A Non-Coding Disease Modifier of Pancreatic Agenesis Identified by Genetic Correction in a Patient-Derived iPSC Line

Summary

GATA6 is a critical regulator of pancreatic development, with heterozygous mutations in this transcription factor being the most common cause of pancreatic agenesis. To study the variability in disease phenotype among individuals harboring these mutations, a patient-induced pluripotent stem cell model was used. Interestingly, GATA6 protein expression remained depressed in pancreatic progenitor cells even after correction of the coding mutation. Screening the regulatory regions of the GATA6 gene in these patient cells and 32 additional agenesis patients revealed a higher minor allele frequency of a SNP 3’ of the GATA6 coding sequence. Introduction of this minor allele SNP by genome editing confirmed its functionality in depressing GATA6 expression and the efficiency of pancreas differentiation. This work highlights a possible genetic modifier contributing to pancreatic agenesis and demonstrates the usefulness of using patient-induced pluripotent stem cells for targeted discovery and validation of non-coding gene variants affecting gene expression and disease penetrance.

Graphical Abstract

eTOC Blurb

Kishore et. al. describe using a patient iPSC line to discover and model the impact of a non-coding variant on disease penetrance. The minor allele variant of a SNP downstream of GATA6 was enriched in patients with pancreas agenesis and was shown to depress GATA6 expression in pancreas progenitors.

Introduction

Pancreatic agenesis (PA) is a rare developmental disorder in which patients exhibit severe exocrine insufficiency and present with neonatal diabetes. Haploinsufficiency of the transcription factor GATA6, caused by heterozygous mutations, is the most common cause of PA (Lango Allen et al., 2012). Interestingly, as is common with many disorders caused by haploinsufficiency, patients with GATA6 heterozygous mutations display a large phenotypic variability, with the pancreatic phenotype ranging from PA, to adult-onset diabetes, to absence of diabetes (Shaw-Smith et al., 2014; Shi et al., 2017). This phenotypic variability can be stochastic, due to mosaicism, and/or result from secondary disease-modifying genes that function in the GATA6 pathway during development (De Franco, Shaw-Smith, Sarah E Flanagan, et al., 2013; Shi et al., 2017; Tiyaboonchai et al., 2017). Another possible explanation for disease penetrance could be differences in the non-coding region of the GATA6 locus that regulate its expression during development (Rodríguez-Seguí, Akerman and Ferrer, 2012; Yorifuji et al., 2012; Catli et al., 2013; Suzuki et al., 2014; Yu et al., 2014; Chao et al., 2015; Stanescu et al., 2015).

The increasing importance of non-coding regulatory regions in development and disease have been recognized fairly recently (Thurman et al., 2012; Gordon and Lyonnet, 2014; Zhang and Lupski, 2015). The non-coding variants residing in these regulatory regions may play a pivotal role in understanding the incomplete penetrance of haploinsufficiency related disorders (Mcclellan and King, 2010; Brewer et al., 2014; Castel et al., 2018). The advent of clustered regularly interspaced short palindromic repeats (CRISPR)-CRISPR-associated 9 (Cas9) gene editing has proven to be an invaluable tool to test the effect of disease modifying single nucleotide polymorphisms (SNPs) on gene expression in specific tissue associated cell lines and diseases (Hsu, Lander and Zhang, 2014; Chen et al., 2018; Wang et al., 2018). However, studies testing the effect of SNPs on gene expression during embryonic development have been limited (Soldner et al., 2016; Pashos et al., 2017).

Human pluripotent stem cells (hPSCs) offer a unique model to study developmental genetic disorders such as PA, where mouse models do not completely phenocopy the human disease (Koutsourakis et al., 1999; Zhao et al., 2005; Carrasco et al., 2012; Xuan et al., 2012). The ability to have clonal populations and isogenic cell lines, where only a single mutation or SNP differentiates two cell lines, makes them ideal for determining the function of a given genomic change on development or disease (Musunuru, 2013). In this study, we used PA patient data and isogenic hPSCs to investigate the role of the non-coding SNP rs12953985, which lies approximately 8kb downstream of the GATA6 gene. Our study identifies a possible genetic modifier contributing to the pancreatic agenesis phenotype in patients with GATA6 mutations.

Results

GATA6 haploinsufficiency leads to a pancreas progenitor defect and a switch in cell fate

Differentiation protocols to generate pancreas β-like cells guide hPSCs through defined stages of development including SOX17+ definitive endoderm (DE), PDX1+ posterior foregut (PFG) and PDX1+/NKX6.1+ pancreas progenitor (PP) stages (Figure 1A) (Pagliuca et al., 2014; Rezania et al., 2014; Nostro et al., 2015). Recently, studies using genome edited embryonic stem cell (ESC) lines have shown that heterozygous mutations in GATA6 lead to a defect at the pancreas progenitor (PP) stage (Shi et al., 2017). Given the variability in the types and locations of mutations in GATA6, the use of iPSC lines derived from PA patients can aid in understanding the effects of different mutations. We used a previously described PA patient iPSC line (iPS^+/mut) and its genetically matched control line (iPS^+/+) to verify the phenotype at the PP stage of differentiation (Stanescu et al., 2015; Tiyaboonchai et al., 2017). This patient had a heterozygous 4 base pair duplication (c.dup606–609) in exon 2 of GATA6 leading to a premature STOP codon (Figure S1A) (Stanescu et al., 2015). An identical mutation was generated in an ESC line (Mel^+/mut) to compare this mutational defect in two different genetic backgrounds (Figures S1A, S1B and Table S1). Flow cytometry was used to measure GATA6 protein levels and examine PDX1+ and NKX6.1+ expression at the DE, PFG and PP stages of differentiation. As expected, all lines differentiated normally to the PFG stage assayed by the percent PDX1+ cells (Figure 1B). The mutant lines (iPS^+/mut and Mel^+/mut) generated fewer PDX1+/NKX6.1+ PP cells compared to their genetically identical controls (iPS^+/+ and Mel^+/+) (Figures 1C, 1D and S1D) and had lower GATA6 protein expression at the PFG and PP stages of differentiation (Figure 1G).

Figure1. GATA6 haploinsufficiency leads to a pancreas progenitor defect and a switch in cell fate.

(A) Schematic of pancreas differentiation protocol from hPSCs. The cell types and key markers at each stage are shown. hPS, undifferentiated hPSC; DE, definitive endoderm; PFG, posterior foregut; PP, pancreas progenitor; β, β-like cells.

(B-H) Data from Mel^+/+, Mel^+/mut, iPS^+/+ and iPS^+/mut cell lines.

(B) Flow cytometry quantification of %PDX1+ cells at the PFG stage.

(C) Representative flow cytometry dot plot for SOX2 and NKX6.1 co-staining gated on PDX1+ cells at the PP stage.

(D-E) Flow cytometry quantification at the PP stage. (D) %PDX1+/NKX6.1+ double positive cells and (E) %PDX1+/SOX2+ double positive cells.

(F) Representative flow cytometry histograms for GATA6 compared to Isotype control at the DE, PFG and PP stages for the Mel^+/+ and iPS^+/+ lines.

(G) Mean fluorescence Intensity (MFI) of GATA6 relative to MFI of Isotype at the DE, PFG and PP stages.

(H) qRT-PCR analysis of key pancreatic and stomach development genes relative to the housekeeping gene TBP and normalized to Mel^+/+ at the PP stage (n = 4).

All data represented as Mean +/− SEM. Ordinary one-way ANOVA with multiple comparisons were used for statistics. *pvalue <0.05, **pvalue <0.01, ***pvalue <0.001, ****pvalue <0.0001.

In mouse models of PA caused by loss of GATA4 and GATA6, the SOX2+ stomach and CDX2+ intestinal domains extend into the pancreatic domain (Xuan and Sussel, 2016). To identify whether the GATA6 mutation resulted in increased expression of stomach or intestinal genes at the PP stage, SOX2+ and CDX2+ cells in the PDX1+ population were measured (Figures S1D and S1E). We found that reduced expression of GATA6 at the PFG and PP stages led to increased SOX2+ cells at the PP stage (Figures 1C, 1D and 1E). SOX2 was not expressed at the PFG stage and CDX2 protein was not detected during any stage of the differentiation (Figure S1E).

A consistent observation was that the patient iPS^+/+ line generated PDX1+/NKX6.1+ PP cells with far less efficiency compared to the Mel^+/+ line. In fact, the iPS^+/+ line had a similar efficiency to the Mel^+/mut line (Figures 1C, 1D and Table S3). While it is not surprising that different cell lines have different baseline efficiencies of differentiation, GATA6 protein expression in these lines displayed an interesting pattern. The expression of GATA6 in the iPS^+/+ and Mel^+/+ lines was identical at the DE stage, but ~30% decreased at the PFG and PP stages in the iPS^+/+ line compared to the Mel^+/+ line. As in PP development, GATA6 protein levels were comparable between the iPS^+/+ and Mel^+/mut lines with the iPS^+/mut line being much lower (Figure 1G and Table S2). Gene expression levels of GATA4, which is regulated by GATA6, showed a similar pattern (Figure 1H) (Lorberbaum and Sussel, 2017; Tiyaboonchai et al., 2017) as did gene expression levels of GATA6, NKX6.1 and PDX1 (Figure 1H). Other pancreas specific genes such as MNX1, PTF1A, NKX2.2, ISL1, NGN3 and NEUROD1 were also found to be downregulated in a similar manner. SOX2 and IRX2, genes involved in stomach development, were found to be upregulated with an inverse expression pattern to GATA6 (Figure 1H). These data suggest that the PA patient iPSC line may carry an additional disease modifier that regulates GATA6 expression specifically during pancreas specification.

The PA patient iPSCs have the minor allele SNP variant rs12953985

The finding that GATA6 expression was lower in PA patient-derived pancreatic progenitor cells, even when the coding mutation was corrected, led us to the hypothesis that a non-coding variant in the regulatory region of GATA6 may act in conjunction with a coding mutation resulting in the PA phenotype. We identified 6 regulatory regions in a 200kb window surrounding GATA6 that were enriched for H3K4Me1 at the PP stage and bind at least two transcription factors that are known regulators of pancreas development at this stage of development (Figure S2A) (Weedon et al., 2014). The only difference between the patient and Mel1 lines was within a 3’ region ~8kb downstream of GATA6 (designated R5) (Figure S2C). The patient line was homozygous for the minor allele A of SNP rs12953985 while the Mel1 line was homozygous for the major allele G variant (Figure S2C). We tested the entire R5 region in enhancer luciferase assays and found activity at the PFG stage, but not the DE stage of differentiation (Figure S2D). Recently published ATAC-seq data also showed accessibility of the chromatin in region R5 only during the PFG stage (Figure S2E) (Lee et al., 2019), and the homozygous minor allele variant of rs12953985 also correlated with lower GATA6 RNA levels in the pituitary and testis when analyzed on GtexEQTL (Figure S2F). This region is conserved in primates but not in rodents (Figure S2H).

The minor allele frequency of the SNP rs12953985 is enriched in PA patients

To confirm our finding in a larger PA patient cohort, the SNP was analyzed in four PA patients having GATA6 mutations, that were referred to the University of Chicago monogenic diabetes registry for genetic testing (Figure S1F and Table S4). By DNA sequencing, all four patients carried the minor allele with two heterozygous for A/G and two homozygous for the A/A minor allele variant. The overall allele frequency in this group was 75% [95%CI=24.5%] (Figures 2A, S1F and Table S4). Sequence analysis of eight wild type hPSC lines resulted in six homozygous for the G/G major allele while only two were heterozygous for A/G (Figure S1F).

Figure2: The minor allele frequency of SNP rs12953985 is higher in PA patients.

(A) rs12953985 minor allele frequency in individuals with PA from the Chicago, Exeter and combined cohorts compared to GnomAD European database. Plots show the frequency of the minor allele A and the 95% confidence intervals.

(B) 4 pedigrees from the Exeter cohort, where the GATA6 mutation was inherited from a parent with a mild or no pancreatic phenotype and the minor allele variant of the SNP was inherited in trans from the other parent. Boxes elaborate the allelic presence of GATA6 coding mutations and rs12953985 genotypes. In the pedigrees, a circle represents female, and a square represents male. Colors indicate phenotype of the patients as follows: Black, PA; Red, Heart Defects; Grey, Adult-onset diabetes; Olive, Small pancreas with normal islets.

We sought to replicate these findings in a cohort of 36 patients of European ancestry (28 with PA, 2 with transient neonatal diabetes, 4 with diabetes diagnosed in adulthood, and 2 with congenital heart defects but no diabetes) heterozygous for GATA6 mutations referred to the Exeter Molecular Genetics laboratory (Table S4). The minor allele, A, was again found to be more frequent among patients with PA compared to the non-PA group (allele frequency 42.3% [95%CI=13%] vs 22.2% [95%CI=19.2%]) (Figure 2A and Table S4). The A allele was present in 25/33 individuals with GATA6 PA tested in the two cohorts for a cumulative allele frequency of 48.5% [95%CI=11.9%] (Figure 2A and Table S4). This is higher than the allele frequency among patients with the GATA6 mutation without PA (27.8% [95%CI=20.7%]) and the frequency among Europeans in the GnomAD database (33.5% [95%CI=0.75%], ChiSquare p= 0.013) (Figure 2A).

Phasing of the GATA6 mutation and the A allele of the SNP in 10 families (8 homozygotes AA and two heterozygotes GA with an inherited GATA6 mutation) showed that the A variant was on the opposite allele of the mutation in 7 cases with PA and only one case without PA (diagnosed with diabetes at 12 years) (Table S4). Interestingly, in four pedigrees with multiple individuals harboring GATA6 mutations, the A allele was in trans with the coding mutation in all the probands with PA while 4 family members without PA either carried the A allele in cis with the coding mutation or were homozygous GG (Figure 2B). The most recent pregnancy for Family Exeter_04 carried a pathogenic GATA6 mutation and was homozygous GG for the SNP. Severe congenital heart defects were detected prenatally leading to a termination of pregnancy. At post-mortem the fetus was found to have a small pancreas with normal islet architecture (Yau et al., 2017). The severity of the pancreatic phenotype in this individual is therefore impossible to assess. The Odds Ratio for carrying the A allele in trans with the mutation suggested that this allele may strongly increase the risk of pancreatic agenesis [OR=7.78, 95%CI=0.8–76.1, p-value=0.0779], however replication in larger PA and non-PA cohorts is needed to confirm this result. These results suggest that the SNP A allele, when found in trans with the coding mutation, is associated with increased risk of PA.

The minor allele variant of rs12953985 lowers GATA6 expression during pancreas specification

To add functional evidence that this SNP influences GATA6 expression and pancreas development, the minor allele variant of the SNP was introduced into the Mel^+/+ and Mel^+/mut lines (Figure S2G and Table S1). The isogenic cell lines, Mel^{+/+ | G/G}, Mel^{+/mut | G/G}, Mel^{+/+ | A/A} and Mel^{+/mut | A/A}, were differentiated to measure GATA6 protein levels and the efficiency of pancreas differentiation. The Mel^{+/+ | G/G} line had the highest GATA6 expression at the PFG and PP stages (Figures 3A, 3B and Table S2). Both, the Mel^{+/mut | G/G} and Mel^{+/+ | A/A} lines, showed ~25% reduction in GATA6 expression while the Mel^{+/mut |A/A} had the lowest GATA6 expression with ~50% reduction at the PFG and PP stages (Figures 3A, 3B and Table S2). This reduction in GATA6 protein expression was verified by western blot (Figure S3B). At the PP stage, the efficiency of generating NKX6.1+ cells or SOX2+ cells was determined (Figures 3C, 3D, 3E, 3F and Table S3). The Mel^{+/+ | A/A} and the Mel^{+/mut | G/G} lines had a lower efficiency of generating NKX6.1+ cells with an increase in SOX2+ cells, while the Mel^{+/mut | A/A} line generated the fewest NKX6.1+ cells and the most SOX2+ cells. These results were also verified using immunofluorescence (Figure S4A). RNA expression of key pancreatic genes GATA4, SOX2, NKX6.1, PDX1, MNX1, PTF1A, NKX2.2, ISL1, NGN3 and NEUROD1, were lower in the lines with lower GATA6 expression (Figure 3G). Conversely, SOX2 and IRX2 RNA expression were higher in these lines suggesting a switch in the cell fate of PDX1+ cells at the PP stage (Figure 3G). Previous studies have shown that GATA6 heterozygous mutations lead to defects in generating C-peptide positive cells at the β-like cell stage (Shi et al., 2017; Tiyaboonchai et al., 2017). All four lines were differentiated to the β-like cell stage and as expected, CPEP+ cells from the Mel^{+/mut | G/G} line were reduced (Figure 3H and 3I). The minor allele variant line harboring the coding mutation (Mel^{+/mut | A/A}) demonstrated a statistically significant decrease in CPEP+ cells compared to its major allele variant counterpart (Mel^{+/mut | G/G})(Figure 3H and 3I). We also detected a decrease in the percentage of SST+ and GCG+ cells at this stage (Figures S3C and S3D). The expression of key β cell genes, islet hormone genes and genes related to β cell functionality were also lower in the lines with lower GATA6 levels (Figure 3J). These experiments were replicated using cells of a different genetic background (CHOPWT6 iPS) with similar results (Figure S3E–J).

Figure3: The minor allele variant of rs12953985 affects pancreas development.

(A-J) Data from Mel^{+/+ | G/G}, Mel^{+/+ | A/A}, Mel^{+/mut | G/G}, Mel^{+/mut | A/A} lines.

(A) Representative flow cytometry histograms for GATA6 compared to Isotype control at the PFG stage.

(B) Mean fluorescence Intensity (MFI) of GATA6 relative to MFI of Isotype at the DE, PFG and PP stages.

(C) Flow cytometry quantification of %PDX1+ cells at the PFG stage.

(D) Representative flow cytometry dot plot for SOX2 and NKX6.1 co-staining at the PP stage.

(E-F) Flow cytometry quantification at the PP stage. (E) %PDX1+/NKX6.1+ double positive cells and (F) %PDX1+/SOX2+ double positive cells.

(G) qRT-PCR analysis of key pancreatic and stomach development genes relative to the housekeeping gene TBP and normalized to Mel^{+/+ | G/G} at the PP stage.

(H) Representative flow cytometry dot plot for C-peptide and forward scatter (FSC) at the β stage.

(I) Quantification of %C-peptide+ cells at the β-like stage. Unpaired two-tailed t-tests were used for statistics. *pvalue <0.05, **pvalue <0.01, ***pvalue <0.001.

(J) qRT-PCR analysis of key pancreatic β cell and islet signature genes relative to the housekeeping gene TBP and normalized to Mel^{+/+ | G/G} at the β-like stage.

All data represented as Mean +/− SEM. Ordinary one-way ANOVA with multiple comparisons were used for statistics. *pvalue <0.05, **pvalue <0.01, ***pvalue <0.001, ****pvalue <0.0001.

The minor allele variant of rs12953985 disrupts RORα binding

To determine if the minor allele variant of rs12953985 disrupted any transcription factor binding sites, we used CIS-BP (Weirauch et al., 2014) and identified a RORα binding domain that could potentially be disrupted by the G>A variant (Figure 4A). Recent bioinformatics data have suggested a possible role for RORα in human pancreas development (Jennings et al., 2017). Robust RORA RNA expression was detected at the DE stage with slightly lower expression at the PFG and PP stages while the ROR family members, RORB and RORC were not expressed (Figure S4A and data not shown). Immunofluorescence showed RORα protein expression at both DE and PFG stages (Figure S4B). To determine RORα binding efficiency in the presence of the G or A variant, ChIP-qPCR was performed at the PFG stage (Figure 4B). The lines with the minor allele A variant bound RORα with lower efficiency compared to the major allele G variant while positive and negative control regions in all four lines were not impacted (Figure 4B). These findings suggest that RORα binds at the 3’ regulatory region, R5, specifically in the presence of the major allele G variant of rs12953985.

Figure4: RORα regulates GATA6 during pancreas development.

(A) Sequence of R5 around rs12953985 (nucleotide in red) and the consensus RORα binding motif from JASPAR.

(B) ChIP-qRTPCR for RORα normalized to IgG control at R5 around rs12953985 and regulatory regions of PDX1, FOXN1, SPARC and GATA4 at the PFG stage for the Mel^{+/+ | G/G}, Mel^{+/+ | A/A}, Mel^{+/mut | G/G}, Mel^{+/mut | A/A} cell lines.

(C) Sequence of R5 around rs12953985. Changes to the minor allele variant are highlighted in red. The green line represents the guide RNA used for targeting with the PAM sequence highlighted in green.

(D-I) Data from iPS^+/+|A/A and iPS^{+/+|cons/cons} lines.

(D) Representative flow cytometry histograms for GATA6 compared to Isotype control at the PFG and PP stages.

(E) Mean fluorescence Intensity (MFI) of GATA6 relative to MFI of Isotype at the PFG and PP stages.

(F) Western blotting for GATA6 and loading control β-Actin at the PFG and PP stages.

(G) Representative flow cytometry dot plot for SOX2 and NKX6.1 co-staining at the PP stage.

(H) Quantification of %PDX1+/NKX6.1+ double positive cells at the PP stage.

(I) Quantification of %PDX1+/SOX2+ double positive cells at the PP stage. All data represented as Mean +/− SEM. Ordinary one-way ANOVA with multiple comparisons were used for statistics. *pvalue <0.05, **pvalue <0.01, ***pvalue <0.001, ****pvalue <0.0001.

By using a known inverse agonist of RORα, SR1001, that represses activity (Figure S4C), we hypothesized that this compound should selectively reduce GATA6 expression only in the presence of the major allele G variant because the minor allele A variant disrupts RORα binding and should not be affected. (Solt and Burris, 2012). By adding SR1001 48 hours prior to the PFG stage, a significant reduction in GATA6 protein expression at the PFG stage in the line was observed with lines having the G allele while those with the A allele were unaffected (Figures S4D and Table S2). The SR1001 treated cultures had a decrease in NKX6.1+ cells carrying the G allele (Figures S4E and Table S3). SR1001 had similar effects in cells of the CHOPWT6 background (Figures S4F and S4G).

These findings were supported by using siRNA to knockdown RORα during pancreas specification. RORα or scrambled siRNAs were transfected 48 hours prior to the PFG stage and cells were harvested at the PFG stage (Figure S4H). RORα protein expression was measured using immunofluorescence and RNA expression was measured by qRT-PCR after transfection with siRNAs (Figures S4I and S4J). A significant reduction in GATA6 protein expression was observed at the PFG stage in the line with the G alleles while those with A alleles were unaffected (Figure S4K). Together, these data suggest that RORα regulates GATA6 during pancreas development via binding at rs12953985.

Modifying the minor allele variant to a consensus RORα binding site rescues GATA6 expression and pancreas differentiation

To confirm the effect of RORα on GATA6 expression, a RORα binding site was introduced into the patient iPS^+/+ line. Initial attempts to generate a single A to G base pair change were unsuccessful due to indel formation (data not shown). To bypass this technical hurdle, a consensus RORα motif was used in which two base changes to the original sequence were introduced enabling the generation of modified clones without indels (iPS^{+/+ | cons/cons})(Table S1 and Figure 4C). The iPS^{+/+ | cons/cons} cells were differentiated and examined at the PFG and PP stages. The introduced RORα consensus motif led to an ~30% increase in GATA6 protein levels at the PFG and PP stages of differentiation (Figures 4D, 4E, 4F and Table S2) and increased NKX6.1+ cells were observed at the PP stage (Figures 4G, 4H and Table S3). These data confirm that a functioning RORα binding site at rs12953985 in the patient iPSC line enhanced the efficiency of pancreas differentiation.

Discussion

Using genome edited isogenic human patient-derived PSC lines, we have shown that the non-coding SNP, rs12953985, in conjunction with a heterozygous GATA6 mutation reduced the efficiency of generating pancreatic progenitor cells in vitro. The frequency of the minor allele of the SNP was enriched among patients with PA and found to be present in trans with the mutation in seven PA patients. Since GATA6 heterozygous mutations are known to have variable clinical penetrance, the presence of this non-coding variant provides one possible mechanistic explanation for the more severe pancreatic agenesis phenotype. Our analysis of human genetics data was limited by the cohort mainly including patients with PA (33 versus 10 patients with diabetes or congenital heart disease) caused by de novo mutations (the mutation was inherited from a heterozygous parent in 4 cases) (Figure 2B and Table S4). Further studies in larger cohorts are needed to assess the effect size of the A allele of SNP rs12953985 on the pancreatic agenesis risk and to identify other contributing factors.

We found that the minor allele variant of SNP rs12953985 leads to the disruption of a RORα binding site, confirming a recent study implicating RORα in regulating key pancreatic transcription factors (Jennings et al., 2017). A more detailed analysis is needed to identify direct targets and pathways regulated by RORα during pancreas development. We do not rule out the possibility that other factors may differentially bind to this site as recent studies have shown that other nuclear receptors can also bind as monomers to the RGGTCA motif (Penvose et al., 2019). Of particular interest is an orphan nuclear receptor, NR5A2, which has been shown to play a role in pancreas development and the maintenance of pancreatic exocrine identity (Hale et al., 2014).

We observed that GATA6 mutant PDX1+ PFG cells develop into SOX2 expressing cells at the expense of NKX6.1 expression. It has been well established that the PDX1+/SOX2+ domain in the developing mouse endoderm at E10.5 gives rise to the antral portion of the stomach (Willet and Mills, 2016; McCracken et al., 2017). Experiments in mouse models deficient in GATA6 and GATA4 have shown that pancreatic lineage cells in the dorsal pancreatic endoderm switch to a stomach identity by expressing SOX2 (Xuan and Sussel, 2016). Additionally, in a patient suffering from dorsal PA, imaging showed the presence of stomach and bowel loops in the distal pancreatic bed (Sandip et al., 2016). Taken together, these results suggest that GATA6 haploinsufficiency leads to a fate switch in the PDX1+ population from NKX6.1+ pancreas progenitors to SOX2+ antral stomach progenitors.

In summary, we highlight the use of hPSCs to identify and validate genome variants that contribute to disease penetrance in heterozygous genetic disorders. By analyzing regulatory regions of causative genes from patients with these disorders, this methodology has the potential to reveal important sites regulating gene expression which would not be possible with typical population-based studies. This could be due to tissue or developmental specificity of regulatory regions and rarity of patients with particular genetic diseases. Finally, the ability to validate a given non-coding variant by genome engineering of hPSCs coupled with in vitro differentiation is critical to confirm the functional consequence of these variants and offer a platform for further investigation.

STAR Methods

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Paul Gadue, The Children’s Hospital of Philadelphia, CTRB 5012, 3501 Civic Center Blvd, Philadelphia, PA 19104; gaduep@email.chop.edu; Phone: +1 267 426 960

Materials Availability

The genome edited stem cell lines are available through the lead contact, Dr. Paul Gadue.

Data and Code Availability

This study analyzed the following datasets. Raw read fastq files for PDX1 ChIPseq GSE58686. H3K4Me1, Input, HNF6, FOXA2 and PDX1 ChIPseq from E-MATB-1990. ATAC-seq raw reads from GSE114101.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

hPSC lines

The Mel1 ES cells were obtained from Ed Stanley and Andrew Elafanty at the Murdoch Children’s Research Institute(Micallef et al., 2012). The CHOP.Panagenesis1 (patient IPS+/indel) IPSC cells were generated from a lymphoblastoid cell line by reprogramming using episomal vectors by the Stem Cell core at the Children’s Hospital of Philadelphia. The CHOPWT6 iPSC line was also reprogrammed by the Stem cell core at the Children’s Hospital of Philadelphia and was previously published as the WTBM1–8 iPSC line (Sullivan et al., 2014).

Monogenic Diabetes Registry

Patients Chicago#1–5 and N6320 were consented for the Monogenic Diabetes Registry, an IRB-approved longitudinal study of people with monogenic diabetes housed at the University of Chicago. Over 3,000 participants are currently enrolled in the Registry and other associated studies. Information about the Monogenic Diabetes Registry can be found at www.monogenicdiabetes.org.

GATA6 testing at the Exeter Molecular Genetics laboratory

The probands of Exeter families 1 to 31 had been referred to Exeter for genetic testing. The GATA6 gene (NM_005257.4) was tested either by PCR followed by Sanger sequencing as previously described (De Franco et al., 2013) or by targeted next generation sequencing of all the known monogenic diabetes genes (Ellard et al., 2013). Clinical information was provided by the referring clinicians via a neonatal diabetes request form (available at www.diabetesgenes.org) or from clinical notes. The study was conducted in accordance with the Declaration of Helsinki principles with informed parental consent given on behalf of children.

METHOD DETAILS

hPSC culture

hPSC cell lines were cultured on 0.1% gelatin and irradiated mouse embryonic fibroblast (MEF) feeder cells in DMEM/F12 supplemented with 2mM of glutamine, 15% Knockout Serum Replacement (KSR), 1X NEAA, penicillin/streptomycin, 0.1mM β-mercaptoethanol and 10ng/ml of bFGF. The medium was changed every day. Cells were passaged when they reach 80% confluence, approximately every 4 days, using TrypLE at a 1:6 ratio. In all hPSC cultures, 5 μM Rho-associated protein kinase (ROCK) inhibitor Y-27632 (Selleck Chemicals, #S1049) was only added into the culture media for ~18 hours when passaging or thawing hPSCs.

Generation of genetically modified hPSCs mediated by CRISPR-CAS9

The 19bp gRNA’s of interest were cloned into the vector (Addgene, #41824) using In-Fusion HD (Clontech Cat No. 639647). PSCs were plated onto gelatin coated 6-well plates with MEF’s 24 hours prior to transfection and were transfected at 40% to 60% confluency. 0.5μg CAS9-GFP (Addgene, plasmid #44719), 0.5μg gRNA, 0.5 μg of ssODN and 3μL Lipofectamine Stem reagent was diluted separately in 50μL DMEM-F12 and gently mixed together. The mixture was incubated at room temperature for 10 minutes and added dropwise into 1 well of cells. 18 to 24 hours post transfection, cells were harvested with TrypLE and cell sorted for GFP positive cells. These cells were plated at low density (~1000–2000 cells / 10cm dish) in human ES cell maintenance media with 5 μM ROCK inhibitor (Cayman chemicals) onto a 1:3 matrigel coated 10cm tissue culture dishes containing MEFs. Approximately 14 to 20 days later single colonies were manually picked and screened. To screen for positive clones, genomic DNA was extracted from the clones by resuspending the cells in 20 μL of Accutaq PCR buffer (Sigma-Aldrich) with 0.1 μg/mL proteinase K (Qiagen) and incubated at 55°C for 60 minutes followed by 95°C for 10 minutes. Cell debris was spun down and 5 μL of supernatant was used for PCR. The PCR products were analyzed by 3.5% agarose gel electrophoresis and sequencing. Heterozygous mutants were confirmed by sub-cloning PCR products using the TOPO TA cloning kit (Life Technologies) sub-cloning of cells to exclude potential contamination of cells with different genotypes (e.g., mixing of homozygous mutant cells with WT cells). All cell lines, primers, guide RNA’s and single stranded DNA oligos used for genome editing are listed in the Table S5.

Pancreatic differentiation from hPSC’s

In all differentiation assays, mutants were analyzed in parallel with isogenic WT controls. hPSC’s were passaged onto 1:30 matrigel coated 6 well plates using TryplE and ROCK inhibitor. Cells were fed with hPSC media every day until they reached ~90% confluency. Pancreas differentiation was initiated on day 0 with RPMI media supplemented with 3μM Chir99021 and 100μg/ml Activin A. On day 1 media was changed to RPMI with 100μg/ml Activin A, 0.3μM Chir99021 and 5μg/ml bFGF. Day 2 was SFD with 100μg/ml Activin A. Cells were harvested on day 3 (DE stage) for flow cytometry analysis and RNA collection using 0.25% Trypsin for 5 minutes. From days 3 to 5 cells were fed with DMEM-F12 containing 0.25mM ascorbic acid, 50ng/ml FGF7 and 1.25μM IWP2. Day 6–8 media contained DMEM high glucose (5g/L) supplemented with 1:100 B27 without RA, 1X glutamax, 0.25mM ascorbic acid, 1:200 ITS-X, 50ng/ml FGF7, 0.5μM SANT-1, 1μM Retinoic Acid, 100nM LDN-193189 and 500nM Phorbol. For experiments with SR1001, the RORα inverse agonist, 1μM of SR1001 was added to the media from day 6–8. Cells were harvested on day 8 (PFG stage) for flow cytometry analysis and RNA collection. Media for days 9–11 consisted of DMEM high glucose (5g/L) supplemented with 1:100 B27 without RA, 1X glutamax, 0.25mM ascorbic acid, 1:200 ITS-X, 2ng/ml FGF7, 0.5μM SANT-1, 0.1μM Retinoic Acid, 200nM LDN-193189 and 250nM Phorbol. Cells were harvested on day 11 (PP stage) for flow cytometry analysis and RNA collection. From days 12–14 the media was changed to MCDB131 supplemented with 20mM glucose, 2% FBS, 1X Glutamax, 1:200 ITS-X, 10ug/ml Heparin, 10uM Zinc sulfate, 0.5μM SANT-1, 0.05μM Retinoic Acid, 200nM LDN-193189, 1μM T3 and 10μM ALK5i II. From day 14–28 cells were fed every other day with media that contained MCDB131 with 20mM glucose, 2% FBS, 1X Glutamax, 1:200 ITS-X, 10ug/ml Heparin, 10uM Zinc sulfate, 200nM LDN-193189, 1μM T3, 10μM ALK5i II and 100nM GSIS XX. From day 29–40 cells were fed every other day after with media that contained MCDB131 with 20mM glucose, 2% FBS, 1X Glutamax, 1:200 ITS-X, 10μg/ml Heparin, 10uM Zinc sulfate, 1μM T3, 10μM ALK5i II, 1mM N-acetyl cysteine, 10μM Trolox and 2μM R428. Cells were harvested on day 40 (β-like stage) for flow cytometry analysis and RNA collection.

Flow Cytometry

Single cell suspensions were prepared by treating cells with 0.25% Trypsin/EDTA for 3 to 5 minutes. For intracellular staining, cells were fixed with 1.6% paraformaldehyde (Electron Microscopy Science) for 30 minutes at 37°C. Cells were washed, permeabilized and stained with 1X saponin buffer (Biolegend). Primary antibodies were diluted to the appropriate concentrations in 100uL of saponin buffer and cells were stained for 30 minutes at room temperature. Samples were washed using 100uL saponin twice and incubated for 30min using the appropriated secondary antibody. Following the staining, cells were resuspended in FACS buffer (DPBS with 0.1% BSA and 0.1% sodium azide). All samples were run on a FACSCantos II or Cytoflex flow cytometer (Becton Dickinson) and analyzed using FlowJo (Treestar) software program.

RNA isolation and cDNA synthesis

Cells were lysed using Lysis buffer provided with the PureLink RNA Micro Kit (Invitrogen Cat No 12183–016) and stored at −80c. To harvest RNA, samples were thawed out at 4c and RNA was extracted using the PureLink RNA Micro Kit following the manufacturer’s instructions. 14μl of RNAse free water was used to resuspend the isolated RNA. cDNA was produced using the SuperScript™ III First-Strand Synthesis System kit (Invitrogen). Quantitative PCR was carried out on a LightCycler 480 II with SYBR select master mix (Invitrogen). For all experiments, TBP (Veazey and Golding, 2011) was used as a housekeeping gene to determine relative gene expression levels. Gene expression levels were then divided by wild type levels for better graphical representation. All primers used for qPCR are in the Table S6.

Immunofluorescence staining

At days 2, 7 and 10 of the differentiation, cells were harvested by incubating with 0.25% Trypsin for 5 mins at 37c. These cells were spun down at 1200rpm for 3 mins in PBS and plated onto 1:3 matrigel coated glass coverslips in the appropriate differentiation media with 5 μM ROCK inhibitor. Cells were fixed the next day at DE (day 3), PFG (day 8) or PP (day 11) stages. For the NKX6.1 and SOX2 stains cells were fixed in 4% PFA in PBS for 15 minutes at room temperature. For the RORα staining, cells were fixed at −20c for 20 mins using cold methanol. Fixed cells were then washed 5–8 times with PBS on ice. Fixed cells were blocked for one hour (5% normal goat serum, 0.3% TritonX-100 in DPBS) and stained in primary antibody in staining buffer (1% BSA, 0.3% TritonX-100 in DPBS) overnight at 4°C. After washing with 3 times PBS for 5 mins each, cells were stained in secondary antibody in staining buffer for 2 hours at room temperature. After washing with 3 times PBS for 5 mins each, cells were stained with Hoescth diluted in PBS for 15 mins. Slides were viewed under a Leica DMI 4000B microscope and digital images were captured with Leica Application Suite software.

Enhancer cloning and luciferase reporter assays

The 3’ regulatory region, R5, was PCR amplified from genomic DNA of the Mel^{+/+ | G/G} cell line with Phusion High-Fidelity DNA Polymerase (New England BioLabs) (primers in TableS2). These PCR products were cleaned up using NucleoSpin® Gel and PCR Clean-up kit (Machery-nagel) and cloned into a pGL4.23[luc2/minP] vector backbone (Promega) using In-FUSION HD. Correct cloning was assessed by Sanger sequencing and restriction enzyme digestion. DNA was prepared with the HiSpeed Plasmid Maxi Kit (Qiagen). At days 2 and 7 of the differentiation, cells were harvested by incubating with 0.25% Trypsin for 5 mins at 37c. These cells were spun down at 1200rpm for 3 mins in PBS and plated onto 1:30 matrigel coated 12 well plates in the appropriate differentiation media with 5 μM ROCK inhibitor. These cells were then transfected on the same day with either 1ug of pGL4.23-R5 vector or empty pGL4.23 vector and 4ng of Renilla normalizer control pGL4.75[hRluc/CMV] (Promega) using Lipofectamine Stem reagent (Invitrogen) in Opti-MEM (Gibco) according to the manufacturer’s’ instructions. Luciferase activity was measured 24 h after transfection (at the DE and PFG stages) with the Dual-Luciferase Reporter Assay System (Promega). Firefly luciferase activity was normalized to Renilla luciferase activity and then to the amount of empty pGL4.23[luc2/minP] vector backbone.

ChIP-qPCR

PP cells (1x10⁷ cells) were harvested using 0.25% Trypsin and cross-linked in 1% formaldehyde in PBS by shaking for 10 min at room temperature. The cross-linking reaction was stopped by the addition of glycine to a final concentration of 125 mM and shaking for 5 min at room temperature. Cross-linked cells were washed 3X with ice cold PBS and pelleted by spinning at 2000rpm for 5 min at 4c. For chromatin fragmentation, cells were resuspended in 1ml cell lysis buffer (10 mM Tris-HCl (pH 8.0), 10 mM NaCl, 0.2% NP-40) with fresh protease inhibitor and PMSF for 10 min on ice. Cells were spun down and resuspended in 1ml nuclei lysis buffer (50mM Tris-HCL (pH 8.1), 10mM EDTA, 1%SDS) with fresh protease inhibitor and PMSF and sonicated in a Covaris S220 sonicator with a duty cycle of 2%, a peak incident power of 105 W and 200 cycles per burst for 20 min. The fragmented chromatin was diluted 1:2 in IP dilution buffer (20 mM Tris-HCl (pH 8.1), 2 mM EDTA, 150mM NaCl, 1% Triton X-100, 0.01% SDS, Protease Inhibitors) and directly used for immunoprecipitation. Samples were precleared with 5ug IgG isotype, Protein G agarose beads for 2 hours at 4c. 180ul of the supernatant was saved as Input DNA. The rest of the supernatant were split equally into 2 1ml tubes (5 X 10⁶ cell each) for incubation with 10ug RORα or IgG control antibody. These antibodies were pre bound to Protein G agarose beads overnight at 4c. Beads were then washed 5 times using (1.) IP wash 1(20mM Tris-HCL (Ph 8.1), 2mM EDTA, 50mM NaCl, 1% Triton X-100, 0.1%SDS), (2 and 3.) High salt buffer (20mM Tris-HCL (Ph 8.1), 2mM EDTA, 500mM NaCl, 1% Triton X-100, 0.01%SDS), (3.) IP wash 2 (10mM Tris-HCL (Ph 8.1), 1mM EDTA, 0.25M LiCl, 1% NP-40, 1% deoxycholic acid), and (4 and 5.) TE buffer (10 mM Tris-HCl (pH 8.0), 1 mM EDTA). Subsequently, protein-DNA complexes were eluted from the beads in Elution-Buffer (100 mM NaHCO3 and 1% SDS) at 65c for 20 min. Cross-links were reversed at 65c overnight and incubated with ProteinaseK and RNase for 2 hours at 55c. DNA was extracted using 400 μl of phenol/chloroform/isoamyl alcohol by vortexing then centrifuged at 14,000 rpm for 5 min at room temperature. The aqueous layer containing pulled down genomic DNA was transferred to fresh 1.5 ml Eppendorf tubes. 16uL of 5M NaCl, 1.5uL of 20mg/mL glycogen and 800μl of 100% ethanol were added to the samples, which were then vortexed. The samples were next incubated overnight at −20°C to precipitate the DNA. Precipit ated DNA was pelleted by centrifuging at 14,000 rpm for 30 min at 4°C. The DNA pellet was then washed with ice-cold 70% ethanol and centrifuged at 14,000 rpm for 5 min at 4°C. The DNA was resuspended with 100ul TE buffer and 1.5ul was used for each qPCR reaction.

In order to identify positive control regions for the ChIP, we looked for RORα binding motifs within active regulatory regions of known pancreas specific transcription factors, such as GATA4, PDX1 and SOX9. We also tested gene expression of SPARC and NR1D2 at the PFG stage. RORα has been demonstrated to bind to the regulatory regions of SPARC and NR1D2 in HepG2 cells (Chauvet et al., 2011). We detected RORα binding at the PFG stage in a previously described regulatory region of SPARC as well as at a region proximal to GATA4 (Figure 4D).

Chipseq data analysis

Raw read fastq files for PDX1 ChIPseq (GSE58686 Teo et. al.), H3K4Me1, Input, HNF6, FOXA2 and PDX1 ChIPseq (E-MATB-1990 Weedon et. al., 2014) were downloaded. ATAC-seq raw read were downloaded from GSE114101 (Lee et al., 2019). Next, reads were aligned to hg19 genome using bowtie2 (2.2.6) with only 1 mismatch allowed per read and only 1 alignment per read. These sam files were converted to bam format using samtools1.6. Next the bam files were converted to bedgraph format using bedtools2.27.1. We used MACS2 for peak calling with a p value cut-off of 1e-5. These peaks were verified by performing motif analysis using the findMotifsGenome.pl command on HOMER. Bedgraph files for Input, H3K4Me1 and peaks for PDX1, HNF6 and FOXA2 were loaded onto Intergrative Genomic Viewer (IGV v2.3) for visualization.

Western Blot

PFG and PP cells (1x10⁷ cells) were harvested using 0.25% Trypsin, washed twice with PBS, pelleted and stored at −80c. Protein from cell pellet was quantified using Pierce™ BCA Protein Assay Kit (Thermo Fischer scientific, cat No. 23227). 20ug of protein was loaded onto a 4–12% Bis-Tris SDS-polyacrylamide gel (Invitrogen). Samples were transferred into a PVDF membrane (Thermo Fisher) and membrane blocking was performed using 2% nonfat dry milk. The membrane was stained in primary antibody diluted in 2% nonfat dry milk overnight at 4°C. After washing 3 times with 1X PBS-T for 5 mins each, the membrane was stained in secondary antibody diluted in 2% nonfat dry milk for 1 hour at room temperature. The membrane was washed 3 times with 1X PBS-T for 5 mins each. HRP was detected using Pierce™ TMB Substrate Kit and membrane was exposed to HyBlot CL autoradiography film (Denville Scientific) to visualize the protein band.

SiRNA knockdown of RORα

Human RORα DsiRNAs (hs.Ri.RORA.13.2) and Scrambled negative control DsiRNA were obtained from Integrated DNA Technologies (IDT). At day6 of the differentiation, cells were harvested by incubating with 0.25% Trypsin for 5 mins at 37c. These cells were spun down at 1200rpm for 3 mins in PBS and plated onto 1:30 matrigel coated 12 well plates at a 1:1 ratio in the appropriate differentiation media with 5 μM ROCK inhibitor. 10nM of the control scrambled siRNA or RORα siRNA (13.2) were added onto these cells at the same time with Lipofectamine RNAi MAX (Invitrogen) following the recommended procedure. SiRNAs were removed the next day by replacing with fresh media. Transfected cells were harvested and examined for knockdown efficiency 48 hrs after transfection (PFG stage) for RNA level by qRT-PCR and protein level by immunofluorescence. Cells were also fixed for intracellular flow cytometry as described.

Experimental replication

In the figures, every data point represents a biological replicate. Each biological replicate is a new differentiation from pluripotent stem cells. From every biological replicate, cells were harvested at different stages of the differentiation. All data is represented as Mean +/− SEM.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analysis was performed using the GraphPad Prism software. The results are expressed as the mean ± standard error of the mean. Ordinary one-way ANOVA with multiple comparisons were used for statistics with correction for multiple comparisons using statistical hypothesis testing perform using Tukey. An unpaired two-tailed Student’s t-test with the assumption of same SD was performed for the enhancer luciferase assays. An unpaired two-tailed Student’s t-test performed for the quantifying c-peptide+ cells at the β-like stage. In figures * P<0.05, **P<0.01, ***P<0.001 and ****P<0.0001. Statistical analysis on the patient cohort data was performed using R. Chi-square and Fisher Exact probability tests were used to compare the frequency of the minor allele of the SNP in different groups. Odds ratios were calculated to estimate the risk of the minor allele A in the PA versus non PA groups.

Supplementary Material

1. Table S4 related to Figure 2:

Table of pancreatic agenesis patients in this study with their GATA6 mutations, mode of inheritance, pancreas phenotypes, rs12953985 genotypes, phasing and country of origin indicated.

Key Resources Table:

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit anti-GATA6	Cell Signaling Technology	5851S
Biotinylated goat anti PDX-1/IPF1	R&D Systems	BAF2419
Mouse IgG1 anti-NKX6.1	DSHB	F55A10
Rat anti-Somatostatin	Santa Cruz	sc-47706
Mouse IgG1 anti-Glucagon	Sigma-Aldrich	G2654-.2ML
Rabbit anti-SOX2	Cell Signaling Technology	3579
Mouse IgG1 anti-SOX2	Biolegend	656102
Mouse IgG1 anti-SOX17	BD Pharmingen	561590
Rabbit anti-C-Peptide	Cell Signaling	4593S
Rabbit anti-CDX2	Abcam	ab76541
Hoechst 33342 solution	Thermo Fischer Scientific	62249
Rabbit anti-ROR alpha	Abcam	ab60134
Goat anti-mouse IgG1-488	Jackson Immunoresearch	115-545-205
Goat anti-mouse IgG1-PE	Jackson Immunoresearch	115-115-205
Goat anti-mouse IgG1-647	Jackson Immunoresearch	115-605-205
Goat anti-rabbit alexa 647	Invitrogen	A21245
Goat anti-rabbit IgG-PE	Jackson Immunoresearch	111-116-144
Donkey anti-mouse IgG alexa647	Jackson Immunoresearch	715-605-150
Donkey anti-rabbit IgG-PE	Jackson Immunoresearch	711-116-152
Streptavidin, Pacific Blue conjugate	Thermo Fischer Scientific	S11222
Goat anti Rat alexa 647	Thermo Fischer scientific	A21247
Chemicals, Peptides, and Recombinant Proteins
Iscove’s DMEM	Corning	10-016-CV
DMEM/F12	Corning	10-092-CV
2-β mercapthoethanol – 55mM	Invitrogen	21985023
NEAA – 10 mM	Invitrogen	11140050
Recombinant human bFGF	R & D Systems	233-FB/CF
Y-27632 (ROCK inhibitor)	R&D systems	1254/50
Fetal Bovine Serum	Tissue Culture Biologicals	101
Knock-out Serum Replacement	Invitrogen	10828-028
TRYPLE Express w/ Phenol Red	Invitrogen	12605010
Pen/Strep 100X	Mediatech	MT30-002-CI
L-Glutamine	Mediatech	MT25-005-CI
Gelatin	Sigma	G1890
Activin A	R&D	338-AC
CHIR99021	Tocris	4423
bFGF	Thermo Fischer Scientific	PHG0263
FGF7	R&D	251-KG-050/CF
SANT-1	Sigma-Aldrich	S4572
Retinoic Acid	Sigma-Aldrich	R2625
LDN-193189	STEMGENT	04-0074
Phorbol 12-myristate 13-acetate	Tocris	1201
Heparin	Sigma-Aldrich	H3149
Zinc Sulfate	Sigma	1724769
T3 (250 mg)	Sigma-Aldrich	T6397
ALK5 inhibitor	Enzo Life Sciences	ALX-270-445
GSIXX	Calbiochem	565789
N-Acetyl-L-cysteine	Sigma-Aldrich	A9165
Trolox	EMD	648471
SR1001	Tocris	4868
R428	Selleck Chemicals	S2841
Critical Commercial Assays
Dual-Luciferase® Reporter Assay System	Promega	E1910
Invitrogen PureLink RNA Micro Kit	Invitrogen	12183-016
Gibson Assembly® Master Mix	NEB	E2611S
NucleoSpin® Gel and PCR Clean-up	Machery-nagel	#740609.250
HiSpeed Plasmid Maxi Kit	Qiagen	12663
SuperScript™ III First-Strand Synthesis System	Invitrogen	18080051
In-Fusion HD cloning kit	Clonetech	639647
Experimental Models: Cell Lines
Provided in Table S1
Oligonucleotides
Guide RNA’s, ssODN’s, primers and siRNA’s	This paper	Table S5
Primers for quantitative RT-PCR	This paper	Table S6
Recombinant DNA
pGL4.23[luc2/minP]	Promega	E8411
pGL4.75[hRluc/CMV]	Promega	E6931
pCas9_GFP	pCas9_GFP was a gift from Kiran Musunuru (Addgene plasmid # 44719; http://n2t.net/addgene:44719; RRID:Addgene_44719)	Addgene #44719
gRNA empty vector	gRNA_Cloning Vector was a gift from George Church (Addgene plasmid # 41824; http://n2t.net/addgene:41824; RRID:Addgene_41824)	Addgene #41824
Software and Algorithms
FlowJo	Ashland	https://www.flowjo.com/solutions/flowjo/downloads
GraphPad Prism	GraphPad Software	https://www.graphpad.com/support/faqid/%201952
ApE plasmid Editor	M. Wayne Davis	http://biologylabs.utorgensen/wayned/ap
Leica Application Suite X	Leica	https://www.leica-stems.com/products/microscope-e/details/product/leica-las-x-ls/
Bowtie2	n/a	http://bowtie-bio.sourceforge.net/bowtie2/index.shtml
Samtools 1.6	n/a	http://samtools.sourceforge.net/
Bedtools 2.27.1	n/a	https://bedtools.readthedocs.io/en/latest/
MACS2	n/a	http://liulab.dfci.harvard.edu/MACS/
HOMER	n/a	http://homer.ucsd.edu/homer/
Intergrative Genomic Viewer (IGV v2.3)	n/a	http://software.broadinstitute.org/software/igv/
SnapGene	n/a	https://www.snapgene.com/
Other
Monogenic Diabetes Registry	n/a	www.monogenicdiabetes.org

Highlights.

• A disease modifying SNP was discovered using a pancreas agenesis iPSC line.

• This non-coding SNP is enriched in patients with GATA6 induced pancreas agenesis.

• The A variant of SNP rs12953985 depresses GATA6 expression in pancreas precursors.