Introduction

Type 2 diabetes mellitus is a complex disease that results from the interplay between adverse health behaviour and genetic risk factors [1]. More than 20 validated common type 2 diabetes susceptibility loci have been identified, of which the majority have been shown to associate with pancreatic islet cell dysfunction, insulin resistance or obesity [2–7]. The rs7903146 transcription factor 7-like 2 (TCF7L2) T allele has shown the strongest association with type 2 diabetes to date, with a per allele odds ratio of 1.4 [8]. However, despite major efforts to elucidate the underlying mechanisms whereby this variant increases the risk of type 2 diabetes, its functional impact and its specific role in disease pathogenesis are still unclear.

TCF7L2 spans a ~216 kb region on chromosome 10, and consists of 17 exons [9, 10]. Tissue-specific splicing patterns have been identified; the pancreatic islet and colon display a unique splice isoform containing exons 4 and 15 [10, 11]. Rs7903146 is located between exons 4 and 5, and may influence the splicing pattern of TCF7L2 [10].

TCF7L2 is a transcription factor that is part of the Wnt/β-catenin signalling pathway. This is a ubiquitous signalling pathway that ultimately activates transcriptional programmes. Thus, Wnt signalling may promote mechanisms such as cell proliferation, tissue expansion and differentiation [12].

As a consequence of the central function of TCF7L2, several mechanisms have been suggested to explain the effect of variants in TCF7L2 on the development of type 2 diabetes. One hypothesis is based on the involvement of TCF7L2 in the gut L and K cells in the transcription of GCG, encoding glucagon-like peptide-1 (GLP-1), [13] and of GIP [14], respectively, connecting TCF7L2 to the entero-insular axis and GCG transcription in alpha cells encoding glucagon [15]. Patients with type 2 diabetes have impaired insulin responses following a meal or an oral glucose stimulus and defective incretin secretion has been shown to contribute to this [16]. It has, however, also been suggested that a defect in glucose sensing of the beta cell could impair glucose-mediated insulin secretion, possibly leading to increased glucagon levels [17].

However, previous studies have failed to detect reduced secretion of GLP-1 following an oral glucose load in T allele carriers of the TCF7L2 rs7903146 variant [18–20], and an increased level of glucagon [21].

A second hypothesis is based on in silico analysis suggesting that the rs7903146 T allele of TCF7L2 might act through decreased transcription of PCSK1 and PCSK2 [22]. These genes encode convertases involved in the processing of GLP-1, GLP-2 and gastric inhibitory polypeptide (GIP) in the intestinal L and K cells, and of glucagon in the alpha cells and proinsulin in the beta cells of the pancreas. In support of the latter hypothesis, several studies have shown increased proinsulin levels as well as increased proinsulin/insulin ratios in TCF7L2 risk allele carriers [22–26].

An additional hypothesis is based on the finding that pancreatic islets from type 2 diabetic patients had a lower levels of glucagon-like peptide-1 receptor (GLP-1R) and gastric inhibitory polypeptide receptor (GIP-R) and did not contain detectable amounts of TCF7L2 protein by immunofluorescence staining. Furthermore, downregulation of GLP-1R and GIP-R was seen after treatment with small interfering TCF7L2 in isolated human pancreatic islets [27]. Several studies have observed impaired GLP-1-induced insulin secretion [19, 21] and a diminished incretin effect among rs7903146 T allele carriers [28] without identifying reduced levels of GLP-1 and GIP [18], supporting the existence of a defect in the transcription of incretin receptors leaving the beta cell resistant to incretin hormones.

In contrast, other studies have reported islet-specific upregulation of TCF7L2 [29] and increased levels of TCF7L2 mRNA [28] in type 2 diabetic patients. The latter study also found a positive correlation between the mRNA levels of TCF7L2 and the insulin gene; however, in this study a lower in vitro glucose-stimulated insulin release was found, suggesting that TCF7L2 influences post-transcriptional events of insulin production [28]. Thus, despite the intensive investigation of the actions whereby variants within TCF7L2 increase the risk of type 2 diabetes, the pathogenetic contribution of this locus to increased type 2 diabetes risk is far from clear.

The aim of this study was to examine the impact of the TCF7L2 rs7903146 on incretin secretion, incretin action and insulin secretion following a standardised meal challenge test in white healthy adults with TCF7L2 rs7903146 low-risk CC and high-risk TT genotypes, respectively.

Methods

Study participants

This study was carried out in 62 men recruited from the Inter99 population-based study [30]. Thirty-one were homozygous for the TCF7L2 rs7903146 high-risk T allele and 31 were age- and BMI-matched individuals, homozygous for the low-risk TCF7L2 C allele (Table 1). They were all glucose tolerant following a standard OGTT when examined in the Inter99 study 7 and 2 years prior to participation in the present study. All individuals agreed to participate and gave oral and written consent. The study was approved by the Copenhagen County Ethical Committee and was conducted according to the principles of the Helsinki Declaration.

Table 1 Anthropometric and biochemical description of 62 individuals included according to TCF7L2 rs7903146 CC and TT genotype
Full size table

Genotyping

Genotyping of TCF7L2 rs7903146 was performed in duplicate in all participants as previously described by Helgason and colleagues [8] and by Rung and colleagues [7]. No mismatches were observed.

Standardised meal challenge

The participants consumed a test meal consisting of 50 g white bread, 50 g black bread, 10 g butter, 40 g cheese, 20 g sugar-free jam and 200 ml milk (34% fat, 47% carbohydrate, 19% protein), comprising a total of 2,370 kJ (566 kcal). The meal was consumed within 15 min. Arterialised venous blood was drawn 20, 10 and 0 min before and 15, 30, 45, 60, 75, 90, 105, 120, 135, 150, 180, 210 and 240 min after ingestion of the meal. Blood was sampled into fluoride tubes for plasma glucose analysis and into tubes containing heparin or EDTA (6 pmol/l) plus aprotinin (500 kallikrein inhibitor units/ml blood; Trasylol, Bayer, Leverkusen, Germany) and a specific dipeptidyl peptidase-4 inhibitor (valine-pyrrolidide, 0.01 mmol/l final concentration; Novo Nordisk, Bagsværd, Denmark) for hormone analysis.

Biochemical analyses

Blood samples were drawn after a 12 h overnight fast. Plasma glucose was analysed by a glucose oxidase method (Granutest, Merck, Darmstadt, Germany) with a detection limit of 0.11 mmol/l and intra- and interassay coefficients of variation of <0.8 and <1.4%, respectively. Serum C-peptide was measured using a time-resolved fluoroimmunoassay with the AutoDELFIA C-peptide kit, having a detection limit of 5 pmol/l and intra- and interassay coefficients of variation of <4.7% and <6.4%, respectively. Serum insulin (excluding des[31, 32] and intact proinsulin) was measured using the AutoDELFIA insulin kit (Perkin-Elmer, Wallac, Turku, Finland) with a detection limit of 3 pmol/l and with intra- and interassay coefficients of variation of <3.2% and <4.5%, respectively. Proinsulin was measured using a sandwich ELISA that measures intact proinsulin as well as the four metabolites: split(32-33), des(31,32), split(65-66) and des(64,65)-proinsulin. The detection limit is 0.3 pmol/l and intra- and interassay coefficients of variation are <6.2% and <8.7%, respectively. This assay has no cross-reactivity with insulin, C-peptide, IGF-I or IGF-II. Total plasma COOH-terminal amidated GLP-1 was assayed by radioimmunoassay using polyclonal antiserum 89390 (raised in rabbits). This antibody has an absolute requirement for the amidated C-terminus of GLP-1, and thus does not cross-react with C-terminally truncated metabolites or with glycine-extended forms; it therefore mainly reacts with GLP-1 of intestinal origin. The assay has a detection limit of ~1 pmol/l and an ED50 of 25 pmol/l. Intra- and interassay coefficients of variation are <6% and <15%, respectively [31]. Plasma levels of intact (NH2-terminal) GIP were assayed by radioimmunoassay using polyclonal antiserum 98171 (raised in rabbits) that is NH2-terminally directed and does not recognise NH2-terminally truncated peptides. The detection limit is <5 pmol/l with an ED50 of 48 pmol/l, and intra- and interassay coefficients of variation were <6% and <15%, respectively [32]. Plasma GLP-2 was measured by radioimmunoassay employing antiserum 92160 and standards of human GLP-2 (proglucagon 126–158, Novo Nordisk, Bagsværd, Denmark) and monoiodinated Tyr-12 GLP-2, specific activity >70 MBq/nmol. The antiserum was directed against the N-terminus of GLP-2 and therefore measures only fully processed active GLP-2 of intestinal origin [33]. The plasma glucagon radioimmunoassay was directed against the C-terminus of the glucagon molecule (antibody 4305) and therefore mainly measures glucagon of pancreatic origin [34].

Statistical analyses

A paired t test was applied to test for differences between genotype pairs. For these analyses measures of serum insulin were logarithmically transformed prior to analysis. A linear regression model was applied in order to perform analyses adjusted for fasting plasma glucose, age and BMI. In the linear regression model measures of serum insulin were log-transformed prior to analysis. A p value of less than 0.05 was considered statistically significant.

Statistical power

The sample size determination is based on a two-sample two-sided t test at 5% significance level. A total of 2 × 30 participants will provide a power of 90% power to detect a difference of 280 pmol/kg (about 24%) in AUC for insulin secretion rate (ISR) between the two groups by the use of a paired t test. By using a regression model we have a power of 90% power to detect an allele-specific difference of 55% when estimated by the previously described method for calculating power [5]. This corresponds to a difference of 57 mmol/l × min AUC for glucose, 2,203 pmol/l×min AUC for proinsulin, 66,294 pmol/l×min AUC for C-peptide, 2,799 pmol/l×min AUC for GIP, 1,425 pmol/l×min AUC for GLP-1, 795 pmol/l×min AUC for GLP-2 and 639 ng/l×min AUC for glucagon between the two groups.

Calculations of physiological measures

The AUCs were calculated by the trapezoidal method, and incremental AUCs (IAUCs) were calculated by subtraction of the fasting values. ISRs (pmol kg−1 min−1) were estimated from measured serum C-peptide concentrations using deconvolution [35]. This mathematical operation calculates the secretion rate based on pre-defined C-peptide kinetic parameters from each individual’s weight, height, age, sex and clinical status (glucose tolerance and obesity status) determined in a population-based study [35, 36]. The software ISEC, which implements all these factors [37], was applied for the present study.

Results

The study was performed in a matched design where TCF7L2 rs7903146 CC allele carriers were selected in order to match TT allele carriers for age and BMI. The two groups were similar in relation to measures of age, body composition and HOMA-insulin resistance (IR; Table 1).

Levels of plasma glucose, serum insulin, serum proinsulin, serum C-peptide, plasma glucagon, intact plasma GIP, total plasma GLP-1 and plasma GLP-2 during a meal test were compared between the CC group and the TT group (Fig. 1 and Electronic supplementary material [ESM] Fig. 1). Preprandial fasting measures of plasma glucose were found to be significantly lower among the TT carriers, with inter-pair differences of 0.31 mmol/l (95% CI 0.13–0.50 mmol/l) as well as levels at 15, 30 and 240 min postprandially with inter-pair differences of 0.36 mmol l−1 (95% CI 0.08–0.6), 0.60 mmol/l (95% CI 0.2–1.05) and 0.17 mmol/l (95% CI 0.02–0.3), respectively. In contrast, significantly higher levels of GLP-2 were observed at 20 min preprandial (inter-pair difference 5.6 pmol/l [95% CI 2.3–8.9]) and 10 min preprandial (inter-pair difference 4.8 pmol/l [95% CI 1.22–8.46]) among TT allele carriers. In addition, postprandial plasma GLP-2 at 60 min (inter-pair difference 4.6 pmol/l [95% CI 0.02–9.3]) was significantly elevated among TT carriers. None of the remaining measures of glucose homeostasis was significantly different using a t test.

Fig. 1
figure 1

Measures of plasma glucose (a), serum proinsulin (b), serum C-peptide (c), serum insulin (d), plasma GIP (e), and plasma GLP-1 (f) in 62 carriers of TCF7L2 rs7903146 CC and TT genotypes, respectively, who underwent a standardised meal test. *p < 0.05 and **p < 0.01 obtained from a paired t test; † p < 0.05 and †† p < 0.01 obtained from a linear regression model. Red line, TT genotype carriers; black line, CC genotype carriers

Full size image

Also fasting levels of total plasma cholesterol, plasma triacylglycerols, plasma HDL-cholesterol, plasma LDL-cholesterol and plasma VLDL-cholesterol were investigated; however, we did not find any significant differences in these lipids between the two groups.

As the fasting plasma glucose level among T allele carriers was significantly lower than that in CC allele carriers, we performed linear regression, adjusting for fasting plasma glucose as well as age and BMI. Adjusting for this confounder resulted in the detection of significantly higher preprandial levels of serum proinsulin, serum C-peptide, plasma GIP and plasma GLP-2 among carriers of the risk allele (Fig. 1 and ESM Fig. 1). However, all analyses were also performed including adjustments for HOMA-IR, which resulted in the associations with proinsulin and C-peptide disappearing even though the variant did not significantly affect this estimate of insulin resistance.

Furthermore, postprandial levels of plasma glucose, serum proinsulin, serum C-peptide and plasma GIP were significantly elevated among TT homozygous individuals (Fig. 1 and ESM Fig. 1).

The AUC and IAUC for glucose, insulin, proinsulin, C-peptide, glucagon, GIP, GLP-1 and GLP-2 were calculated (Table 2). TT carriers had a significantly higher IAUC and AUC for glucose, AUC and IAUC for GIP, AUC and IAUC for proinsulin and AUC for C-peptide compared with CC carriers (Table 2).

Table 2 AUC results for 62 carriers of the TCF7L2 rs7903146 CC or TT genotype who underwent a standardised meal test
Full size table

One way of assessing whether incomplete processing of proinsulin could be involved in disease pathogenesis is to examine differences in proinsulin AUC/insulin AUC ratios between the two groups. However, we did not observe any significant difference (Table 2).

ISRs were calculated. Fasting levels of ISR were significantly elevated among carriers of the TT genotype as well as levels at 90 and 105 min postprandially. However, we did not find any genotypic influence on AUC or IAUC for the ISR.

Discussion

Carriers of the rs7903146 TCF7L2 diabetes risk T allele present an increased IAUC for plasma glucose in spite of a significantly lower fasting plasma glucose level among these individuals. Additionally, we found that the AUCs for circulating levels of proinsulin, C-peptide and GIP following a meal challenge were increased in the risk T allele carriers. However, the association with C-peptide was not consistent; thus, the nominally significant association may be a chance finding.

It was unexpected that we observed a decrease in fasting plasma glucose level among TT genotype carriers. This finding may be due to the design of the study, whereby middle-aged glucose-tolerant TT genotype men, who may have compensatory mechanisms that decrease their risk of developing type 2 diabetes, were studied. If individuals with impaired glucose regulation or overt type 2 diabetes had been included in the study, it is possible that a more obvious beta cell secretory defect may have been unmasked. The present study also only included men, and we may have had a different outcome had the study also included women.

Previous studies have reported either no association between variants in TCF7L2 and impaired insulin sensitivity [21, 38–43] or an association with increased insulin sensitivity [19, 44]. Thus, the elevated plasma glucose levels in the postprandial state are not likely to be due to impaired insulin action in the peripheral tissues. Therefore, we hypothesise that the elevated postprandial glycaemia is mediated by an impaired pancreatic beta cell function.

In line with previous studies [19, 22], the linear regression modelling revealed significantly elevated serum proinsulin levels during the meal test. Yet we did not see any significant differences in insulin levels, insulin/proinsulin ratio or proinsulin when we adjusted for HOMA-IR. Therefore, this may suggest that the rs7903146 T allele of TCF7L2 partly acts through decreased transcription of PCSK1 and PCSK2 and thereby via decreased processing of key hormones in glucose metabolism [22]. However, this association may also be a result of increased insulin secretion which, as previously mentioned, may be a consequence of including only glucose-tolerant individuals.

We did not find significantly reduced levels of plasma GLP-1, GLP-2 or GIP during the standardised meal test in carriers of the risk T allele. These findings are also in line with previous studies [20, 21, 28]. In fact, in the fasting state, we observed an increase in the plasma levels of GLP-2 (p = 0.02). GLP-2 is expressed in equimolar ratios with GLP-1 in entero-endocrine cells, yet GLP-2 is not known to possess major insulinotropic effects [45]. Rather, GLP-2 has glucagonotropic effects [45] and has effects on nutrient absorption and intestinal mucosal functionality [46]. Thus, there is no clear relationship between the observed decrease in fasting glucose and the increased GLP-2 in the fasting state.

Increased levels of plasma GIP during the meal test were seen among TCF7L2 TT carriers. This may be an adaptive response, possibly to overcome a reduction in GIP-R expression. Previously, impaired incretin effects among T allele carriers have been demonstrated [19, 28]. Furthermore, reduced insulinotropic effects of both GLP-1 and GIP among individuals with the rs7903146 risk allele based on a hyperglycaemic clamp with co-infusion of GLP-1 and GIP have been demonstrated [21]. Moreover, a recent study among 18 non-diabetic individuals undergoing an OGTT found a reduction in the incretin effect as well as a 50% reduction in beta cell responsiveness to glucose [20]. Thus, the increased level of GIP observed may be a response to reduced beta cell responsiveness among TT allele carriers. Interestingly, it has recently been suggested that GIP may have a more important role in type 2 diabetes pathogenesis than previously recognised, supported by observations that patients undergoing bariatric surgery experience resolution of type 2 diabetes as well as blunted postprandial GIP secretion [47]. Thus, plasma GIP concentrations are not decreased in type 2 diabetes, as observed for GLP-1, but rather, GIP might be hypersecreted [48] while high-fat-fed animals with decreased GIP actions are protected against hyperglycaemia and impaired glucose tolerance [49]. Bearing this in mind, there is increasing evidence to suggest that TCF7L2 T variants partly mediate their effect through reduced production of GLP-1R and the GIP-R in the pancreatic beta cell [27], possibly by alterations in alternative splicing actions [10, 11]. Thus, we suggest that one of the physiological consequences of having the rs7903146 TT genotype is elevated levels of GIP, possibly arising as a result of downregulated beta cell GIP-R.

Based on the established effect of the T allele on the development of type 2 diabetes, it may be surprising that we did not detect a more severe effect of having the TCF7L2 T allele on measures of glucose homeostasis. This may, however, be a consequence of including only 62 individuals. Furthermore, none of the presented data was corrected for multiple testing. Even though some of the tests are not independent, it should be kept in mind that this multiple testing is prone to generating statistical type 1 errors.

In conclusion, our data indicate that multiple defects in glucose homeostasis are present among glucose-tolerant men carrying the high-risk TCF7L2 rs7903146 TT genotype. These individuals have increased levels of plasma glucose in response to a meal challenge despite having lower fasting plasma glucose levels, decreased processing of proinsulin and elevated GIP secretion.