Common intronic variation within the gene encoding transcription factor 7-like 2 (TCF7L2) is now considered to be definitively associated with type 2 diabetes (T2D). Since our first report of this association in 2006 (1), independent investigators have readily replicated this finding in all the main ethnic groups; in addition, from the first genome-wide association study (GWAS) of T2D in Caucasians (2) onwards, the strongest association has been consistently with the TCF7L2 locus. Indeed, a meta-analysis of published studies worldwide as early as 2007 estimated a pooled odds ratio of 1.46 (with an impressive P = 5.4 × 10−140) (3), making it one of the most statistically significant genetic findings in T2D to date.
Despite knowing that this association has been beyond doubt for over 6 years, the mechanism through which TCF7L2 exerts its effect on T2D is still very unclear, thus making the work by Kaminska et al. (4) in this current issue of Diabetes so timely.
TCF7L2 is a high-mobility group box-containing transcription factor and operates at the last key stage of the canonical Wnt signaling transduction cascade, regulating the expression of a set of target genes. Moreover, tissue-specific isoforms have been reported (5,6), with the transcript being heavily alternatively spliced at the 3′ terminus, with the 5′ end being the most stable region among isoforms.
Following the discovery of this genetic association, it was initially speculated that TCF7L2 operated in tandem with insulin to influence blood glucose homeostasis through the alteration of levels of glucagon-like peptide 1 in the gut (1). However, other studies have shown that the TCF7L2 genetic variation is associated with increased TCF7L2 expression and decreased insulin secretion, possibly implicating the pancreatic β-cell (7,8), although an indirect effect on insulin secretion by TCF7L2 action in another tissue cannot be ruled out. Indeed, from one of our previous studies of alternative splicing of TCF7L2 in metabolically relevant tissues and cell lines, we suggested that variants in TCF7L2 correlate with an insulin-suppressed isoform retaining exon 13 and 13a in subcutaneous adipose (5).
Following on from this latter observation, Kaminska et al. elected to investigate the mRNA expression pattern of alternative splicing in the context of weight loss following Roux-en-Y gastric bypass surgery. Interestingly, they observed that expression levels of a specific spliced isoform, which they termed a “short mRNA variant” i.e., lacking exons 12, 13, and 13a, was reduced both in subcutaneous fat and in liver following surgery. In addition, this short isoform was more common among T2D cases than in normal glucose obese subjects or in the general population. Moreover, there was a positive correlation with this isoform and fasting glucose levels in nondiabetic subjects, which is in line with individuals with both impaired glucose tolerance and harboring the at-risk TCF7L2 variant being at a greater risk of T2D (7); in addition there was also association with high levels of free fatty acids in serum during hyperinsulinemia. This final point is in agreement with our observation (5) in that the short form in suppressed by insulin, suggesting impaired insulin action at this site.
What is most intriguing is that TCF7L2 genotype status is not correlated with these isoform observations. This is not entirely surprising, as it is clear that genetic variation at this locus is not associated with obesity. After all there have been many GWASs of BMI and obesity in both adults and children, and they have never detected an association with this locus. That said, the fact that more than one study shows there is an isoform correlation with various metrics of metabolic activity in adipose suggests that this is a key gene for insulin action within this tissue, but its mechanism of action appears not to be dictated to any great degree by genetic variants harbored within the gene. Thus, it is conceivable that the TCF7L2 locus exerts its effect on both insulin production (based on genetic association studies of T2D) and on insulin action (based on isoform studies in key tissues).
Although many genetic associations have been robustly uncovered with GWAS for various complex traits over recent years using proxy single nucleotide polymorphisms, only in a handful of instances has the actual underlying causal variant been determined. This is where the T2D TCF7L2 association has an advantage, as the causal variant is widely thought to be identified i.e., the work with multiple ethnicities has distilled down the association to a single variant, rs7903146, in intron 3. It is still far from clear if rs7903146 impacts alternative splicing directly, with existing studies—including the current one—being underpowered statistically to elucidate this possible connection. What is also sorely lacking is a comparable study in pancreatic islets, but of course obtaining sufficient sample size in this context is highly challenging. Therefore, as it is still far from clear where the primary sites of action are with respect to the observed genetic association, the ideal scenario would be to conduct relatively large-scale parallel studies of isoform distributions in at least adipose tissue, the gut, and pancreatic islets obtained from individuals of both known TCF7L2 genotype and known disease status.
The case for a TCF7L2 role in metabolism only continues to strengthen, with our own chromatin immunoprecipitation sequencing work suggesting the genes bound by TCF7L2 are strongly enriched in pathway categories related to metabolism (9). The case is further strengthened by an emerging metabolic role in cancer; for instance, the key 8q24 locus associated with a number of cancers likely exerts its effect through mutation of an upstream TCF7L2-binding element driving the transcription of the MYC gene (10,11).
Although many more studies are required beyond that of Kaminska et al., TCF7L2 remains a key and intriguing factor in the arena of metabolic traits. Given the important need for effective translational research in T2D to enable potentially more specific and individualized treatments to be used for the disease, it is clear that there is great promise in understanding the mechanism of action of this key gene product.
See accompanying original article, p. .
ACKNOWLEDGMENTS
No potential conflicts of interest relevant to this article were reported.
