Specific hunger- and satiety-induced tuning of guinea pig enteric nerve activity

Abstract

Although hunger and satiety are mainly centrally regulated, there is convincing evidence that also gastrointestinal motor activity and hormone fluctuations significantly contribute to appetite signalling. In this study, we investigated how motility and enteric nerve activity are set by fasting and feeding. By means of video-imaging, we tested whether peristaltic activity differs in ex vivo preparations from fasted and re-fed guinea pigs. Ca²⁺ imaging was used to investigate whether the feeding state directly alters neuronal activity, either occurring spontaneously or evoked by (an)orexigenic signalling molecules. We found that pressure-induced (2 cmH₂O) peristaltic activity occurs at a higher frequency in ileal segments from re-fed animals (re-fed versus fasted, 6.12 ± 0.22 vs. 4.84 ± 0.52 waves min⁻¹, P = 0.028), even in vitro hours after death. Myenteric neuronal responses were tuned to the feeding status, since neurons in tissues from re-fed animals remained hyper-responsive to high K⁺-evoked depolarization (P < 0.001) and anorexigenic molecules (P < 0.001), while being less responsive to orexigenic ghrelin (P = 0.013). This illustrates that the feeding status remains ‘imprinted’ex vivo. We were able to reproduce this feeding state-related memory in vitro and found humoral feeding state-related factors to be implicated. Although the molecular link with hyperactivity is not entirely elucidated yet, glucose-dependent pathways are clearly involved in tuning neuronal excitability. We conclude that a bistable memory system that tunes neuronal responses to fasting and re-feeding is present in the enteric nervous system, increasing responses to depolarization and anorexigenic molecules in the re-fed state, while decreasing responses to orexigenic ghrelin. Unlike the hypothalamus, where specific cell populations sensitive to either orexigenic or anorexigenic molecules exist, the enteric feeding state-related memory system is present at the functional level of receptor signalling rather than confined to specific neuron subtypes.

Key points

• Acute changes in glycaemia can have substantial effects on gastro-intestinal motor function.

• A feeding state-related bistable memory system has been previously described in neurons of the hypothalamus.

• We found that peristaltic bowel movements are more pronounced in intestinal segments from re-fed compared to fasted animals and neuronal responses in the myenteric plexus are tuned to the feeding state, which illustrates that the feeding status remains imprinted ex vivo.

• This feeding state-related memory can be reproduced in vitro and humoral factors as well as glucose-dependent pathways are involved.

• Pharmacological manipulation of feeding state-specific signalling in enteric neurons can be of therapeutic importance in the treatment of motility-related feeding disorders.

Introduction

The gastro-intestinal (GI) tract processes ingested food and enables uptake of nutrients and removal of indigestible remnants. The digestive function, which is controlled by a complex interplay between the brain and neuromuscular and hormonal GI elements, not only comprises key elements like propulsion, secretion and absorption, but also contributes to hunger- and satiety-related signalling (Thorens & Larsen, 2004; Grundy, 2006; Camilleri & Grudell, 2007; Maljaars et al. 2007). Hormonal and neuronal signals arising from the GI tract play an important role in satiety and satiation. During food intake, hunger is replaced by a sensation of fullness (satiation) and does not return (satiety) until the organism runs short of nutrients or energy (Cummings & Overduin, 2007). Abnormalities in this satiation or satiety process may result in clinically relevant syndromes such as obesity, or impaired food intake and weight loss (Rigaud et al. 1995; Tack et al. 2001).

The regulation of hunger and satiety is a complex process. Orexigenic signalling molecules, like ghrelin, and anorexigenic molecules, like cholecystokinin (CCK) and serotonin (5-HT), wax and wane over time and their relative plasma levels are associated with the occurrence of hunger, satiety and satiation (Cummings & Overduin, 2007). Many of these hunger-related molecules are produced by enteroendocrine cells in the mucosa of the GI system and signal to the hypothalamus to regulate food intake (Cummings & Overduin, 2007). Although this multifactorial process is mainly regulated centrally, there is convincing evidence that also GI hormone fluctuations and motor activity significantly contribute to appetite signalling. The rumbling sounds produced by the strong stomach contractions during gastric phase III of the migrating motor complex are a commonly perceived link between hunger and GI activity and are accompanied by elevated ghrelin levels in rats (Taniguchi et al. 2008) and peak hunger sensations in man (Sepple & Read, 1989).

Interdigestive and food intake-specific motility has been studied both in humans and animal models. These patterns obviously differ, but it is not clear whether this results solely from a concerted action of luminal stimulation and hormone release, or whether also enteric nerve activity itself is intrinsically altered, even after removing luminal contents or hormone circulation. Such a neuronal involvement could have important implications, since this would mean that by targeting the enteric nervous system (ENS) and its activity, feedback control towards food intake could possibly be provided.

With this study, we addressed this issue by testing two hypotheses: (1) peristaltic activity differs between ex vivo small intestinal segments taken from fasted and re-fed guinea pigs. (2) The feeding state alters enteric neuronal activity and determines the specific responses to orexigenic and anorexigenic signalling molecules.

We indeed found that peristaltic activity in ileal segments from re-fed animals is more pronounced, and that myenteric neuronal responses differ according to the feeding status. Neurons are hyper-responsive to depolarization and anorexigenic molecules in the re-fed state, while being less responsive to orexigenic ghrelin. These findings illustrate that the feeding status remains ‘imprinted’ex vivo, a condition that could be reproduced in vitro.

Methods

Ethical approval

Animals were killed by a sharp blow to the cranium resulting in concussion of the brain, followed by immediate exsanguination from the carotid vessels. These procedures have been performed according to the regulations of the Animal Ethics Committee of KU Leuven, University of Leuven (approval: P88-2007 and P133-2011).

Animals

To investigate alterations due to feeding status, male guinea pigs (Cavia porcellus, 250–700 g) were fasted overnight (20 h). One group of animals was re-fed with standard lab chow (ssniff® Ms-H) 3 h prior to killing, while the others were killed immediately after the fasting period. Throughout the study these two groups will be referred to respectively as re-fed and fasted. All animals were killed at the same time of day (10.00 h) in order to limit diurnal variations.

Video-imaging

Ileal segments from fasted and re-fed guinea pigs were suspended in an organ bath (Fig. 1B) (±30 min after death) filled with Krebs solution (in mm: 120.9 NaCl, 5.9 KCl, 1.2 MgCl₂, 2.5 CaCl₂, 1.2 NaH₂PO₄, 14.4 NaHCO₃, 11.5 glucose) kept at 37°C and continuously bubbled with 95% O₂–5% CO₂ (pH 7.4). Intestines were allowed to equilibrate for 30 min, after which a control movie (0 cmH₂O) was recorded. The intraluminal pressure was subsequently elevated (at the oral side) with 1 and 2 cmH₂O and both times imaged for 90 s (1 min interval between recordings) (Fig. 1A) (Hennig et al. 1999).

Figure 1. Organ bath experiments.

A, ileal segments from fasted and re-fed guinea pigs were suspended in an organ bath. Recordings were made at rest and after elevation of the intraluminal pressure (1 and 2 cmH₂O). B, top to bottom: schematic representation of the organ bath set-up, an intestinal segment and a spatiotemporal map depicting the diameter of the intestinal segment (bar: 1 cm). C, frequency of peristaltic waves in ileal segments from fasted and re-fed guinea pigs. D, propagation velocity of the peristaltic waves. *P < 0.05.

All movies (90 s, 10 Hz frame rate) were recorded with and sampled on a Sensoray 611 PCI Frame Grabber (Sensoray, Tigard, OR, USA). Images were read into Igor Pro (Wavemetrics, Lake Oswego, OR, USA) and analysed using custom-written algorithms. The bowel's edges were determined and the width computed and mapped over time (Fig. 1B). From the generated spatiotemporal maps, the peristaltic wave frequency and propagation speed were determined. The frequency of intestinal peristaltic waves in re-fed and ad libitum fed (data not shown) guinea pigs was not different, but the re-fed group showed less inter-animal variability and was therefore better suited to compare the differences between fasted and fed states.

Ca²⁺ imaging

The Ca²⁺ imaging experiments were performed on guinea pig longitudinal muscle–myenteric plexus (LMMP) preparations (ileum and duodenum) and primary myenteric neuron cultures prepared from guinea pig ileum (Vanden Berghe et al. 2008).

Tissue preparations

Guinea pig ileum or duodenum was removed and pinned out in a Sylgard-lined Petri dish. Pieces of LMMP were prepared in cold Krebs solution (continuously bubbled with 95% O₂–5% CO₂; pH 7.4) by removing the mucosa, submucosa and circular muscle layer. Tissue samples (±1 cm²) were stretched over a small inox ring and immobilized by a matched rubber O-ring, after which they were loaded with 1 μm Fluo-4 AM (Molecular Probes/Invitrogen, Merelbeke, Belgium) (room temperature (RT), 30 min) in Krebs solution supplemented with cremophor EL (0.01%, Fluka). All further tissue experiments were done in Krebs solution.

Cell cultures

Cultured myenteric neurons were prepared from adult guinea pig ileum (Vanden Berghe et al. 2008). LMMP segments were digested in a protease–collagenase solution (20 min, 37 °C). The suspension was spun at 500 g, pellets were resuspended and individual ganglia were plated onto glass coverslips. Cultures were kept at 37°C (5% CO₂) and medium (Medium 199 enriched with 10% fetal bovine serum (FBS), 50 ng ml⁻¹ nerve growth factor (NGF), 100 U ml⁻¹ penicillin, 100 μg ml⁻¹ streptomycin, 30 mm glucose) was changed every 3 days. To model different feeding states in culture, enteric neurons were also grown in medium in which FBS was replaced by serum taken from fasted or re-fed guinea pigs (pooled from 7–8 animals – 3/20 h incubation). Since glycaemia was quite different between the two groups (fasted, 122.5 ± 4.0 mg dl⁻¹ (N = 16); re-fed, 321.8 ± 16.6 mg dl⁻¹ (N = 17); P < 0.001), culture media with glucose concentrations matching the fasted or re-fed guinea pig serum were also used (20 h incubation). The cultures were ready to be used after 7–12 days. Thirty minutes prior to the experiment, the cultured myenteric neurons were loaded with 5 μm Fluo-4 AM (RT, 30 min) in Hepes solution (in mm: 148 NaCl, 5 KCl, 1 MgCl₂, 10 glucose, 10 Hepes, 2 CaCl₂; pH adjusted to 7.4 with NaOH). All further cell experiments were done in Hepes solution.

Live imaging

After loading with Ca²⁺ indicator, tissues (immobilized over metal rings) or cells (grown on glass coverslips) were rinsed and transferred to a coverglass bottom chamber mounted on the microscope stage. During the experiment, a constant flow of Krebs/Hepes solution (1 ml min⁻¹) was maintained via a gravity-fed electronic valve system that allowed switching (in less than 1 s) between standard and drug-containing solutions. Three minutes prior to the experiment, neurons were identified by depolarization following a 5 s exposure to a high K⁺ solution (75 mm K⁺) (Vanden Berghe et al. 2000). During the Ca²⁺ imaging experiments, tissues/cultured neurons were stimulated with drug (20 s exposure) or serum (160 s exposure) containing solutions (RT), after which amplitudes of the evoked Ca²⁺ transients and numbers of responders were determined.

Images were recorded using an inverted Zeiss Axiovert 200M microscope (Carl Zeiss, Oberkochen, Germany) with TILL Poly V light source (TILL Photonics, Gräfelfing, Germany). Fluo-4 was excited at 475 nm and images were recorded at 525/50 nm on a cooled CCD-camera (PCO Sensicam-QE, Kelheim, Germany) using TillVisION (TILL Photonics).

Data analysis

All image analyses were performed with custom-written routines in Igor Pro. First, although tissues were mechanically restrained, some residual movement had to be corrected for using algorithms in Igor Pro (Gallego et al. 2008; Boesmans et al. 2009). Regions of interest were drawn, after which average Ca²⁺ signal intensity was calculated, normalized to the initial Fluo-4 values and reported as F/F₀. Neurons were judged ‘active’ when they displayed at least one Ca²⁺ spike – defined if the signal rose above baseline plus 5 times the intrinsic noise (standard deviation) – during the recording. The peak amplitudes and percentages of responders were compared between fasted and re-fed groups.

Real time PCR

Real time PCR experiments were performed on small intestinal guinea pig tissue (after removal of the mucosa) and cell cultures to monitor the expression of the growth hormone secretagogue receptor 1a (GHS-R1a), the cholecystokinin 1 receptor (CCK-1R) and the serotonin receptor type 3 (5-HT₃-R) in relation to the housekeeping genes glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and/or tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, zeta polypeptide (YWHAZ).

Total RNA was extracted (tissue: Trizol method; cell culture: Qiagen ‘RNeasy Mini kit’ (Qiagen, Venlo, The Netherlands)) and purified (‘high pure RNA isolation kit’, Roche), after which reverse transcriptase (Superscript II) was used to obtain cDNA. The latter served as a template for the PCR reaction (45 amplification cycles: 95°C (10 s), 60°C (15 s), 72°C (15 s)). Specific guinea pig primers were designed (GHS-R1a: forward: CTCGGGGCTGCTCACCGTCA, reverse: CGGTGAGGCAGAAGACGGGC; CCK-1R: forward: AGGCGAGGATGGACGTGGTAGA, reverse: AAGGCCGGGGCCGATCCAAG; 5-HT₃-R: forward: CCCGAGGACTTTGACAACAT, reverse: TCACCTTGATGCCGAACATA; GAPDH: forward: TGCTTTCATGTCTGGCAAAG, reverse: CTTGCCGTGGGTAGAATCAT; YWHAZ: forward: CAAGCATACCAAGAAGCATTTGA, reverse: GGGCCAGACCCAGTCTGA) and the Light-Cycler 480 SYBR Green I Master mix was used to run the qPCR reaction on a Lightcycler 480 system (Roche Diagnostics, Mannheim, Germany). An inter-run calibrator was used and standard curves were created to obtain PCR efficiencies. Relative expression levels were calculated with the LightCycler 480 software, expressed relative to GAPDH and/or YWHAZ, and corrected for inter-run variability.

Solutions and drugs

All cell culture-related products, Fluo-4 AM, salts, and products for qPCR were from Molecular Probes (Invitrogen, Merelbeke, Belgium), except for NGF, which was from Alomone Labs (Jerusalem, Israel), and the ‘high pure RNA isolation kit’ from Roche. Serotonin and Triton X-100 were from Sigma-Aldrich. Ghrelin and CCK-8 were obtained from Tocris (Bristol, UK) and PolyPeptide Laboratories (San Diego, CA, USA) (NeoMPS), respectively.

Data presentation and statistics

Unless stated otherwise, all data are presented as means ± SEM. The n values are the numbers of cells analysed for each group and the N values report the numbers of animals used per group. Distributions were tested for normality with a Kolmogorov–Smirnov test and depending on their distribution, data from different conditions were compared with an ANOVA or Mann–Whitney U test. When proportions were compared, a χ² test was performed. Differences were considered significant if P < 0.05. Statistical analysis was performed with Statistica (StatSoft, Tulsa, OK, USA).

Results

Motility patterns in the fasted and re-fed state

Pressure stimuli (1 and 2 cmH₂O) were used to elicit peristaltic waves in ileal small intestinal segments of fasted and re-fed guinea pigs. Only occasionally, a peristaltic wave appeared at rest (0 cmH₂O) (Fig. 1C). Depending on the stimulus amplitude, a higher frequency of peristaltic waves was observed, and this effect was significantly more pronounced in segments from re-fed guinea pigs (waves min⁻¹; 0 cmH₂O: fasted, 0.18 ± 0.09; re-fed, 0.84 ± 0.28; 1 cmH₂O: fasted, 2.81 ± 0.58; re-fed, 3.88 ± 0.48; 2 cmH₂O: fasted, 4.84 ± 0.52; re-fed, 6.12 ± 0.22; P = 0.028) (Fig. 1C). We also compared the propagation velocity of the peristaltic waves in the fasted and re-fed condition. All waves propagated at approximately the same speed, regardless of the pressure stimulus, and no difference was found between fasted and re-fed animals (mm s⁻¹: 0 cmH₂O: fasted, 16.16 ± 1.63; re-fed, 20.99 ± 0.71; 1 cmH₂O: fasted, 24.03 ± 3.72; re-fed, 23.58 ± 1.36; 2 cmH₂O: fasted, 24.68 ± 5.19; re-fed, 21.96 ± 2.03) (Fig. 1D). The intestinal segments that did not respond to any of the stimuli were excluded from the analysis (fasted, N = 3/22; re-fed, N = 2/21).

Do neuronal responses differ according to the feeding status?

Enteric neurons in tissues from fasted and re-fed animals respond differently to ghrelin, CCK-8 and serotonin

Ca²⁺ imaging experiments were conducted on LMMP preparations from fasted and re-fed guinea pigs (N = 3–4 animals per group) to investigate whether neuronal responses differ according to the feeding status, even in an in vitro situation. To test this hypothesis, three molecules (ghrelin, CCK and 5-HT) were used, which – due to their (an)orexigenic nature – are likely candidates to unveil potential feeding-related differences (Fig. 2A).

Figure 2. Ca²⁺ imaging experiments in LMMP preparations of guinea pig small intestine.

A, ileal LMMP preparations from fasted and re-fed guinea pigs were exposed to a high K⁺ solution (75 mm, 5 s) to identify viable neurons, after which the tissues were stimulated with solutions containing (an)orexigenic molecules (20 s). B, representative fluorescence images of a typical myenteric ganglion loaded with the Ca²⁺ indicator Fluo-4. Top, at rest (left) and after exposure to high K⁺ (right); bottom, after exposure to ghrelin (left) and CCK-8 (right). The white arrowheads indicate a neuron that responds to both ghrelin and CCK-8, while the grey arrowheads point to a neuron that responds to ghrelin but not to CCK-8 (bar: 20 μm). C, amplitudes of the calcium transients evoked by ghrelin (10⁻⁷ m), CCK-8 (10⁻⁷ m) and 5-HT (10⁻⁵ m) in fasted and re-fed ileal tissue. D, percentage of responders after exposure to ghrelin, CCK-8 and 5-HT. E, amplitudes of the high K⁺ (75 mm)-evoked Ca²⁺ peaks. *P = 0.01, **P < 0.001.

The amplitudes of the Ca²⁺ transients (F/F₀) evoked by the hunger-related ghrelin (10⁻⁷ m) (fasted, 1.12 ± 0.02; re-fed, 1.07 ± 0.02; P = 0.013; n = 20–22) were higher in ileal tissue from fasted compared to re-fed guinea pigs, but no differences were observed between the response amplitudes to the satiety-related CCK-8 (10⁻⁷ m; n = 28–83) or 5-HT (10⁻⁵ m; n = 52–67) (Fig. 2B-C). As for percentages of responding neurons, no difference was found for ghrelin or 5-HT, but in the re-fed state, a higher percentage of neurons responded to CCK-8 (fasted, 23.5%; re-fed, 47.7%; P < 0.001) (Fig. 2D). We also investigated whether specific neuronal populations in the duodenum responded to either ghrelin or CCK-8 and found that in fasted animals, the largest fraction (37.4%) was activated by both molecules, while only smaller groups of independent responders to ghrelin and CCK-8 (13.4 and 14.1%, respectively) were present. These proportions were similar in re-fed animals (42.6%, 11.9% and 16.7%).

Interestingly, the high K⁺-evoked Ca²⁺ transients (F/F₀) in ileum were also higher in the re-fed state (75 mm K⁺: fasted, 1.46 ± 0.02; re-fed, 1.63 ± 0.02; P < 0.001; n = 241–311; N = 7 animals/group) (Fig. 2E), and a similar difference was observed in duodenal myenteric neurons (fasted, 1.85 ± 0.02; re-fed, 2.03 ± 0.02; P < 0.001; n = 359–462; N = 3 animals/group). Similarly, also electrical fibre tract stimulation in tissues from re-fed animals caused significantly more neurons to respond, and responses were again higher than in fasted animals (data not shown).

qPCR

The relative mRNA expression of GHS-R1a, CCK-1R and 5-HT₃-R was determined and compared between fasted and re-fed guinea pigs (N = 5–7), but no differences were observed in expression levels relative to GAPDH.

Fasted and re-fed sera affect enteric nerve signalling differently

To investigate whether circulating humoral factors could impinge feeding state-related effects on enteric neurons, we treated cultured myenteric neurons with serum from fasted and re-fed animals. We investigated the direct effect of the sera on neuronal Ca²⁺ signalling on the one hand (Fig. 3A), and their prolonged effect after incubation (3/20 h) with either fasted or re-fed serum on the other hand (Fig. 4A and 5A).

Figure 3. Acute effects of fasted and re-fed serum on primary myenteric neuron cultures from guinea pig ileum.

A, Ca²⁺-signals were recorded from guinea pig myenteric neuron cultures while exposing them to fasted or re-fed guinea pig serum (10% in Hepes solution). The acute serum-evoked Ca²⁺-peaks and the subsequent spontaneous activity events were evaluated (bar: 5% (ΔF/F₀) vs. 20 s). B, rise in relative Fluo-4 fluorescence (ΔF/F₀) after serum-application, plotted against time. C, acute serum-evoked rise in Ca²⁺-fluorescence. D, amplitudes of the spontaneously evoked Ca²⁺-transients after acute exposure to fasted/re-fed guinea pig serum. E, percentage of spontaneously active neurons. *P = 0.01, **P < 0.001.

Figure 4. Spontaneous activity events in myenteric neuron cultures after prolonged serum-exposure.

A, outline of the experimental protocol in which spontaneous activity was monitored in cultures after prolonged (3/20 h) exposure to fasted or re-fed guinea pig serum (bar: 5% (ΔF/F₀) vs. 20 s). B, amplitudes of the spontaneously evoked Ca²⁺-transients after incubation (3 h/20 h) with serum from fasted/re-fed guinea pigs. C, percentage of spontaneously active neurons. *P < 0.01, **P = 0.001, ***P < 0.001.

Figure 5. Ca²⁺ imaging experiments in myenteric neuron cultures after long-term serum exposure.

A, cell cultures from guinea pig ileum incubated with fasted/re-fed guinea pig serum (20 h) were first exposed to a high K⁺ solution (20/75 mm, 5 s), after which the tissues were stimulated with solutions containing (an)orexigenic molecules (20 s). B, representative fluorescence images of cultured myenteric neurons loaded with Fluo-4: in rest, and after exposure to high K⁺ and serotonin (left to right; bar: 20 μm). C, amplitudes of the Ca²⁺-transients evoked by ghrelin, CCK-8 and 5-HT. D, percentage of responders after exposure to the (an)orexigenic molecules. E, amplitudes of the K⁺-evoked Ca²⁺ transients in neurons incubated with fasted/re-fed serum. F, typical rise in relative Fluo-4 fluorescence (ΔF/F₀) after high K⁺, ghrelin, CCK-8 and 5-HT application, plotted against time. *P < 0.05, **P < 0.01, ***P < 0.001.

Acute effects of serum exposure

Ca²⁺ signals were recorded from guinea pig myenteric neuron cultures while acutely exposing them to serum from fasted or re-fed guinea pigs (10% in Hepes solution). The amplitude of the acute serum-evoked Ca²⁺ peaks (F/F₀) was significantly higher for the re-fed serum (fasted, 1.17 ± 0.01; re-fed, 1.19 ± 0.01; P < 0.001; n = 134–194 from N = 3 animals; N (serum samples) = 7–8) (Fig. 3B and C). After the initial peak (∼1 min after start of serum exposure), several spontaneous activity events were observed (Boesmans et al. 2009). Both the percentage of spontaneously active neurons (fasted, 7.1%; re-fed, 36.0%; P < 0.001) (Fig. 3D) and the amplitude of the spontaneous Ca²⁺-peaks (F/F₀) (fasted, 1.02 ± 0.01; re-fed, 1.03 ± 0.00; P = 0.013) (Fig. 3E) were significantly higher after exposure to re-fed serum (n = 111–112 from N = 6 animals), although Ca²⁺-spike frequency was unaltered.

Long-term effects of serum exposure

Spontaneous activity in nerve networks. We also monitored the spontaneous activity in cultures after prolonged exposure to 10% fasted or re-fed guinea pig serum (3 h: n = 46–102 from N = 5 animals; 20 h: n = 81–85 from N = 4–5 animals, N (serum samples) = 7–8) (Fig. 4A). The amplitude (F/F₀) of these spontaneous Ca²⁺ peaks differed between the 20 h re-fed and fasted serum treatments (1.07 ± 0.01 vs. 1.02 ± 0.00; P = 0.001), but not for the shorter 3 h incubation (re-fed, 1.02 ± 0.00; fasted, 1.03 ± 0.00; P = 0.220) (Fig. 4B). The percentage of active neurons was significantly higher (up to 5-fold, depending on exposure time) in cultures treated with re-fed serum (3 h: 21.7%vs. 5.9%; P = 0.004; 20 h: 25.9%vs. 5.0%; P < 0.001 re-fed vs. fasted) (Fig. 4C).

Effects of ghrelin, CCK-8 and serotonin. After prolonged serum exposure (20 h; 10% serum in culture medium; N = 4–5 animals), we found that ghrelin (10⁻⁷ m) evoked significantly higher responses (F/F₀) in cells incubated with fasted serum (fasted, 1.12 ± 0.02; re-fed, 1.06 ± 0.01; P = 0.023; n = 29–47), whereas serotonin (10⁻⁵ m) evoked higher amplitudes in neurons incubated with re-fed serum (fasted, 1.41 ± 0.02; re-fed, 1.52 ± 0.03; P = 0.005; n = 156–176). No difference was observed in the amplitudes of the CCK-8-responses (10⁻⁷ m; n = 157–159), nor in the percentages of responders (Fig. 5B–F).

No difference was found for the Ca²⁺ responses to 75 mm K⁺ (n = 359–390 from N = 9 animals). However, since 75 mm K⁺ is a supra-maximal stimulus mainly designed for identification purposes, we also used a more subtle depolarizing stimulus (20 mm K⁺), which, again, evoked a significantly higher response (F/F₀) in cells treated with re-fed serum (fasted, 1.09 ± 0.01; re-fed, 1.24 ± 0.01; P < 0.001; n = 41–156 from N = 1 animal per group) (Fig. 5E). To test whether the increased Ca²⁺ response could be due to a slight de- or hyperpolarization prior to the stimulus, we preincubated a set of untreated neurons (n = 575–714 from N = 2 animals) with 3 or 7 mm K⁺ instead of the normal 5 mm to induce an estimated 5 mV hyper- or depolarization, respectively. No changes in the responses to 20 mm K⁺ were found in the hyperpolarizing condition, but the slight depolarization (7 mm K⁺) did evoke a significantly higher response (F/F₀) to 20 mm K⁺ (1.24 ± 0.01 vs. 1.19 ± 0.01; P < 0.001).

qPCR. No differences were observed in the mRNA expression of GHS-R1a, CCK-1R, or 5-HT₃-R relative to GAPDH or YWHAZ in myenteric neuron cultures after prolonged exposure to 10% fasted or re-fed guinea pig serum (20 h, N = 4).

Effects of glucose adaptation in the culture medium

Since glucose levels of fasted and re-fed guinea pig serum differed substantially (see Methods), we tested whether these glucose concentrations alone could be causing differences in neuronal excitability. Therefore, 20 h prior to the Ca²⁺ imaging experiments, the glucose concentration in the culture medium was adjusted to match either the fasted or re-fed serum (Fig. 6A).

Figure 6. Ca²⁺ imaging experiments in cultured myenteric neurons after long-term exposure to glucose levels matching the fasted and re-fed serum.

A, outline of the experimental protocol in which, after a 20 h exposure to fasted/re-fed glucose levels, neurons were exposed to high K⁺ (75 mm, 5 s) and (an)orexigenic molecules (20 s). B, amplitudes of the ghrelin-, CCK-8- and 5-HT-evoked Ca²⁺ transients in neurons after exposure to glucose levels matching the fasted/re-fed serum. C, percentage of responders after exposure to the (an)orexigenic molecules. D, amplitudes of the high K⁺-evoked Ca²⁺ transients. *P < 0.01, **P < 0.001.

We found that the responses (F/F₀) evoked by ghrelin (10⁻⁷ m) were significantly higher in cells pre-incubated with fasted glucose medium (1.13 ± 0.01 vs. 1.06 ± 0.01; P < 0.001; n = 33–58). Conversely, neurons incubated with re-fed glucose medium displayed higher responses to CCK-8 (10⁻⁷ m) (fasted, 1.11 ± 0.01; re-fed, 1.17 ± 0.01; P = 0.005; n = 114–173) and serotonin (10⁻⁵ m) (fasted, 1.23 ± 0.01; re-fed, 1.31 ± 0.02; P = 0.006; n = 183–228) (Fig. 6B), as well as to high K⁺ (75 mm) (fasted, 1.63 ± 0.02; re-fed, 1.77 ± 0.02; P < 0.001; n = 192–243) (Fig. 6D). Also the percentage of responders to ghrelin was higher after incubation with fasted glucose medium (30.2 ± 3.3 vs. 13.6 ± 2.2 in re-fed glucose medium; P < 0.001), although no differences were observed for CCK-8- or 5-HT-responders (Fig. 6C) (N = 3 animals).

qPCR

The mRNA-expression of GHS-R1a, CCK-1R, and 5-HT₃-R relative to GAPDH or YWHAZ was determined in cultures after prolonged exposure to fasted or re-fed glucose-levels (20 h, N = 3), but no differences were found.

Discussion

In this study, we investigated the intrinsic effect of fasting and feeding on intestinal peristalsis and enteric nerve activity.

We found that pressure-induced peristaltic waves occur at a higher frequency in intestinal segments taken from re-fed guinea pigs, even hours after killing of the animal. However, compared to the dramatic differences in fed and fasted motility patterns in vivo, the fasted-re-fed differences observed here are fairly subtle. We believe the reason for this is to be found in the strong stimulus used to evoke peristaltic reflexes in our organ bath experiments. Especially in the fasted state, this pressure stimulus is high compared to the in vivo situation where pressure stimuli are obviously absent in fasted animals. Therefore, the peristaltic activity observed in the fasted guinea pig segments is most probably an overestimation that obscures the true motility differences between the fasted and re-fed state in vitro. In a first set of experiments (data not shown) also ad libitum fed guinea pigs were used, but no differences in peristaltic wave frequency were found between ad libitum fed and re-fed animals. Due to the strict fasting and feeding schedule, we considered that the re-fed animals were a better ‘controlled’ group, which was illustrated by the smaller inter-animal variability. Moreover, since in this study, we were particularly interested in the effects of satiety after hunger, we choose to perform the remainder of the study on the fasted and re-fed groups only.

Ca²⁺ imaging experiments revealed that feeding state-related differences are also present at the cellular level since neuronal responses to orexigenic ghrelin were more prominent in fasted animals, whereas responses to anorexigenic CCK-8 were more pronounced in the re-fed state. The amplitudes of depolarization-evoked Ca²⁺ peaks were consistently higher in re-fed animals, suggestive of a hyper-responsive state.

These findings indicate that the feeding status stays imprinted ex vivo, since both peristalsis and neuronal Ca²⁺ signalling remained altered hours after killing.

Unlike the hypothalamus, where specific populations are sensitive to either orexigenic or anorexigenic molecules, we found no evidence for specific populations of responders. The largest fraction (∼ 40%) of responding neurons in the ENS appeared to have receptors for both ghrelin and CCK-8. As such, the ENS seems to be more similar to the vagal neurons of the nodose ganglion, which are also more promiscuous in their responses to CCK and ghrelin, as over 75% of the neurons in rat nodose ganglia expressing CCK-1 receptors also express GHS-1a receptors (Burdyga et al. 2006). Interestingly, despite the general increase in neuronal Ca²⁺ signalling in the re-fed state, responses to ghrelin were consistently lower, while in the fasted state, only ghrelin responses were enhanced. This indicates that in the ENS, rather than specific populations of neurons, selected signalling pathways are specifically tuned according to the feeding state. Opposing actions of ghrelin and CCK have also been described in vagal afferent neurons as ghrelin inhibits and CCK stimulates vagal afferent discharge. Also the expression of the cannabinoid receptor 1 (CB-1) and the melanin-concentrating hormone (MCH)-1 receptor by vagal afferent neurons is subject to differential modulation by ghrelin and CCK as the expression is increased by energy restriction and decreased again upon refeeding or CCK administration, whereas ghrelin administration prevented the refeeding-induced downregulation (Burdyga et al. 2006).

To investigate the factors responsible for altering the response state, we first tested whether circulating humoral factors could impinge feeding state-related effects on enteric neurons. Enteric neurons in culture acutely exposed to serum from re-fed animals displayed a higher immediate Ca²⁺ peak compared to those exposed to fasted serum, and also the percentage of spontaneously active neurons was significantly higher, whether the serum was added acutely or during a pre-incubation period (3 or 20 h). Moreover, similar to the tissue experiments, we also found that orexigenic ghrelin induced higher responses in neurons pre-incubated with fasted serum, while, conversely, anorexigenic serotonin displayed significantly higher responses after re-fed serum. These effects remained even after the serum had been removed, indicating that, also in naive cultures, a sort of memory can be impinged by serum exposure.

Since glycaemia in re-fed guinea pigs was almost three times as high as in fasted guinea pigs, which is a lot higher and longer-lasting than in humans, we tested whether the prolonged effects of these glucose concentrations alone could be causing these differences in enteric nerve signalling. We found that significantly more neurons with higher responses to ghrelin were present after incubation with fasted glucose medium, whereas in neurons incubated with re-fed glucose medium, CCK-8 and serotonin, as well as high K⁺, evoked higher responses (Fig. 7). These experiments indicate that glucose concentrations indeed account – at least in part – for the prolonged changes that enteric neurons undergo.

Figure 7. Schematic summary of the results.

The green and red squares represent responses that were significantly higher or lower, respectively. The grey squares indicate that no statistical differences were found. Note that all the red squares in the re-fed column are associated with ghrelin experiments, while the green squares are referring either to generalized increase in activity (peristalsis, spontaneous activity, response to depolarization) or to anorexigenic CCK-8 and 5-HT.

The molecular link between glucose – probably in combination with other factors – and neuronal responsiveness is not yet clear, but it is known that most guinea pig enteric neurons are capable of sensing glucose and are excited by either increases or decreases in extracellular glucose (Liu et al. 1999). These concentration-dependent responses may explain why acute changes in glycaemia can have substantial effects on GI motor function (MacGregor et al. 1976; Liu et al. 1999). Marked hyperglycaemia appears to affect every region of the GI tract. During diabetes, gastric emptying as well as oesophageal and small intestinal transit can be significantly delayed (Rayner et al. 2001). Although these GI motor symptoms have previously been attributed to a generalized autonomic neuropathy, glucose-evoked changes in the activity of enteric neurons may also play a role. Smaller, transient elevations of blood glucose that remain within the normal postprandial range also influence gut function and may be important in the regulation of gut motility. As such, blood glucose elevations towards the upper end of the physiological range decrease duodenal compliance while increasing the stimulation of duodenal waves by balloon distention (Lingenfelser et al. 1999; Rayner et al. 2001). At the neuronal level, several mechanisms have been described by which cells detect extracellular glucose. Liu et al. postulate that glucose-sensing in enteric neurons might be mediated by ATP-sensitive K⁺-channels, as is the case in the ventromedial hypothalamus, pancreatic β-cells, and glucose-excited neurons in rat nodose ganglia (Liu et al. 1999; Grabauskas et al. 2010), while Diez-Sampedro et al. (2003) suggest a more direct role of glucose through SGLT-3, a member of the sodium–glucose cotransporter gene family. Most of the above studies address the acute effects of glucose, but little is known about sustained effects, which last for minutes or hours. Intraluminal glucose was recently shown to cause phosphorylation of Ca²⁺–calmodulin-dependent protein kinase II (pCaMKII) in myenteric neurons and enterochromaffin cells of the rat (Vincent et al. 2011). Glucose effects may live on via sustained protein phosphorylation and still have an impact on enteric neurons hours after exposure. Recent evidence also indicates that other luminal factors can quickly modulate neuronal function and alter protein expression. The short chain fatty acid butyrate for example can increase neuronal excitability rapidly and over short-term periods, as well as increase the proportion of choline acetyltransferase-immunoreactive myenteric neurons in rat colon (Soret et al. 2010).

The generalized enhancement of Ca²⁺ responses to depolarization and anorexigenic molecules in the re-fed state is likely to be due to enhanced excitability as our data indicate that a small depolarization would induce a similar effect in terms of Ca²⁺ response amplitude. What exactly underlies this change in excitability is currently not known. Changes in expression of Ca²⁺ channels or exchangers, or altered channel conductivity or exchanger activity might play a role, but this needs further investigation. The exceptional elevated ghrelin responses in the fasted state on the other hand could be due to activation of AMP-activated protein kinase (AMPK) under conditions of energy depletion – as is the case in hypothalamic AgRP-neurons. Yang et al. (2011) show that an AMPK-mediated feedback loop upregulates activity in presynaptic terminals for at least 5 h and suggest that this serves as a memory for ghrelin in the form of sustained synaptic activity onto AgRP-neurons. Furthermore, Shah et al. (2011) found that activation of AMPK positively regulates insulin signalling and glucose uptake in neuronal cells.

In conclusion, we found that peristaltic activity occurs at a higher frequency in intestines taken from re-fed animals, and the myenteric neurons involved were hyper-responsive to depolarization and anorexigenic molecules, while being less responsive to orexigenic ghrelin. Conversely, after fasting responses to ghrelin were enhanced, which suggests that the feeding state alters specific receptor signalling pathways to cause prolonged changes in activity. This feeding state-related memory system can be reproduced in vitro and humoral feeding state-related factors are implicated. Although the molecular link with differential hyperresponsiveness is not entirely elucidated yet, we found that a glucose-dependent pathway is at least in part involved in tuning neuronal excitability, which could be of therapeutic importance in the treatment of motility-related feeding disorders.

Glossary

AMPK

AMP-activated protein kinase

CB-1

cannabinoid receptor 1

CCK

cholecystokinin

CCK-1R

cholecystokinin 1 receptor

ENS

enteric nervous system

GHS-R1a

growth hormone secretagogue receptor 1a

GI

gastrointestinal

5-HT

serotonin

5-HT₃-R

serotonin receptor type 3

LMMP

longitudinal muscle–myenteric plexus

MCH-1

melanin-concentrating hormone 1 receptor

NGF

nerve growth factor

pCaMKII

Ca²⁺/calmodulin dependent protein kinase II

YWHAZ

tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, zeta polypeptide

Author contributions

Study concept and design, acquisition of data, analysis and interpretation of data, drafting and editing of the manuscript were done by L.R. and P.V.B. Statistical analysis was performed by L.R. W.B. added significantly to the methodology and study design. J.T., I.D. and W.B. provided scientific advice and critically revised the manuscript. M.D. contributed to data acquisition and manuscript revision. P.V.B. wrote the analysis software; J.T. and P.V.B. obtained funding. The authors disclose no conflicts of interest.