Recessive mutations in a distal PTF1A enhancer cause isolated pancreatic agenesis

Abstract

The contribution of cis-regulatory mutations to human disease remains poorly understood. Whole-genome sequencing can identify all noncoding variants, yet the discrimination of causal regulatory mutations represents a formidable challenge. We used epigenomic annotation in human embryonic stem cell (hESC)-derived pancreatic progenitor cells to guide the interpretation of whole-genome sequences from individuals with isolated pancreatic agenesis. This analysis uncovered six different recessive mutations in a previously uncharacterized ~400-bp sequence located 25 kb downstream of PTF1A (encoding pancreas-specific transcription factor 1a) in ten families with pancreatic agenesis. We show that this region acts as a developmental enhancer of PTF1A and that the mutations abolish enhancer activity. These mutations are the most common cause of isolated pancreatic agenesis. Integrating genome sequencing and epigenomic annotation in a disease-relevant cell type can thus uncover new noncoding elements underlying human development and disease.

Figure 1. Epigenome annotation of variants from genome sequencing identifies a shared variant in a putative enhancer element

Variant identified by whole genome sequencing, plus an additional five variants in patients with pancreatic agenesis map to a 25 Kb region downstream of PTF1A, which contains a single candidate pancreatic progenitor-specific enhancer within a highly conserved 400bp element. The top panel depicts ChIP-seq density plots for the enhancer mark H3K4me1, the second and third show occupancy for FOXA2 and PDX1. A broad panel of embryonic and adult human tissues do not show active chromatin marks in this region (Supplementary Figure 5). Vertebrate Conservation and mammalian conservation tracks (as measured by the GERP Score) tracks illustrate the high conservation of this element. The red line depicts the approximate location of a 7.6 kb deletion in this region, and red arrows indicate point mutations, using the final 3 digits from the hg19 coordinates as labels (referring to positions 23508305A>G, 23508363A>G, 23508365A>G, 23508437A>G and 23508446A>C on chromosome 10 respectively).

Figure 2. Families with mutations in the PTF1A enhancer.

Black filled individuals have isolated pancreatic agenesis. Striped filled individual had pancreatic agenesis with intrahepatic cholestatic failure from which he died. Dark grey symbols are patients with exocrine insufficiency and young onset diabetes (age at diagnosis < 22 years). * Whole genome sequenced individual. DNA from the parents in family 2, 5, 6 and 8 was not available.

Figure 3. Pancreas agenesis mutations disrupt the function of a transcriptional enhancer that is specifically active in pancreatic progenitors

(A) Reporter assays for the novel enhancer in hESC-derived pancreatic endoderm progenitors, several adult pancreatic exocrine transformed cells (Panc-1, 266-6, AR42J) and HeLa cells. Transcriptional enhancer activity was only observed in hESC-derived pancreatic progenitor cells, and it was disrupted by all 5 mutations. Asterisks indicate Student’s t test P<0.05 for comparisons of mutant vs. wild type enhancers in progenitors. (B) Chromosome conformation capture shows that the newly identified enhancer interacts directly with PTF1A promoter in hES cell-derived pancreatic progenitor cells. The viewpoint at PTF1A is signaled with an arrow, and the approximate regions that were tested for interaction with the viewpoint (C1, C2, C3, and P, for PTF1A promoter) are shown in (C). C4 represents a control region in an unrelated locus. (C) Pancreas agenesis mutations target critical residues in predicted binding sites for pancreatic regulators FOXA2 and PDX1. (D) Electrophoretic mobility shift assay showing high affinity, sequence-specific interaction of FOXA2 with a double stranded oligonucleotide containing the wild type (WT) sequence, but not the 363 A>G mutation (MUT). The retardation signal was suppressed by unlabelled consensus high-affinity binding site for FOXA2, and supershifted with antibodies recognizing FOXA2. Black triangles represent competition gradients of 30 and 100-fold cold probe excess. Binding activity is shown for MIN6 cells and dissected pancreatic buds from e12.5 mouse embryos. Probe refers to the radioactively labelled probed used in binding assays. (E) Electrophoretic mobility shift assay showing sequence-specific interaction of PDX1 with oligonucleotide containing the wild type (WT) sequence, but not the 446 A>C mutation (MUT). The retardation signal was suppressed by unlabelled consensus high-affinity binding site for PDX1, and supershifted with antibodies recognizing PDX1. Competition was performed with 100-fold cold probe excess. Binding activity is shown for MIN6 cells and pancreatic buds from e11.5 mouse embryos.