Single-Cell Transcriptome Profiling of Human Pancreatic Islets in Health and Type 2 Diabetes

Abstract

Hormone-secreting cells within pancreatic islets of Langerhans play important roles in metabolic homeostasis and disease. However, their transcriptional characterization is still incomplete. Here, we sequenced the transcriptomes of thousands of human islet cells from healthy and type 2 diabetic donors. We could define specific genetic programs for each individual endocrine and exocrine cell type, even for rare δ, γ, ε, and stellate cells, and revealed subpopulations of α, β, and acinar cells. Intriguingly, δ cells expressed several important receptors, indicating an unrecognized importance of these cells in integrating paracrine and systemic metabolic signals. Genes previously associated with obesity or diabetes were found to correlate with BMI. Finally, comparing healthy and T2D transcriptomes in a cell-type resolved manner uncovered candidates for future functional studies. Altogether, our analyses demonstrate the utility of the generated single-cell gene expression resource.

Graphical abstract

Figure 1

Single-Cell Transcriptome Analyses of Human Pancreas (A) Table of donor information (HbA1c, glycated hemoglobin). (B) Projection of all cells (n = 2,209) onto two dimensions using t-SNE based on the expression values (log₂RPKM) of the 1,000 genes with highest biological variation across cells. (C) Expression (log₂RPKM) of exocrine marker genes (PRSS1 for acinar and KRT19 for ductal cells) overlaid onto the 2D t-SNE as shown in (B). (D) Boxplots displaying the expression levels in the seven obtained clusters of marker genes for MHC class II antigen-presenting cells (CD86), mast cells (TPSAB1), pancreatic stellate cells (COL1A2), and endothelial cells (PLVAP). Median and mean are shown as a line and circle, respectively. Edges of each box indicate the 25th and 75th percentiles. Bars extend to extreme data points and outliers are plotted as gray dots. (E) Two-dimensional t-SNE projection of the endocrine cells (n = 1,554) based on the expression values (log₂RPKM) of the 500 genes with highest biological variation across endocrine cells. The obtained clusters were assigned to the endocrine cell types based on the hormone expression levels in (F). Colors correspond to cell types and shadings indicate donors. Healthy and T2D cells are marked with circles and triangles, respectively. (F) t-SNE representation of cells as in (E) illustrating the expression of the five endocrine hormones: GCG, INS, PPY, SST, and GHRL. Color scale is according to log₂RPKM values, with white and red colors corresponding to minimum (zero) and maximum (log₂RPKM = 21) expression, respectively. (G) Bar graphs showing the percentage of cells classified into cell types, per donor.

Figure 2

Characterization of Endocrine Cell Transcriptomes (A) Boxplots showing the number of genes detected in each cell type (expression threshold, RPKM ≥ 1). Median and mean are shown as a line and circle, respectively. Edges of each box indicate the 25th and 75th percentiles. Bars extend to extreme data points and outliers are plotted as gray dots. (B) Percentage of all mRNAs in respect to the total transcriptome in each cell type (Table S1), with genes ranked according to the expression magnitude in descending order (x axis). (C) Table with the number of significantly enriched genes in α, β, γ, δ, acinar, and ductal cells. (D) Heatmap with expression distributions for the top 25 enriched genes in each of the four endocrine cell types (α, β, γ, and δ cells). The genes were selected based on the magnitude of expression range among the four endocrine cell types. The expression profiles of the ten donors are shown separately and for each endocrine and exocrine cell type, with labels indicating cell type and donor (top). Colors correspond to standardized log₂ expression values, where each cell in the heatmap contains the distribution of values across the cells for each cell type and donor. (E) Boxplots with the expression levels of selected cell-type enriched genes identified in the differential expression analysis: FAP, DPP4, and GPR119 for α cells; LINC01099 for β cells; and LEPR and GHSR for δ cells. Gene expression is shown for α, β, γ, δ, ε, acinar, and ductal cells, with color shadings representing different donors. Median and mean are shown as a line and circle, respectively. Edges of each box indicate the 25th and 75th percentiles. Bars extend to extreme data points and outliers are plotted as gray dots. (F and G) Single-molecule RNA FISH on pancreatic tissue section. (F, left) A representative islet (donor H5) co-stained with GCG (green), FAP (red), and DAPI (blue). (G, left) Islet (donor H5) co-stained with SST (green), LEPR (red), and DAPI (blue). (F and G, right) Zoom-in on merged or individual channels. Quantification and additional images in Table S3. Scale bar represents 25 μm.

Figure 3

Cell-Type-Specific Expression (A) Heatmap showing ε cell enriched gene expression (rows) across the seven cell types for the ten donors (columns). Labels indicating the cell type and donor are shown on the top. Colors in the heatmap correspond to standardized log₂ expression values, where each cell in the heatmap contains the distribution of values across the cells for each cell type and donor. (B) Heatmap with expression levels of transcription factors (rows) across the cells from the six cell types for the ten donors (columns). Blue and red colors correspond to minimum (zero) and maximum (log₂RPKM = 12) expression, respectively. (C) Heatmap with expression of significantly enriched genes (rows) in acinar and ductal cells across the six cell types for the ten donors (columns). The genes were selected based on the magnitude of expression range between the two exocrine cell types. The heatmap was generated as in (A). (D) Heatmap showing the differentially expressed (DE) genes (columns) between the pancreatic stellate (PSCs) and the endothelial cells (rows). The results were obtained using single-cell differential expression (SCDE), and only the genes with a log₂ fold change in expression of at least seven between the two groups are shown. The colors in the heatmap correspond to standardized log₂ expression values, where each cell in the heatmap contains the distribution of values across the cells of each cell type.

Figure 4

Uncovering Subtypes of Endocrine and Exocrine Cells (A) (Left) Two-dimensional t-SNE representation of all α cells (n = 886, 10 donors) using donor-normalized expression values of the 500 most variable genes in α cells. Colored according to cluster assignments. (Right) Heatmap illustrating the top DE genes (columns) between the two α cell clusters (rows). The colors in the heatmap correspond to standardized log₂ expression values, where each cell in heatmap contains the distribution of values across the cells in each cluster. (B) (Left) Two-dimensional t-SNE representation of β cells (n = 270, 10 donors) using donor-normalized expression values of the 50 most variable genes in β cells. (Right) t-SNE representation of cells, colored according to expression (log₂RPKM) of DE genes among the five clusters. (C) (Left) Two-dimensional t-SNE projection of acinar cells (n = 185, 10 donors) using donor-normalized expression values of the 100 most variable genes in acinar cells. (Right) Heatmaps illustrating the top DE genes (columns) per cell cluster (rows). The heatmap was generated as in (A).

Figure 5

Gene Expression Correlates to Donor Physiological Characteristics (A) Heatmap of gene expression associated with BMI in α cells (n = 417, healthy male donors). Genes (rows) with a positive or negative correlation of at least 0.7 in magnitude are ranked in descending order of coefficient. Cells (columns) are ordered in descending order of BMI. Blue and red colors correspond to minimum (zero) and maximum (log₂RPKM = 14) expression, respectively. The corresponding Spearman’s correlation coefficients computed for the genes in each cell type (based on the cells from healthy male donors) are displayed on the right of the heatmap. (B) Scatterplots showing the expression levels of four genes with a robust correlation with BMI. The fitted curve in each scatter shows the linear regression between the expression and BMI, where m is the slope and b the y-intercept. (C) Scatterplots of Spearman’s correlations of genes toward BMI computed based on overall expression in donors (x axis) or for the specific indicated cell type (y axis). Colored are the genes with an absolute correlation coefficient of at least 0.6 within a particular cell type that is also at least 10% higher in respect to the correlation computed using all cells.

Figure 6

Altered Gene Expression in Cells from T2D Individuals (A) Bar graphs showing the number of DE genes between cells from healthy and T2D donors per cell type. (B) Heatmaps with the DE genes between healthy and T2D α (left) and β cells (right). Labels indicating the disease status and sex of the cells are shown on the top. Colors in the heatmap correspond to standardized log₂ expression values, where each cell in the heatmap contains the distribution of values across the cells of each group. (C) Heatmap of the enriched gene sets (using GSEA) within the α, β, γ, δ, acinar, and ductal cell types for healthy and T2D groups. The heatmap is colored according to the adjusted p values (−log₁₀), with red and blue colors corresponding to enrichments among genes up- and downregulated in T2D, respectively. Sets with no significant enrichment are indicated with white color.

Figure 7

Functional Analysis of GLP1R (A) Boxplot showing expression of GLP1R in α, β, γ, δ, ε, acinar, and ductal cells, with color shadings representing different donors. Median and mean are shown as a line and circle, respectively. Edges of each box indicate the 25th and 75th percentiles. Bars extend to extreme data points and outliers are plotted as gray dots. (B) Immunohistochemistry of pancreatic tissue for INS (brown) and GCG (red) (top left and zoom-in on top right), or GLP1R (brown) (lower left and zoom-in on lower right). (C) (Left) FACS analysis of islet cells stained with fluorescently conjugated GLP1R antagonist. Gate shows fraction of the total islet cells positive for GLP1R. (Right) Pie charts with the percentage of α, β, δ, and other cells in FACS analyses of total islet cells (upper) or the GLP1R+ cellular fraction (lower), labeled with INS, GCG, and SST antibodies. (D) Dot plots showing GSIS of human islets from one healthy (left) and one T2D (right) donor treated with 16.7 mM glucose ± 10 nM Exenatide (Ex4).