Synthetic G protein‐coupled bile acid receptor agonists and bile acids act via basolateral receptors in ileal and colonic mucosa

The G protein‐coupled bile acid (BA) receptor, GPBA (previously named TGR5), mediates BA gastrointestinal (GI) activities. Our aim was to elucidate the mucosal and motility responses to selective GPBA agonists compared with conjugated BA (eg, taurodeoxycholate, TDCA) in mouse and human colon.

also known as the ileal BA transporter, IBAT), while the organic solute transporter (Ostα/Ostβ) is responsible for BA movement across basolateral membranes, delivering conjugated BAs into the lamina propria.
BAs that remain in the lumen (~5%) can be deconjugated, oxidized, and dehydroxylated by microbial enzymes. The resultant secondary BAs (eg, deoxycholic acid, DCA; and lithocholic acid, LCA) are thought to be absorbed passively across the colonic epithelial lining. 2 Conjugated and deconjugated BAs differentially alter motility, contributing to "ileal brake" that enhances nutrient absorption in the upper GI tract, while in the colon, they increase motility, cause defaecation, 3 and are laxative. The prokinetic effect is probably linked to increased epithelial electrolyte and fluid secretion 4,5 most likely via sensory enteric neurons activated by 5-hydroxytryptamine (5-HT). 3 In contrast, the mucosal effects of selective G protein-coupled BA receptor (GPBA) agonism are not well understood. Abnormalities in BA delivery to the distal bowel are known to result in GI dysfunction. Increased colonic BA delivery, for example, following bariatric surgery or reduced ileal absorption as a consequence of inflammatory bowel disease, is associated with diarrhoea, 6,7 while reduced colonic BA delivery (eg, as a consequence of BA sequestrants to treat lipid disorders) is associated with constipation. 2,8 Recently, BAs have been identified as chemosensory molecules with important roles in regulating lipid, glucose, and energy metabolism. 9 They exhibit affinity for two receptors, the membrane-located GPBA (previously known as TGR5 10 ) and the nuclear farnesoid X receptor (FXR 11 ). Both receptors are expressed by nutrient-sensing enteroendocrine L cells, which predominate in mammalian distal ileum and colon, 12 although GPBAR1 expression appears to be consistent along the length of the murine and human GI tract. 13 L cells in the distal ileum and colon release the satiety hormone, peptide YY (PYY) and the incretin, glucagon-like peptide 1 (GLP-1). Another incretin, glucose-dependent insulinotropic peptide (GIP) is a major product secreted by proximal small intestinal K and L cells in response to nutrient metabolites and BAs. [14][15][16] Synthetic, high-affinity GPBA agonists have been shown to stimulate GLP-1, 17,18 PYY, and neurotensin release from colonic organoids. 19 They also lower blood glucose in rodent models, 20,21 and GPBA remains a target with therapeutic potential for treating metabolic diseases. 9 Unexpectedly, GPBA is trafficked predominantly to basolateral epithelial membranes, 17,21-23 and therefore, BA or drug absorption is a necessary step for efficacy at this receptor. Consequently, lumen-restricted GPBA ligands are likely to be less efficacious than readily absorbed ligands. Systemic BA signaling, for example, to taurodeoxycholate (TDCA), increases L cell cAMP i and Ca 2+ , notably to a greater extent than either selective GPBA agonist (GP-A) or FXR agonist, GW4064. 17 GP-A causes GLP-1 release preferentially via basolateral GPBA, 17 and in preliminary studies, we observed more rapid PYYmediated responses when GP-A was applied to mucosa basolaterally. 22 BA-mediated PYY release has been widely demonstrated in rodent GI tract 14,21,[24][25][26] and human colon following luminal DCA. 23,27,28 The elevation in circulating PYY and GLP-1 following ileal BA absorption indicates that L cell GPBA signaling has antiobesity (PYY-mediated) and type 2 antidiabetic (GLP-1-mediated) potential.
Thus, our initial aim was to establish the similarities and differences in mucosal mechanisms stimulated by a GPBA-selective agonist compared with a BA such as TDCA, focusing on ileal and colonic mucosae from the mouse and including human colonic mucosa. The time dependence of luminal vs basolateral TDCA responses (at a low concentration, 100 µmol/L) was performed using two proven GPBA synthetic agonists (GP-A, 17 and Merck V 18 ), each administered to ileal or colonic preparations with intact submucosal innervation, from wild-type (WT), PYY−/− or GPBA−/− mice. Further aims were to establish (a) whether luminal GPBA responses required ASBT in ileal and colonic mucosae, and (b) to determine whether the GPBA response was glucose-dependent, as is the case for other L cell mechanisms. [29][30][31][32][33] Finally, we investigated the neuronal and non-neuronal mediators of GPBA-specific mucosal signaling and motility in vitro, in WT, PYY−/−, and GPBA−/− colon in order to better understand the paracrine consequences of selective GPBA stimulation.
PYY, exendin-4, and exendin (9-39) were purchased from Generon, and VIP and ω-conotoxin GVIA were from AnaSpec. Merck V was a generous donation to TWS from Merck & Co and is the same agonist utilized by Hauge et al. 18 All other chemicals were purchased from Sigma-Aldrich. BIBO3304, BIIE0246, RS39604, tropisetron, and the GPBA small molecule agonists were each dissolved in neat DMSO at 1 mmol/L. BAs were dissolved in water except for LCA, which was

Key Points
• Bile acids exert multiple effects on gut function at high concentrations. The availability of selective synthetic agonists for the G protein-coupled receptor GPBA (previously known as TGR5) now offers the opportunity to better understand this receptor's signaling capacity specifically.
• Synthetic GPBA agonism involved basolaterally located receptors and endogenous PYY, 5-HT, and cholinergic mucosal mechanisms. Luminal TDCA required ASBT to cross the epithelium and access GPBA. Agonists also slowed natural fecal pellet movement via GLP-1 and NO inhibitory mechanisms.
• GPBA senses BAs present in the lamina propria and enables coincident antisecretory and antimotile paracrine effects. soluble in 95% ethanol. All peptide stocks were in aqueous solution, frozen, and stored at −20°C, undergoing one freeze-thaw cycle only.
The changes in Isc are primarily due to altered electrogenic Cl − secretion as indicated by previous studies using rat 39 and mouse mucosae. 22 each administered apically at 2 mmol/L), based upon data from rat intestine showing significant BA-mediated PYY release. 25 TDCA was chosen as the BA for further investigation in WT and knockout tissues, and its activity was compared with two synthetic GPBA agonists, Merck V and GP-A (utilized previously 17,22 ). Throughout these experiments, we recorded absolute changes in Isc from the point of drug addition.
To study the underlying mechanisms of GPBA agonism, phar-

| Natural fecal pellet transit
The movement of natural fecal pellets along the colon was performed in vitro as described previously. 22,38 Briefly, the entire colon

| Statistical analyses
Responses are expressed as the mean ± 1 SEM from pooled individual preparations taken from different animals. Quoted means were the responses observed at either a given time point for time-course studies or were the peak responses within 5-15 minutes following drug administration. Data from in vitro colonic motility studies were expressed as a percentage of total colon length. GraphPad Prism (version 7.03, GraphPad Software) was used to establish statistical differences, using one-way ANOVA with either Tukey's or Dunnett's post-tests as appropriate for multiple data groups, and Student's t test for single comparisons.

| Basal electrical parameters of mucosae from different murine GI areas and human colon
Basal levels of Isc and TER values in different intestinal tissues are compared in Table S1. WT terminal ileum mucosa exhibited significantly higher Isc values than WT jejunum or distal colon. WT proximal colon Isc levels were higher than those of the distal colon.

| Tissue survey, sidedness, and time courses of Isc responses to Merck V or TDCA in WT and GPBA−/− tissues
An initial survey of responses to the synthetic GPBA agonist Merck V or to TDCA was undertaken using pairs of adjacent preparations from the duodenum, jejunum, terminal ileum, proximal colon, and distal colon. In tissues exposed to vehicle (0.1% DMSO was the vehicle for Merck V), the Isc changed transiently after apical addition (by 0.8 ± 2.0 µA/cm 2 (n = 5) and, by −0.5 ± 0.3 µA/cm 2 (n = 3) after VIP in descending colon), returning to the original Isc level within 10 minutes. Water was the vehicle for bile salts, and it had no effect on Isc ( Figure S1). Apical Merck V induced significant monophasic reductions in Isc (ie, antisecretory responses) that were larger in the terminal ileum, proximal, and distal colon ( Figure 1A).
Concentration dependence was observed for responses to apical and basolateral Merck V, GP-A, and TDCA ( Figure 1G,H).
The synthetic agonists were similarly efficacious on either mucosal surface ( Figure 1G). Basolateral TDCA activity, however, was biphasic, in contrast to monophasic apical TDCA responses, at 0. Biphasic responses were observed with each BA added to WT jejunum and colon, with an immediate increase in Isc (1° response) followed by a slower (2°) decrease in Isc ( Figure S1A,B). The most consistent difference between mucosae from these genotypes was a loss of the reductions in Isc (the 2° response) in PYY−/− tissues. As a consequence, the 1° increases in Isc were amplified to DCA, hyodeoxycholic acid (HDCA), and TDCA in PYY−/− tissues ( Figure S1B).
In addition, DCA, CA, and HDCA decreased basal Isc levels in human colon ( Figure S1C). Since TDCA exhibited significant efficacy at 100 μmol/L in mouse colon, where L cell storage of PYY and enterocyte Y1 receptor signaling is most robust, this concentration of BA was used in subsequent experiments.

| Apical TDCA responses require ASBT activity in mouse distal ileum and colon
ASBT is expressed predominantly in the terminal ileum on apical surfaces, and we used the inhibitor GSK2330672 to test whether apical TDCA or GPBA agonism was dependent on transporter activity. TDCA responses were investigated in mucosae under basal or secretagogue-stimulated (ie, with VIP) conditions. In both situations, GSK2330672 significantly inhibited apical TDCA responses in terminal ileum ( Figure 2A) and distal colon mucosa ( Figure 2B).
GSK2330672 had no effect on basolateral TDCA responses in either region ( Figure 2C,D). This indicates that GPBA is located predominantly on basolateral surfaces and that apical BAs are absorbed rapidly across the epithelium via ASBT, before activating basolateral GPBA. In contrast, responses to apical Merck V or GP-A were unaffected by GSK2330672 ( Figure 2E). In human colon mucosa, from n values as shown and expressed as mean ± 1 SEM. *P < .05, **P < .01, ***P < .001 in C-F (Student's t test); and in H; + P < .05, ++ P < .01, +++ P < .001 for differences between basolateral TDCA responses, compared to apical TDCA in GPBA−/− or WT colon, by ANOVA with Tukey's post-test basolateral TDCA responses were biphasic (as seen in mouse tissue) while apical activity was monophasic, with partial sensitivity to the ASBT inhibitor, although this was not statistically significant (P = .15; Figure 2F). To investigate the non-GPBA mechanisms underpinning basolateral TDCA responses, we pretreated GPBA−/− colon with different Ca 2+ channel antagonists (ie, CdCl 2 , a non-selective blocker; nifedipine, the L-type inhibitor; or the P/Q-type blocker, ω-conotoxin). Nifedipine or CdCl 2 reduced TDCA responses significantly, while ω-conotoxin had no effect ( Figure 3G  . Bars are the mean ± 1 SEM from n values in parentheses. *P < .05, **P < .01, and ***P < .001, using one-way ANOVA with Dunnett's post-test the basolateral TDCA response ( Figure 4C) and the apical BA antisecretory response ( Figure 4D) were each inhibited significantly by apical mannitol, indicating the glucose dependence of these signals.

| Glucose dependence of Merck V and TDCA colonic Isc responses
However, the 1° response to basolateral TDCA was not significantly reduced following removal of apical glucose. As expected, removal of basolateral glucose had no effect on Merck V or TDCA responses, or upon phloridzin's electrogenic effect, as SGLT1 is located on apical surfaces.

| Merck V slows natural fecal pellet progression and requires GLP-1
Fecal pellet movement was unaltered by Merck V in WT colon ( Figure 5A), but it was reduced in PYY−/− colons ( Figure 5B Figure 5D). In contrast with the inhibitory GPBA-mediated effect, TDCA (at 100 µmol/L) was pro-motile in WT colon, significantly so ( Figure 5E). The faster transit rate of PYY−/− colons was unaltered by TDCA (at 100 µmol/L or 1 mmol/L; Figure 5F). Finally, in GPBA+/+ colons Merck V reduced, and TDCA increased transit ( Figure 5G; as seen in the WT cohort; Figure 5A,E) although not statistically significant (P = .07, one-way ANOVA). Control GPBA−/− transit was slightly faster than WT transit (not significantly; Figure 5H), and Merck V or TDCA were inactive ( Figure 5H) in these knockout colons. The only significant difference was between the transit observed for Merck V in WT vs GPBA−/− colon, indicating Merck V's requirement for GPBA expression.

| D ISCUSS I ON
BAs stimulate incretin hormone GLP-1 17,23,28 and GIP secretion. 14,23 Plasma taurine-and glycine-conjugated BAs peak at concentra- The mucosal GPBA antisecretory response observed in WT tissues with Merck V (and another synthetic ligand, GP-A 17,22 ) or, with TDCA added to either mucosal surface, was both found to be PYYmediated, and this was consistent in mouse and human mucosae.
However, basolateral TDCA exerted additional mucosal activities in mouse tissue (discussed below, and Figure 6). Antisecretory responses were observed with 10 µmol/L TDCA, which is in the region of postprandial total plasma BA levels 42 and thus is physiologically relevant. Higher TDCA concentrations (>300 µmol/L) elevated Isc levels in a GPBA-independent manner, so we selected ~ EC 50 concentrations (ie, 100 µmol/L TDCA or 300 nmol/L Merck V) in subsequent investigations. Interestingly, although the potencies of Merck V and GP-A were similar, their efficacies were blunted compared with the less potent TDCA, added to either surface. This difference in efficacy may be linked to the selectivity of the former ligands relative to several co-incident BA activities 3,45,46 some of which may amplify epithelial signaling and/or directly activate the apical cystic fibrosis transmembrane conductance regulator. 47 The latter activity may contribute to the increase in Isc observed after apical application of a high concentration (2 mmol/L) of TDCA or other BAs that was more prominent in PYY−/− compared with WT colonic and jejunal mouse mucosae. It is not clear why monophasic (rather than biphasic) Isc responses were observed in human colon to apical DCA, CA, or HDCA, but it should be noted that only three specimens contributed to this dataset.
Establishing the sidedness of mucosal BA signaling was important. Comparison of Merck V (or GP-A) with TDCA responses added to either surface revealed marked differences in agonist kinetics. Apical Merck V responses were slower to peak than apical TDCA responses, and both activities were slower than basolateral Merck V activity, indicating that GPBAs are preferentially basolateral and the synthetic agonist's passage across the epithelium was slower than that of the BA. The greater efficacy of basolateral Merck V and TDCA antisecretory responses also indicates preferential localization of GPBA on basolateral L cell surfaces, as suggested previously. 17,21,22,48 Basolateral GPBA agonism was not only more rapid, it was also biphasic, particularly for TDCA (>300 µmol/L) and at the highest concentration used (10 µmol/L) of Merck V or GP-A. Furthermore, apical ileal and colonic TDCA responses required transport via ASBT ( Figure 6). We showed that ASBT inhibition (with GSK2336702) attenuated the apical TDCA response but had no effect on Merck V or GP-A activity as these small molecules do not utilize ASBT but can diffuse passively across mucosae. Our findings concur with the basolateral signaling preference of GPBA-induced GLP-1 release from mouse distal ileum and perfused rat intestine. 17  and LCA), we observed consistent endogenous PYY involvement.
The ASBT-insensitive activity of TDCA-induced PYY responses may be similar to those described by Christiansen et al. 28 It is clear, however, that conjugated BAs are absorbed swiftly via ASBT across ileal and colonic mucosae.
Responses to Merck V or GP-A (added to either surface) or apical TDCA were absent from GPBA−/− tissues, indicating a common GPBA-specific antisecretory mechanism in WT tissues.
This response was also absent in PYY−/− colon, and from WT and importantly human colon mucosa, pretreated with PYY-Y 1 and Y 2 antagonists. Thus in WT mouse and human colon, endogenous PYY mediates GPBA antisecretory signaling ( Figure 6). In contrast, the rapid Isc increase after basolateral TDCA administration to GPBA−/− colon was GPBA-independent. This nifedipine-sensitive BA signal appears to utilize L-type Ca 2+ channels, plus a small component of P/Q-type Ca 2+ channel activity. This broader BA pharmacology matches that seen in ileal organoids, where TDCA exerted significant effect on L-type and P/Q-type Ca 2+ channels causing GLP-1 release. 44 Together, these studies show that TDCA-induced Ca 2+ channel signaling is significant in The GPBA-PYY response was glucose-sensitive, irrespective of whether Merck V or TDCA was used and this requirement for glucose complements studies with primary cultures. 17 Interestingly, the initial increases in Isc to basolateral TDCA were not significantly glucose-sensitive, implicating an alternative non-L cell mechanism.
Our mechanistic findings also complement those in diabetic human F I G U R E 6 Basolateral GPBA signaling mechanisms initiated by the selective agonist Merck V and bile acid TDCA, in WT colonic mucosal preparations with intact submucosal innervation. Our data indicate the transepithelial transport of TDCA via apical ASBT (and basolateral Ostα-β into the lamina propria in the descending colon and ileum. TDCA or basolateral administered Merck V each co-activate GPBA present on L cells (causing PYY release), EC cells (causing 5-HT release), submucosal cholinergic neurons (causing ACh release), and L-type Ca 2+ channels (not shown, but present on the majority of epithelial basolateral membranes). Other intracellular and membrane mechanisms are included for pharmacological context, including 5-HT 3 and 5-HT 4 receptor-induced CGRP signaling from submucosal sensory neurons (as also described in 3 ). Local PYY activity (via epithelial Y 1 and neural Y 2 receptors 38 ) together with epithelial muscarinic (M 1 and M 3 ) receptors and VIP-induced epithelial Cl − -secretion via basolateral VPAC and their modulation of apical CFTR channels (that mediate the changes in Isc measured) are included. ASBT and SGLT1 are present on apical epithelial surfaces but in the interests of clarity each transporter is shown on a single epithelial or L cell luminal surface, respectively volunteers who exhibited significantly raised levels of PYY, GLP-1, and insulin, alongside reduced food intake and plasma glucose levels following intrarectal administration of taurocholate. 51 We also observed the antisecretory GPBA response was not amplified by the DPPIV inhibition indicating that endogenous PYY and consequent Y 1 signaling is potentially maximal. The basolateral BA response was also insensitive to GLP-1 receptor antagonism, indicating limited GLP-1 mediation in the mouse distal colon. This may however be due to the GLP-1 response in descending colon being much smaller than that of the proximal colon. 43 The GLP-1 mediation of Merck V's inhibitory effect upon fecal pellet movement was clear and complements studies showing GPBA-mediated GLP-1 secretion, 17,18,28,46,52 the peptide's well described antimotile GI effects and inhibition of gastric emptying. 53,54 In WT and PYY−/− colons, we found Merck V slowed fecal pellet transit, which was reversed by blockers of GLP-1 and NO signaling. Enhanced delivery of BA to the colon as observed following ASBT inhibitor administration can result in diarrhoea, 55 and this pathophysiological prokinetic effect (when luminal BA levels can reach high, mmol/L concentrations) most likely involves 5-HT release from EC cells, 56 and possibly also neurons, 44 CGRP release from intrinsic sensory neurons (both submucosal and myenteric IPANs) resulting in the stimulation of peristalsis, 3 alongside coincident electrolyte secretion (for the mucosal GPBA mechanisms, see Figure 6). 5-HT, CGRP, and cholinergic activity exert robust epithelial anion secretion, the former via 5-HT 4 and CGRP via epithelial CGRP receptors. 43 We established that a combination of 5-HT 4 and cholinergic activities was responsible for the mucosal secretory response to basolateral TDCA administration in the mouse distal colon. Therefore, the EC cell (5-HT 4 -mediated) and L cell (PYY, with GLP-1) mechanisms occur coincidentally upon GPBA stimulation in the colon, an area where the two cell types account for the overwhelming majority of colonic enteroendocrine cells. 12 In conclusion, our results highlight the complexity of acute GPBA signaling within the ileal and colonic mucosa; luminal BAs require ASBT-mediated absorption in both GI regions prior to receptor activation ( Figure 6). GPBA therefore senses predominantly systemic, rather than luminal BAs, alongside which deconjugated BAs are able to diffuse passively across the epithelium (not shown in Figure 6) and thus access basolateral GPBA. GPBA agonism results in L cell, EC cell hormone, and neurotransmitter release, with consequent local antisecretory (PYY-mediated), antimotility (GLP-1-mediated), and even transient pro-secretory (5-HT-mediated) activities. The resolution of these coincident pathways contributes to our understanding of GPBA signaling and cautions against future luminally restricted GPBA agonists (with fewer systemic side effects) that may lack efficacy because they cannot readily reach their basolateral receptor targets when the epithelium is intact. TWS is funded by the Novo Nordisk Foundation, while IRT is supported by the BBSRC (BB/N006763/1 to HC).

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

AUTH O R CO NTR I B UTI O N
HC designed the research with input from TWS. IRT performed the experiments, and all three authors contributed to the writing of the paper. All authors reviewed the final manuscript.