To the Editor: Monogenic forms of beta cell dysfunction in childhood and young adulthood are rare disorders with genetically heterogeneous aetiologies [1, 2]. They include neonatal diabetes mellitus (~1:300,000 newborn infants), maturity onset diabetes of the young (MODY; ~1–2% of non-autoimmune diabetes) and congenital hypoglycaemia of infancy (HI; ~1:50,000 newborn infants) [1, 2]. At present these disorders are known to involve more than ten genes that are highly expressed in pancreatic beta cells [1–3]. Causal mutations resulting in severely impaired beta cell function and inadequate insulin secretion have been implicated in both early onset diabetes and HI, as has been described for HNF4A (hepatocyte nuclear factor 4, alpha), GCK (glucokinase) or the ATP-sensitive K+ channel genes (KCNJ11, ABCC8) [2, 3].

Glucose-6-phosphatase catalytic subunit 2 (G6PC2), also known as islet specific glucose-6-phosphatase related protein (IGRP), is a glycoprotein embedded in the endoplasmic reticulum membrane [4]. G6PC2 belongs to the glucose-6-phosphatase family that includes the glucose-6-phosphatase catalytic subunit (G6PC), which catalyses the hydrolysis of glucose 6-phosphate to release endogenous glucose from the gluconeogenic tissues, mainly the liver [5]. The expression of G6PC2 is highly specific to pancreatic islets and beta cells [6]. Thus, G6PC2 has been proposed to modulate the beta cell glycolytic pathway and glucose-stimulated insulin secretion by antagonising the activity of glucokinase, the major beta cell glucose sensor [6]. Despite 50% amino acid sequence identity with G6PC, G6PC2 shows little or no enzymatic activity on glucose 6-phosphate hydrolysis [4]. However, we have recently shown that a frequent single nucleotide polymorphism (SNP) within the third intron of G6PC2 is strongly associated with fasting plasma glucose levels and insulin release indices in several independent European populations [6]. Another study independently confirmed our findings and reported several SNPs that had strong effects on fasting plasma glucose levels in the same locus [7]. In a Chinese population, Hu and colleagues identified a novel G6PC2 SNP rs16856187 that confers risk for type 2 diabetes [8]. These genetic findings in humans are consistent with G6pc2 invalidation that results in a decrease of approximately 13% in fasting glucose concentrations in mice [9].

Inactivated heterozygous mutations of GCK that reduce the glucose input into the glycolytic pathway are responsible for MODY type 2; when homozygous, the mutations are responsible for a few cases of permanent neonatal diabetes mellitus. However, gain-of-function mutations that increase glucose metabolism in the beta cell are associated with a persistent phenotype of HI sensitive to diazoxide [1–3]. In order to evaluate whether gain- or loss-of-function mutations in G6PC2 may be a cause of monogenic forms of beta cell dysfunction in humans, we screened 88 patients for mutations of G6PC2. These patients had genetically unexplained non-autoimmune early infancy diabetes (EID) (either transient [n = 12] or permanent [n = 32]), or MODY (n = 16) or the inverse phenotype of congenital HI (n = 28) (Table 1). The patients were recruited by the French Network for the Study of Neonatal Diabetes Mellitus, the CNRS-UMR8090 Unit (Lille, France) and the Department of Metabolic Disorders of Necker Hospital (Paris, France). The study was approved by the local ethics committees, and parents provided written informed consent for the genetic testing of their children. Details of the previous genetic testing in these patients are shown in full in Table 1. G6PC2 is located on human chromosome 2q24, and encodes a 355 amino acid protein (MIM# 608058). The genomic sequences were analysed in 11 fragments spanning the promoter and 5′-untranslated regions (UTR) (from nucleotide c.−1100), all five exons, the exon–intron boundaries and the 3′-UTR (up to c.*1921) (primer sequences and PCR conditions are given in Table 1 of the Electronic supplementary material [ESM]). Amplicons were sequenced on a 3730xl DNA Analyser (Applied Biosystems, Foster City, CA, USA) using a standard protocol. Electrophoregram reads were assembled and analysed with the Variant Reporter software (Applied Biosystems).

Table 1 Clinical characteristics and previous gene screening of patients with unexplained permanent or transient EID, MODY or persistent HI
Full size table

Analysis of the genomic sequence of G6PC2 revealed 19 rare (minor allele frequency ≤0.05) and nine frequent (minor allele frequency >0.05) variants (Table 2). All variants were present in the heterozygous state. They included four exonic and five intronic variants; the other extragenic variants were located in the promoter region and in the 3′-UTR. All four coding variants were located in exon 5 of G6PC2 (Table 2). Three of them were non-synonymous and had been described previously, with an assigned rs number from the dbSNP build 129 database (p.Y207S-rs2232323, p.V219L-rs492594 and p.S342C-rs2232328). The synonymous variant (c.699G>A-p.L233L) had not yet been described: it was found in one MODY patient, but was not present in 294 control chromosomes. We investigated the proband’s family and observed no obvious segregation between this novel coding variant and diabetes. Ten novel rare (minor allele frequency <0.05) but non-coding variants were also identified (Table 2). The c.−752C>T variant located in the promoter region was found in one transient EID patient, and was not present in the control group. This sequence variant maps to a non-conserved genomic region across species, according to the UCSC/Penn State Bioinformatics (www.bx.psu.edu/miller_lab/, accessed 1 July 2008). In silico analysis did not predict any transcription factor binding site in the vicinity of c.−752C>T (Genomatix software: www.genomatix.de/, accessed 1 July 2008). Taken together, these data did not support further investigation of the role of c.−752C>T variant. The two new intronic variants (c.327+91G>C and c.328–147C>T) were identified in one and two permanent EID patients respectively, but not in the control group. These intronic variants are located in non-conserved regions across species. Finally, seven novel variants were found in the 3′ UTR (Table 2). No putative microRNA binding sites were predicted in their vicinity according to the TargetScan software (www.targetscan.org/, accessed 1 July 2008).

Table 2 G6PC2 gene variants identified in the patients with unexplained EID, MODY or persistent HI, and control individuals
Full size table

Our results do not support the notion that G6PC2 disruption may be a common cause of monogenic forms of beta cell dysfunction in humans, at least in our patient cohorts. These findings differ markedly from large epidemiology studies that showed strong associations between common DNA variations at the G6PC2 locus and fasting glucose level [6, 7], or even type 2 diabetes in the Chinese population [8]. Such divergent data between the aetiologies of monogenic beta cell disorders and polygenic type 2 diabetes have already been highlighted in previous studies, which did not show that causal mutations in several genes (such as TCF7L2, HHEX or SLC30A8) led to monogenic beta cell dysfunction [10–12], although these gene loci were confirmed to confer risk strongly for type 2 diabetes. Nonetheless, the approach of screening genes known to be involved in polygenic type 2 diabetes for mutations in rare monogenic diabetes has been remarkably fruitful, as exemplified by the pharmacogenetic impact of KCNJ11 mutations [13, 14].

Following our study based on three limited patient cohorts, albeit representative of rare monogenic forms of beta cell dysfunction, it would obviously be informative to test the association between all the novel variants and fasting glucose level or multifactorial forms of type 2 diabetes at the population level. Finally, further molecular studies are still required to clarify the function of G6PC2 in the pancreatic beta cell, particularly with regard to a role in the glucose phosphorylation pathway.