Case report: maternal mosaicism resulting in inheritance of a novel GATA6 mutation causing pancreatic agenesis and neonatal diabetes mellitus

Although type 1 diabetes mellitus accounts for the vast majority of pediatric diabetes, monogenic forms account for up to 4% of cases [1, 2]. The latter are due to single gene defects affecting pancreatic β-cell function, development or survival and include neonatal diabetes mellitus (NDM). NDM is defined as diabetes occurring in the first six months of life and occurs with an estimated frequency of approximately 1:100 000 live births [2‐5]. Both transient and permanent forms of NDM exist (Table 1). Although clinically indistinguishable from each other at presentation, most transient cases resolve at a median age of 12 weeks but relapse in 50–60% [6, 7]. Different genetic loci and genes are associated with transient versus permanent NDM. Methylation abnormalities at chromosome 6q24 account for two thirds of transient cases, while in permanent NDM, activating mutations in ABCC8 and KCNJ11 are the most common genetic defects in outbred populations and EIF2AK3 is the most frequent in consanguineous groups [3, 7]. KCNJ11 and ABCC8 encode the KIR6.2 and SUR1 subunits of the K_ATP channel, respectively, which links glucose metabolism to insulin secretion. EIF2AK3 encodes a kinase involved in regulating the endoplasmic reticulum stress response to misfolded proteins [8, 9]. Although the mechanism causing diabetes is unclear, defects in EIF2AK3 may cause ER stress in the β-cell from misfolded proteins due to the high demand for insulin secretion, eventually leading to β-cell apoptosis. Alternatively, diabetes may be linked to reduced β-cell proliferation with abnormal insulin trafficking and secretion, as observed in EIF2AK3 knockout mice [9].

Permanent NDM can also result from pancreatic agenesis or hypoplasia, which has been linked to mutations in transcription factors important for pancreatic and β-cell development. The most common of these is GATA6 [10]. Other causes include mutations in PDX1 and the PTF1A enhancer causing isolated abnormal pancreas development, while mutations in PTF1A, RFX6, HNF1B and GATA4 are associated with both pancreatic and extra-pancreatic abnormalities (Table 2) [11‐16]. GATA6 is one of a family of evolutionarily conserved transcription factors recognizing the A/T-GATA-A/G consensus sequence, with key roles in the development and differentiation of multiple cell lineages and tissues [17, 18]. Heterozygous inactivating mutations in GATA6 were identified through a whole exome sequencing strategy in 15/27 (56%) individuals with pancreatic agenesis, defined as NDM requiring insulin treatment and exocrine pancreatic insufficiency requiring enzyme replacement. This is a syndromic form of NDM associated with extra-pancreatic features including cardiac, hepatobiliary, gastrointestinal, neurocognitive and other endocrine involvement. Since its initial discovery, the phenotypic spectrum of GATA6 haploinsufficiency in humans has proven to be more diverse than initially appreciated, and although the initial cases were de novo, several instances of inherited GATA6 mutations have now been described [19‐21]. This report describes three siblings with inherited GATA6 haploinsufficiency and the rare finding of parental mosaicism.

The proband was born at 37 weeks gestation after a pregnancy complicated by severe intrauterine growth restriction, reflected in the birth weight of 1.74 kg (<3rd percentile, −4 standard deviations). He was the first child of non-consanguineous parents of French Canadian background. Hyperglycemia occurred within the first 12 h of life and neonatal diabetes mellitus was diagnosed. An initial ultrasound visualized only the head of the pancreas, suggesting pancreatic hypoplasia. He was placed on subcutaneous insulin therapy with adequate glycemic control. Moderate valvular pulmonary stenosis and a patent ductus arteriosus (PDA) were also diagnosed and valvuloplasty was performed in the neonatal period.

At several months of age, recurrent episodes of hypoglycemia with decreasing insulin requirements began to occur. He was diagnosed with pancreatic exocrine insufficiency (PI) based on the recurrent hypoglycemia, poor weight gain and low stool fecal elastase, leading to initiation of pancreatic enzyme replacement. At 8 months of age, he was readmitted for severe failure to thrive and hypoglycemia secondary to inadequately treated PI, heart failure from the persistent PDA and a previously undiagnosed congenital diaphragmatic hernia. Intestinal malrotation was also noted. His medical management was optimized and he underwent repair of the PDA, diaphragmatic hernia and intestinal malrotation. At the time of surgery, no pancreatic tissue or gallbladder could be identified, demonstrating both pancreatic and gallbladder agenesis.

Despite improvement in his nutritional and overall health status, he continued to have episodes of early morning hypoglycemia, presumably due to deficient glucagon secretion. This was managed with feeds of uncooked cornstarch to provide a source of slowly-released glucose while on insulin injections and, when transitioned to pump therapy, with low basal insulin rates. Control of both the diabetes and exocrine insufficiency have been adequate with no evidence of malabsorption. There has been no clinical evidence of other endocrine dysfunction and thyroid function testing has been normal. He has also demonstrated appropriate neurocognitive development as of 9 years of age. Previous testing of the PDX1 gene, a recognized cause of pancreatic agenesis (Table 2) as well as broader genetic analysis through the 1,000,000 Single Nucleotide Polymorphism project were unrevealing.

A second pregnancy was terminated at 20 weeks due to complex congenital heart disease. Examination of the abortus revealed a male fetus with tetralogy of Fallot, congenital diaphragmatic hernia, severe pancreatic hypoplasia, gallbladder agenesis and intestinal non-rotation (Fig. 1, individual IIB). Subsequently, two male infants were born. The youngest child was found to have a PDA requiring procedural intervention, left pulmonary artery stenosis, and moderate atrial septal defect (Fig. 1, individual IID). Neither the remaining sibling nor either parent have any clinical evidence of diabetes, pancreatic exocrine insufficiency or cardiac disease.

Given the similarity between the proband’s spectrum of congenital anomalies and the clinical phenotype of GATA6 mutations, Sanger sequencing of exons 2–7 of the GATA6 gene was performed in peripheral leukocytes. A novel c.635_660del frameshift mutation (p.Pro212fs) was identified in the proband. The mutation is located in exon 2, within the transcriptional activation domain, and is predicted to result in inclusion of a premature stop codon and likely degradation of the gene transcript by nonsense-mediated decay (Fig. 2). The mother, the abortus and the sibling with the congenital heart defect were also found to carry the mutation (Fig. 1). Notably, the mother, who has no clinical evidence of diabetes, including gestational diabetes or cardiac disease, was found to be mosaic for the mutation (Fig. 1, individual IA). This was confirmed and quantified by next-generation sequencing, revealing 11% mosaicism in peripheral blood. The father and the remaining sibling’s testing were negative.

Histological analysis of the pancreas from the abortus (Fig. 1, individual IIB) demonstrated well-formed islets despite severe pancreatic hypoplasia (Fig. 3). The majority of islet cells stained positive for insulin and were surrounded by smaller populations of glucagon and somatostatin positive cells, demonstrating normal islet morphology.

We describe three siblings heterozygous for the same GATA6 mutation inherited via parental mosaicism, yet with widely different manifestations, further defining the clinical phenotype associated with this rare disease. Since the initial report of heterozygous GATA6 mutations as the most common cause of pancreatic agenesis or hypoplasia, a much greater degree of phenotypic variability has been recognized for both pancreatic and extra-pancreatic features [10, 19]. Diabetes mellitus is the most common pancreatic feature in 98% of cases (Table 3). Although the majority of cases present with NDM, in others diabetes develops later in life or has yet to occur by adulthood. Similarly, although clinical PI has been reported in 81% of cases, subclinical PI and normal exocrine function have also been described (Table 3) [10, 19, 23]. Of the extra-pancreatic features, cardiac defects are the most common, in 88% of cases, although gastrointestinal, hepatobiliary, neurodevelopmental and other endocrine involvement have also been reported (Table 3) [10, 19‐21, 23‐29]. The family described here illustrates this clinical variability. The proband has NDM and PI, with significant cardiothoracic, gastrointestinal and biliary involvement. A similar phenotype was seen in the second affected sibling whereas in contrast, the third sibling has isolated cardiac involvement. Thus, GATA6 haploinsufficiency needs to be strongly considered as a potential cause in cases of diabetes secondary to abnormal pancreas development and, if confirmed, a thorough, multi-system assessment, particularly cardiac, should be performed to evaluate for other abnormalities. Consideration should also be given to screening for glycemic abnormalities in those whom diabetes has not yet developed.

Although most cases of GATA6 mutations are de novo, several cases of dominant inheritance have been reported [10, 19]. The family described here is notable for being amongst these inherited cases and moreover, is only the second report of parental mosaicism [19]. In the previous report, the parent was known to have a congenital heart defect, whereas in our case the mother has no known cardiac abnormalities or pancreatic phenotype. Presumably differing levels of tissue mosaicism is key. Since three out of four offspring inherited the mutation, the level of mosaicism is likely significantly higher in the germline compared to other tissues in our case given the absence of other system involvement and almost certainly higher than the 11% detected in peripheral blood. Accordingly, testing of parents and other potentially affected family members should be performed, even in individuals without an obvious clinical phenotype.

The mechanism underlying the variable expressivity of GATA6 haploinsufficiency remains unclear with no obvious genotype-phenotype relationship [19]. Histologically, grossly preserved islet morphology was observed, which contrasts with murine models demonstrating abnormal morphogenesis with failure of progenitor cells to differentiate into endocrine and exocrine lineages (Fig. 3). However, humans appear to have greater sensitivity to GATA6 gene dosage as both Gata4 alleles in addition to at least one Gata6 allele must be inactivated in mice to cause pancreatic abnormalities [30, 31]. Although not explored in this report, at a molecular level, there are several potential mechanisms. In vitro studies have shown that missense mutations affecting highly conserved zinc finger residues result in failure to bind target promoters in luciferase reporter promoter assays [10, 19, 27]. Given the location of the novel frameshift mutation described here in the transactivation domain, there could be an impact on downstream genetic targets through altered binding of co-regulatory factors. However, the most likely explanation is that the introduction of a premature termination codon results in a reduction in GATA6 mRNA transcript levels through nonsense-mediated decay, as previously described [26]. Variable efficiency of this decay in different tissues could contribute to the phenotypic variability. Modifier genes, epigenetic mechanisms and environmental factors represent other potential mechanisms that could also contribute to the variability associated with GATA6 heterozygous mutations.

Heterozygous mutations in the GATA6 transcription factor gene are the most common cause of neonatal diabetes due to pancreatic agenesis. We describe three siblings heterozygous for the same GATA6 mutation inherited via parental mosaicism, a rare occurrence. Their markedly different manifestations illustrate both the importance of GATA6 for pancreas development as well as the phenotypic variability of GATA6 haploinsufficiency. Elucidating the mechanisms underlying the phenotype and its variability will improve our understanding of pancreas development and may provide insights into the mechanisms underlying other forms of abnormal pancreas development and neonatal diabetes.