Skip to main content
. 2021 May 19;10:e67776. doi: 10.7554/eLife.67776

Figure 3. Comparison of ductal clusters 0 vs. 2, 4 vs. 1, and 4 vs. 4:Dmbt1+Ly6d+.

(A) The cluster dendrogram created using dims (used to define the cluster) shows the Euclidean relationships between clusters. The tree is calculated in the principal component analysis space. The genes used to define the tree were set as the variable features of the object. (B) Pearson’s correlation calculated using average gene expression is depicted. (C) Stacked violin plots show five differentially expressed genes (DEGs) sharing similar expression patterns in clusters 0 and 2. (D) The dot plot shows all nine DEGs found when comparing cluster 0 vs. 2. (E) The top eight altered pathways from ingenuity pathways analysis (IPA) comparing cluster 0 vs. 2 are depicted. (F) Stacked violin plots show five DEGs sharing similar expression patterns in clusters 4 and 1. (G) The dot plot shows the top 20 DEGs ranked by fold change when comparing cluster 4 vs. 1. (H) The top eight deregulated pathways from IPA comparing cluster 4 vs. 1 are depicted. (I) Stacked violin plots of five DEGs sharing similar expression patterns in clusters 4:Dmbt1+Ly6d+ and 4. (J) The dot plot shows the top 20 DEGs ranked by fold change when comparing clusters 4:Dmbt1+Ly6d+ and 4. (K) The top eight changed pathways from IPA comparing clusters 4:Dmbt1+Ly6d+ and 4 are depicted.

Figure 3—source data 1. Differentially expressed genes comparing duct cluster 0 vs. 2, duct cluster 4 vs. 1, and duct cluster 4-Dbmt1+Ly6d+ vs. 4.
Figure 3—source data 2. Ingenuity pathways analysis results comparing duct cluster 0 vs. 2, duct cluster 4 vs. 1, and duct cluster 4-Dbmt1+Ly6d+ vs. 4.
Figure 3—source data 3. Ingenuity pathways analysis upstream regulator analysis comparing duct cluster 0 vs. 2, duct cluster 4 vs. 1, and duct cluster 4-Dbmt1+Ly6d+ vs. 4.

Figure 3.

Figure 3—figure supplement 1. Alignment to an adult murine hepatic biliary epithelial cell (BEC) dataset.

Figure 3—figure supplement 1.

(A) Uniform Manifold Approximation and Projection (UMAP) showing alignment of adult murine hepatic BECs (blue) to our murine intrapancreatic bile duct cells (red) and pancreatobiliary cells (green). (B) Clustering of merged datasets defines five clusters. (C) Intrapancreatic bile duct cells in DBA+ duct cluster 3 are primarily located within the merged clusters 0 and 1, and pancreatobiliary cells in DBA+ duct cluster 4 are primarily located within the merged clusters 1 and 2. The heatmap shows the percent of cells from our clusters 3 and 4 within each of the merged clusters 0–4. (D) Feature plots depict the 75th percentile and higher of cells expressing the published gene signatures of hepatic BEC subpopulation A and B, respectively. (E) Cells in clusters 1, 2, and 4 have the strongest enrichment for subpopulation A genes, while cells in clusters 0, 1, and 3 have the strongest enrichment for subpopulation B genes in the merged dataset. (F) Dual violin plots show expression of the ductal marker Sox9 and the YAP1 targets Cyr61, Ankrd1, and Gadd45b in the merged clusters. (G) Dot plot shows expression of hepatic BEC subpopulation A and B genes, analyzed by t-SNE, in Figure 1D of Pepe-Mooney et al., 2019 in our murine pancreas DBA+ duct clusters 0–5.
Figure 3—figure supplement 1—source data 1. Number of cells and average number of genes and transcripts for merged BEC–DBA+ duct clusters 3 and 4 dataset. BEC: biliary epithelial cell.
Figure 3—figure supplement 1—source data 2. Differentially expressed genes in merged BEC–DBA+ duct clusters 3 and 4 dataset. BEC: biliary epithelial cell.
Figure 3—figure supplement 2. Analysis of pancreas duct cells during development.

Figure 3—figure supplement 2.

(A, E, I) Uniform Manifold Approximation and Projections (UMAPs) depict ductal clusters evident at E12.5, E14.5, and E17.5, respectively. (B, F, J) Cluster dendrograms created using dims (used to define the cluster) show the Euclidean relationships between clusters at E12.5, E14.5, and E17.5, respectively. (C, G, K) Dot plots show expression of the top five genes defining adult C57BL/6J duct subpopulations in developmental biology samples at E12.5, E14.5, and E17.5, respectively. (D, H, L) Fraction of cells similar to adult C57BL/6J duct subpopulations (based on gene expression) at E12.5, E14.5, and E17.5, respectively. The developmental model shows cells that are transitioning between defined adult clusters. (M, N) Feature plots of genes that characterize subpopulations of mouse and human duct cells in Baron et al., 2016. (O) Feature plot showing Fth1 expression, which typifies a human pancreas duct subpopulation in Grün et al., 2016.
Figure 3—figure supplement 3. Comparison of DBA+ lectin sorted mouse pancreas duct subpopulations to ALK3+ human pancreas duct subpopulations.

Figure 3—figure supplement 3.

(A–E) Aggregated expression of control feature sets shown in panel (A) was subtracted from the average expression levels of differentially expressed genes (DEGs) for each cluster 0–5 on a single-cell level to determine the AddModuleScore comparing each DBA+ pancreas ductal cluster to ALK3+ human pancreas clusters. Panel (D) shows the number of DEGs in murine DBA+ pancreas duct clusters 0–5 that have a human homolog and could be used in this comparison. (F) Bmpr1a is expressed in a subset of murine DBA+ pancreas duct cells. (G) Stacked violin plots depict expression of centroacinar/terminal ductal cell markers Hes1, Aldh1b1, Aldh1a1, and Aldh1a7 in DBA+ pancreas duct clusters 0–5. (H) Immunostaining identifies ducto-acinar cells in murine pancreas. Yellow arrows point to CPA1 or α-amylase-positive murine ductal cells. Similar to other murine ductal cell markers (Figure 1—figure supplement 1G), DBA lectin also shows heterogenous expression in murine pancreatic duct cells. Blue arrows point to a DBA lectin-negative duct cell. The scale bar is 50 µm.