Targeting extra-oral bitter taste receptors modulates gastrointestinal motility with effects on satiation

Bitter taste receptors (TAS2Rs) are present in extra-oral tissues, including gut endocrine cells. This study explored the presence and mechanism of action of TAS2R agonists on gut smooth muscle in vitro and investigated functional effects of intra-gastric administration of TAS2R agonists on gastric motility and satiation. TAS2Rs and taste signalling elements were expressed in smooth muscle tissue along the mouse gut and in human gastric smooth muscle cells (hGSMC). Bitter tastants induced concentration and region-dependent contractility changes in mouse intestinal muscle strips. Contractions induced by denatonium benzoate (DB) in gastric fundus were mediated via increases in intracellular Ca2+ release and extracellular Ca2+-influx, partially masked by a hyperpolarizing K+-efflux. Intra-gastric administration of DB in mice induced a TAS2R-dependent delay in gastric emptying. In hGSMC, bitter compounds evoked Ca2+-rises and increased ERK-phosphorylation. Healthy volunteers showed an impaired fundic relaxation in response to nutrient infusion and a decreased nutrient volume tolerance and increased satiation during an oral nutrient challenge test after intra-gastric DB administration. These findings suggest a potential role for intestinal TAS2Rs as therapeutic targets to alter gastrointestinal motility and hence to interfere with hunger signalling.


Results
TAS2Rs and their downstream signalling molecules are expressed in mouse gut smooth muscle. The expression of a selection of different bitter taste receptors and taste signalling elements was investigated in muscle strip preparations of mouse fundus, antrum, duodenum and colon via RT-PCR. No expression was shown for mTAS2R118 (salicin) and mTAS2R138 (phenylthiocarbamide (PTC)), while mRNA for mTAS2R108 (DB), mTAS2R135 (DB) and mTAS2R137 (chloroquine) was demonstrated throughout the gut with mouse tongue epithelium as positive control (Fig. 1a). Candidate TAS2R-coupled G-proteins such as α -gustducin and its corresponding β 3 γ 13 -subunits, α -transducin, Gα i1, Gα i2, Gα i3, were also expressed ( Fig. 1b) [15][16][17] . The expression of phospholipase C β 2 and TRPM5 confirmed the presence of the full canonical bitter taste signalling pathway. Bitter taste receptor agonists induce concentration-dependent changes in smooth muscle contractility.
Effects of DB on smooth muscle contractility. In mouse fundic muscle strips, DB induced a concentration-dependent tonic contraction, which was maximal at 100 μ M and amounted 12 ± 1% of the maximal response to ACh (P < 0.01) (Fig. 1c). At higher concentrations (1 mM), DB relaxed muscle strips strongly, up to 83 ± 17% of the maximal relaxation attained by nitroglycerine (NG) (P < 0.001).
To study the involvement of membrane K + channels in the effect of DB on muscle contractility, fundic muscle strips were pretreated with blockers of different types of K + channels. Blockade of large conductance Ca 2+ -activated K + (BK Ca ) channels by iberiotoxin (IbTx, 100 nM), intermediate conductance Ca 2+ -activated K + (IK Ca ) channels by TRAM-34 (1 μ M) and small conductance Ca 2+ -activated K + (SK Ca ) channels by apamin (500 nM) did not affect the contractile response to DB (Fig. 1d). In contrast, pretreatment with the promiscuous K + channel blocker charybdotoxin (CbTX, 100 nM) resulted, surprisingly, in an increased tonic contraction in response to DB (28 ± 2% of response to ACh) (P < 0.001), without affecting the relaxation at higher concentrations (Fig. 1c,d). Spikes of phasic contractions in fundic muscle strips (Fig. 1c) are a characteristic of CbTX-treatment and do not involve DB treatment.
In the antrum (Fig. 2a) and the colon (Fig. 2b), a similar contractile pattern was observed with DB as in the fundus. In the antrum, DB reached a maximal increase in phasic activity, characteristic for antral smooth muscle, of 37 ± 7% (P < 0.001) and a maximal tonic contraction of 19 ± 3% (P < 0.001) in the colon, both at 50 μ M. Pretreatment with CbTX resulted in a 2-fold and 1.4-fold increase in contractility in the antrum and colon respectively. At higher concentrations, DB inhibited the phasic activity in the antrum (P < 0.01) and the tonic contraction in the colon. Effects of other bitter tastants on smooth muscle contractility. Chloroquine induced a significant increase (34 ± 3%, P < 0.001) in phasic activity of the antrum, followed by an inhibition at higher concentrations (P < 0.001) (Fig. 2c). No significant tonic contractions were observed in fundus or colon, but chloroquine relaxed (90% relative to NG, P < 0.001) fundic muscle at 500 μ M (P < 0.001) ( Fig. S1 and 2d). Phenylthiocarbamide did not induce a contraction in any of the tissues tested but relaxed fundic muscle strips, up to 51 ± 10% at 5 mM (P < 0.05 Supplementary Fig. S1) and completely inhibited phasic activity in the antrum (Fig. 2c, P < 0.001). PTC had no effect in the colon (Fig. 2d). Salicin did not influence contractility in any of the tissues tested (Fig. S1, 2c,d).
Contractions in response to bitter tastants are mediated via TAS2R activation and the canonical taste signalling pathway. The DB-induced contraction was further characterized in fundic muscle strips and was shown to be muscle-specific, since blockade of neurotransmission with tetrodotoxin (3 μ M) did not influence the contraction ( Table 1).

Relaxations in response to higher concentrations of bitter tastants do not involve the canonical taste pathway, but are PKC-dependent.
None of the previously tested blockers influenced the relaxations at higher concentrations. Neither did inhibition of cAMP/cGMP signalling with SQ-22536 (cAMP, 100 μ M) or ODQ (cGMP, 10 μ M). In contrast, the general PKC-inhibitor calphostine C (1 μ M; P < 0.001) inhibited both DB-induced contractions and relaxations.
Activation of the PKC-ERK signalling pathway was demonstrated by western blotting. Stimulation of hGSMC with 5 mM DB resulted in an increased phosphorylation of both Erk1 (P < 0.05) and Erk2 (P = 0.05) (Fig. 5f).

DB inhibits gastric accommodation in healthy volunteers.
All volunteers (n = 12) completed the study without reporting adverse effects. Baseline intra-gastric pressure (IGP) prior to nutrient drink infusion was not different between placebo and DB. The course of the drop in IGP during nutrient infusion differed significantly (treatment: P = 0.01; treatment x time: P = 0.09) between placebo and DB treatment (Fig. 6a). IGP at nadir (placebo: − 4.575 vs DB: − 3.104 mmHg) tended (P = 0.06) to be more elevated during the DB condition but especially the return from the drop in IGP at nadir was faster after DB.

DB increases satiation scores during liquid meal intake in healthy volunteers. After DB
administration subjects (n = 13) felt satiated earlier (64 ± 6 min) compared with placebo treatment (74 ± 8 min; P = 0.01). Under vehicle conditions subjects ingested 998 ± 133 ml, while after DB treatment this was decreased to 869 ± 96 ml (P = 0.03). None of the other symptom ratings differed between DB and placebo (details not shown).

Discussion
This study provided for the first time evidence that functional bitter taste receptors are present on smooth muscle cells of the human and mouse gut. Bitter tastants induced contractions in vitro, while intra-gastric administration of DB resulted in a TAS2R mediated delay in gastric emptying in mice. In humans, DB impaired gastric accommodation in response to intra-gastric nutrient infusion and decreased both time to reach full satiation and ingested nutrient volume in a nutrient drinking test. The presence of bitter taste receptors on both endocrine and smooth muscle cells in the gut suggests that bitter tastants might be used as novel stimuli to alter gastrointestinal motility and hence hunger signalling in man.
The contractile pattern induced by bitter tastants was region specific and agonist selective. DB induced a contraction in all tissues tested, followed by a relaxation at higher concentrations in fundus and antrum. Chloroquine followed largely the same pattern in the antrum, but was less potent in evoking contractions in fundus and colon. PTC only relaxed gut smooth muscle, while salicin had no effect on contractility. The discrepancy between these agonists might be a consequence of a differential bitter taste receptor expression in mouse gut smooth muscle. Both substances that to some extent cause contractions have a potential receptor present, but their ability to induce a response might differ depending on the sensitivity and number of TAS2Rs that are expressed. Alternatively, different TAS2Rs might elicit differential responses upon activation, explaining distinct responses for different bitter agonists. It needs to be emphasized that bitter compounds activate various TAS2Rs in different concentration ranges 18 . For instance in humans, DB can interact with 8 different TAS2Rs, chloroquine with 4 TAS2Rs of which two in common with DB and PTC does only interact with one receptor. In humans 80% of the 25 TAS2R subtypes have been de-orphaned, in mice only 6% of the 33 TAS2R subtypes 19 . The expression of a selection of mTAS2Rs was examined, based on homology to hTAS2Rs activated by the compounds used in the present study 20 . Identifying actual TAS2Rs involved in the effects of bitter tastants on contractility is hampered by a lack of molecular tools. Reliable antibodies for TAS2Rs are unavailable and antagonists are scarce. TAS2R expression was confirmed by RT-PCR and their function studied using probenecid, the only commercially available molecule described to block a subset of TAS2Rs (namely hTAS2R16, − 38 and − 43 but not hTAS2R31) 21 . Probenecid reduced but did not abolish the DB-induced contractions in the mouse fundus suggesting that probenecid does not block all DB-activated TAS2Rs or acts with a different potency.
Experiments with muscle strips from α -gustducin −/− mice indicate that the taste receptor-associated G-protein is involved in the observed effects of bitter tastants on gut contractility. However, since the contractions are not fully abolished, other G-proteins are probably in play as well. Indeed, TAS2Rs have been shown to couple to several G-proteins, including transducin and the inhibitory G-protein-subunits Gα i [15][16][17] , all expressed in gut smooth muscle. The taste receptor-associated cation channel TRPM5 seems, despite its expression in gut smooth muscle, not to be involved in the effects of bitter tastants on contractility.
The contraction induced by bitter tastants involves [Ca 2+ ] i increases and Ca 2+ -sensitization (summarized in Fig. 6b). The increase in [Ca 2+ ] i is mediated by the activation of the canonical taste signalling pathway and consequently depolarization of the cell, leading to further Ca 2+ -influx from the extracellular space. Rises in [Ca 2+ ] i in response to bitter tastants have been described in airway smooth muscle, but are being claimed to be part of the relaxation mechanism 14 on the one hand, or have been considered too low to induce a contraction on the other hand 22 . A recent paper however described Ca 2+ mobilization in the pulmonary artery in response to various bitter tastants, including DB, and a TAS2R-mediated vasoconstriction 23 . We are the first to describe a small contraction in response to bitter tastant-induced Ca 2+ -rises in mouse gut smooth muscle, enhanced by incubation with a promiscuous K + -channel blocker, charybdotoxin. This suggests that a K + -efflux, causing hyperpolarization, is being induced by bitter taste receptor agonists, inhibiting the contraction, possibly by inhibiting L-type Ca 2+ -channel activation 14 . The Ca 2+ -sensitizing effect of bitter tastants is mediated via activation of ROCK, known to lead to inhibition of myosin light chain phosphatase 24 .
Bitter tastants induced a relaxation or inhibition of contractile activity at higher concentrations in several regions of the gastrointestinal tract. The mechanism behind the relaxation induced by DB in the mouse fundus could not be fully elucidated, but appears to be TAS2R-independent. Involvement of BK Ca 2+ -activated K + -channels 14 or a direct effect of the β γ -subunits of the G-protein on L-type Ca 2+ -channels, as described in airway smooth muscle 22 , could not be confirmed. However, a general PKC-inhibitor abolished the relaxations, possibly due to a direct attenuating effect of PKC on L-type Ca 2+ -channels, as reported for the inhibitory effects of bile acids on gut smooth muscle 25 .
Bitter tastants also induced functional effects in vivo. Intra-gastric administration of DB or PTC delayed gastric emptying in mice. Moreover, the effect of DB but not of PTC could be blocked by probenecid. This confirms that the effect of DB is TAS2R mediated, while that of PTC is not and is in agreement with the finding that mice are insensitive to the bitter taste of PTC 26 and is further substantiated by the absence of mRNA expression for mTAS2R138 in the mouse gut. Some bitter agonists (chloroquine, salicin) are absorbed via the lumen and can gain access to the smooth muscle via the bloodstream 27,28 . Although benzoate is readily absorbed it is unclear whether this is also the case for denatonium benzoate 29 . Therefore, it needs to be emphasized that in vivo other modes of action for bitter agonists besides effects on smooth muscle cells are probably in play as well. For instance, DB receptors are expressed in the rat brain 11 and also the release of gut peptides will play a part in the overall in vivo effect of bitter administration on gastric emptying 4 . A delay in gastric emptying correlating with a decrease in food intake in response to bitter tastants has previously been reported, and was shown to be independent of CCK or GLP-1 release 10,30 .
A primary cell culture of human gastric smooth muscle cells was used to determine whether the observed effects of bitter tastants in mouse smooth muscle strips were also representative for human smooth muscle. Bitter taste receptors, homologous to the ones identified in mouse tissue were found to be expressed, as well as α -gustducin and α -transducin. Stimulation of human gastric smooth muscle cells with bitter tastants evoked [Ca 2+ ] i rises and phosphorylation of ERK, and thus suggest the presence of a mechanism analogous to that described in mouse tissue.
Bitter agonists also induced functional effects in healthy volunteers, since intra-gastric administration of DB inhibited the drop in fundic tone in response to nutrient infusion. The functional relevance of this response was confirmed during a satiety drinking test, where intra-gastric administration of DB was associated with decreased nutrient volume tolerance and earlier satiation without inducing adverse symptoms. It still remains to be determined whether the effects described in our pre-clinical models are responsible for the observations in humans.
The actual physiological meaning of the dose-dependent changes in contractility in response to bitter tastants is not fully clear. Impaired accommodation, measured with the barostat or IGP monitoring, is associated with increased satiety and unexplained weight loss in functional dyspepsia 31,32 . Together with a possible inhibition of gastric emptying this may represent a system of highly orchestrated interactions that together with the aversive reactions elicited by bitter in the lingual system prevent toxic bitter compounds from being ingested and absorbed into the circulation. In the colon, an increased contractility, in combination with increased ionic secretions 33 might be effective to expel pathogenic bacteria from the body. Interaction of bacterial quorum sensing molecules with TAS2Rs has already been shown in the airways 34 .
The current observations indicate potential therapeutic applications for bitter tastants in different disorders. The effect of bitter tastants on gastric accommodation or gastric emptying and hence on satiation or satiety might encompass a therapeutic potential in the treatment of obesity. The observed effects on contractility in the lower gut suggest possibilities in the treatment of ileus or constipation with bitter agonists. The regional specificity of a selective bitter agonist might make it possible to target specific regions. However, bitter tastants affect other tissues than mere the smooth muscle, such as the mucosal endocrine cells to affect hormone release 10 , and our observations in vivo are probably a summation of these effects. Long-term studies with bitter agonists in animals of disease might shed light on the therapeutic potential of TAS2Rs in the gut.

Materials and Methods
Animals. Wild type C57BL/6 (Janvier), α -gust −/− (kindly provided by R. Margolskee, Monell Chemical Senses center) and TRPM5 −/− (kindly provided by K. Talavera, KU Leuven) mice were kept in the animal facility in 14-10 light-dark cycle with ad libitum access to chow and water. All experiments were approved by and conducted in accordance to the regulations of the ethical committee on animal experiments of the University of Leuven.
Reverse transcription (RT) PCR. Total RNA was extracted from gut smooth muscle tissue and human gastric smooth muscle cells/lung tissue using the RNeasy fibrous tissue mini kit or RNeasy mini kit (Qiagen), respectively, and treated with the Turbo DNA-free kit (Ambion, Life technologies) to exclude genomic DNA contamination before reverse transcription using Superscript II Reverse Transcriptase (Invitrogen, Life technologies). 5 different samples per tissue were analyzed in triplicate, originating from 5 different mice. The RT-PCR reaction was performed as described previously 35 and PCR-products were analyzed with agarose gel electrophoresis. The primers used are summarized in Supplementary Table S1. Amplicons for the TAS2Rs, α -gustducin and α -transducin in mouse fundic smooth muscle and hGSMC were extracted from the agarose gel with the Qiaquick gel extraction kit (Qiagen) and sequenced (LGC genomics) to confirm primer specificity. Amplicons of TAS2Rs that were not expressed in the target tissues were sequenced from the positive controls, mouse tongue and human lung.
In vitro contractility studies in smooth muscle strips. Mucosal-free smooth muscle strips from fundus of WT, α -gust −/− and TRPM5 −/− and antrum and colon of WT were suspended along their circular axis in a Krebs-filled tissue bath. Tension was measured isometrically in response to increasing concentrations (10 μ M-10 mM) of the bitter agonists DB, chloroquine, salicin or PTC, in the presence and absence of different pharmacological blockers (described in supplementary materials) added 30 min before bitter stimulation 36 . 13 C Octanoic acid breath test for gastric emptying. Gastric emptying was measured as previously described 35 . Mice (n = 12) were gavaged with either DB (60 μ mol/kg) or PTC (30 μ mol/kg) 10 or water, as control, 30 minutes prior to start of the measurement. Probenecid (50 mg/kg) or vehicle was injected intraperitoneally, 15 min before gavage. Control experiments were run during the first 4 weeks until stable gastric emptying times were obtained. Mice were then assigned, using a cross-over design, to the following treatments: vehicle+ water, vehicle+ bitter, probenecid+ water and probenecid+ bitter. Between each treatment a wash-out period of one week was introduced.

Gastric accommodation and satiation drinking test in humans.
Detailed protocols of these experiments can be found in supplementary materials. In short, in a first protocol, healthy volunteers (n = 12, 5 males, mean age: 30.6 ± 2.7; mean BMI: 23.8 ± 1.2) underwent two intra-gastric pressure (IGP) studies, at least one week apart, with administration of saline or DB in a single-blind fashion. After an overnight fast, a high-resolution solid-state manometer system (Manoscan 360, Sierra Scientific Instruments) and nutrient catheter were positioned through the nose in the stomach. Following a stabilization period of at least 15 minutes, either 1 μ mol/kg DB or vehicle was infused through the nutrient catheter, followed by a 5 ml water flush. After another 30 minutes, nutrient drink (Nutridrink, Nutricia; 630 KJ) was infused (60 ml/min) directly in the stomach until maximum satiation. Epigastric sensations (fullness, nausea, bloating, belching, epigastric pain and abdominal cramps) were scored every 5 min using a 100 mm visual analogue scale.
In a second protocol, volunteers received an intra-gastric administration of saline or DB in an identical manner as in the first protocol. Thirty minutes after drug administration, subjects were requested to maintain intake of the liquid meal (Nutridrink, Nutricia) at a fixed rate until full satiation. All procedures were approved by and conducted in accordance to the guidelines of the Ethics Committee of the Leuven University Hospital.

Statistical analysis.
Results are presented as mean ± SEM. Muscle tension in muscle strips in response to bitter agonists (gF/mm 2 ) was calculated as the median tension for each dose (fundus and colon) or the difference of the 99.5 and 2 percentile (antrum). The latter is done to reflect changes in the phasic activity, characteristic for antral muscle strips. The effects of different doses of bitter agonists on muscle tension (% of maximal contraction in response to ACh) were analyzed using one-way ANOVA, followed by planned comparisons (STATISTICA 12; Statsoft). Effects of pharmacological agents on basal muscle tension (% ACh) were analyzed using a paired student's t-test. Effects of pharmacological agents on concentration-dependent contractions were analyzed with repeated measures ANOVA, comparing to representative, non-treated strips from the same animal. Effects of pharmacological agents on maximal relaxation were analyzed comparing maximal relaxation in response to treatment to the maximal relaxation in response to 10 μ M NG with a paired student's t-test. Differences in gastric emptying were analyzed with two-way ANOVA. Effects of bitter tastants on Ca 2+ -release and ERK-phosphorylation in hGSMCs were analyzed with one-way ANOVA and an unpaired student's t-test respectively. Differences in intra-gastric pressures between placebo and DB treatment in healthy volunteers were determined using a repeated measures mixed model analysis (SAS), while effects on maximal satiation time, nutrient volume intake and epigastric symptoms were analyzed with a paired student's t-test. Significance was accepted at the 5% level.