Secretin release after Roux-en-Y Gastric Bypass reveals a population of glucose-sensitive S-cells in distal small intestine

Objective Gastrointestinal hormones contribute to the beneficial effects of Roux-en-Y gastric bypass surgery (RYGB) on glycemic control. Secretin is secreted from duodenal S-cells in response to low luminal pH, but it is unknown whether its secretion is altered after RYGB and if secretin contributes to the post-operative improvement in glycemic control. We hypothesized that secretin secretion increases after RYGB as a result of the diversion of nutrients to more distal parts of the small intestine, and thereby affects islet hormone release. Methods A specific secretin radioimmunoassay was developed, evaluated biochemically, and used to quantify plasma concentrations of secretin in 13 obese individuals before, 1 week after and 3 months after RYGB. Distribution of secretin and its receptor was assessed by RNA-sequencing, mass-spectrometry and in situ hybridization in human and rat tissues. Isolated, perfused rat intestine and pancreas were used to explore the molecular mechanism underlying glucose-induced secretin secretion and to study direct effects of secretin on glucagon, insulin and somatostatin secretion. Secretin was administered alone or in combination with GLP-1 to non-sedated rats to evaluate effects on glucose regulation. Results Plasma postprandial secretin was more than doubled in humans after RYGB (P<0.001). The distal small intestine harbored secretin expressing cells in both rats and humans. Glucose increased secretion of secretin in a sodium-glucose co-transporter dependent manner when administered to the distal part but not into the proximal part of the rat small intestine. Secretin stimulated somatostatin secretion (fold change: 1.59, P<0.05) from the perfused rat pancreas but affected neither insulin (P=0.2) nor glucagon (P=0.97) secretion. When administered to rats in vivo, insulin secretion was attenuated and glucagon secretion increased (P=0.04), while blood glucose peak time was delayed (from 15 min to 45 min) and gastric emptying time prolonged (P=0.004). Conclusion Glucose-sensing secretin cells located in the distal part of the small intestine may contribute to increased plasma concentrations observed after RYGB. The metabolic role of the distal S-cells warrants further studies.

Gastrointestinal hormones, including glucagon-like peptide-1 (GLP-1), are currently used for treatment of metabolic diseases and endogenous GLP-1 contributes to the glucose lowering and weight reducing effect of Roux-en-Y gastric bypass (RYGB) (11,12). Whereas it is well established that GLP-1 secretion is increased after RYGB in humans (13), it is unknown if secretin secretion is increased.
Glucose is a powerful stimulus for GLP-1 secretion (14) and is rapidly absorbed in the upper part of the small intestine, therefore intraluminal concentrations are low in the distal part of the small intestine. We hypothesized that delivery of glucose to the distal small intestine, as seen after the diversion following RYGB, stimulates secretin secretion by luminal-sensing mechanisms. To address this, we measured plasma secretin before and after RYGB. Furthermore, we isolated and perfused the upper or lower half of the rat small intestine and studied the effect of intra-luminal glucose on secretin secretion as well as the molecular sensors involved.
To investigate potential effects of secretin on blood glucose and secretion of islet hormones, we administered secretin subcutaneously to conscious rats and perfused, in separate experiments, the rat pancreas, testing the effects of secretin on glucagon, insulin and somatostatin secretion.

Ethical Approvals
Human studies: Written informed consent was obtained from all study participants, and the study was

Peptides
GLP-1 7-36NH2, rat and human secretin were obtained from Bachem (cat no. 4030663, cat no. 4037181 and cat no. 4031250, Bubendorf, Switzerland). Radioactive labeled rat and human secretin was obtained from Phoenix Pharmaceuticals, Inc (cat no. T-067-06 and T-067-07, CA, USA). Development and evaluation of a secretion radioimmunoassay are described in Supplementary Materials 1.

Mixed meal tests before and after Roux-en-Y Gastric Bypass surgery in obese subjects
Plasma obtained during a standardized liquid mixed-meal test (MMT) before, one week and three months after RYGB from 13 obese subjects (type 2 diabetes; n=4, impaired glucose tolerance; n=3, normal glucose tolerance; n=6) was analyzed. Glucose tolerance was determined by standard OGTT preoperatively. All samples were from a study by Martinussen et al as previously described (15).

Distribution of secretin and GLP-1 along the gastrointestinal tract in humans and rats
Mass-spectrometry based detection was used to assess distribution of secretin and GLP-1 (for comparison) in human intestinal tissue as described previously (16). The secretin and GLP-1 profiles are presented as peak area divided by tissue weight.
Whole wall tissue biopsies (~1 cm) of esophagus, ventricle, duodenum, proximal jejunum, distal ileum and colon were collected from non-fasted rats (anatomical definitions are listed in Supplementary Table   1). Peptides in tissue biopsies were extracted using trifluoroacetic acid (Supplementary Materials 2), and immunohistochemical staining of secretin-positive cells was performed on paraffin embedded tissue samples using anti-secretin (5585-3), a generous gift from Professor Jan Fahrenkrug (Supplementary Materials 3).

Animal experiments
Male Wistar rats (200-250 g) were obtained from Janvier (Le Genest-Saint-Isle, France) and housed two to four rats per cage. Rats were allowed one week of acclimatization and kept on a 12 h light/dark cycle with ad libitum access to water and standard chow.

Isolated perfused rat small intestine and pancreas
Perfusion was performed using a single pass perfusion system (UP100, Hugo Sachs Harvard Apparatus, Germany). The rat small intestine and pancreas were surgically isolated (as described in Supplementary Materials 4). Each protocol started with a baseline period followed by addition of various test substances.

In vivo experiments in rats
Experiments were carried out on 2 occasions on fasted rats (300±13 g) just before their nocturnal feeding period (5:00 PM). Rats were divided into weight-matched groups (n=8/group). At -10 minutes, 200 μL tail blood was collected into pre-chilled EDTA-coated capillary tubes (catalog no. 200 K3E, Microvette; Sarstedt, Nümbrecht, Germany) and instantly transferred onto ice. At -5 minutes, 300 uL test solution was injected subcutaneously. Test solutions were isotonic saline, (negative control) or peptides prepared in isotonic saline: secretin (30 nmol/kg), GLP-1 (30 nmol/kg) or secretin+GLP-1 (both 30 nmol/kg). At 0 minutes, a bolus of D-glucose (2 g/kg) and acetaminophen (100 mg/kg), prepared in isotonic saline, was given orally. Rats from the same cage received different treatments. Blood was collected at times -10, -5, 5, 15, 30, 45, 60 and 90 min. Glucose was measured immediately after blood collection, while the remainder of the samples were instantly transferred onto ice and centrifuged (1,650 g, 4°C, 10 min) within half an hour to obtain plasma. Plasma was transferred to Eppendorf tubes, immediately frozen on dry ice and stored at −20°C until analysis.

Biochemical measurements of perfusion effluents, rat blood and rat plasma
Perfusion effluents: Insulin was measured with antiserum (code no. 2006-3) which cross-reacts with rodent insulin I and II (17). Glucagon was measured with an antibody directed against the C-terminus (code no. 4305) as previously described (18). Total GLP-1 (the sum of 7-36NH2 , 9-36NH2 and potential mid-terminal cleaved fragments) was measured using a C-terminal specific radioimmunoassay targeting amidated forms (code no. 89390) (19). Somatostatin was measured using a side-viewing antibody (code no. 1758-5), detecting all bioactive forms of somatostatin (20,21).

RNA sequencing of human islets
Publicly available RNAseq dataset (from GSE85241, GSE81608 and E-MTAB-5061 (23-25)) were obtained. Average Reads Per Kilobase Million (RPKM) values for secretin receptor (SCTR) and GLP-1 receptor (GLP-1R) were uploaded to the Jupyter Notebook (http://jupyter.org/). Data were then log2 transformed and mean expression levels were calculated in alpha, beta and delta cells, respectively. We excluded individuals with diabetes from these analyses. For further details about the donors, isolation of cells and RNA-sequencing methods please see the original studies (23)(24)(25).
In situ hybridization on rat pancreases is described in Supplementary Materials 5.

Statistical analysis
Clinical samples and in vivo data: Plasma concentrations of hormones were evaluated using area under the curve analysis and statistical testing by one-way ANOVA followed by Holm-Sidak multiple comparisons test. A mixed-effect model was applied in order to test the effect of treatment and time on the variables.

Development of a sensitive secretin assay
We generated titer curves for three different antibodies against human secretin (structure in Figure 1A) using two radioactive iodine-labeled secretin peptides (termed tracer). Based on binding characteristics, the antibody named 5595-3 was selected for further testing and for preparation of calibrator curves (Supplementary Figure 1). By sequencing alignment, we found that human and rat secretin differ at position 14-16 (human: REG vs. rat: QDS) ( Figure 1A) but no species variation was found within the antibody's epitope (position 18-27) ( Figure 1A). For rat samples, we therefore used the same antibody and tracer. For calibrating purposes, we included the rat isoform of secretin as calibrator control.
Recovery of human secretin added to human pooled plasma was calculated to 71±11% (mean±SD) with a lower limit of detection of 1 pmol/L and a dynamic range from 2.5 pmol/L to 80 pmol/L (Supplementary Figure 1) using solvent phase-extraction (70% ethanol). We did not observe crossreactivity (not significantly different compared to background) towards CCK, glucagon, glicentin, GLP-1, human insulin, oxyntomodulin, neurotensin and PYY at concentrations up to 300 pmol/L.

Comparison of Commercial ELISAs
The two commercially available ELISAs had low recoveries of human secretin of 15%±8% (mean±SD) in assay buffer. A Bland-Altman analysis (comparing the concentrations measured using the two ELISAs to the in-house developed RIA) showed acceptable (less than 2SD) degree of variation for concentrations above 20 pmol/L, however, for physiological plasma concentrations of secretin (<20pmol/L) the two commercial ELISAs were inadequate (Supplementary Figure 2).
Having developed a specific and sensitive secretin radioimmunoassay, we next investigated whether postprandial plasma secretin concentrations are elevated after RYGB surgery.

Roux-en-Y Gastric Bypass surgery increases meal-induced secretin release in obese individuals
Fasting plasma concentrations of secretin were not significantly different after comparison with before RYGB ( Figure 1B, P>0.45), whereas postprandial plasma secretin concentrations in response to a liquid meal test were significantly increased three months after compared with before RYGB (P<0.05), reaching a 2-3 fold higher peak value, and total AUCs were approximately doubled (tAUC0-180min: 569±475 vs. 956±360 pmol/L x min) ( Figure 1C). Concentrations were also increased one week after RYGB but to a more modest degree and total AUCs did not differ from pre-operative AUCs (P=0.67).

Distribution of secretin in humans and rats
Using publicly available RNA-seq data from 37 different human tissues (26) expression profiles of secretin showed a maximum in the duodenum, as expected. However, high levels were also found in the more distal parts of the small intestine ( Figure 1D), consistent with previous reports on secretin distribution in other species (27,28). Mass-spectrometry analysis on human gastrointestinal tissue revealed that secretin concentrations were highest in the duodenum as well as in jejunal biopsies ( Figure   1E). Concentrations of GLP-1 7-36NH2 were included as positive controls. GLP-1 concentration was highest in the distal part of the gastrointestinal tract (ileum, colon and rectum) ( Figure 1E) in line with previous reports (29).
To examine if the distribution of secretin in rats was similar to humans, we measured the concentration of extractable secretin from esophagus to colon in rats (Supplementary Table 1). Secretin was not detectable in esophagus and the ventricle ( Figure 1F, n=7) but in the duodenum and in the distal ileum, secretin was found at comparable concentrations ~20-30 pmol/g tissue ( Figure 1F, n=7). Concentrations of GLP-1 increased from duodenum to colon with the highest concentration in the distal ileum (~60 pmol/g tissue) ( Figure 1F, n= 7) in line with a previous report (30). Immunohistochemical stainings of secretin ( Figure 1G+H) showed no staining in esophagus, ventricle and colon and high intensity in proximal jejunum and distal ileum, consistent with the measured extractable concentrations.

Secretin responses from the proximal and distal small intestine using the isolated perfused rat intestine model
Luminal  Figure 2B). Our data therefore suggests that cell depolarization is involved in secretin release from the perfused rat small intestine and furthermore that our perfusion model reflects the physiology of secretin secretion. We therefore decided to evaluate potential differences in HCl and glucose-induced secretin secretion in the proximal and distal small intestine using this model. GLP-1 secretion was measured as this is a well validated control in this experimental setup (31,32).

Differential sensing of glucose by secretin producing cells in the proximal and distal small intestine
Secretion of GLP-1 in response to both glucose and HCl infusion was likewise significantly higher from the distal small intestine compared to the proximal small intestine ( Figure 2E+F), which is consistent with high extractable concentrations of GLP-1 in the distal small intestine ( Figure 1F).

Glucose induced secretin secretion from the distal small intestine is mediated by sodium-glucose co-transporter
Inhibition of SGLT with the competitive SGLT1/2 inhibitor, phloridzin, eliminated glucose-stimulated secretin response from the distal rat small intestine (baseline-subtracted values; Glucose: 1.9±0.7 vs. Glucose + Phloridzin: 0±0.2 pmol/L, P<0.05, n=6) ( Figure 2G+H), indicating that sodium-coupled glucose absorption is involved in the mechanism of glucose stimulated secretin secretion from the distal rat small intestine.

Secretin delays blood glucose peak time and increases plasma glucagon in non-sedated rats
Blood glucose concentrations were similar across groups at baseline (P>0.71). Peak times for blood glucose were prolonged from 15 min (saline group) to 30 min (GLP-1), 45 min (secretin), and 60 min (secretin+GLP-1) ( Figure 3A) but the incremental area under the curve (iAUC0-90min) was not significantly different between groups (P>0.77, Figure 3B). Gastric emptying, reflected by changes in plasma acetaminophen concentrations, was markedly prolonged in secretin treated rats compared to saline and GLP-1 treated groups (P<0.05, Figure 3C+D). GLP-1 had significant effect on plasma acetaminophen at the initial time points compared to saline (P<0.05).
Plasma concentrations of insulin and C-peptide mirrored blood glucose levels ( Figure 3E+G). However, in rats receiving both secretin and GLP-1, insulin and C-peptide peaked 5 minutes after the oral glucose gavage at which time point no change in blood glucose levels were observed ( Figure 3A). When minutes, levels were similar to those in the control group. Glucagon n-AUCs were lower in the group receiving GLP-1 compared to saline group (P=0.08) and to the secretin-treated group (P<0.05) ( Figure   3J).

Pancreatic delta cells express the secretin receptor (Sctr) and exogenous secretin increases somatostatin secretion from the perfused rat pancreas
To evaluate the potential metabolic effects of increased secretin release, we initially assessed tissue specific enrichment of the Sctr. We found that aside from the known expression of secretin receptors in the duodenum, pancreatic tissue contained the secretin receptor transcript ( Figure 4A). However, since these values represent both expression in exocrine acinar/ductular tissue and islets, we next investigated if Sctr was present in human islets. Sctr was expressed in delta-cells and detectable at lower levels in alpha and beta-cells ( Figure 4B). Sctr expression were compared to Glp-1r profiles, which, expectedly, showed high expression in the beta cells and, to a lesser extent, in the delta cells ( Figure 4B). To investigate the translational relevance of rat islets to humans in regard to Sctr expression, we used an in situ hybridization approach. The Sctr was primarily found to be expressed in pancreatic delta-cells and to a lesser extent in the alpha and beta-cells of the rat ( Figure 4C).
To examine whether secretin has an insulinotropic effect on the rat pancreas, we perfused the rat pancreas Glucagon secretion was low, due to the high glucose concentration in the perfusion buffer, and neither secretin nor GLP-1 infusions led to significant changes in its secretion (P>0.18, n=6) ( Figure 4H+I).

Discussion
Here, we demonstrate that postprandial plasma concentrations of secretin are increased after RYGB surgery in obese individuals. This may result from the anatomical rearrangement of the small intestine which diverts luminal nutrients, like glucose, to more distal sites of the small intestine. Increased villus length after RYGB may also contribute the observed changes in plasma secretin concentrations (38).
Tissue levels of secretin have been reported to be up-regulated in a rodent model of RYGB (39) but postoperative plasma concentrations of secretin have to our knowledge not been reported previously.
Furthermore, we show that secretin, besides in the duodenum, is expressed in the distal small intestine in humans and rats, and using a physiologically relevant model: the perfused rat intestine (31), we demonstrate that glucose is a strong stimulus for secretin release from the distal but not from the proximal rat small intestine. Using an inhibitor of SGLT1/2 activity (phloridzin) (34) we found that the molecular mechanism responsible for glucose-induced secretin secretion involves glucose absorption through the SGLT. SGLT-2 inhibition has gained major clinical interest due to its glucose lowering effect and reduced risk of cardiovascular disease (40). One may therefore speculate that patients treated with nonselective SGLT inhibitors would have an altered secretin secretion profile, similar to what has been reported for GLP-1 (41) but this remains to be explored. Previous attempts to show glucose-induced secretin secretion have been negative (7,(42)(43)(44), probably because in these studies glucose was administered to the proximal rather than the distal part of the small intestine.
Glucose-stimulated secretin release from the distal small intestine is probably not of major physiological importance in un-operated humans since glucose mainly is absorbed in the proximal part of the small intestine. However, after intake of large carbohydrate rich meals, some intraluminal glucose may reach more distal parts of the small intestine (45), and secretion may under these circumstances play a role together with GLP-1 in improving glycemic control by inhibition of gastric emptying. This putative effect is supported by our in vivo data and consistent with previous findings (2)(3)(4).
Intraluminal acidification in the distal small intestine resulted in a larger secretory secretin response compared to the proximal intestine. The physiological relevance of this finding is not clear. The acidity of the gastric chyme entering the upper part of the small intestine is rapidly neutralized by a mixture of bile, mucosal -and pancreatic bicarbonate secretion, but the stimulatory effect of low pH in the distal small intestine on secretin-secreting cells may reflect a safety mechanism to reduce gastric emptying.
The fact that also GLP-1 secretion was increased, may point to a more general effect, however, where hydrogen ions formed during digestion and perhaps fermentation in the mucosa micromilieu, stimulate ileal endocrine cells to activate the ileal brake. Further studies are required to investigate this new unexpected observation.
The observations regarding the effects of secretin on pancreatic islets are conflicting (6,7,(46)(47)(48)(49)(50). Our data support that physiological levels of secretin regulate pancreatic secretion of somatostatin which is consistent with Sctr expression in rat and human delta-cells. Although Sctr was also expressed by betacells, secretin did not affect insulin secretion in the perfused rat pancreas whereas in vivo secretin actually led to lowering of plasma insulin levels early after an OGTT. The underlying reason for this awaits further investigation but this may be related to prolonged gastric emptying time and to an increased secretion of somatostatin which through paracrine effects may have overruled potential direct stimulatory effects of secretin on insulin secretion. Glucagon secretion was also not affected but this may be due to the already low levels when the pancreas was perfused in hyperglycemic conditions, thereby restricting the capability to detect further inhibition of alpha cells. However, to our surprise, there was a shortlasting increase in glucagon secretion after secretin injection in vivo, which was uninfluenced by the glucagonostatic effect of GLP-1 (51) and sympathetic stress (as this was not observed in the control group), as observed when similar doses of both peptides were injected. The underlying reason(s) is not clear.

Conclusion
Our study expands the current knowledge regarding secretin physiology and suggest that the RYGBrelated increase in plasma secretin concentrations is mediated by nutrients reaching S-cells located in the distal small intestine. The physiological role of glucose-sensing S-cells warrant further studies but given secretin's potential effects on islet secretion and whole-body metabolism (52) it may be speculated that secretin contribute to the metabolic effects of RYGB.

Conflicts of interest
No conflicts of interest, financial or otherwise, are declared by the authors.

Acknowledgement
The study was supported by a grant from the European Research Council (grant no. 695069) to JJH and by an Excellence Emerging Investigator Grant-Endocrinology and Metabolism (NNF19OC0055001) to NJWA.