obtained simultaneously. This innovation can be leveraged as an important screening tool sensitive to interactions between genes and their microenvironments. Validation of whether observed human islet transcriptomic changes are functionally linked to impaired insulin secretion is necessary. The analyses by Bacos et al.1 of individuals not previously diagnosed with type 2 diabetes mellitus in the Lund University Diabetes Center islet glycated hemoglobin (HbA1c) cohort showed that expression of one-third of the identified DEGs associated linearly with HbA1c level. These results suggest that changes in the expression of these genes might precede type 2 diabetes mellitus diagnosis and could potentially contribute to the development of type 2 diabetes mellitus. DEG outcomes depend on the stringency of HbA1c-level range criteria for stratification. Thus, a subset of individuals in the study by Bacos et al.1 who were not considered to have type 2 diabetes mellitus based on their having an HbA1c level <42 mmol/mol (6.0%) could have been prediabetic. Indeed, bioinformatic, genetic and epigenetic analyses have shown that several DEGs identified in the study by Bacos et al.1 had an altered chromatin state or deoxyribonucleic acid methylation; and single-nucleotide polymorphisms associated with these DEGs could have affected type 2 diabetes mellitus etiology and type 2 diabetes mellitus-related metabolic traits. Analyses of public rodent in vivo datasets (see the International Mouse Phenotyping Consortium) have shown that mouse strains deficient in some DEGs had impaired glucose homeostasis and altered body composition. However, it should be clarified whether metabolic defects in such mice are due to islet and/or peripheral effects. The aforementioned analyses suggested that many DEGs identified by Bacos et al.1 might be associated with insulin secretion dysfunction, but direct evidence for this possibility is needed. Functional genomic approaches – wherein the impact of modifying the expression of implicated genes is explored through mimicking type 2 diabetes mellitus-associated changes of DEGs in human islets – are usually required to test DEG effects on insulin secretion. Of the 11 top-ranked and/or critical DEGs that Bacos et al.1 selected for functional validation, six were confirmed to have a functional relationship with perturbed insulin secretion, including three DEGs with increased expression (PAX5, NEFL and PCOLCE2) and three with decreased expression (OPRD1, CHL1 and SLC2A2). Because the manipulated expressions of these DEGs in islet cells differ from their expression profiles in the islets of individuals with type 2 diabetes mellitus, their actual roles in insulin secretion need to be further explored. Partial functional validation experiments carried out with rat clonal β-cells, rather than with human islets, cannot reflect the normal cellular environment of human islets. Thus, despite the difficulty associated with obtaining human pancreata, it is necessary to observe the impacts of these genes' differential expression on the activities of primary human islets. An attractive strategy for developing type 2 diabetes mellitus preventative interventions and treatments is to screen small molecule compounds informed by the prioritization of highly promising genes. Bacos et al.1 confirmed that PAX5 overexpression had a particularly strong suppressive effect on glucose-stimulated insulin secretion. PAX5, a member of the paired box transcription factor family, is an essential transcription factor for B-lymphocyte lineage identity determination during lymphoid differentiation, which has yet to be studied in β-cells3. Potential mechanisms of PAX5's attenuating effect on insulin secretion to be examined include impaired mitochondrial function and β-cell loss in Bacos et al.'s study1, and other mechanisms need to be further explored. Importantly, bioinformatics analysis further suggested that PAX5 might regulate the transcription of many type 2 diabetes mellitus-associated DEGs, including some critical DEGs, such as SCL2A2. Notwithstanding, multiple hypotheses derived from the discovery of type 2 diabetes mellitus candidate gene PAX5 should be examined to potentially inform the development of precision therapeutics. Thus far, analyses have principally considered the effects of individual DEGs in isolation from the effects of other genes and environmental factors. The PAX5 gene showed a high prevalence of somatic mutation, with alterations observed in a few cases4. Thus, it would be timely for attention to be turned to elucidating gene–gene and gene–environment interactions in both rodent and human studies. PAX5 effects should be examined in multiple islet cell types beyond β cell-acting variants, including in α, δ and PP cells. Additionally, influences of PAX5 in peripheral tissues – such as in the liver, skeletal muscle and adipose tissue, which are important in insulin resistance development – should be clarified (Figure 1). SCL2A2 is another gene of particular interest. It encodes GLUT2 (glucose transporter 2), the main glucose transporter in β-cells. Although the effects of SCL2A2 expression on glucose-stimulated insulin secretion function are well established in type 2 diabetes mellitus rodents5, SCL2A2 involvement in human islets has been debated. Bacos et al.1 provided evidence showing that SCL2A2 might also play an important role in insulin secretion in human islets. Notably, potential roles in type 2 diabetes mellitus islets of DEGs whose manipulation did not alter β-cell number or insulin secretion function cannot be ruled out. These genes' effects on other metabolic defects should be further examined. In addition, there are many DEGs reported by Bacos et al.1 that are strongly associated with type 2 diabetes mellitus, but have not been functionally validated. DEGs beyond the top-ranked DEGs should be examined in future studies. The work of Bacos et al.1 provides a valuable resource for obtaining a deeper understanding of type 2 diabetes mellitus pathophysiology and decoding type 2 diabetes mellitus-associated transcriptomic changes that underlie the functional decay of pancreatic islet β-cell insulin secretion. The findings provide novel insights that might be useful for future genomic-based prediction, prevention and treatment of type 2 diabetes mellitus. Notably, the mechanisms by which gene expression changes affect pancreatic islet pathophysiology remain to be clarified, and the clinical implications of these changes should be evaluated judiciously. The authors declare no conflict of interest.

Delineating the Transcriptional Atlas for Impaired Insulin Secretion: A Window into Type 2 Diabetes Pathophysiology