Ghrelin inhibits autonomic response to gastric distension in rats by acting on vagal pathway

Ghrelin is the only orexigenic peptide currently known and a potent prokinetic by promoting gastric motility but novel insights suggest that its role extends beyond satiety regulation. Whereas ghrelin was shown to provide somatic and colonic antinociception, its impact on gastric sensitivity is unknown even though stomach is a major ghrelin secreting tissue. Autonomic response to gastric mechanosensitivity was estimated by measuring blood pressure variation as a surrogate marker in response to gastric distension (GD) before and after ghrelin (or vehicle) administration. Involvement of spinal and vagal pathways in the ghrelin effect was studied by performing celiac ganglionectomy and subdiaphragmatic vagotomy respectively and by evaluating the expression of phosphorylated extracellular-regulated kinase 1/2 (p-ERK1/2) in dorsal root and nodose ganglia. Finally the phenotype of Ghrelin receptor expressing neurons within the nodose ganglia was determined by in situ hybridization and immunofluorescence. Ghrelin reduced blood pressure variation in response to GD except in vagotomized rats. Phosphorylated-ERK1/2 levels indicated that ghrelin reduced neuronal activation induced by GD in nodose ganglion. The effect of ghrelin on gastric mechanosensitivity was abolished by pre-treatment with antagonist [D-Lys3]-GHRP-6 (0.3 mg/kg i.v.). Immunofluorescence staining highlights the colocalization of Ghrelin receptor with ASIC3 and TRPV1 within gastric neurons of nodose ganglion. Ghrelin administration reduced autonomic response to gastric distension. This effect likely involved the Ghrelin receptor and vagal pathways.

Ghrelin is a 28-amino acid peptide produced by endocrine cells of the oxyntic gastric mucosa and, to a lesser extent, by hypothalamus, pituitary and the intestine 1,2 . This hormone exerts its action by binding the G protein-coupled ghrelin receptor expressed both in central nervous system and peripheral organs 2 . Acylation of the 3 rd serine confers to ghrelin an orexigenic activity that induces a subsequent increase in food intake following either central or peripheral administration 2 . Moreover ghrelin displays prokinetic properties by promoting propulsive activity of the stomach 2,3 . Taken into consideration the orexigenic and prokinetic actions of ghrelin, analogs have been recently tested to treat gastric-dysmotility disorders like in diabetic gastroparesis 4,5 .
For instance; the Ghrelin receptor agonist, relamorelin, has recently been studied in a phase 2 clinical trial enrolling diabetic gastroparesis patients 6 . Twelve weeks administration of 10, 30 or 100 µg of relamorelin resulted in a reduction of diabetic gastroparesis cardinal symptoms. However, symptomatic efficacy was not correlated with acceleration of gastric emptying induced by relamorelin. Concomitantly, several prokinetics failed to improve gastroparesis symptoms despite gastric emptying acceleration [7][8][9] . Those data suggest that gastric emptying acceleration is not a necessary condition for overall symptom improvement and the symptomatic efficacy of Ghrelin receptor agonist may imply additional mechanisms. Interestingly, in the aforementioned study, relamorelin reduced epigastric pain 6 . This may suggest that ghrelin pathway may be involved in visceral nociception. To date, although ghrelin has been shown to reduce somatic [10][11][12][13] and colonic 14 pain, its effect on gastric nociception remains uninvestigated whereas the stomach is a major ghrelin-producing site. Owing to the potent effect of ghrelin on gastric motility 3 , studying its impact on gastric mechanosensitivity in response to isobaric distension is challenging. This may explain why such effect has not been yet investigated. Therefore the aim of this study was

Results
Ghrelin decreases the variation of Bp in response to GD through activation of Ghrelin receptors. We first validated the influence of atropine in the anesthetized rat model on the variation of BP in response to GD. The variation of BP in response to graded GD was similar before and after i.v. atropine injection (1 mg/kg) followed by i.v. saline injection. (Fig. 1A). Using this model, we investigated the effect of acylated ghrelin on the variation of BP in response to GD. A decrease in the variation of BP was observed after i.v. ghrelin injection (30 µg/kg) compared to the saline injected group. This decrease yielded 34% at 40 mmHg (p < 0.01) and 60 mmHg of distension (p < 0.01) and 33% at 80 mmHg of distension (p < 0.01; Fig. 1B). By contrast, Figure 1. Effect of acylated ghrelin on the variation of blood pressure (BP). (A,B) Variation of BP in response to graded gastric distension (GD) of 20, 40, 60, 80 mmHg. A first set of GD (baseline) was followed by intravenous (iv) injection of vehicle (saline; n = 10, A) or acylated ghrelin at 30 µg/kg (n = 10, B), then a second set of GD was performed 5 minutes after injection. (C) Ten minutes before ghrelin administration the Ghrelin receptor antagonist [D-Lys 3 ]-GHRP-6 was intravenously administered (n = 8). Data are expressed as mean ± SEM. **p < 0.01. pretreatment with [D-Lys 3 ]-GHRP-6 (0.3 mg/kg i.v.) abolished the effect of ghrelin on the variation of blood pressure in response to gastric distension (Fig. 1C). This effect appears to be specific to gastric stimulation and do not involve an alteration of vascular function since mean arterial pressure (MAP) monitored before the 2 distension sets and before and after ghrelin (or vehicle) intravenous administration did not differ between groups (Supp. Fig. 1).
Effect of ghrelin on gastric mechanosensitivity involves preferentially vagal pathway. In this experiment we performed subdiaphragmatic vagotomy or celiac ganglionectomy in order to disrupt vagal and splanchnic innervation of the stomach respectively. The effect of ghrelin (30 µg/kg i.v.) was completely abolished in vagotomized rats ( Fig. 2A) while it was preserved in rats with splanchnectomy (Reduction of 47% at 40 mmHg (p < 0.05), 42% at 60 mmHg (p < 0.05) and 48% at 80 mmHg (p < 0.01); Fig. 2B). It is of note that both vagotomy and splanchnetomy minimized the variation in blood pressure in response to gastric distension compared to naïve animals used in the first experiment (Supp. Fig. 2). This observation validates the effective surgical denervation of the stomach on gastric sensitivity. The effect of ghrelin in vagotomised rats unlikely involves an alteration in vascular function since mean arterial pressure measured before the 2 distension sets and before and after ghrelin administration is similar between groups (Supp. Fig. 1).
Effect of the different treatments on gastric compliance. In order to check the effect of ghrelin and [D-Lys 3 ]-GHRP-6 on the intensity of gastric stimulation we compared gastric compliance between the 2 nd and the 1 st distension sets in saline, ghrelin and ghrelin + [D-Lys 3 ]-GHRP-6 treated groups. Although ghrelin treatment significantly increased balloon volume during the 2 nd distension set for the 20 mmHg distending pressure compared to other groups, changes in air volume were similar for the others distending pressures suggesting that the intensity of gastric stimulation is not affected by the treatments (Fig. 3A). Gastric compliance was also compared between vagotomised and naïve rats for the 1 st distension set. No difference in intra-gastric balloon volume was observed between groups meaning that the surgery had no impact on the intensity of stimulation (Fig. 3B). www.nature.com/scientificreports www.nature.com/scientificreports/ Effect of ghrelin on neuronal activation. Repeated nociceptive gastric distensions increased p-ERK1/2/ ERK1/2 ratio compared to non-distended rats in nodose ganglion (2.09 ± 0.15 vs. 0.86 ± 0.30, p < 0.05) but not in T8-T9 DRGs (Fig. 4A,B). Ghrelin administration (30 µg/kg i.v.) restored p-ERK1/2/ERK1/2 ratio to basal level in nodose ganglion compared to rats treated with vehicle (0.74 ± 0.28 vs. 2.09 ± 0.15, p < 0.01).

Discussion
In the present report, we showed that acylated ghrelin decreased the variation of BP in response to GD, without change in gastric compliance, in anesthetized rats. This effect was prevented by pre-treatment with the Ghrelin receptor antagonist [D-Lys 3 ]-GHRP-6. Reduction of the autonomic response to GD following ghrelin systemic administration involved a vagal pathway as demonstrated by the lack of effect observed in vagotomized rats and the decrease in neuronal activation, estimated by phosphorylated ERK1/2 expression, in nodose ganglion of rats treated with ghrelin. In this structure a subset of gastric neurons expressing Ghrelin receptors also expressed ASIC3 or TRPV1 ion channels that have previously been shown as key modulators of gastric mechanosensitivity 15 . Taken together, our data support a role for acylated ghrelin in the modulation of gastric mechanosensitivity to distension through the involvement of Ghrelin receptor and vagal afferents.
The antinociceptive or antihypersensitive action of ghrelin was previously reported in somatic pain 10-13 and colonic distension models 14,16 . In the stomach, the role of ghrelin on gastroesophageal vagal afferents excitability has already been investigated on mouse ex vivo preparation by decreasing the response of tension receptors 17 . This inhibition involved Ghrelin receptor since it was reversed by [D-Lys 3 ]-GHRP-6. In line with this study our results show a reduction of BP variation, used as a surrogate marker of gastric sensitivity in response to graded distensions, after ghrelin administration. Assessment of gastric sensitivity in vivo remains challenging since accessibility of the stomach for balloon insertion requires invasive surgical technique and nociceptive response necessitate viscero-somatic 18 or viscero-visceral 19 reflex recordings either in awake or anesthetized animals. We chose to work on a model anesthetized with thiobutabarbital because of its minimal impact on cardiovascular system and it provides a sufficient long-lasting anesthesia to perform the whole procedure 20 . In response to gastric distension we recorded changes in BP as a viscero-visceral reflex. This cardiovascular reflex implies a complex interplay between parasympathetic vagal and sympathetic splanchnic fibers. A recent report established a prominent role for vagal afferents in the trigger of the cardiovascular reflex using low pressure distension whereas splanchnic nerves are preferentially involved during high pressure distension 19 . Our data indicate a more pronounced effect of ghrelin at 40, 60 and 80 mmHg. Therefore, pharmacological action of ghrelin was expected to recruit splanchnic rather than vagal afferent. We demonstrated that intact vagal innervation is necessary for ghrelin to exert its action and the peptide prevents vagal neuronal activation induced by gastric distension in rats. Nevertheless the aforementioned study determined that an intact vagus nerve was essential for the complete eliciting of the cardiovascular reflex even during high pressure distension. This finding suggests a necessary crosstalk between vagal and splanchnic afferents to modulate cardiovascular responses to gastric distension with high intensity and ghrelin, by primary acting on vagal inputs, could modulate splanchnic regulation of the reflex. On the other hand, despite the effect of ghrelin was still observed in splanchnectomized rats, we cannot rule out its potential effect on spinal afferents. Indeed, Ghrelin receptor has been shown to be expressed in embryonic chick DRG 21 and unpublished data obtained in our laboratory demonstrated that it is also expressed within T9-T10 DRG in rats. Moreover a recent report investing the effect of ghrelin in a rat model of colonic hypersensitivity pointed out an inhibited expression of TRPV1 channel in DRG neurons after ghrelin treatment 22 . Our data did not demonstrate such involvement of splanchnic pathway since we did not observe any effect of gastric distension on neuronal activation in DRG neurons and, as a consequence, no modulation by ghrelin administration. This lack of effect of gastric distension may be explained by the methodology we used. Compared to a previous study that demonstrated increased expression of p-ERK1/2 in retro-labeled gastric DRG neurons following gastric distensions by immunofluorescence 22 , we chose western blotting analysis. Since DRG afferent neurons innervate not only viscera but also somatic and articular territories, the signal could have been drowned into non-specific afferent messages.
The inhibitory effect of ghrelin observed in our work clearly involved Ghrelin receptor since pre-treatment with the antagonist [D-Lys 3 ]-GHRP-6 prevented this effect. The involvement of Ghrelin receptor in the effect of acylated ghrelin has been clearly established in gastric motility or food intake 23,24 . However, its specific involvement in the sensitive action of ghrelin remains debated. Indeed, divergent observations have been made regarding the effect of ghrelin on somatic sensitivity.  17 . To date, whether ghrelin acts through a paracrine or an endocrine pathway to modulate GD-sensitive afferent fibers is not known. In the present work, we used higher dose of acylated ghrelin compared to the level of physiological circulating levels. In fact, the half-life of acylated ghrelin in the blood is relatively short (<30 min) since acylated ghrelin is degraded by plasma esterase 25 . Ghrelin receptor has been found in nerve afferent endings in the stomach where receptors are transported to the periphery 26,27 . Moreover, the blood to brain transport of acylated ghrelin appears to be highly regulated 28 . Therefore, the concentration of acylated ghrelin is likely to be much higher at the periphery where afferent fibers would be an obvious site of action.
As suggested by immunofluorescence analysis, Ghrelin receptor colocalizes with TRPV1 or ASIC3 channels within gastric afferents of the nodose ganglion. Although these data do not allow us to conclude about the mechanism of action of ghrelin in vagal afferents it indicates that the ghrelin receptor is expressed in sensory fibers that also express key modulators of gastric mechanosensitivity. Indeed, vagal afferents from mice lacking ASIC3 and TRPV1 display blunted responses to distension on in vitro preparation compared to wild-type controls with www.nature.com/scientificreports www.nature.com/scientificreports/ a more pronounced effect of TRPV1 deletion 15 . It was previously shown that Ghrelin receptor can modulate the activity of ion channels and particularly of voltage-gated calcium channels Cav2.1 and Cav2.2 in rat and mouse hypothalamic neurons either by reducing the channel density at the membrane via the constitutive activity of the ghrelin receptor or by altering Cav2 gating via the binding of ghrelin to the receptor 29 . However, further experiments, using knock out strategies are warranted to assess the potential interplay between Ghrelin receptor and TRPV1 or ASIC3. In this particular case, pharmacological strategies blocking either TRPV1 or ASIC3 channels may not be relevant since it would result in an overall decreased gastric mechanosensation and thus would not allow to observe antinociceptive effect of ghrelin.
The therapeutic interest of ghrelin has been highlighted by numerous animals and human studies with beneficial effects in gastroparesis and postoperative ileus by promoting gastric emptying (for review) 30 . Interestingly ghrelin and others Ghrelin receptor agonists have also been shown to relieve nausea and vomiting in animal models and patients under chemotherapy 31 or with diabetic gastroparesis 32 . Although Ghrelin receptor agonists appear to be a new promising pharmacological target to treat nausea and vomiting, the effect of a chronic administration on symptoms remains unknown. As described earlier, Ghrelin receptor agonist like relamorelin alleviates gastroparetic symptoms and accelerates gastric emptying but not necessarily in a concomitant fashion. On the other hand, visceral hypersensitivity is associated with functional dyspepsia, which shares clinical features with gastroparesis, including epigastric pain. It can therefore be speculated that Ghrelin receptor agonists acts on gastric nociceptive pathways, although this remains to be investigated in patients. Therefore, and in accordance with this previous report, our data demonstrate that ghrelin, by acting on Ghrelin receptor on vagal afferents could modulate gastric sensitivity and may be a therapeutic option of interest to treat gastroparetic patients with visceral hypersensitivity.

conclusion
We showed in the present work the ability of acylated ghrelin to decrease the autonomic response to gastric distension likely through the Ghrelin receptors and vagal afferents. This offers thus a new therapeutic target to treat symptoms related to the alteration of gastric sensory processing that includes nausea, vomiting, or pain. Further work both on animal models and in patients is therefore warranted to further explore the role of Ghrelin receptor in visceral pain processing. Gastric sensitivity measurement in anesthetized rats. Gastric sensitivity was assessed using the variation of the arterial blood pressure (BP) in response to GD 33 . The BP was measured continuously in anesthetized animals using a perfused catheter (NaCl 0.9%; heparin 0.3%) introduced into the left carotid and connected to a pressure transducer (Millar Instruments, Houston, TX, USA). Pressure signal was recorded using Spike2 software (Cambridge Electronic Design Limited, Cambridge, United Kingdom) and processed by applying a DC remove (time constant: 20 s) and a smoothing (time constant: 5 s). Blood pressure variation was quantified by measuring the modulus of the processed signal during the 20 s of each distension.

Gastric distension in rats.
A spherical infinitely compliant distension balloon (diameter: 3 cm; maximum volume 12 mL in rats) was made using a polyethylene bag attached to a tube in polyethylene (Dutscher, Brumath, France) drilled in its extremity. The balloon was inserted in non-fasted anaesthetized rats through an incision at tip of the proximal stomach. Prior to balloon insertion, the gastric content was removed by gentle suction/reflux of warmed (32 °C) saline solution in order to mix gastric content until the stomach was emptied. The balloon was then connected to an electronic barostat (G&J Electronics Inc, Toronto, Canada) to perform isobaric graded GD. Distension paradigm consisted in balloon inflation from 20 to 80 mmHg by 20 mmHg increments in rats. Each distension lasted 20 seconds, was repeated twice and separated by a 4 min interval from the next (Fig. 6) 20,34 . experimental protocols. Experiment 1: Effect of acylated ghrelin on variation of BP to graded GD in anesthetized rats. In anesthetized rats, a first set of graded GD was performed. Fifteen minutes later, atropine (1 mg/kg) was injected intravenously (i.v.) to prevent acylated ghrelin stimulation of gastric motility that would have jeopardized isobaric distension of the stomach. Five minutes after atropine injection, acylated ghrelin (30 µg/kg; Phoenix Pharmaceuticals, Burlingame, CA) or saline was injected i.v. (n = 10/group). A second series of GD was started 5 min later. In one experiment, [D-Lys 3 ]-GHRP-6 (Phoenix Pharmaceuticals, Burlingame, CA), a Ghrelin receptor antagonist, was injected (0.3 mg/kg i.v.) 10 min before acylated ghrelin injection (n = 8). The dose of peptides were selected accordingly to our previous report as the maximal effective dose to accelerate gastric emptying 3 (Fig. 6).
Experiment 2: Characterization of the neural pathway involved in the effect of acylated ghrelin on gastric mechanosensitivity. In a second experiment vagal or spinal denervation of the stomach was performed prior to the first set of gastric distension by subdiaphragmatic vagotomy (n = 10) or celiac ganglionectomy (n = 6), respectively. Then the same protocol as described in experiment 1 was performed (Fig. 6).
In situ hybridization. Total RNA from hypothalamus was reverse transcribed to generate a 559-bp DNA fragment of Ghrelin receptor. The reverse transcription product was amplified by PCR using the following primers: forward primer, 5′-TGTGGTGGTGTTTGCTTTCATCC-3′; and reverse primer, 5′-CCTGCTGTGGGTATGAGTTGT-3′ (IDT). The amplified 559-bp PCR product was subcloned into the pGEM-T vector, and the resulting plasmid was linearized and transcribed with T7 or SP6 RNA polymerase to generate antisense or sense probes, respectively. The probes were labeled by incorporation of digoxigenin (Dig)-11-UTP (Roche Diagnostics, Meylan, France).
Gastric neurons retro labeling. Laparotomy was performed under Ketamine (10mg/kg)/Xylazine (75 mg/kg) anesthesia in rats (n = 3). Ten to 12 injections of 0.5 to 1.5 µL of Fluorogold 4% in NaCl 0.9% equally distributed between fundus and antrum were realized and injection sites were thoroughly rinsed with warm NaCl 0.9% to prevent retrotracer spreading to adjacent structures. Muscular layer was closed with 6-0 absorbable sutures and the skin with 4-0 non-absorbable sutures (Ethicon). After 7 days recovery rats were terminally anesthetized with intraperitoneal pentobarbital injection and transcardially perfused with 75 mL warm NaCl 0.9% then 500 mL paraformaldehyde 4% in PBS. Left and right nodose ganglia were harvested, post-fixed overnight at 4 °C in 4% PFA and 10 µm sections were performed using a cryostat before mounting on slides.