Disordered Control of Intestinal Sweet Taste Receptor Expression and Glucose Absorption in Type 2 Diabetes

Fourteen healthy subjects and 13 patients with type 2 diabetes were studied in randomized, crossover fashion. The mean duration of known diabetes in the latter group was 5 ± 1 years, HbA_1c was 6.3 ± 0.2% (45 ± 2 mmol/mol), and all were free of significant comorbidities and managed by diet alone. The protocol was approved by the Human Research Ethics Committee of the Royal Adelaide Hospital and conducted in accordance with the Declaration of Helsinki as revised in 2000. Each subject provided written informed consent.

Each subject attended the laboratory at 0830 h after an overnight fast of 12 h for solids and 10 h for liquids. An intravenous cannula was inserted for blood sampling, and subjects consumed a glucose drink (75 g glucose dissolved in water to 300 mL, labeled with 150 mg ¹³C acetate) within 5 min (T = −5 to 0 min). Blood was sampled at T = −5, 30, 60, 120, and 180 min to measure blood glucose by a glucometer (Medisense Precision QID; Abbott Laboratories, Bedford, MA). Breath samples were collected before and every 5 min after oral glucose during the first hour, and every 15 min for a further 2 h to measure ¹³CO₂ concentrations by isotope ratio mass spectrometer (ABCA 2020; Europa Scientific, Crewe, U.K.). The gastric half-emptying time was calculated using the formula of Ghoos et al. (26). Gastrointestinal symptoms were assessed by a standard questionnaire (maximum score, 27), as previously described (27). Autonomic nerve function was assessed in the type 2 diabetes patients using standardized cardiovascular reflex tests, with a score ≥3 (of a maximum of 6) indicating autonomic dysfunction (28).

After the screening visit, each subject was studied twice, separated by at least a week, with female subjects studied exclusively during the follicular phrase of the menstrual cycle to limit variations in gut hormone concentrations (29). Subjects attended the laboratory at 0830 h after an overnight fast, and an insulin/glucose clamp was established to achieve euglycemia (∼5 mmol/L) or hyperglycemia (∼12 mmol/L) (30). A 50-mL intravenous bolus of 25% glucose (Baxter Healthcare, Old Toongabbie, NSW, Australia) was administered on the hyperglycemic day, and 0.9% saline (Baxter Healthcare) on the euglycemic day, over 1 min each, followed by continuous infusion of the same solution starting at 150 mL/h and adjusted according to blood glucose measurements every 5 min on the hyperglycemic day or remaining at 150 mL/h on the euglycemic day. On the euglycemic day, 25% dextrose was infused intravenously if the blood glucose concentration fell below 5 mmol/L. In addition, 100 IU of insulin (Actrapid; Novo Nordisk, Baulkham Hills, NSW, Australia), in 500 mL 4% succinylated gelatin solution (Gelofusine; B. Braun Australia, Bella Vista, NSW, Australia), was infused intravenously at a variable rate to maintain euglycemia. Once blood glucose concentrations were stable for 30 min (12.3 ± 0.2 mmol/L on the hyperglycemic day or 5.2 ± 0.2 mmol/L on the euglycemic day), a small diameter video endoscope (GIF-XP160; Olympus, Tokyo, Japan) was passed via an anesthetized nostril into the second part of the duodenum, from which mucosal biopsy specimens were collected using standard biopsy forceps and placed into RNAlater (Qiagen, Sydney, NSW, Australia) or 4% paraformaldehyde for 2 h. At T = 0 an intraduodenal infusion containing 30 g glucose and 3 g glucose absorption marker 3-O-methyglucose (3-OMG; Sigma-Aldrich, St. Louis, MO) was commenced via the biopsy channel of the endoscope, and continued for 30 min (1 g/min; 4 kcal/min). At T = 10 and T = 30 min, additional biopsy specimens were collected. Blood samples (20 mL) were taken every 10 min over 1 h to determine concentrations of 3-OMG, C-peptide, GLP-1, and GIP.

Plasma total GLP-1 concentrations were measured by radioimmunoassay (GLPIT-36HK; Millipore, Billerica, MA) with sensitivity of 3 pmol/L and intra- and interassay coefficients of variation (CV) of 4.2% and 10.5%. Total plasma GIP was measured by radioimmunoassay as previously reported, with sensitivity of 2 pmol/L and intra- and interassay CV of 6.1% and 15.4%, respectively (31). Plasma C-peptide concentrations were measured by ELISA (10-1136-01; Mercodia, Uppsala, Sweden), with sensitivity of 15 pmol/L and intra- and interassay CV of 3.6% and 3.3%. Serum 3-OMG concentrations were measured by liquid chromatography and mass spectrometry, with sensitivity of 10 pmol/L (32).

RNA was extracted from tissues using an RNeasy Mini kit (Qiagen), following the manufacturer’s instructions, and RNA yield and quality were determined using a NanoDrop (NanoDrop Technologies, Wilmington, DE). Quantitative real-time RT-PCR was then used to determine the absolute expression of sweet taste molecules. Validated human primers for T1R2, α-gustducin, and TRPM5 were used as primer assays (QuantiTect, Qiagen). T1R3 primers were designed using Primer 3.0 software (Applied Biosystems, Foster City, CA) based on target sequences obtained from the National Center for Biotechnology Information nucleotide database (Table 1). Absolute standard curves were generated by including known copy number standards in RT-PCR for each target (Table 2), as described (13). RT-PCR was performed on a Chromo4 (MJ Research, Waltham, MA) real-time instrument attached to a PTC-200 Peltierthermal cycler (MJ Research) using a QuantiTect SYBR Green one-step RT-PCR kit (Qiagen) according to the manufacturer’s specifications, as previously described (13). Each assay was performed in triplicate and included internal no-template and no-RT controls. All replicates were averaged for final mRNA copy number, which was expressed as copies/50 ng of total RNA.

Fixed tissues were cryoprotected (30% sucrose in PBS), embedded in cryomolds, and frozen before sectioning at 6–10 μm (Cryocut 1800; Leica Biosystems, Nussloch, Germany) and thaw mounting onto gelatin-coated slides. Immunoreactivity was detected using rabbit T1R2 primary (H90, 1:400, SC-50305; Santa Cruz Biotechnology, Santa Cruz, CA), goat GLP-1 primary (1:400, SC-7782; Santa Cruz Biotechnology), monoclonal 5-hydroxytryptamine (5-HT; 1:1000, M0758; Dako Australia, Victoria, Australia), and GIP primary antibodies (1:800, AB30679; Abcam). All were visualized using species-specific secondary antibodies conjugated to Alexa Fluor dyes (1:200 in PBS-Tween 20) as previously described (12,13). Antigen retrieval (S1700; Dako) was performed for T1R2 according to the manufacturer’s instructions. Nucleated epithelial cells immunopositive for individual targets were counted per square millimeter of high-power field and averaged over at least 10 intact transverse sections per subject.

The incremental area under the curve (iAUC) for 3-OMG, GLP-1, and GIP concentrations was calculated using the trapezoidal rule (33) and analyzed by one-factor ANOVA using Prism software (version 6.0; GraphPad Software Inc., La Jolla, CA). These variables were also assessed using repeated-measures ANOVA, with treatment and time as factors. Post hoc comparisons, adjusted for multiple comparisons by Holm-Sidak’s correction, were performed if ANOVAs showed significant effects. One-way ANOVA, with Holm-Sidak’s post hoc test, was used to compare differences in duodenal levels of STR transcripts between healthy subjects and type 2 diabetic patients. Relationships between transcript expression and other factors were evaluated by the Pearson correlation coefficient (r). We calculated that 12 subjects had 80% power to detect a one-third difference in duodenal T1R2 expression in paired studies (α = 0.05), compared with control (13). P values ≤ 0.05 were considered statistically significant. Data are expressed as mean ± SEM.

All subjects tolerated the study well. The patients with type 2 diabetes were older than the healthy subjects, but gastrointestinal symptom scores, BMI, and gastric emptying of glucose did not differ (Table 3). Five type 2 diabetic patients had autonomic dysfunction, but none had evidence of peripheral neuropathy, nephropathy, retinopathy, or macrovascular complications. As expected, blood glucose concentrations were higher in type 2 diabetic patients during fasting and after oral glucose (P < 0.05; Fig. 1A).

Transcripts for T1R2, T1R3, α-gustducin, and TRPM5 were readily detected in duodenal biopsy specimens by quantitative RT-PCR. TRPM5 was the most abundant STR transcript in all subjects, with lower levels of α-gustducin and much lower levels of T1R2 and T1R3; T1R2 was the least expressed transcript (Fig. 2A). TRPM5 transcript levels in healthy subjects during euglycemia were 34 ± 8-fold higher than those of T1R2 (P < 0.001), whereas α-gustducin levels were 22 ± 7-fold higher (P < 0.05) and T1R3 levels were 12 ± 5-fold higher.

Fasting expression of STR transcripts was unaffected by the glycemic state in health or type 2 diabetes and did not differ between the groups (Fig. 2B–E).

Owing to intersubject variability in STR expression, responses to luminal glucose were evaluated as changes from baseline. During euglycemia, T1R2 transcript levels increased in response to duodenal glucose infusion in health and in type 2 diabetes after 30 min (+5.9 × 10⁴ and +5.8 × 10⁴ copies; Fig. 3A). During hyperglycemia, T1R2 transcript levels decreased in healthy subjects after 30 min (−1.4 × 10⁴ copies) but increased in type 2 diabetic patients (+6.9 × 10⁵ copies), so that levels in health were lower at 30 min during hyperglycemia than euglycemia and lower than in type 2 diabetic patients during either glycemic state (subject × time interactions P < 0.01 for each). Levels of T1R3, α-gustducin, and TRPM5 transcript, in contrast, did not significantly change in response to luminal glucose under either glycemic condition (Fig. 3B–D).

Fasting plasma GLP-1 concentrations did not differ between health and type 2 diabetes and were not acutely affected by the glycemic state. Plasma GLP-1 increased in response to duodenal glucose infusion in all groups (P < 0.001; Fig. 1C), with higher concentrations evident in type 2 diabetic patients at 40 min irrespective of glycemic status (subject × time interactions P < 0.01) and at 50 min during euglycemia compared with healthy subjects (subject × time interactions P < 0.05). The iAUC for GLP-1 was higher in type 2 diabetic patients during euglycemia and hyperglycemia compared with healthy subjects (P < 0.05 each; Table 4).

Fasting plasma GIP concentrations did not differ between healthy subjects and type 2 diabetic patients and were not acutely affected by the glycemic state. Plasma GIP increased in response to duodenal glucose infusion in both groups (P < 0.001; Fig. 1D), with higher GIP concentrations evident in type 2 diabetic patients at 40 min irrespective of glycemic status and higher concentrations during euglycemia at 20 and 50 min compared with healthy subjects (subject × time interaction P < 0.05). The iAUC was higher in type 2 diabetic patients during euglycemia compared with healthy subjects (P < 0.05; Table 4).

Fasting C-peptide concentrations were higher during hyperglycemia than during euglycemia in healthy subjects (P < 0.001; Fig. 1E) but not in type 2 diabetic patients. C-peptide concentrations increased in response to duodenal glucose infusion in both groups during hyperglycemia (subject × time interaction P < 0.05; iAUC P < 0.001) but not during euglycemia (Fig. 1E and Table 4). C-peptide concentrations during hyperglycemia were higher in healthy subjects than in type 2 diabetic patients throughout the glucose infusion (subject × time interaction P < 0.05; iAUC P < 0.001).

Serum 3-OMG concentrations increased over time in all groups but were higher at 60 min in type 2 diabetic patients during hyperglycemia than in any other group (subject × time interaction P < 0.001; Fig. 1F). The iAUC for 3-OMG was higher in type 2 diabetic patients and in healthy subjects during hyperglycemia than during euglycemia (P < 0.05; Table 4).

Immunolabeling for T1R2 was evident in single cells dispersed throughout the mucosal epithelium in healthy subjects and type 2 diabetic patients (Fig. 4). Immunopositive cells showed a homogenous distribution of the label throughout the cytoplasm, were largely open or “flask-shaped,” and were found with equal frequency within villi or crypts. In dual-labeling experiments in healthy subjects, 19 ± 11% of T1R2-labeled duodenal cells coexpressed GLP-1, whereas 13 ± 8% of L cells coexpressed T1R2 (Fig. 4A). In a similar manner, 15 ± 10% of T1R2-labeled duodenal cells coexpressed GIP, whereas 12 ± 8% of K cells coexpressed T1R2 (Fig. 4B). Separate populations of T1R2-labeled cells coexpressed 5-HT (31 ± 6%), whereas 5 ± 1% of enterochromaffin (EC) cells coexpressed T1R2 in healthy subjects (Fig. 4C). During fasting, an equivalent number of T1R2 immunopositive cells were evident in healthy subjects and in type 2 diabetic patients, under euglycemia or hyperglycemia, and the number did not change during the duodenal glucose infusion. Similarly, the proportion of cells immunopositive for GLP-1, GIP, and 5-HT did not differ between healthy subjects and type 2 diabetic patients or with glycemic state or exposure to luminal glucose, although a trend for increased L cells in fasting type 2 diabetic patients was evident (data not shown; P = 0.07).

Absolute copy numbers of STR transcripts during fasting and after the 30-min glucose infusion did not correlate with age, sex, BMI, symptom score, or gastric half-emptying time in either group, and in type 2 diabetic patients, they were not related to duration of diabetes, HbA_1c, autonomic dysfunction, or symptom score. In contrast, the change in T1R2 transcript level after luminal glucose exposure correlated with the iAUC for 3-OMG in healthy subjects during euglycemia (r = 0.73, P < 0.05), and the change in TRPM5 transcript level with plasma GLP-1 concentrations at 30 min (r = 0.62, P < 0.05) in the same group. Changes in T1R2 (r = 0.78, P < 0.01) and T1R3 transcript levels (r = 0.59, P < 0.05) in type 2 diabetic patients during hyperglycemia also correlated with plasma GIP concentrations at 30 min, and the change in T1R2 correlated with the iAUC for GIP (r = 0.69, P < 0.05).

This study is the first to define changes in expression of intestinal STR transcripts in healthy humans and patients with type 2 diabetes in response to acute changes in systemic and luminal glucose. We have shown that absolute levels of STR transcripts are unaffected by acute variations in glycemia during fasting in either group but that T1R2 expression increases upon exposure to luminal glucose during euglycemia. In contrast, T1R2 expression decreases markedly in response to luminal glucose during hyperglycemia in health but increases under the same conditions in type 2 diabetes. Type 2 diabetic patients also exhibit increased glucose absorption during acute hyperglycemia compared with healthy subjects, suggesting that dysregulated expression of intestinal STRs can perpetuate postprandial hyperglycemia in this group.

We confirmed our previous observation that fasting STR transcript levels are similar in health and in type 2 diabetes irrespective of age, sex, or BMI (13). Although we previously observed that levels of STR transcript were inversely related to fasting blood glucose concentrations in unselected type 2 diabetic patients presenting for endoscopy, we have now established unequivocally that acute changes in glycemia do not influence fasting intestinal STR expression in health or in “well-controlled” type 2 diabetes. The apparent discrepancy in these observations may reflect the effects of more longstanding hyperglycemia or differences in the duration of fasting in the earlier cross-sectional study. We have now shown that the intestinal STR system is, in contrast, highly responsive to the presence of luminal glucose, with rapid and reciprocal regulation of T1R2 transcripts in health, depending on the prevailing blood glucose concentration. Comparable changes were evident in T1R3 and TRPM5 transcript levels, although these were not statistically significant. Increased intersubject variability seen for T1R3 and TRPM5 transcript levels may be due to their expression in additional populations of intestinal cells tuned to detect other taste modalities and, therefore, unresponsive to luminal and/or systemic glucose.

Healthy subjects who displayed the largest glucose-induced increase in duodenal T1R2 transcript levels during euglycemia had the highest plasma concentrations of the glucose absorption marker 3-OMG. Because SGLT-1 is responsible for the active transport of luminal 3-OMG, our findings support a role of intestinal T1R2 signals in the regulation of glucose absorption via SGLT-1. Indeed, intestinal STR activation has been shown to upregulate SGLT-1 transcript, apical protein, and function in a number of species (17,18). Accordingly, reciprocal regulation of T1R2 in human health may increase SGLT-1 function at euglycemia to facilitate glucose absorption and reduce SGLT-1 function during hyperglycemia to limit postprandial glycemic excursion. However, despite a reduction in T1R2 transcript after luminal glucose exposure during hyperglycemia, our healthy subjects still displayed greater rates of glucose absorption than during euglycemia, which might be accounted for by changes in SGLT-1 lagging behind those in T1R2. Our finding that plasma 3-OMG concentrations were elevated in type 2 diabetic patients during hyperglycemia is in keeping with the concept that SGLT-1 transporter capacity was maintained, or increased, in the presence of luminal glucose under these conditions. In fact, even small changes in SGLT-1 may increase this risk, because type 2 diabetic patients are reported to have up to fourfold higher levels of transcript, protein, and function of this transporter at baseline compared with healthy control subjects (24). We note that an increased level of facilitated glucose transport via the basolateral glucose transporter GLUT2 may have contributed to plasma levels of 3-OMG in the current study; however, the role of STR signals to direct the apical insertion of GLUT2 in enterocytes appears to be limited to rodents (25,34).

The link between STR stimulation and incretin hormone release in healthy humans is not clear. Most in vivo studies indicate that acute administration of nonnutritive sweeteners does not trigger incretin secretion in humans or rodents (35–37). Nonetheless, we observed that subsets of duodenal L cells, K cells, and EC cells were immunopositive for T1R2, in accord with previous reports (4,12,14). Together with positive associations between luminal glucose-induced changes in some STR transcripts and measures of GLP-1 and GIP secretion in the current study, it remains possible that STRs do have a regulatory role in gut hormone release. The inhibition of glucose-induced GLP-1 secretion in healthy humans by the STR blocker, lactisole (15), supports this concept. We also recognize that STR signals may serve autocrine and/or paracrine functions within the intestinal mucosa that are not reflected in circulating gut hormone concentrations; the latter appear to be a blunt marker for local concentrations of GLP-1 (38). There is also a large body of evidence indicating that the intestinotrophic gut peptide, glucagon-like peptide 2 (GLP-2), coreleased from L cells with GLP-1, is a powerful local stimulus to increase intestinal glucose transport via SGLT-1 and GLUT2 in rodents and in patients with short bowel syndrome (39–41). Importantly, GLP-2 release has recently been revealed as STR-dependent in animals and in a human enteroendocrine cell line (42,43), highlighting an important link between STRs and GLP-2 in the regulation of intestinal glucose transport.

Reports concerning postprandial incretin hormone release in patients with type 2 diabetes have been inconsistent, with plasma GLP-1 concentrations after a mixed meal being either reduced (22) or intact (44), although such studies are potentially confounded by failure to control for differences in the rate of gastric emptying, which is frequently delayed in longstanding diabetes or during acute hyperglycemia (20). Our observation that GLP-1 and GIP responses to a standardized rate of duodenal glucose infusion were maintained, and indeed increased, in type 2 diabetic patients, supports our previous findings (45) and is in keeping with the trend for increased L-cell density in these patients in the current study and a report of an increased density of L cells and mixed L/K cells in the duodenum of well-controlled type 2 diabetic patients (46). There is now strong evidence that SGLT-1 transport is a key stimulus for release of GLP-1 and GIP, which occurs even after exposure to nonmetabolized SGLT-1 substrates and is inhibited by pharmacological blockade or genetic ablation of SGLT-1 in rodents (47–49). Therefore, increased SGLT-1 capacity could explain enhanced glucose-induced GLP-1 and GIP responses in our type 2 diabetic patients. Any deficiency in the incretin effect in type 2 diabetes is likely to be explained by impaired β-cell function rather than by deficient incretin hormone secretion (45,50), and indeed, defective C-peptide responses in our type 2 diabetic patients during hyperglycemia support this assertion. Acute hyperglycemia had no effect on GLP-1 or GIP secretion, as noted previously (51,52). Although SGLT-1 transport appears a major determinant of GLP-1 and GIP release, other transporters (49,53) or signaling pathways (54) may also be involved, so increased glucose absorptive capacity during hyperglycemia may not necessarily result in enhanced GLP-1 or GIP concentrations.

Our study had a number of limitations. Transcriptional regulation of intestinal T1R2 occurred rapidly in humans, but we did not quantify changes in STR protein in parallel due to ethical considerations on the additional biopsy specimens required. However, similarly rapid changes in these proteins after glucose or sucralose exposure are known to occur in apical membrane vesicles of rat jejunum (25). We did not assess effects on SGLT-1 transcript or protein here, although measures of glucose absorption with 3-OMG reflect, in large part, SGLT-1 function as the primary intestinal glucose transporter in humans. There was considerable interindividual variability in baseline expression of intestinal STR transcripts, so that our study was insufficiently powered to detect relationships between absolute transcript levels and concentrations of gut hormones and 3-OMG. Our 3-OMG measurements were limited to 60 min, and differences between groups or glycemic states might have become more marked after this point. The duodenal glucose infusion was also relatively brief, being limited by the tolerability of unsedated endoscopy. Our type 2 diabetic patients had relatively good glycemic control, and more marked differences from health might be observed in patients with a higher HbA_1c. The type 2 diabetic patients were older than the healthy control subjects, although we have not previously shown any age-related differences in postprandial GLP-1 responses (55).

In conclusion, we have shown that the intestinal STR system is reciprocally regulated in the presence of luminal glucose according to glycemic status in health but not in type 2 diabetes. In the latter, T1R2 dysregulation potentially increases the risk of postprandial hyperglycemia, but the intestinal STR system appears unlikely to be a major determinant of circulating GLP-1 or GIP concentrations in humans.

This study was supported by grants awarded to R.L.Y. by the National Health and Medical Research Council (NHMRC) of Australia (grant no. 627127) and to R.L.Y., M.H., and C.K.R. from Diabetes Australia. T.W. was supported by an NHMRC Overseas Clinical Postdoctoral Training Fellowship (grant no. 519349)

No potential conflicts of interest relevant to this article were reported.

R.L.Y. and C.K.R. conceived, designed, and supervised the study, obtained funding, acquired data, undertook statistical analyses and interpreted data, and drafted and critically reviewed the manuscript. B.C. and N.J.I. acquired data and provided technical support. J.M., J.K., and T.W. assisted in study design, acquired data, and critically reviewed the manuscript. M.H. designed the study, interpreted data, and critically reviewed the manuscript. R.L.Y. and C.K.R. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank the Endoscopy Unit staff, Department of Gastroenterology and Hepatology, Royal Adelaide Hospital, for their assistance with the study; Dr. Kate Sutherland, University of Sydney, for contributing to the molecular assays in this study; and Kylie Lange, National Health and NHMRC Centre of Research Excellence in Translating Nutritional Science to Good Health, University of Adelaide, for professional biostatistical support.

Preliminary accounts of this study were presented at the Digestive Disease Week meeting of the American Gastroenterological Association, Chicago, IL, 7–11 May 2011; and at the Joint International Neurogastroenterology & Motility Meeting, Bologna, Italy, 6–8 September 2012.