INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Recent studies on the pharmacology of gastro-esophageal vagal afferents have revealed important roles for cholecystokinin (CCK), 5-hydroxytryptamine (5-HT₃), GABA_B,  opioid, and purinergic (P_2X) receptors (Blackshaw and Grundy 1990, 1993a; Ozaki et al. 2000; Page and Blackshaw 1999; Page et al. 2000). In vitro preparations of tissue from small mammals provide highly controllable conditions for pharmacological studies of this kind. There are, however, many other targets, such as ion channels, intracellular messengers, cytokines, and neurotrophins, some of which are inaccessible using pharmacological tools. Understanding more about how these factors influence visceral afferents depends on the use of transgenic or knockout mice. An in vitro approach in knockout mice for studying cutaneous afferents has provided valuable information on a range of mechanisms influencing their function and structure (Carroll et al. 1998; Caterina et al. 2000; Price et al. 2000). However, for meaningful comparisons to be made in evaluating the roles of these mechanisms, it is first necessary to establish a normative inventory of the properties of wild-type mouse afferents (Koltzenburg et al. 1997). It is also important to compare the mouse model to another more established model of visceral afferent function. Recent studies of the anatomy of murine gastroduodenal afferents has verified their similarity with those in other species (Fox et al. 2000b) and provided preliminary evidence for the role of neurotrophin-4 in their development (Fox et al. 2000a). Electrophysiological studies in mice are completely lacking, which prompted us to embark on the present study.

In vivo electrophysiological studies have demonstrated the existence of two major functional classes of upper gastrointestinal vagal afferent endings (see Cervero 1994; Grundy and Scratcherd 1989, for reviews): tension receptors are exclusively sensitive to muscular contraction and distension; mucosal receptors are insensitive to muscular stimuli but respond to mucosal stroking, various luminal chemical stimuli, and drugs. Three populations of specific chemoreceptors have been postulated (Jeanningros 1982; Mei 1978; Melone 1986), but their mechanical sensitivity may have been overlooked. In vitro studies of upper gastrointestinal vagal afferents have revealed an additional class of mechanoreceptor in ferretsthe tension/mucosal receptor. This has properties of both classes described in the preceding text (Page and Blackshaw 1998).

The present study was undertaken to provide a comparison of the mechanosensory properties of mouse gastro-esophageal vagal afferent fibers with those in an established ferret model (Blackshaw et al. 2000; Page and Blackshaw 1998, 1999; Page et al. 2000) and to embark on investigation of aspects of chemosensitivity to physiological, pathophysiological, and pharmacological stimuli in mouse afferents.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

All studies were performed in accordance with the guidelines of the Animal Ethics Committees of the Royal Adelaide Hospital and Institute for Medical and Veterinary Science, Adelaide, Australia.

General

IN VITRO MOUSE GASTRO-ESOPHAGEAL AFFERENT PREPARATION. Female mice [C57/BL6 (n = 55): 20-30 g body wt] were killed by CO₂ inhalation and cervical dislocation. The stomach, esophagus with attached vagal nerves, heart, and lungs were removed and placed in modified Krebs' solution of the following composition (mM): 118.1 NaCl, 4.7 KCl, 25.1 NaHCO₃, 1.3 NaH₂PO₄, 1.2 MgSO₄.7H₂O, 1.5 CaCl₂, 1.0 citric acid, and 11.1 glucose, bubbled with 95% O₂-5% CO₂. The temperature was maintained at 4°C during dissection to prevent metabolic degradation. After further dissection the preparation was opened out longitudinally along the esophagus and greater curve of the stomach. The dorsal aspect of the stomach was removed completely to enable the tissue to be pinned flat in the organ bath with a straight edge. The tissue was then pinned mucosa side up in a perspex chamber (dimensions: 6.0 × 2.5 × 1.2 cm) and perfused at a rate of 11-12 ml/min with carbogenated Krebs' bicarbonate buffer solution maintained at 34°C. A sliding wall with a small "mouse hole" for the vagus nerves to pass through was moved into position so that the nerves extended into a second round chamber (dimensions: 3.7 cm diam, 1.2 cm deep) where they were laid on a mirror and bathed in paraffin oil. Under a dissecting microscope, the nerve sheath was gently peeled back to expose the nerve trunk. Using fine forceps, nerve fibers were teased apart into 8-12 bundles, then one by one, the small nerve bundles were placed onto a platinum recording electrode. A reference electrode rested on the mirror in a small pool of Krebs' solution.

FERRET GASTRO-ESOPHAGEAL AFFERENT PREPARATION. Ferrets (0.5-1.0 kg body wt) were deeply anesthetized with pentobarbitone sodium (50 mg/kg ip), and the thorax was opened by a midline incision. The ferret was then killed and exsanguinated. The preparation was then placed in an organ bath in a similar manner to the mouse preparation. This preparation has been described in detail previously (Page and Blackshaw 1998).

Characterization of gastro-esophageal vagal afferent properties

Location of receptive fields along the esophagus and stomach was determined by mechanical stimulation with either a blunt glass rod or a brush. The mechanical stimulus-response functions were determined using calibrated von Frey hairs. Tension response curves were also obtained for each receptive field; the curves were used in combination with von Frey responses to classify afferents. Tension was applied via a claw made from bent dissection pins attached with thread to a pulley and cantilever system. To balance the cantilever, weights were placed on the opposite side. The claw was always hooked to the edge of the stomach or esophagus adjacent to the receptive field under investigation. All afferents tested that were tension sensitive responded to stretch in both the longitudinal and circular direction, and no afferents were observed that responded only to longitudinal stretch. For this reason and also to provide optimum control, stretch was applied in a circular direction.

Chemosensitivity of mouse gastro-esophageal vagal afferents was determined after mechanical thresholds had been established in a total of 39 fibers using similar methods to our prior in vitro study in the ferret (Page and Blackshaw 1998). It is important to note that chemosensitivity of one fiber only per experiment was evaluated to avoid bias due to sensitization or desensitization. In all experiments, the mechanical sensitivity of receptive fields was checked between each drug application to ensure continued viability of the unit under investigation. Further application of other drugs did not occur if a certain drug affected the sensitivity of the receptive field to mechanical stimulation. After removal of the drug from around the receptive field, the afferents were allowed to return to a normal baseline level of activity. Five minutes of normal baseline activity was maintained before the addition of another drug.

Data recording and analysis

Afferent neural activity was amplified with a biological amplifier (BA 1, JRAK, Melbourne, Australia) and scaling amplifier (SA 1, JRAK,), filtered (F1 filter, JRAK), and monitored using an oscilloscope (Yokogawa Tokyo, DL 1200A). Single units were discriminated on the basis of action potential shape, duration, and amplitude using Spike 2 software (Cambridge Electronic Design, Cambridge UK). It was very rare with the size of the strands we used to have more than 3 units/nerve strand; however, when this occurred, the strand was split further to reduce the number of units. If there were only 2 or 3 units but the shape of the action potentials of each unit was similar (making them difficult to discriminate using spike 2 software), then the stand was split to try and separate the units. All data were recorded on magnetic tape and analyzed off-line using a personal computer (Compaq Armada M700). Peristimulus time histograms and discharge traces were displayed using Spike 2 software. Data are expressed as means ± SE with n = number of individual afferents in all instances. Differences between stimulus-response curves were evaluated using two-way ANOVA. Differences were considered significant if P < 0.05. A response to a chemical stimulus was scored when a 10% increase in discharge frequency occurred above a steady baseline after completing application of chemical.

Drugs

Stock solutions of all drugs were kept frozen and diluted to their final concentration in Krebs' solution on the day of the experiment. 5-Hydroxytryptamine and ,-methylene ATP were obtained from Sigma (Sydney, Australia). Ferret bile was collected from the gall bladder of anesthetized ferrets used for other studies within our laboratory.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Mechanical properties of mouse gastro-esophageal vagal afferent fibers

Two types of mechanosensitive fibers were observed using this in vitro preparation (Fig. 1): those responding to mucosal stroking but not circular tension (mucosal receptors; n = 27) and those responding to mucosal stroking and circular tension (tension receptors; n = 39). Mucosal receptors did not respond to tension, whereas tension receptors gave a graded increase in afferent discharge with increases in load (Fig. 1, A, B, and Ciii). Both mucosal and tension receptors responded to mucosal stroking with calibrated von Frey hairs with forces as small as 10 mg (Fig. 1, A, B, and C, i and ii). However, this stimulus caused observable distortion of the underlying layers in which we supposed the tension receptors were located. The responses of gastric mucosal receptors (n = 5) to mucosal stroking were significantly greater (P < 0.0001 for impulses evoked per stroke and P = 0.0073 for maximum instantaneous frequency per stroke; 2-way ANOVA) than responses of esophageal mucosal receptors (n = 11; data not illustrated). The responses of esophageal tension receptors (n = 15) to circular tension tended to be larger than responses of gastric tension receptors (n = 3; data not illustrated). Receptive fields of mucosal and tension receptors were randomly distributed over the esophagus and stomach but predominantly distal to the point at which the vagus was separated from the esophagus. These observations indicate that fibers course up or down the esophagus before exiting. The receptive fields were small (<0.5 mm diam) and distinct.

View larger version (21K): [in this window] [in a new window]   Fig. 1. Response of mouse gastro-esophageal vagal afferents to circular tension and mucosal stroking with calibrated von Frey hairs. A: typical responses of a mucosal receptor to mucosal stroking (left; raw traces) and circular tension (right; spike frequency). B: typical responses of a tension receptor to mucosal stroking (left; raw traces) and circular tension (right; spike frequency). C: group data. Discharge rate is shown on the y axis and applied force on the x axis.  or , mucosal receptor responses;  or , tension receptor responses (n  16). Mucosal receptors were insensitive to graded tension. The mucosal and tension receptor responses to circular tension were significantly different (P < 0.0001; 2-way ANOVA). Both mucosal and tension receptors were sensitive to stroking with calibrated von Frey hairs. The number of impulses evoked per stroke and the maximum instantaneous frequency (Hz) per stroke in tension receptors were significantly greater than in mucosal receptors (P < 0.05 and P < 0.01, respectively; 2-way ANOVA).

In general, mucosal receptors did not show marked resting activity, although spontaneous discharge was evident in nine fibers. Spontaneous activities of mucosal (median: 0.3 imp/s, interquartile range: 3.1, n = 19) and tension receptors (median: 2.5 imp/s, interquartile range: 3, n = 25) were significantly different (P = 0.01; using a Mann-Whitney U test). Spontaneous activity of the mucosal receptors showed no apparent rhythmicity, whereas the tension receptors often fired spontaneously with a constant regular rhythm comparable to the spontaneous rhythm of ferret tension-sensitive afferents (Page and Blackshaw 1998).

Comparison of mechanical sensitivity of gastro-esophageal vagal afferents in the mouse and ferret

Of the 52 ferret gastro-esophageal afferents recorded, 16 were mucosal, 18 were tension/mucosal, and 18 were tension receptors. Three gastric afferents were recordedtwo were mucosal receptors and one was a tension receptor. The response of mouse tension receptors to circular tension was significantly greater than ferret tension/mucosal receptors (P < 0.0001: 2-way ANOVA; Fig. 2A). These were in turn more responsive than ferret tension receptors (P = 0.0013: 2-way ANOVA; Fig. 2A). The response to mucosal stroking of mouse mucosal receptors was significantly greater than that of ferret mucosal (P < 0.0001: 2-way ANOVA; Fig. 2B) and ferret tension/mucosal receptors (P < 0.0001: 2-way ANOVA; Fig. 2B). The responses of ferret mucosal and tension/mucosal receptors to mucosal stroking were also significantly different (P < 0.0001; 2-way ANOVA; Fig. 2B). Analysis of data only from esophageal afferents had no effect on the significance of results when comparing mouse and ferret afferents.

View larger version (20K): [in this window] [in a new window]   Fig. 2. Comparison of the response of ferret and mouse gastro-esophageal vagal afferents to mechanical stimulation. A: response of mouse and ferret gastro-esophageal vagal afferents to circular tension. , mouse tension (n = 18) receptors; , ferret tension (n = 18) receptors; and , ferret tension/mucosal (n = 18) receptors. Discharge rate is shown on the y axis, and applied force on the x axis. The response of mouse tension receptors was significantly greater than the response of ferret tension or tension/mucosal receptors to circular tension (P < 0.0001 in both cases; using 2-way ANOVA). B: response of mouse and ferret gastro-esophageal vagal afferents to mucosal stroking with calibrated von Frey hairs. , mouse mucosal receptors (n = 16); , ferret mucosal receptors (n = 16); and , ferret tension/mucosal receptors (n = 18). Impulses per stroke is shown on the y axis, and applied force using a calibrated von Frey hair is shown on the x axis. The impulses evoked per stroke were significantly greater in mouse mucosal receptors than either ferret mucosal or tension/mucosal receptors (P < 0.0001 in both cases; using 2-way ANOVA).

Chemosensitivity of gastro-esophageal vagal afferents in the mouse

The chemosensitivity of gastro-esophageal vagal afferents in the mouse is illustrated in Table 1 and Fig. 3. There was no difference in latency of response between gastric and esophageal afferents.

View larger version (28K): [in this window] [in a new window]   Fig. 3. Response of mouse gastro-esophageal vagal afferents to ,-methylene ATP (,meATP; 1-1,000 µM) (A), 5-hydroxytryptamine (5-HT; 1-1,000 µM) (B), and ferret bile (1:8 to 1:1 dilution) (C). Dose-response curves of tension receptors (i, n  5) and mucosal receptors (ii, n  4) are illustrated. - - -, the mean spontaneous activity of the afferents. Representative esophageal mucosal receptor responses to ,meATP (1 µM; Aiii), 5-HT (100 µM; Biii), and ferret bile (Ciii) are illustrated. The traces are raw recordings of electrical activity. The responses in this figure are not all from the same fiber.

The response to ,-methylene ATP (1 µM; Fig. 3Aiii) was rapid and the duration of the response short with a cessation of response before washout. Two of the mucosal receptors that responded to ,-methylene ATP (1 µM) were found in the stomach as was one of the nine tension receptors. In all cases (13/13), the response was repeated on second application of ,-methylene ATP. Both tension and mucosal receptors responded in a concentration-dependent manner to application of ,-methylene ATP to the receptive field (Fig. 3A, i and ii).

Only 1 fiber in a total of 11 fibers tested responded to hydrochloric acid (100 mM). This exhibited a similar response profile to that evoked by ,-methylene ATP. This fiber had its receptive field in the stomach.

Responses to 5-hydroxytryptamine were generally prolonged, continuing for a period after washout. There was no desensitization to subsequent application of 5-hydroxytryptamine to the receptive field when tested. All fibers that responded to 5-hydroxytryptamine had receptive fields in the esophagus apart from one mucosal receptor that had its receptive field in the stomach. In nine fibers, the concentration response relationship of the response to 5-HT was evaluated. This showed a threshold of <1 µM and a maximal response 10 µM (Fig. 3B, i and ii). One of the tension receptors, after an initial response to 5-hydroxytryptamine, showed rhythmical bursting activity associated with visible contractions of the esophagus (observed using stereomicroscope at ×40 magnification). This was the only occasion that muscular activity in response to chemical application was observed, and this was not counted as a direct effect.

Unlike the other chemicals tested, bile evoked prolonged responses in 100% of mucosal receptors. Seven of these afferents had receptive fields within the stomach and six were located in the esophagus. Eleven tension receptors that responded to bile had receptive fields in the esophagus, and 1 had its receptive field in the stomach. All responses were reproducible on a second application. Ferret bile had an osmolarity of 285 ± 5.92 mosmol (n = 5) and pH of 7.7 ± 0.2 (n = 5), indicating that the response was not due to a change in osmolarity or pH. Increases in discharge during washout of bile were observed, but these subsided before testing of other chemicals.

Specific chemoreceptors

Twelve afferent fibers were recruited during application of bile to the receptive field and surrounding tissue of another fiber under investigation in 11 experiments. They had no mechanosensitive receptive fields and were not spontaneously active. These afferents remained mechanically insensitive, even at noxious probing forces (>200 g), but responded to repeated exposure to bile in 6/6 cases. An example of a recruited afferent responding to bile is illustrated in Fig. 4. These specific chemosensitive afferents did not respond to ,-methylene ATP (1 µM; 12/12 afferents), 5-hydroxytryptamine (100 µM; 9/9 afferents), or hydrochloric acid (100 mM; 4/4 afferents).

View larger version (27K): [in this window] [in a new window]   Fig. 4. Recruitment of a mechanically insensitive afferent by addition of bile into a ring surrounding a mechanically sensitive afferent. A: response of a mechanically sensitive mucosal receptor to addition of bile (spike frequency). B: recruitment of a mechanically insensitive receptor during the same recording after addition of bile (spike frequency). Insets: the average spike shape of the mucosal receptor (A) and the recruited chemoreceptor unit (B). C: raw trace showing the response to bile of the mechanically sensitive (large amplitude) unit and the mechanically insensitive (small amplitude) unit. D: an expanded section of the raw trace in the boxed area in C. Note the increase in firing rate of the smaller unit about 5 s into the recording.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study provides the first documentation of functional properties of visceral primary afferent fibers in the mouse. Two types of gastro-esophageal afferent fiber were classified according to their sensitivity to mechanical stimuli: those responsive to low intensity mucosal stimuli (10 mg) but not circular tension (mucosal receptors) and those responsive to low intensity mucosal stimuli and circular tension (tension receptors). These are comparable with those found previously in other species. Another type of fiber is described for the first time that did not respond to mechanical stimulation in the form of either circular stretch or mucosal stroking. These fibers were exclusively chemosensitive responding only to bile.

Mechanical sensitivity

Both mouse and ferret tension receptors recorded in this study exhibited slowly adapting responses to circular tension. This is directly comparable with responses previously reported in various species including cats (Iggo 1957), sheep (Iggo 1955), ferrets (Andrews et al. 1980; Page and Blackshaw 1998), and rats (Davison and Clarke 1988). The anatomical correlate of the tension receptor has recently been investigated using a combination of anterograde neuronal tracing and close extracellular recording in the guinea pig (Zagorodnyuk and Brookes 2000). Tension-sensitive receptive fields were found to correspond with intra-ganglionic laminar endings (IGLEs) in myenteric ganglia in the esophagus. We assume, therefore, that the IGLEs described in the mouse by Fox et al. (2000b) would correspond to tension receptive sites in our preparation and are responsible for signaling active contraction and passive distension.

Vagal mucosal afferent sensitivity to light stroking and chemical stimuli is well documented (e.g., Blackshaw and Grundy 1990, 1993a; Clarke and Davison 1978; Cottrell and Iggo 1984b; Davison 1972; Page and Blackshaw 1998). It is likely that the tension-insensitive, mucosal-stroking-sensitive receptors we have recorded in mice are of this type, so we have labeled them mucosal receptors. Mouse mucosal receptors responded in a force-dependent manner to mucosal stroking with calibrated von Frey hairs as did ferret mucosal and tension/mucosal receptors in our previous study (Page and Blackshaw 1998). Responses of mouse afferents were, however, significantly greater than those in ferret. Free endings of vagal nerve fibers have been observed in the esophageal epithelium of several species, including dogs, cats, and monkeys (Clerc and Condamin 1987; Hudson and Cummings 1985; Robles-Chillida et al. 1981) where they are presumed to act as chemoreceptors and tactile mechanoreceptors. We conclude that the two types of vagal afferent fibers that we have encountered (mucosal and tension receptors) correspond to those that have previously been described anatomically in the esophagus and stomach and electrophysiologically in the abdominal viscera.

In the ferret in vitro preparation, we found three types of gastro-esophageal vagal afferent fibers: first, those responsive to low-intensity mucosal stimuli (10 mg) but not circular tension; second, those responsive to circular tension and to high-intensity mucosal stimuli (200 mg); third, those responsive to low-intensity mucosal stimuli (10 mg) and to circular tension (Page and Blackshaw 1998). Therefore the main difference between the mouse in vitro preparation and the ferret is that we were unable to differentiate two subtypes of tension-sensitive afferent. This difference could simply be due to the thinness of the mouse tissue compared with the ferret. All afferents observed in the mouse in vitro preparation responded to stroking with the 10 mg von Frey hair, which visibly distended the muscular layer of the tissue. In fact, tension receptors were significantly more responsive to stroking with calibrated von Frey hairs than mucosal receptors. We did obtain von Frey hairs that when placed on the receptive field at 90° gave a constant force. These von Frey hairs were calibrated as low as 8 mg. Unfortunately, the 8 mg von Frey hair would not penetrate the Krebs meniscus and was extremely difficult to place on the receptive field due to the flow of the Krebs solution, so data were unavailable using this method. Therefore it can be concluded that using the present methods of identification only two populations of mechanosensitive mouse gastro-esophageal vagal afferent fibers are discernable. In addition to the number of action potentials per stroke of the von Frey hair, the maximum instantaneous frequency per stroke was also determined for mouse afferents. This measurement may be sensitive for revealing specific changes in sensitivity of gastro-esophageal vagal afferents in future studies. Frequency analysis revealed that tension receptors fired at a significantly greater rate than mucosal receptors when stroked with calibrated von Frey hairs. The topographical distribution of these two types of afferent was random throughout the esophagus and stomach but predominantly distal to the point of separation of the vagal trunks from the esophagus. This is in agreement with our previous study comparing distribution of tension and mucosal receptors (Page and Blackshaw 1998).

Most of our knowledge about the electrophysiology of visceral afferents has been gained from whole animal experiments using "single-fiber" recording techniques first introduced by Paintal (1953) and Iggo (1955). It is difficult to compare directly data from these experiments to those obtained in the mouse in vitro. Therefore we have compared mouse in vitro data with those from an already established in vitro preparation, the ferret isolated gastro-esophageal vagal afferent preparation. Methodologically, these preparations have only minor differences. The main difference is that when tension is applied across the ferret tissue a single hook is used whereas a claw is used on the mouse preparation. The claw technique was adopted for the mouse preparation because the hook tended to damage the tissue. The load applied to the tissue by the claw was spread over a 0.5 cm width, and this was enough to prevent damage. Despite this spreading of the load, the tension response curve for mouse tension receptors was still steeper than those for ferret tension and tension/mucosal receptors. Thus mouse tension receptors are more sensitive than ferret tension receptors to circular loads. This is likely to be attributable to the viscoelastic properties of tissues from the different species. In addition, the responses of mouse esophageal tension receptors to circular tension was greater than that of mouse gastric tension receptors. Again, a difference in mechano-elastic properties of different muscle types may be responsible for the observed difference in responsiveness of nerves innervating the striated muscle of the esophagus and the smooth muscle of the stomach.

Chemical sensitivity

Examination of the chemosensitivity of mouse mucosal and tension receptors showed that a proportion of both types of fiber responded to one or more of the chemical stimuli applied to the receptive field. The latency and duration of the responses to these chemicals tended to be shorter in mouse afferents compared with ferret afferents previously reported (Page and Blackshaw 1998). However, this may simply be due to the higher permeability of the mouse tissue or lower mucous secretion and not a property of the afferent endings. We chose a range of chemical stimuli to investigate mouse gastro-esophageal vagal afferents to deliver stimuli that may be encountered in the lumen during physiological or mildly pathophysiological conditions. Acid and bile can reflux into the esophagus and stomach causing damage to the mucosal surface (Black et al. 1971; Gillen et al. 1988; Nehra et al. 1999). 5-Hydroxytryptamine may be released from enterochromaffin cells or play a role as an inflammatory mediator after injury (see Blackshaw and Grundy 1993a). A proportion of afferents were chemosensitive to 5-hydroxytryptamine and hydrochloric acid, as previously reported by our group in the ferret esophagus in vitro (Page and Blackshaw 1998; Page et al. 2000). Approximately, half of esophageal tension-sensitive afferents responded to 5-HT. One afferent was discounted from this count on the basis that the bursting activity in response to 5-HT correlated to visible muscular contractions of the esophageal wall. This response to visible muscular contraction was smaller and had a longer latency than the responses of the other tension receptors to 5-HT. Therefore there is a reasonable degree of confidence that the responses are due to direct activation of the receptive field as opposed to a secondary effect due to muscular contraction.

Within the gastrointestinal tract there is an abundance of evidence that ATP acts as a neurotransmitter, being released from either extrinsic sympathetic efferent nerves or from intrinsic enteric neurons (Burnstock 1990; Galligan and Bertrand 1994). In addition, ATP can be released from cells at the site of tissue injury (Burnstock and Wood 1996). ATP and ,-methylene ATP have been shown to excite small intestinal mesenteric afferent nerves (Kirkup et al. 1999), the early phase of the response being due to direct activation of the afferents. A small proportion of mouse gastro-esophageal afferents were excited by the addition of ,-methylene ATP to the mucosal surface surrounding the receptive field. In contrast none of the ferret mucosal gastro-esophageal vagal afferents we recorded were responsive to ,-methylene ATP (Page et al. 2000). They were, however, sensitized by ,-methylene ATP following inflammation. The data, taken together, suggest more direct coupling of purinergic receptors to excitatory mechanisms in mouse afferents than in ferret. The concentration at which responses to ,-methylene ATP were evoked is consisitent with a role for purinergic receptors in signaling tissue injury in the gastro-esophageal region to the CNS.

Bile elicited an increase in discharge in all mouse mucosal receptors tested and >50% of tension receptors tested. This is in excess of the proportion responsive in an in vitro colon preparation where ~50% of mucosal and serosal receptors were excited by bile (Lynn and Blackshaw 1999). The response to bile was not due to a change in osmolarity or pH, therefore one of the bile components must have induced the excitatory response. Bile reflux into the stomach and esophagus is not uncommon and may underlie development of disease (Gillen et al. 1988; Nehra et al. 1999). Our data indicate that bile may trigger reflex events or symptoms via a vagal afferent pathway. A direct action of bile on nerve endings cannot be totally discounted because the response was repeatable and not attenuated as would be expected if supplies of released mediators were being exhausted. However, exposure of the esophagus and stomach to bile may have caused damage to the mucosal surface resulting in release of local inflammatory mediators. These local mediators could be responsible for the excitatory effect of bile. ATP and 5-HT release from damaged cells can be discounted from the possible list of mediators because afferent responses to bile were far more vigorous than responses to ,-methylene ATP or 5-HT. In addition, some of the bile-sensitive afferents did not respond to ,-methylene ATP or 5-HT. Our frequent observation of specific sensitivity to bile may constitute a highly tuned sensory transduction pathway and is under further investigation within our laboratory.

Another group of afferents was discovered when bile was added to the ring around a mechanoreceptive field. This group of afferents did not respond to mechanical stimulation in the form of either mucosal stroking or circular tension but responded in a repeatable manner to bile. These are distinct from the previously described "silent nociceptors" (reviewed by Cervero 1994) in that they remain insensitive to mechanical stimuli after they have been recruited by chemical stimuli. The specific chemoreceptors observed in the present study did not respond to any chemicals other than bile. Afferents that were initially nonmechanosensitive were also observed in the rat colon (Lynn and Blackshaw 1999); however, these afferents responded to more than one chemical stimulus and subsequently became mechanically sensitive. In the present study, the bile-sensitive afferents did not become mechanosensitive even after repeated exposure to bile. Receptors sensitive to luminal perfusion of nutrients such as glucose (Mei 1978) and amino acids (Jeanningros 1982) have been described in the gastrointestinal tract. These were suggested to be specific chemoreceptors, but their possible responsiveness to mucosal stroking was not routinely evaluated. This is probably because experiments used closed-loop perfusion of different regions of the gastrointestinal tract that prevented access for testing of mucosal receptor mechanical sensitivity. In the present study, the tissue was opened out flat, and therefore rigorous testing of mechanical sensitivity was easily achieved. Our findings therefore represent the first description of bile-specific chemoreceptors in any species and possibly the first rigorous description of specific luminal chemoreceptors.

Biophysical properties

Spontaneous activity was present in a subgroup of both types of mouse afferent observed in this study. Analysis of group data on the two types of afferent we encountered in the mouse preparation showed that mucosal receptors had significantly lower rates of resting discharge. This is similar to differences previously reported between mucosal and other classes of fiber in the vagal innervation (Blackshaw and Grundy 1990, 1993a,b; Cottrell and Iggo 1984a,b; Page and Blackshaw 1998). For methodological reasons, we did not determine the conduction velocity of afferents in the present study. Reports have indicated the existence of a mixed population of A and C fibers in sheep (Cottrell and Iggo 1984a), opossum (Sengupta et al. 1989, 1992), and ferret (Andrews and Lang 1982; Page and Blackshaw 1998). There have also been reports in rat (Cervero and Sharkey 1988; Clarke and Davison 1978) and cat (Clerc and Mei 1983) that have shown that gastro-intestinal tension receptors are all C fibers or that only A fibers are present in sheep (Falempin et al. 1978). In our ferret, in vitro preparation mechanical sensitivity of the C and A fibers was not significantly different within the three types of afferent observed (Page and Blackshaw 1998). Therefore although the conduction velocity data would have provided an additional classification criterion for future studies, it is unlikely to be of functional significance to the type of sensory information encoded in the present study.

Conclusions

The present study shows that properties of gastro-esophageal vagal afferents may be studied directly in vitro in the mouse. The mouse isolated gastro-esophageal preparation is ideal for investigating both mechanical and chemical sensitivity of mucosal and tension receptors. Chemosensitivity of mouse tension and mucosal receptors is restricted to a small subpopulation, but these show robust responses consistent with expression of receptors to 5-HT, ATP, low pH, and bile constituents. In addition to two classes of mechanoreceptor, we have also encountered a population of afferents not observed elsewhere that are mechanically insensitive that are recruited by exposure to bile. An advantage of this model for future studies is that mice can be genetically modified for investigation of the roles of specific receptors, transmitters, channels, and trophic factors in mechano- and chemosensitivity.