Integration of single-cell datasets reveals novel transcriptomic signatures of β-cells in human type 2 diabetes

Pancreatic islet beta-cell failure is key to the onset and progression of type 2 diabetes (T2D). The advent of single-cell RNA sequencing (scRNA-seq) has opened the possibility to determine transcriptional signatures specifically relevant for T2D at the beta-cell level. Yet, applications of this technique have been underwhelming, as three independent studies failed to show shared differentially expressed genes in T2D beta-cells. We performed an integrative analysis of the available datasets from these studies to overcome confounding sources of variability and better highlight common T2D beta-cell transcriptomic signatures. After removing low-quality transcriptomes, we retained 3046 single cells expressing 27 931 genes. Cells were integrated to attenuate dataset-specific biases, and clustered into cell type groups. In T2D beta-cells (n = 801), we found 210 upregulated and 16 downregulated genes, identifying key pathways for T2D pathogenesis, including defective insulin secretion, SREBP signaling and oxidative stress. We also compared these results with previous data of human T2D beta-cells from laser capture microdissection and diabetic rat islets, revealing shared beta-cell genes. Overall, the present study encourages the pursuit of single beta-cell RNA-seq analysis, preventing presently identified sources of variability, to identify transcriptomic changes associated with human T2D and underscores specific traits of dysfunctional beta-cells across different models and techniques.