Pancreas agenesis mutations disrupt a lead enhancer controlling a developmental enhancer cluster

Abstract

Sequence variants in cis-acting enhancers are important for polygenic disease, but their role in Mendelian disease is poorly understood. Redundancy between enhancers that regulate the same gene is thought to mitigate the pathogenic impact of enhancer mutations. Recent findings, however, have shown that loss-of-function mutations in a single enhancer near PTF1A cause pancreas agenesis and neonatal diabetes. Using mouse and human genetic models, we show that this enhancer activates an entire PTF1A enhancer cluster in early pancreatic multipotent progenitors. This leading role, therefore, precludes functional redundancy. We further demonstrate that transient expression of PTF1A in multipotent progenitors sets in motion an epigenetic cascade that is required for duct and endocrine differentiation. These findings shed insights into the genome regulatory mechanisms that drive pancreas differentiation. Furthermore, they reveal an enhancer that acts as a regulatory master key and is thus vulnerable to pathogenic loss-of-function mutations.

Keywords: Mendelian disease; NEUROG3; PTF1A; diabetes mellitus; endocrine differentiation; enhancers; non-coding mutations; pancreas development; stem cell differentiation.

Graphical abstract

Figure 1

Ptf1a enhancer deletion in mice causes pancreatic hypoplasia and diabetes (A) Human PTF1A locus and ECR Browser conservation tracks. Sequences with >70% similarity over 100 bp in pairwise alignments are identified by horizontal pink lines on top of each track. The location of the mouse Ptf1a^enhP deletion is shown below. (B and C) (B) Body weight and (C) pancreas weight (expressed as percentage of body weight) in 6- to 11-week-old mice (n = 22 each genotype, Student’s t test p values). (D) Ad libitum glycemia of male mice after weaning (n = 22 each genotype). (E) Basal and post-fed plasma insulin from 7-week-old male mice (n = 7 each genotype) after an overnight fast. (D) and (E) show means ± SEM. Student’s t test. ^∗∗∗p ≤ 0.0001, ^∗∗∗∗p ≤ 0.00001. See also Figure S1.

Figure 2

Ptf1a^enhP controls Ptf1a expression in mouse multipotent pancreatic progenitors (A) HE staining of pancreas from adult control and Ptf1a^enhΔ/enhΔ mice. (B) PTF1A immunofluorescence was preserved in acinar cells from adult Ptf1a^enhΔ/enhΔ pancreas. (C) 3D reconstructions of E11.5 pancreatic buds from in toto immunofluorescence stainings for PTF1A (green), PDX1 (red), and glucagon (GCG, blue). See also Video S1. (D) PTF1A (green) was depleted in dorsal pancreas from E11.5 Ptf1a^enhΔ/enhΔ embryos. PDX1 (red) and NKX6.1 (blue) were co-stained to label MPCs. (E) PTF1A expression in sagittal sections from control and mutant E12.5 neural tube, hypothalamus, cerebellum, and retinal cells. See also Figure S1.

Figure 3

Modeling PTF1A enhancer mutations in human MPCs (A) qRT-PCR of human MPCs for pancreatic progenitor markers (n = 7–13 independent differentiation experiments per genotype, using 6 PTF1A^enhΔ/enhΔ clones—3 lines with 127 bp and 3 lines with 321-bp deletions, see Figure S2B—and 4 PTF1A^+/+ control lines. Graphs show means ± SEM. Mann-Whitney ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001). (B) Quantification of FACS data for PDX1+ NKX6-1+ stage-4 in vitro derived MPCs (n = 8–10 independent differentiation experiments per genotype; ns, not significant). (C) Immunofluorescence of human MPCs (stage 4) shows absence of PTF1A in PTF1A^enhΔ/enhΔ lines, without changes in NKX6-1. See also Figure S2.

Figure 4

PTF1A regulates an evolutionary conserved program in MPCs (A) Differential analysis of H3K27ac at active regulatory regions in human PTF1A^enhΔ/enhΔ MPCs. Regions bound by PTF1A in MPCs are highlighted in pink. (B) Top HOMER de novo motifs of regions bound by PTF1A in MPCs. (C and D) (C) Uniform Manifold Approximation and Projection (UMAP) plots of scATAC and (D) metacell 2D projection of scRNA-seq from E10.5 Ptf1a^enhΔ/enhΔ and Ptf1a^+/+ pancreatic buds. Both identified cells compatible with MPCs and glucagon-expressing cells (Alpha). Cells are colored by cell type (left) or genotype (right). (E) Functional enrichment of 244 genes showing PTF1A-binding or loss of H3K27ac in human mutant cells and differential accessibility or mRNA expression in mouse mutants. (F) Selected PTF1A-regulated genes in human and mouse MPCs. Mesenchymal cells (mesen) are shown as controls. Dot sizes represent adjusted p values, and color shade fold-change in mutant samples. (G–J) Examples of loci showing altered chromatin at PTF1A-bound regions in human mutant cells, and altered chromatin in orthologous or syntenic regions in mouse mutant E10.5 MPCs. Shown are genes involved in endocrinogenesis (NKX2-2, ST18), Notch signaling (HEY1), and cell adhesion (KIRREL2-NPHS1). Mouse tracks show aggregated MPC single-cell chromatin accessibility. See also Figures S3 and S4 and Tables S1, S2, S3, S4, and S5.

Figure 5

PTF1A in MPCs primes endocrine differentiation of mouse bipotent trunk progenitors (A) E12.5-15.5 pancreas showing NKX6-1 (red) in “trunk” bipotent duct-endocrine progenitors, and PTF1A (green) in peripheral pro-acinar cells. White empty arrows point to NKX6-1 negative poorly differentiated trunk cells in Ptf1a^enhΔ/enhΔ pancreas. White solid arrowheads depict PTF1A-positive tip cells in Ptf1a^enhΔ/enhΔ pancreas. (B) NEUROG3+ endocrine progenitors (red) are severely reduced in E13.5 Ptf1a^enhΔ/enhΔ pancreas (see also Figure S5J). (C and D) Insulin (INS), glucagon (GCG), and somatostatin (SOM) immunofluorescence of neonatal (P1) and E18.5 pancreas showed reduced endocrine cells. A representative section from P1 is shown in (C), whereas (D) shows quantifications of the relative pancreas area occupied by each endocrine cell type in E18.5 (n = 6/genotype; ^∗∗p ≤ 0.01, ^∗∗∗Welch’s t-test p ≤ 0.0001). (E) qRT-PCR of endocrine markers in human hPSC-derived beta-like cells (n = 6–8 independent differentiations/genotype, using 6 PTF1A^enhΔ/enhΔ and 4 PTF1A^+/+ control lines). Error bars represent mean ± SEM. Mann-Whitney test, ^∗∗p < 0.01. (F and G) (F) Flow cytometry for C-peptide expressing beta-like cells in differentiated control and mutant S7 stem cell islets (Mann-Whitney test, ^∗∗p < 0.01), and (G) representative FACS plots (n = 6 independent differentiations/genotype). (H) Schematic summarizing the differentiation phenotype in Ptf1a^enhΔ/enhΔ pancreas. See also Figure S5.

Figure 6

PTF1A in MPCs triggers sequential chromatin changes in Neurog3 (A) scATAC UMAP plots of E13.5 Ptf1a^enhΔ/enhΔ and Ptf1a^+/+ pancreas identifies NEUROG3+ endocrine progenitors, pro-acinar progenitors, trunk bipotent progenitors, and mutant-specific trunk null cells. Nuclei are colored by cell type (left) or genotype (right). (B–E) Pseudo-bulk scATAC-seq profiles from E13.5 Ptf1a^+/+ and Ptf1a^enhΔ/enhΔ trunk and trunk null cells. Regions downregulated in trunk null cells (log₂-fold-change < −0.5, binomial test FDR < 0.1) are highlighted in yellow. (B) Depicts Hnf1b and Sox9 loci and (C–E) show reduced accessibility in Ptf1a^enhΔ/enhΔ trunk null cells at indicated sites of endocrine regulatory loci. Profiles in NEUROG3+ cells are shown for comparison. In (E), E13.5 Ptf1a^+/+ trunk progenitors exhibit an active chromatin state at Neurog3 that is similar to NEUROG3+ progenitors, whereas this is abrogated in Ptf1a^enhΔ/enhΔ trunk null cells and is altered at several sites in other Ptf1a^enhΔ/enhΔ trunk cells. (F) Proposed model illustrating sequential steps triggered by PTF1A^enhP activation of PTF1A. PTF1A binds and remodels chromatin at pro-endocrine gene loci in MPCs. Active chromatin states are maintained at endocrine genes such as NEUROG3 in bipotent progenitor trunk cells, enabling full activation of NEUROG3 in endocrine-committed progenitors. PTF1A^enhP deletion prevents this process, causing reduced endocrine differentiation. See also Figure S6.

Figure 7

PTF1A^enhP creates an active enhancer cluster in mouse and human MPCs (A) Regulatory landscape of the human PTF1A locus in PTF1A^+/+ and PTF1A^enhΔ/enhΔ MPCs. Six H3K27ac-enriched putative enhancers and the PTF1A promoter, most of which show strong mediator (MED1) binding, are shaded in gray. All show absent activity in PTF1A^enhΔ/enhΔ MPCs (q < 0.05). ChIP-seq tracks show a MACS2 −log₁₀ p values. (B) scATAC-seq profiles for MPCs from Ptf1a^+/+ and Ptf1a^enhΔ/enhΔ E10.5 pancreatic buds showed chromatin accessibility at Ptf1a and E1-E6 regions orthologous to human enhancers, highlighted in gray. All showed loss in Ptf1a^enhΔ/enhΔ cells (q < 0.1, log₂FC < −0.5). Conservation tracks show multiple alignments between 100 vertebrate species. (C and D) H3K27ac at the PTF1A locus in 2 hPSC-derived pancreatic progenitor datasets (Alvarez-Dominguez et al., 2020; Geusz et al., 2021). Both used a protocol that generates two stages of early pancreatic progenitors: PP1 PDX1+ cells that do not express MPC markers such as NKX6-1, PP2 PDX1+, and NKX6.1+ cells that are comparable with stage 4 MPCs from the current study. In both datasets, H3K27ac enrichment at PTF1A^enhP preceded that of all other enhancers. (E) Summary model illustrating how PTF1A^enhP precedes and activates the enhancer cluster in the PTF1A locus. See also Figure S7.