Roux-en-Y gastric bypass (RYGB) is the most effective treatment of both obesity and type 2 diabetes (1, 2). Interestingly, glucose metabolism is improved in patients with type 2 diabetes within days after surgery indicating causal mechanisms other than weight loss (3, 4). Several factors seem to be involved including an accelerated transit of nutrients to the small intestine and augmented absorption, which leads to altered gut hormone secretion and among these a greatly enhanced (up to 10-fold) secretion of glucagon-like peptide-1 (GLP-1) (5, 6). GLP-1 is an insulinotropic and appetite inhibiting hormone, and in studies involving the GLP-1 receptor antagonist, Exendin 9-39, the GLP-1 response has been shown to be important for the reduction in food intake and improved beta-cell function observed after RYGB (7–9).

The mechanisms driving the increased secretion of GLP-1 are not fully clarified, but bile acids have been suggested to play a role. Apart from their role in lipid digestion and absorption, bile acids are signal molecules that stimulate GLP-1 secretion through activation of the G-protein coupled receptor TGR-5 (10), as demonstrated both in vivo (11) and in vitro (12–17). Moreover, activation of the bile-acid sensitive nuclear receptor FXR has also been suggested to modulate GLP-1 secretion (18, 19). After RYGB, increased concentrations of plasma bile acids are positively correlated to the increase in GLP-1 concentrations (20, 21) and improved glucose tolerance (22, 23) months to years after RYGB. These observations led to the hypothesis that bile acids are involved in the metabolic improvements after RYGB. However, there is a dissociation in the timing of the postoperative changes since total plasma bile acid concentrations are unchanged in the early postoperative period (21, 24), whereas postprandial GLP-1 secretion is enhanced immediately after surgery, concomitantly with increased insulin secretion and improved glucose tolerance (25, 26). Thus, the role of bile acids for the exaggerated postprandial GLP-1 secretion and glucose metabolism after RYGB remains unclear.

We have previously demonstrated that administration of the unconjugated primary bile acid chenodeoxycholic acid (CDCA) [a potent TGR5 and FXR agonist (27, 28)] to RYGB operated subjects stimulates both GLP-1, PYY, neurotensin, glucagon, and insulin secretion even in the absence of nutrient ingestion.

The primary aim of the current study was to expand on this observation of a link between bile acids and GLP-1 secretion by examining the effect of CDCA in a different cohort and during meal-ingestion and, more importantly, to investigate the contribution of endogenous bile acids for meal-stimulated GLP-1 secretion. The latter was done by inhibiting bile acid reabsorption through administration of the bile acid sequestrant colesevelam (COL), which efficiently blocks reabsorption by cross-linking of bile acids in the intestinal lumen leading to increased fecal bile acid excretion (29, 30). As acute bile acid-stimulated GLP-1 secretion has been demonstrated to be mediated by activation of basolaterally located TGR5 receptors on the L-cell (27, 31), blocking reabsorption would be expected to reduce GLP-1 secretion. The ability of COL to attenuate bile acid absorption and decrease GLP-1 secretion was tested by administration of exogenous CDCA with and without COL in the absence of nutrients in a sub-study (study 1) before testing the postprandial responses (study 2). In addition, we investigated the modulatory effects of exogenous (CDCA) and endogenous bile acids (by COL administration) on meal-induced plasma concentrations of glucose, including the risk of hypoglycemia, and other glucoregulatory hormones including insulin, glucagon, CCK, neurotensin and FGF-19. Apart from in the N-cell, neurotensin is co-expressed with GLP-1 in some L-cells (32) and its secretion, like GLP-1, depends on intestinal absorption and receptor activation, which might be modulated by CDCA and COL (27, 32).

We investigated the effect of exogenous CDCA and the importance of endogenous bile acids for postprandial plasma GLP-1 concentrations and for plasma glucose, insulin, glucagon, CCK, neurotensin, and FGF-19 concentrations in individuals that had undergone RYGB by acute administration of CDCA and the bile acid sequestrant COL.

First, we administered CDCA alone or together with COL to RYGB operated participants in the absence of nutrients. This confirmed that CDCA elicits a rise in plasma GLP-1, glucagon, neurotensin, cholecystokinin, and FGF-19 concentrations. This was reported previously both in RYGB-operated (38) and in non-operated subjects after either gastric (15–17) or rectal administration (12, 13). The combined administration of CDCA and COL attenuated both plasma TBA, GLP-1, glucagon, neurotensin, C-peptide, and FGF-19 responses as compared with CDCA alone. Since bile acids stimulate GLP-1 secretion via activation of basolaterally located TGR-5 receptors on the L-cell (27, 28), absorption of bile acids across the gut epithelium is required for stimulation of GLP-1 secretion (11). Accordingly, the attenuated bile acid absorption and GLP-1 secretion after combined administration of CDCA and COL in study 1 is consistent with this notion and confirms findings from un-operated subjects (15, 39). Importantly, these results also demonstrate that bile acid sequestrants provide a suitable experimental tool for investigations of the physiological effects of meal-induced bile acid absorption also in RYGB-operated patients. Interestingly, the rise in glucagon and neurotensin concentrations seen during CDCA was also attenuated after administration of COL. The responses of neurotensin to CDCA likely involve similar mechanisms as for GLP-1 (27), whereas stimulation of glucagon secretion by CDCA probably involves other mechanisms. The presence of TGR-5 receptors on pancreatic beta and alpha cells is debated, and the stimulatory effects of glucagon by CDCA may be either indirectly or of intestinal origin (27, 40, 41). The CDCA induced glucagon response could be specific to RYGB given the well documented postprandial hyperglucagonemia due to de novo synthesis in the small intestine (26, 40).

Of note, the magnitude of the GLP-1 response obtained after CDCA in study 1 and our previous study (38) was accompanied by a small albeit significant increase in C-peptide without changes in glucose concentrations. When plasma glucose concentrations are within the fasting range, the potentiating effect of GLP-1 on insulin secretion is limited per se (42, 43), and therefore the insulinotropic effect of CDCA mediated GLP-1 secretion was expected to be small.

After RYGB, the postprandial GLP-1 secretion is many-fold increased and the contribution of bile acids to this response has been debated but not directly examined previously. Therefore, in study 2 we investigated postprandial GLP-1 concentrations after addition of an exogenous bile acid (CDCA) to a meal or after inhibiting endogenous bile acid reabsorption with COL in RYGB operated subjects. We found that CDCA increased both postprandial TBA concentrations and stimulated overall GLP-1 secretion, indicating that the GLP-1 secreting L-cells were not maximally stimulated by the meal ingestion alone. Also, peak glucagon concentrations were higher during CDCA administration. Meanwhile, both plasma glucose and C-peptide responses were slightly decreased, and no beneficial effect on beta-cell function was found despite the increased GLP-1 secretion. This finding is surprising, but may be explained by lower peak concentrations of plasma glucose after CDCA, which might be mediated by effects of CDCA on intestinal nutrient entry or glucose absorption although time to peak of paracetamol concentrations, reflecting rate of nutrient absorption, was only numerically increased after CDCA. Notable, bile acids have been suggested to be involved in the pathogenesis of postprandial hypoglycemia via the bile acids-GLP-1-insulin-axis or indirectly via stimulation of FGF-19. Subjects developing postprandial hypoglycemia have been shown to have higher postprandial bile acids levels coinciding with augmented GLP-1 and insulin responses during a mixed meal (44). However, the present study does not support a role for acutely altered intestinal bile acids concentrations for the development of postprandial hypoglycemia as no differences in nadir or in time to nadir of plasma glucose concentrations were found between the four days with mixed meal and differences in plasma bile acid concentrations.

Administration of COL together with the meal decreased postprandial plasma concentrations of TBA with ~68%, indicative of reduced postprandial bile acid absorption, as intended. We added COL in two dosage regimens based on the recent findings with altered bile acid circulation (33), and the extra dose of COL was added to secure binding of the bile acids within the gut lumen already before food ingestion. However, the results obtained from the two different dosage regimens of COL were comparable including fasting bile acid concentrations, which suggests that insufficient inhibition of absorption is not a concern with the single dose of COL. The large decrease in TBA concentrations after COL administration did not affect GLP-1 secretion, which suggests that endogenous bile acids do not appear to be quantitatively important for the total postprandial GLP-1 response after RYGB despite the fact that exogenous CDCA can increase plasma GLP-1 concentrations after RYGB. This difference might be driven by differences in plasma TBA concentrations, which were substantially (nine-fold) higher after adding CDCA suggesting that the levels of endogenous BAs stimulated by meal intake did not have a similar stimulatory capacity. Moreover, CDCA is known to be a relatively good agonist for TGR-5, which is not the case for all endogenous bile acids (27). In previous studies the composition of the different bile acids was not changed after surgery (24, 25) contradicting any special impact of CDCA or other subfractions for the metabolic improvements after RYGB.

The immediate effects of the acute interventions in our studies were anticipated to affect glucose metabolism and GLP-1 secretion mainly through TGR-5 activation. However, bile acids may also affect these axes via the FXR pathway although the precise contribution remains controversial and is unlikely to contribute to postprandial responses within the first hour after meal intake, since FXR is a nuclear receptor and responses depend on transcriptional changes (45). FXR activation is supposed to stimulate release of FGF-19, suppress de-novo bile acid synthesis, stimulate glycogen synthesis, reduce hepatic gluconeogenesis, inhibit gall bladder emptying, and may also regulate insulin sensitivity (46, 47). In both our studies we observed an increase in FGF-19 concentrations with peak values occurring after the 120 min postprandial timepoint in line with previous studies (25, 48). The observed increase in FGF19 was greatly enhanced by adding CDCA to the meal confirming a link between CDCA and activation of the FXR pathway (49). GLP-1 concentrations peak within the first postprandial hour after RYGB, why any acute effects of bile acids on postprandial GLP-1 and insulin secretion seem independent of FXR activation. This is consistent with findings in some rodent models where activation of the FXR-pathway does not stimulate GLP-1 secretion and may even decrease secretion (18, 27). However, results from other pre-clinical models have opposing findings with increased GLP-1 secretion (50, 51). In addition, targeted knockout of the FXR receptors in mice eliminated the beneficial effects of sleeve gastrectomy on glucose metabolism (52), but this finding can be unrelated to GLP-1 secretion and may also be procedure or species specific. We observed a tendency of decreased FGF-19 concentrations during both doses of COL and GLP-1 secretion was unchanged on these days as mentioned above. In general, the role of FXR activation after bariatric surgery in humans remains unknown and further studies are needed to determine the contribution of FXR activity for the changed glucose metabolism, gut hormone secretion and insulin sensitivity after surgery. With a more chronic treatment with bile acid sequestrants, results might differ with accumulated effects on glucose metabolism and gut hormone secretion by manipulating both TGR5 and FXR activation with more efficient binding of bile acids. Especially, effects from changed FXR activation would be expected to be more pronounced in line with findings in un-operated subjects with type 2 diabetes (53).

The relation between bile acids and CCK in RYGB operated subjects appears to be complex. In unoperated individuals the effect of CDCA (15, 16) and bile acids (54) on CCK secretion was inconsistent, whereas liquid mixed meal tests after adding bile acid sequestrants were associated with increased CCK secretion (53, 55). In the meal study, CCK responses were unaltered by both CDCA and colesevelam. The discrepancy between findings in RYGB-operated and unoperated individuals might involve the apparent dissociation of the normal physiological CCK-mediated regulation of gallbladder emptying and postprandial bile acid concentrations after RYGB, where a larger pool of the bile acids is located within the small intestine in the fasting state, attenuating the importance of postprandial gallbladder emptying (33).

Overall, our findings do not support a major role for endogenous bile acids for the altered postprandial secretions of gastrointestinal hormones or glucose metabolism in RYGB operated subjects. In the present studies, we were unable to identify changes in GLP-1 or measures of beta-cell function after acute attenuation of postprandial bile acid absorption by the bile acid sequestrant COL, although the conclusions based on these finding may be limited by the acute dosage regimen and the inclusion of subjects without type 2 diabetes. Our study also seems to contrast data suggesting that the FXR pathway is crucial for glucose tolerance after bariatric surgery (52), although studies with longer and more chronic administration of sequestrants are probably needed to establish this.

In conclusion, administration of the bile acid sequestrant COL to RYGB operated subjects effectively reduced CDCA-induced circulating bile acids and GLP-1. Further, CDCA administered during meal intake enhanced GLP-1 secretion and decreased both peak glucose and C-peptide concentrations with unaltered beta-cell function. COL added to a mixed meal effectively blocked the absorption of bile acids but neither affected GLP-1, glucose concentrations nor beta cell function, suggesting a limited role for endogenous bile acids in the acute regulation of postprandial glucose metabolism after RYGB.

The original contributions presented in the study are included in the article. Further inquiries can be directed to the corresponding authors.

The studies involving human participants were reviewed and approved by the Ethical Committee of the Capital Region of Denmark. The patients/participants provided their written informed consent to participate in this study.

IJ: Investigation, formal analysis, and writing—original draft. KB-M: Conceptualization, methodology, and writing—review and editing. VK: Resources and writing—review and editing. SV: Resources and writing—review and editing. NW: Resources and writing—review and editing. TC: Resources and writing—review and editing. RK: Resources and writing—review and editing. JR: Resources and writing—review and editing. JH: Conceptualization, methodology, writing—review and editing, and funding acquisition. SM: Conceptualization, methodology, writing—review and editing, and supervision. MS: Conceptualization, methodology, formal analysis, writing—original draft, and visualization. All authors contributed to the article and approved the submitted version.

This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (grant agreement No 695069-BYPASSWITHOUTSURGERY) and from Amager and Hvidovre Hospital Research Foundation.

RK works for Novo Nordisk A/S, but was at the time of his contribution exclusively employed by University of Copenhagen. TC works for Novo Nordisk A/S and own shares in Novo Nordisk A/S and Zealand Pharma A/S. All other authors declare no conflict of interest in relation to the current manuscript.