Electrophysiological and morphological heterogeneity of rat dorsal vagal neurones which project to specific areas of the gastrointestinal tract

Abstract

1. The electrophysiological properties of rat dorsal motor nucleus of the vagus (DMV) neurones (n = 162) were examined using whole cell patch clamp recordings from brainstem slices. Recordings were made from DMV neurones whose projections to the gastrointestinal tract had been identified by previously applying fluorescent retrograde tracers to the gastric fundus, corpus or antrum/pylorus, or to the duodenum or caecum.

2. The neuronal groups were markedly heterogeneous with respect to several electrophysiological properties. For example, neurones which projected to the fundus had a higher input resistance (400 ± 25 MΩ), a smaller and shorter after-hyperpolarization (16.7 ± 0.49 mV and 63.5 ± 3.9 ms) and a higher frequency of action potential firing (19.3 ± 1.4 action potentials s⁻¹) following injection of depolarizing current (270 pA) when compared with caecum-projecting neurones (302 ± 22 MΩ; 23.5 ± 0.87 mV and 81.1 ± 5.3 ms; 9.7 ± 1.1 action potentials s⁻¹; P < 0.05 for each parameter). Differences between neuronal groups were also apparent with respect to the distribution of several voltage-dependent potassium currents. Inward rectification was present only in caecum-projecting neurones, for example.

3. Neurones (n = 82) were filled with the intracellular stain Neurobiotin allowing post-fixation morphological reconstruction. Neurones projecting to the caecum had the largest cell volume (5238 ± 535 μm³), soma area (489 ± 46 μm²) and soma diameter (24.6 ± 1.24 μm) as well as the largest number of dendritic branch segments (23 ± 2).

4. In summary, these results suggest that DMV neurones are heterogeneous with respect to some electrophysiological as well as some morphological properties and can be divided into subgroups according to their gastrointestinal projections.

The vagus nerve is the source of parasympathetic innervation to the gastrointestinal (GI) tract and plays a major role in the regulation of GI motility as well as gastric acid and pancreatic endocrine/exocrine secretion. The parasympathetic motor innervation to the major portion of the GI tract is provided by neurones in the dorsal motor nucleus of the vagus (DMV) (Gillis et al. 1989). Retrograde tracing experiments have shown the existence of a viscerotopic medio-lateral columnar organization that spans the entire rostro-caudal extent of the rat DMV (Fox & Powley, 1985; Shapiro & Miselis, 1985; Norgren & Smith, 1988). The medial portion of the DMV contains the somata of neurones which project to stomach and proximal intestine, while the more lateral tips of the DMV contain the somata of preganglionic neurones projecting to the distal intestine (Altschuler et al. 1991; Berthoud et al. 1991).

Numerous studies have reported diverse morphological features of rat DMV neurones, mainly in their dendritic extent (Rinaman et al. 1989; Rinaman & Miselis, 1990; Fox & Powley, 1992) but also in their somata size or shape (Zhang et al. 1995; Fogel et al. 1996; Jarvinen & Powley, 1999). Fogel et al. (1996) have recently provided strong evidence for the existence of a relationship between structure and function in the DMV. These authors used in vivo techniques to demonstrate that DMV neurones responsive to gastric or intestinal distension can be classified into four separate morphological groups. A feature of these subgroups is that the extracellular recorded action potentials in all the groups have a different shape (Fogel et al. 1996), which would suggest the possibility of an underlying different arrangement of electrical membrane properties within DMV neurones.

Such diversity in electrophysiological properties is also supported by previous reports of a large array of unevenly distributed membrane currents in DMV neurones (Travagli et al. 1992; Sah & McLachlan, 1992b, 1993; Travagli & Gillis, 1994, 1995; Dean et al. 1997; Bertolino et al. 1997). Travagli and colleagues, among others, have shown that the distribution of the potassium inward rectifier (I_K(IR)), as well as the non-selective cationic current (I_H), is not uniform among DMV neurones (Travagli & Gillis, 1994).

A recent hypothesis proposed that a sensory-motor lattice underlies the control of vago-vagal reflexes, one of the corollaries being that the location of the DMV neurone determines the nature of their electrophysiological as well as morphological characteristics (Powley et al. 1992). Indeed, neurones containing the non-selective cationic current I_H are more likely to be located in the medial portion of the DMV, strongly resembling the distribution of gastric vagal motoneurones (Travagli & Gillis, 1994). Also, the extent and orientation of the DMV neuronal arbor was shown to vary along the rostro-caudal and medio-lateral boundaries of the nucleus (Fox & Powley, 1992). In addition, a recent analysis of DMV neurones correlated their response to stimulation of different regions of the GI tract with their soma-dendritic morphology and concluded that certain features of DMV architecture are indeed associated with specific functions (Fogel et al. 1996).

To this end, the electrophysiological and morphological characteristics of DMV neurones have not yet been linked to a particular motoneurone output. The purpose of this study therefore was to correlate the electrical and morphological properties of DMV neurones with their GI projections.

Preliminary accounts of these data have been presented previously (Travagli et al. 1998).

METHODS

Retrograde tracing

Sprague-Dawley rats (12 days old) of either sex were deeply anaesthetized with a 6% solution of 2-bromo-2-chloro-1,1,1-trifluoroethane (halothane) with air (400-600 ml min⁻¹). The depth of anaesthesia (foot pinch withdrawal reflex) was checked prior to beginning surgery. The abdominal and thoracic areas were shaved and cleaned with 70% ethanol, and a laparotomy performed. During surgery, the head of the rat was placed in a custom-made anaesthetic chamber through which halothane mixed with air was administered.

Crystals of the retrograde tracer DiI (Molecular Probes) were applied to one GI region per rat, either to the gastric fundus, corpus or antrum/pylorus, or to the duodenum (at the level of the bifurcation of the hepatic and pancreatico-duodenal arteries) or the caecum (at the level of the ileo-caecal junction). In order to confine the dye to the application site, the surgical area was embedded in a fast-hardening epoxy compound. The epoxy compound was allowed to dry (3-5 min) before the area was examined visually to ensure that the dye had remained affixed to the organ surface and not made contact with other tissues. The surgical area was then washed with warm sterile saline solution, the excess solution blotted with cotton tips, the wound sutured with 5-0 silk and the animal allowed to recover for 10-15 days. In 15 experiments the retrograde tracer DiO (Molecular Probes) was also applied to a region distinct from that which received application of DiI (DiI on fundus and DiO on corpus, n = 5; DiI on duodenum and DiO on caecum, n = 4; DiI on caecum and DiO on fundus, n = 3; DiI on antrum/pylorus and DiO on corpus, n = 3).

Electrophysiology

The method used for the tissue slice preparation has already been described (Travagli & Gillis, 1994). Briefly, rats were placed in a transparent anaesthetic chamber through which a solution of halothane bubbled with air was passed (see above). When a deep level of anaesthesia was induced (see above), the rat was killed by severing the major blood vessels in the chest, in accordance with the Henry Ford Health Sciences Centre veterinary guidelines. The brainstem was then removed and placed in oxygenated physiological saline at 4°C (see below). Using a vibratome, three to four horizontal slices (200 μm thick) containing the DMV were cut. The slices were incubated for at least 1 h in oxygenated physiological saline at 35 ± 1°C until use.

A single slice was then placed in a custom-made perfusion chamber (volume 500 μl), and kept in place by a nylon mesh. The chamber was maintained at 35°C by continual perfusion with warmed oxygenated physiological saline at a rate of 2.5 ml min⁻¹.

Retrogradely labelled DMV neurones were identified prior to electrophysiological recordings using a Nikon E600-FS microscope equipped with tetramethyl rhodamine isothiocyanate (TRITC) or fluorescein isothiocyanate (FITC) epifluorescent filters. Carbocyanine dyes (such as DiI or DiO) do not cause adverse effects with the brief illuminations used for neuronal identification (Honig & Hume, 1989; Mendelowitz & Kunze, 1991). Once a labelled cell was identified, the neurone's identity was confirmed under brightfield illumination using DIC (Nomarski) optics (see Fig. 1).

Figure 1. Selective labelling of DMV neurones in the horizontal plane.

A, brightfield photomicrograph at low magnification of a brainstem slice. B, fluorescence (TRITC) photomicrograph of the same field of view as in A. Note the medial location of the neurones labelled following application of DiI to the antrum/pylorus. C, brightfield photomicrograph (DIC optics) at higher magnification of the medial DMV. Arrows indicate the 2 intensely fluorescent DiI-filled DMV neurones. Note the presence of unlabelled DMV neurones in the same field. D, fluorescence (TRITC) photomicrograph of the same field of view as in C. Note the three intensely fluorescent neurones (two of which are marked by arrows in C; the neurone at the left of the field of view is out of focus and cannot be observed in C) and the surrounding weakly stained neurones. Scale bars in A and B represent 100 μm; scale bars in C and D represent 25 μm.

Whole cell recordings were performed with patch pipettes (5-8 MΩ) filled with potassium gluconate intracellular solution (see below) using a single electrode voltage clamp amplifier (Axoclamp 2B or Axopatch 1D, Axon Instruments). Recordings were performed only in cells that were unequivocally identified as labelled with DiI. Data were filtered at 2 kHz, digitized via a Digidata 1200C interface (Axon Instruments), acquired and stored on an IBM PC utilizing pCLAMP 6 software (Axon Instruments). Only those recordings having a series resistance (i.e. pipette + access resistance) < 15 MΩ were used. For a neuronal recording to be accepted for measurements, the membrane had to be stable at the holding potential, the action potential evoked following injection of DC current had to have an amplitude of at least 60 mV, and the membrane had to return to baseline at the end of the after-hyperpolarization. Data analysis was performed using pCLAMP 6 software.

Drugs and ion substitutions were applied to the bath via a series of manually operated valves. Results are expressed as means ±s.e.m. Intergroup comparisons were analysed with one-way ANOVA followed by the conservative Bonferroni test for individual post hoc comparisons, Student's paired t test or chi-squared (χ²) test. Significance was defined as P < 0.05.

Morphological reconstructions

At the end of the electrophysiological recordings, prior to removal of the pipette, Neurobiotin (2.5% w/v) was injected into the DMV neurone by passing positive current (0.4 nA, 600 ms on-1200 ms off) for 10 min through the patch pipette. In order to allow the proper morphological reconstruction (i.e. for the purpose of assigning the dendritic branches to the somata of origin) only one cell per side/per slice was recorded and labelled. Following injection of Neurobiotin, the pipette was retrieved from the cell, which was allowed to seal for 10-20 min before overnight fixation at 4°C in Zamboni's fixative (see below). The method used for the morphometric analysis of neurones has already been described (Fogel et al. 1996). Briefly, the slice was cleared of fixative in PBS-TX (see below) and kept at 4°C until incubation in avidin D-horseradish peroxidase (HRP) solution (see below) for 2 h. Following 15 min rinsing in PBS-TX and subsequent incubation for 15-20 min in a PBS solution containing diaminobenzidine, cobalt chloride and nickel sulfate (DAB solution, see below), the slice was incubated finally for 15 min in the presence of 3% H₂O₂. The slice was rinsed in PBS-TX (15 min) before plating onto a gelatin-coated coverslip and air-dried. Prior to microscopic analysis, the slice was dehydrated with graded concentrations of alcohol, cleared in xylene and mounted in Permount.

Three-dimensional reconstructions of individual Neurobiotin-labelled neurones, digitized at a final magnification of ×400, were made using the Eutectic Neuron Tracing System (Eutectic Electronics, Raleigh, NC, USA), a system which allows complete reconstruction of branching structures. The optical and physical compression of the slice that may have occurred was corrected by a subroutine of the Eutectic software which rescaled the section to 200 μm (the original thickness at time of sectioning).

Several morphological features were assessed, classified into categories related to overall size (e.g. total surface area, somatic volume), somatic shape and size (form factor, somatic cross-sectional area), dendritic branching (e.g. number of primary dendrites, total number of branches, highest branching order, average dendritic extent), and indices of connectivity such as spine density. (Form factor (FF) is a measure of circularity for which a value of 100 indicates a perfect circle and 0 indicates a line; FF = 4πa × 100/p², where a = soma area and p = the perimeter of the soma in the horizontal plane.) In order to accept the results of the morphological reconstructions, neurones had to have a medio-lateral and rostro-caudal branch extension of at least 200 μm, no major branches had to be severed during the initial cutting procedures (a severed branch is recognizable by the swelling-like retraction bulb) and the neuronal somata could not have been deformed by the process of retrieval of the patch pipette (this is recognizable by the presence of a short and thick ‘stump’ protruding from the side of the somata).

Solution composition

Krebs solution (mM): 126 NaCl, 25 NaHCO₃, 2.5 KCl, 1.2 MgCl₂, 2.4 CaCl₂, 1.2 NaH₂PO₄ and 11 dextrose, maintained at pH 7.4 by bubbling with 95% O₂-5% CO₂. Intracellular solution (potassium) (mM): 128 potassium gluconate, 10 KCl, 0.3 CaCl₂, 1 MgCl₂, 10 Hepes, 1 EGTA, 2 ATP, 0.25 GTP; adjusted to pH 7.35 with KOH. Zamboni's fixative: 1.6% (w/v) paraformaldehyde, 19 mM KH₂PO₄ and 100 mM Na₂HPO₄·7H₂O in 240 ml saturated picric acid-1600 ml H₂O; adjusted to pH 7.4 with HCl. PBS-TX (mM): 115 NaCl, 75 Na₂HPO₄·7H₂O, 7.5 KH₂PO₄ and 0.15% Triton X-100. Avidin D-HRP solution: 0.002% avidin D-HRP in PBS containing 1% Triton X-100. DAB solution: 0.05% DAB in PBS containing 0.5% gelatin supplemented with 0.025% CoCl₂ and 0.02% NiNH₄SO₄.

Drugs and chemicals

Neurobiotin and avidin D-HRP were purchased from Vector Laboratories Inc. (Burlingame, CA, USA); Permount was purchased from Fisher Scientific (Pittsburgh, PA, USA); DiI and DiO were purchased from Molecular Probes; all other chemicals were purchased from Sigma.

RESULTS

A total of sixteen parameters were assessed in 162 neurones recorded in whole cell configuration. Of these neurones, 82 were sufficiently well filled for morphological analysis to be performed and 22 neurones provided the full spectrum of electrophysiological and morphological parameters.

We will first describe the electrical membrane properties and then the morphological properties of identified DMV neurones.

Neuronal identification

Whole cell recordings were obtained from 43 neurones labelled following placement of DiI on the fundus, 32 from the corpus, 30 from the antrum/pylorus, 30 from the duodenum, and 27 from the caecum.

Figure 1 shows a typical brainstem slice observed under brightfield (A and C) or fluorescent TRITC filters (B and D). Typically, two to ten unequivocally labelled neurones could be identified in each slice among the more numerous unlabelled neurones. In the fifteen experiments in which we placed both DiI and DiO on separate regions of the GI tract, double labelling was never observed (i.e. no neurones were showing the presence of both DiI and DiO), thus arguing in favour of the selectivity of our tracing technique. Recordings were limited to cells showing the brightest and most intense stain, i.e. the cells whose terminal fields were the most likely to be on the site of dye apposition.

Basic membrane properties

Table 1 is a summary of the electrical properties of identified DMV neurones.

Table 1.

Overview of electrophysiological properties

	Fundus (n = 25–42)	Corpus (n = 28–32)	Antrum/pylorus (n = 24–30)	Duodenum (n = 21–30)	Caecum (n = 16–27)
Input resistance (MΩ)	400 ± 25*C,A/P,Ca	299 ± 19*F	291 ± 23*F	330 ± 28	302 ± 22*F
Action potential duration (ms)	3.29 ± 0.11	3.29 ± 0.18	2.89 ± 0.13*Ca	2.78 ± 0.12*Ca	3.53 ± 0.13*A/P,D
AHP amplitude (mV)	16.7 ± 0.49*D,Ca	17.4 ± 0.63*D,Ca	17.2 ± 0.77*D,Ca	22.1 ± 0.72*F,C	23.4 ± 0.87*F,C,A/P
AHP duration (T ms)	63.6 ± 3.93*Ca	64.3 ± 4.25*Ca	49.5 ± 2.76*D,Ca	77.4 ± 5.05*A/P	81.5 ± 5.34*F,C,A/P
No. of action potentials s−1 at 30 pA DC	2.50 ± 0.41	3.35 ± 0.34	3.39 ± 0.34	3.12 ± 0.37	2.50 ± 0.33
No. of action potentials s−1 at 270 pA DC	19.3 ± 1.4*D,Ca	15.6 ± 1.1*Ca	19.9 ± 1.6*D,Ca	12.7 ± 1.1*F,A/P	9.7 ± 1.1*F,C,A/P
IM (% presence)	35%*C,D,Ca	81%*F	55%*D	92%*F,A/P,Ca	67%*F,D
IH (% presence)	47%*C,A/P	75%*F	72%*F	79%*	65%
IK(IR) (% presence)	2.5%*Ca	3.1%*Ca	0%*Ca	0%*Ca	77%*F,C,A/P,D
Islow (% presence)	32%*C,Ca	66%*F,A/P	31%*C,Ca	48%	60%*F,A/P

^*The mean difference is significant at the 0.05 level vs. fundus (F), corpus (C), antrum/pylorus (A/P), duodenum (D) or caecum (Ca). Since not all the cells were used for all the measures, the range of n values have been quoted. For example, some cells received robust synaptic input and the presence of spontaneous excitatory and inhibitory synaptic potentials prevented accurate measurement of the AHP_τ. Furthermore, in some neurones, a second action potential fired before baseline voltage was recovered, again preventing the correct measurement of the AHP_τ. The values quoted for voltage dependent potassium currents refer to the proportion of neurones which displayed that current. The other neurones either did not exhibit that current, or its amplitude was below that of our detection system (10 pA).

The vast majority of DMV neurones recorded in vitro were spontaneously active (143 out of 162, i.e. 88%); the mean firing rate was not different among the neuronal groups (0.96 ± 0.23 spikes s⁻¹).

The input resistance (R_imp) of the neurones was calculated by measuring the instantaneous current displacement (at the end of the stimulus artifact) obtained by stepping the membrane from -50 to -60 mV. As can be noticed from Table 1, the R_imp of fundic neurones (400 ± 25.3 MΩ, range 348-452 MΩ, n = 28) was significantly higher than that of other stomach-projecting neurones (299 ± 19.5 MΩ, range 259-339 MΩ, n = 28 for corpal neurones; 291 ± 23.4 MΩ, range 242-339 MΩ, n = 24 for antral/pyloric neurones) or caecum-projecting neurones (302 ± 22.7 MΩ, range 254-350 MΩ, n = 16; ANOVA F = 3.90, P = 0.005; Bonferroni P < 0.05). No significant differences were observed among the other groups.

Single action potential and after-hyperpolarization properties

Depolarizing current pulses, 5-30 ms long, of intensity sufficient to evoke a single action potential at its offset were injected into DMV neurones held at -55 mV. Differences were not observed in either the action potential threshold (42.1 ± 0.26 mV, n = 159) or amplitude (70.6 ± 0.78 mV, n = 159). Conversely, at threshold, the duration of the action potential of caecum neurones (3.53 ± 0.13 ms, range 3.24-3.81 ms, n = 25) was significantly longer than that of antrum/pylorus neurones (2.89 ± 0.13 ms, range 2.61-3.16 ms, n = 30) and duodenum neurones (2.78 ± 0.12 ms, range 2.53-3.03 ms, n = 29; ANOVA F = 4.505, P = 0.002; Bonferroni P < 0.05). The shape of the action potential and the ensuing membrane after-hyperpolarization (AHP) remained constant within a 30-40 min period after initiation of whole cell recording.

The prominent AHP that occurred at the termination of the action potential in stomach-projecting neurones was significantly smaller and shorter than in intestine-projecting neurones. The mean amplitude of the AHP was similar for stomach-projecting neurones (fundus: 16.7 ± 0.49 mV, range 15.7-17.7 mV, n = 42; corpus: 17.4 ± 0.63 mV, range 16.1-18.6 mV, n = 29; antrum/pylorus: 17.2 ± 0.77 mV, range 15.6-18.7 mV, n = 29). Conversely, the mean amplitude of the AHP was 22.1 ± 0.72 mV (range 20.7- 23.6 mV, n = 29) and 23.4 ± 0.87 (range 21.7-25.2 mV, n = 27) for duodenum- or caecum-projecting neurones, respectively (AHP amplitude: ANOVA F = 20.770, P < 0.001; Bonferroni P < 0.05).

The constant of decay (τ) of the AHP was fitted by a single exponential equation. Again, the stomach-projecting neurones showed a faster decay for the AHP compared with intestine-projecting neurones, i.e. duodenum- and caecum-projecting neurones. Specifically, the τ was 63.6 ± 3.93 ms (range 55.7-71.6 ms, n = 33), 64.3 ± 4.25 ms (range 55.6- 72.9 ms, n = 32) and 49.5 ± 2.76 ms (range 43.8-55.2 ms, n = 28) for fundus, corpus and antrum/pylorus neurones, respectively. All of these values are significantly lower than those obtained from caecum-projecting neurones (mean 81.5 ± 5.34 ms, range 70.1-92.2 ms; n = 23; ANOVA F = 8.002, P < 0.001; Bonferroni P < 0.05) with the antrum/pylorus-projecting neurones additionally differing significantly from duodenum-projecting neurones (mean 77.4 ± 5.05 ms, range 67.0-87.9 ms, n = 25). Representative traces are depicted in Fig. 2A.

Figure 2. DMV neurones show heterogeneity in the after-hyperpolarization.

A, single action potentials were evoked in neurones projecting to the corpus, antrum/pylorus or caecum. Neurones were current clamped at a holding potential of -55 mV before injection of a short (30 ms) depolarizing current pulse of intensity sufficient to evoke a single action potential at the offset of the current pulse. Note that while the amplitude and duration of the action potentials were similar in all three neurones, the caecum-projecting neurone had a much larger and longer-lasting after-hyperpolarization. B, effects of apamin and charybdotoxin on the amplitude and duration of the after-hyperpolarization following the firing of a single action potential. Action potentials were evoked as described above. Charybdotoxin (50 nM; left panel) did not have any effect on either the amplitude or duration of the after-hyperpolarization. Apamin (100 nM; right panel) did not significantly affect the duration of the action potential, but reduced both the amplitude and duration of the after-hyperpolarization.

Given that apamin-insensitive but TEA- and charybdotoxin-sensitive calcium-dependent potassium currents have been suggested to contribute to the action potential repolarization of rat DMV neurones (Sah & McLachlan, 1992b), we used apamin (100 nM), TEA (0.4 mM) and charybdotoxin (CbTX; 40 nM) to investigate the contribution of calcium-dependent potassium currents to the action potential repolarization of DMV neurones.

Data obtained from the neurones investigated were qualitatively similar regardless of their target projection and have thus been pooled. Perfusion with TEA increased the duration of the action potential (168 ± 16% of control, n = 5; paired t test P < 0.05) but affected neither the amplitude of the AHP (99.8 ± 6% of control, n = 5; n.s.) nor its decay (113 ± 14% of control, n = 5; n.s.) (data not shown). CbTX did not affect the duration of the action potential (130 ± 22% of control, n = 4; n.s.), the amplitude of the AHP (95.5 ± 5% of control, n = 4; n.s.) or its decay (84 ± 19% of control, n = 4; n.s.; see Fig. 2B). Apamin did not affect the duration of the action potential (131 ± 14% of control, n = 6; n.s.); however, it decreased the amplitude of the AHP (65.5 ± 6% of control, n = 6; P < 0.05) as well as its rate of decay (47.3 ± 7% of control, n = 6; P < 0.05) (Fig. 2B).

Since the absolute amplitude of the AHP varied dramatically within the neuronal groups, we investigated the firing rate of DMV neurones following 400 ms-long depolarizations at different frequencies. DMV neurones were held at -55 mV and step depolarized every 5 s by injection of 30-270 pA of current in 60 pA increments. Again, stomach-projecting neurones, i.e. the ones with smaller and shorter AHP, displayed faster firing rates at higher frequencies of stimulation than intestine-projecting neurones, i.e. the ones with larger and longer AHP (Bonferroni P < 0.05; Fig. 3 and Table 1).

Figure 3. DMV neurones show diversity in their frequency-response curves.

A, representative traces showing repetitive action potentials evoked following injection of a 400 ms long current pulse (270 pA). Note the faster frequency of firing in the neurone projecting to the antrum/pylorus (upper panel) compared with the caecum-projecting neurone (lower panel). The full extent of the action potential is not represented due to software-sampling limitations. B, frequency-response curves for DMV neurones projecting to various regions of the gastrointestinal tract. Note that neurones projecting to the duodenum and caecum fire fewer action potentials at higher frequencies of stimulation than neurones projecting to the gastric fundus, corpus or antrum/pylorus. Holding potential = -55 mV. Asterisks indicate statistically significant differences vs. stomach-projecting neurones.

Fast transient outward potassium current, I_A

The inactivation curve for I_A was constructed using hyperpolarizing steps (400 ms duration) from a holding potential of -50 mV in 10 mV increments to -120 mV and repolarization to -50 mV. The resultant values were normalized (I_max= 1), averaged (Fig. 4C) and plotted to construct the inactivation curve from which we calculated the activation potential, the I_max for inactivation removal and the half-inactivation potential (V_½) (Travagli et al. 1992). We also measured the kinetics of decay of I_A at -90 mV by fitting the trace with a single exponential equation. Perfusion with the antagonist 4-aminopyridine (4-AP; 5 mM) abolished I_A (n = 3). I_A was present in all of the 146 cells investigated. While no significant differences were found with regard to the maximal amplitude of the I_A, differences were apparent in the V_½ and in the kinetics of decay. In detail, the V_½ was 78 ± 2 mV (n = 13), 75 ± 6 mV (n = 4), and 77 ± 3 mV (n = 18) for fundus-, corpus- and antrum/pylorus-projecting neurones, respectively, compared with 71 ± 2 mV (n = 12) for caecum-projecting neurones (Bonferroni P < 0.05). Conversely, the caecum-projecting neurones had a slower decay constant (measured at -90 mV) compared with corpus- and antrum/pylorus-projecting neurones (290 ± 39 ms compared with 116 ± 19 ms and 170 ± 24 ms, respectively) (ANOVA F = 3.695, P = 0.011; Bonferroni P < 0.05; Fig. 4).

Figure 4. DMV neurones have different types of voltage activated potassium currents.

A, representative traces showing the presence of a fast transient potassium current (I_A) in caecum- and corpus-projecting neurones. Note the slower decay of I_A in the caecum-projecting neurone. The holding potential was -50 mV and the membrane potential was stepped in -10 mV increments to -120 mV before being step-depolarized back to -50 mV. I_A was seen as the outward current elicited upon restoration of the holding potential to -50 mV. The traces at hyperpolarized potentials have been erased. B, representative traces showing two different types of voltage activated potassium current elicited by the same voltage protocol as in A. Perfusion with 4-AP (5 mM) abolished I_A leaving a slower component unaffected. C, plot of the average inactivation curves for I_A in the different DMV neuronal subtypes. Note that the half-inactivation potential (V_½) is different in the neuronal subgroups. Asterisks indicate statistically significant differences vs. stomach-projecting neurones.

Using the same protocol as for the characterization of the I_A inactivation, a second type of voltage activated current was observed (I_slow; Fig. 4B). I_slow was 4-AP insensitive, had a very slow constant of decay and was activated upon membrane repolarization from hyperpolarized potentials. Interestingly, the distribution of I_slow differed significantly among the neuronal groups, in fact corpus- and caecum-projecting neurones had a significantly higher incidence of I_slow (21 out of 32 neurones and 12 out of 20 neurones, respectively) when compared with fundus- and antrum/pylorus-projecting neurones (13 out of 40 and 9 out of 29, respectively; χ² test P < 0.05; Table 1).

Other voltage-dependent currents, I_H, I_K(IR) and I_M

The voltage protocols used for identification of I_K(IR) and I_H were identical: neurones were voltage clamped at -50 mV and 10 mV incremental steps (1 s duration) were performed every 10 s to reach the final V_hold of -120 mV. These currents have already been described and characterized thoroughly in DMV neurones (Travagli & Gillis, 1994), and thus their ionic characterization will not be discussed further except to say that they were antagonized by extracellular perfusion with barium (1 mM; I_K(IR)n = 5) or caesium (2 mM; I_K(IR) and I_H, n = 5 and n = 3, respectively). To study I_M, neurones were voltage clamped at -40 mV and 10 mV incremental steps (1 s duration) were performed every 10 s to reach the final V_hold of -110 mV. I_M had a reversal potential close to our experimental potassium reversal potential (E_K) and was antagonized by extracellular perfusion with barium (1 mM; n = 5).

Interestingly, the distribution of I_K(IR), I_H and I_M differed significantly among the neuronal groups. In detail, caecum-projecting neurones were the only neuronal subgroup showing the presence of I_K(IR) (16 out of 21; χ² test P < 0.05vs. all the other groups; in the fundus- and corpus-projecting neurones 1 cell out of 40 and 1 cell out of 32 showed the presence of I_K(IR), respectively; Table 1). Regarding I_H, although the peak amplitude did not differ among the neuronal groups, it is significantly less prevalent in fundus-projecting neurones (15 out of 32 cells) when compared with any other neuronal groups (24 out of 32, 21 out of 29, 19 out of 24 and 13 out of 20 cells for corpus-, antrum/pylorus-, duodenum- and caecum-projecting neurones, respectively; χ² test P < 0.05; Table 1). As with I_H, measurements of I_M did not reveal significant differences in the amplitude measured upon stepping the membrane from -40 mV to -50 mV. On the other hand, the presence of I_M was significantly lower in fundus-projecting neurones (14 out of 40 cells) when compared with corpus-, duodenum or caecum-projecting neurones (26 out of 32, 22 out of 24, 14 out of 21 cells, respectively; χ² test P < 0.05) and significantly higher in duodenum-projecting neurones when compared with either fundus-, caecum- or antrum/pylorus-projecting neurones (χ² test P < 0.05; Table 1).

Morphology

The morphological analysis was based on 82 out of 159 cells that were chosen using the criteria outlined earlier (see Methods). Of these neurones, 18 neurones were labelled following placement of DiI on the fundus, 19 from the corpus, 14 from the antrum/pylorus, 19 from the duodenum, and 12 from the caecum. A summary of the morphological data for each neuronal subgroup is shown in Table 2.

Table 2.

Overview of morphological properties

	Fundus (n = 16–18)	Corpus (n = 16–19)	Antrum/pylorus (n = 12–14)	Duodenum (n = 15–19)	Caecum (n = 11–12)
Soma volume (μm3)	2944 ± 369*D,Ca	3450 ± 354*Ca	3208 ± 431*Ca	4669 ± 407*F	5238 ± 535*F,C,A/P
Soma area (μm2)	264 ± 19*C,D,Ca	385 ± 29*F	330 ± 21*Ca	364 ± 23*F,Ca	489 ± 46*F,A/P,D
Soma diameter (μm)	18.3 ± 0.64*C,D,Ca	21.8 ± 0.85*F	20.3 ± 0.66*Ca	21.3 ± 0.71*F	24.6 ± 1.24*F,A/P
Form factor	74 ± 4	80 ± 2	81 ± 2	80 ± 2	84 ± 2
No. of segments	17.1 ± 1.3*D,Ca	21.5 ± 1.0	18.0 ± 1.7	23.7 ± 1.8*F	23.5 ± 2.3*F
Branch order	4.4 ± 0.4	5.0 ± 0.3	4.7 ± 0.3	5.2 ± 1.3	4.4 ± 0.4
Dendritic x plane (μm)	317 ± 40	419 ± 44	404 ± 33	434 ± 36	417 ± 45
Dendritic y plane (μm)	302 ± 33	344 ± 42	302 ± 39	312 ± 45	383 ± 49
Segment length (μm)	113 ± 8	114 ± 5	106 ± 9	116 ± 8	117 ± 10

^*The mean difference is significant at the 0.05 level vs. fundus (F), corpus (C), antrum/pylorus (A/P), duodenum (D) or caecum (Ca). Since not all the cells were used for all the measures, the range of n values have been quoted. For example, although cells with obviously truncated dendritic trees were excluded from this study, some cells had extensive, but incomplete, dendritic arbors. For such neurones, values related to the number of segments, branch order and length of dendritic plane were excluded.

Figure 5A shows the location of the DMV neurones utilized for the morphological study. The relative medio-lateral position occupied by the neuronal subgroups in this study corresponds to that already reported (Altschuler et al. 1991; Berthoud et al. 1991), i.e. stomach-projecting neurones were in medial columns and intestine-projecting neurones were in lateral portions of the DMV, providing further support for the selectivity of our labelling technique. Figure 5B shows computer reconstructions of representative neurones from the five neuronal populations investigated.

Figure 5. Localization and morphology of identified DMV neurones.

A, location of DMV neurones in the horizontal plane. Note that stomach-projecting neurones (filled symbols) are located in more medial positions when compared with intestine-projecting neurones (open symbols). B, computer-aided reconstruction of five representative neurones projecting to the various targets. Note that the DMV neurone projecting to the fundus has a smaller soma diameter when compared with the other neurones depicted. See Table 2.

Differences in morphology were observed for two cellular features, soma size and the number of dendritic branch segments. We found that the DMV neurones that projected to the fundus had a smaller soma area (264 ± 19 μm², horizontal plane) than the DMV neurones that projected to the corpus (385 ± 29 μm²), duodenum (364 ± 23 μm²) and caecum (489 ± 46 μm²; ANOVA F = 7.962, P < 0.001; Bonferroni P < 0.05 for all comparisons). Conversely, the neurones that project to the caecum had the largest somata (24.62 ± 1.24 μm) in the data set compared with the fundus (18.3 ± 0.64 μm) and antrum/pylorus (20.3 ± 0.66 μm; ANOVA F = 7.017, P < 0.001; Bonferroni P < 0.05). The soma diameter generally reflected the differences noted in the soma area comparisons, with the exception that the difference between the soma diameter of the neurones projecting to the duodenum and the neurones projecting to the caecum was not significant. The soma volume comparisons (which the Eutectic system derives from the soma diameter and area measures) also produced results that were similar to the soma area observations, with minor exceptions.

The second major morphological feature that was distinctive within our sample of DMV neurones was the number of dendritic branch segments. Our results demonstrated that the neurones that projected to the fundus had fewer dendritic branches (17.1 ± 1.3) than the neurones that projected to the duodenum (23.7 ± 1.8) and caecum (23.5 ± 2.3; ANOVA F = 3.484, P = 0.12; Bonferroni P < 0.05).

Correlation of membrane properties and morphology

In order to determine whether the differences in electrophysiological properties measured between the neuronal subgroups were determined as a consequence of neuronal size, we examined fundus- and caecum-projecting neurones whose cell sizes were the most similar to each other, that is to say, the largest fundus- and smallest caecum-projecting neurones. Fundic and caecal neurones were chosen because they were the most different from each other as well as the farthest apart in the GI axis. In these selected cells, soma diameter and area as well as total cell volume were 21.5 ± 0.5 and 22.7 ± 0.6 μm, 363 ± 18 and 407 ± 21 μm², 4132 ± 922 and 4456 ± 481 μm³ for fundus- and caecum-projecting neurones, respectively (n = 5; grouped t test n.s.). Conversely, significant differences were still observed in the amplitude of the AHP (16.2 ± 0.5 and 24.1 ± 1.4 mV for fundus- and caecum-projecting neurones, respectively; n = 5; P < 0.05), in the frequency response to DC current injection (20.0 ± 0.3 and 7.5 ± 0.7 spikes s⁻¹ for fundus- and caecum-projecting neurones, respectively; n = 5; P < 0.05) and in the R_imp (390 ± 10 and 358 ± 14 MΩ for fundus- and caecum-projecting neurones, respectively; n = 5; P < 0.05). Hence, it would appear that the differences observed were independent of neuronal size but reflected intrinsic differences in membrane properties.

However, differences in neuronal size among the subgroups are apparent when one considers each entire subpopulation. In order to investigate whether there were correlations between electrophysiological and morphological parameters, we pooled twenty-two neurones that provided the full array of all the parameters measured (both electrophysiological and morphological). Figure 6 shows 3-D scatter plots that demonstrate a linear correlation between the frequency of firing upon injection of 270 pA DC current, the kinetics of I_A decay at -90 mV (right panel) or the amplitude of the AHP (left panel) and the soma area. These parameters were chosen because they were generated with independent protocols. The frequency of firing, the soma area and the AHP amplitude were inversely correlated for all 22 neurones. The parameters were fitted by the equation y = 18.2+ (-0.01z) + (-0.49x), where y = frequency of firing, x = AHP amplitude and z = soma area; r²= 0.65. The frequency of firing, the soma area, and the kinetics of I_A decay were also inversely correlated for all 22 neurones. The parameters were fitted by the equation y = 17.8+ (-0.02z) + (-0.02x), where y = frequency of firing, x = kinetics of I_A decay and z = soma area; r² = 0.62.

Figure 6. 3-D scatter plot correlating physiological with morphological properties.

A, 3-D scatter plot that relates the frequency of firing following injection of 270 pA DC current, amplitude of the after-hyperpolarization (AHP) and the somatic area. r²= 0.65. B, 3-D scatter plot that relates the frequency of firing (following 270 pA DC current injection), kinetics of I_A decay at -90 mV (I_Aτ) and the somatic area. r²= 0.62. Twenty-two cells have been pooled as for their stomach or intestine projections (filled symbols and open symbols, respectively). The correlation coefficient relates to the pooled data.

DISCUSSION

In this study we have shown that the DMV is composed of heterogeneous neuronal subpopulations which can be distinguished based on their electrophysiological and morphological properties; these characteristics can be correlated to the neuronal GI projections. While using the current clamp configuration, DMV neurones can be rather easily identified as projecting to either the stomach (fundus, corpus or antrum/pylorus) or to the intestine (duodenum or caecum) by the characteristic shape and duration of the AHP. Conversely, when using the voltage clamp configuration, apart from the presence of I_K(IR) and the slower decay of I_A, there are no electrophysiological properties characteristic of the different neuronal groups that would indicate the gastrointestinal area they innervate.

Electrophysiological differences in DMV neuronal subgroups

Among the diverse families of voltage-dependent currents, potassium currents such as the I_K(IR), the I_A, the I_M, the I_K(Ca)s (which comprise I_C and I_AHP), and the non-selective cationic hyperpolarization-activated current (I_H) are the ones that help set the resting potential, repolarize the cell, hyperpolarize the cell and shape action potentials. All of these currents are present in various subpopulations of DMV neurones.

I_H is present in all the neuronal subpopulations although fundus-projecting neurones have the lowest incidence. Although I_H has been identified in various neuronal systems as contributing to pacemaker activity (DiFrancesco, 1991), we have previously shown that this is not the case in the DMV (Travagli & Gillis, 1994). We speculate that the role of I_H in rat DMV neurones could be to decrease or limit the efficacy of inhibitory inputs such as 5-hydroxytryptamine (5-HT) or noradrenaline which utilize potassium as a current carrier (Fukuda et al. 1987; Wang et al. 1997).

I_M is found in all the neuronal subpopulations although the incidence is lowest in fundus- and antral/pyloric-projecting neurones. Activation of muscarinic receptors by acetylcholine causes a depolarization of DMV neurones (Travagli et al. 1992), which is likely to be due to inhibition of I_M. Hence, it would be predicted that fundic and antral/pyloric neurones would be less open to modulation by acetylcholine than other GI-projecting DMV neurones.

We have shown that a current (I_slow) with previously described characteristics (Sah & McLachlan, 1992a) is present in our DMV neuronal subpopulation. Because I_slow requires hyperpolarization of the membrane below -70 mV for its inactivation to be removed, it cannot play a role at normal resting membrane potentials. If, however, a cell is hyperpolarized following either the release of inhibitory neurotransmitters or the action potential AHP, I_slow would be activated upon the ensuing depolarization and slow the subsequent frequency of firing. The incidence of I_slow is found to be significantly higher in caecum-projecting neurones where the prominent AHP reaches potentials at which the inactivation of I_slow would be removed. The role of I_slow might be to limit the firing frequency; indeed, caecal neurones have a slower frequency response than gastric neurones.

We have shown that all DMV neurones have a fast transient outward current identified as I_A by its voltage dependency and sensitivity to 4-AP. Though no significant differences are observed among the neuronal subgroups in the absolute amplitude of I_A, its kinetics of decay and half-activation potential differed. In fact, intestine-projecting neurones show the slowest kinetics of decay and the least negative half-inactivation potential. I_A acts to increase the interspike interval, thus grading smoothly the rate of action potential firing, and its inactivation is removed following membrane hyperpolarization. We speculate that the inhibition that follows the release of inhibitory neurotransmitters or action potential AHP would have a more dramatic and longer lasting effect on DMV intestinal neurones because of the more readily achieved removal of I_A inactivation and its longer duration.

We have observed that the potassium inward rectifier (I_K(IR)) is present primarily in caecum-projecting neurones. I_K(IR) helps to stabilize the membrane potential favouring the entry of potassium ions and inward (depolarizing) current during membrane hyperpolarization thus modulating the degree of hyperpolarization in response to inhibitory stimuli.

To summarize our electrophysiological data, caecal neurones are the only ones that show the presence of I_K(IR), have the largest and slowest AHP, and are more likely to have I_slow. It is not surprising then that they have the slowest frequency response. On the other hand, corpal neurones have a relatively fast I_A and very high incidence of I_H while antral/pyloric neurones have a very short action potential and high occurrence of I_H making these neurones more likely to respond to depolarization with a faster frequency response. Duodenal neurones, instead, combine two opposite characteristics, a large and slow AHP (which would reduce their excitability) and a very large I_H (which would increase their excitability/reduce their inhibition). The final result is a frequency response intermediate between the fast gastric and the slow caecal neurones. The distinctive characteristic of fundic neurones is their elevated input resistance. Though no evident current can be associated with this high input resistance, it is interesting that these neurones show the lowest probability of having I_M and thus are the ones more likely to be modulated by neurotransmitters other than acetylcholine.

Although thorough investigations of the mechanism of action of putative neurotransmitters have not been conducted in DMV, recent studies have reported diverse responses to their exogenous applications (reviewed by Krowicki & Hornby, 1995). As recently pointed out by Llinas (1988, 1990), the intrinsic electrophysiological properties of cells govern the dynamics of neuronal networks. In fact, even if the synaptic connectivity and the type of neurotransmitter outputs are identical, neurones with different complements of voltage-dependent currents will respond differently to the same challenge. For example, both in vivo and in vitro studies have shown that in DMV responses to thyrotrophin-releasing hormone are not uniform (Travagli et al. 1992) and might be related to the neuronal function, i.e. secretion vs. motility (McCann et al. 1989; Chi et al. 1996). The response to pharmacological stimuli of DMV neurones is even more complex when 5-HT is analysed. Studies conducted by various groups have shown that discrete subsets of DMV neurones show diverse responses to exogenous application of 5-HT through actions at different receptor subtypes (Travagli & Gillis, 1995; Yoneda & Tache, 1995; Albert et al. 1996; Wang et al. 1996, 1997, 1998). It is proposed that such non-uniformity of response of DMV neurones to application of exogenous neurotransmitters may reflect differences in their complement of voltage-dependent currents as well as receptor binding sites.

Morphological differences in DMV neuronal subgroups

Previous reports have indicated that GI-projecting neurones exhibit a viscerotopic pattern of localization within the DMV (Fox & Powley, 1985; Shapiro & Miselis, 1985; Norgren & Smith, 1988; Berthoud et al. 1991; Altschuler et al. 1991, 1993). It has also been previously observed that the size of rat DMV neurones is non-uniform, with the largest neurones being located in the lateral DMV and the smallest in the medial DMV (Fox & Powley, 1992). We confirmed the selectivity of our labelling technique by the observation that intestine-projecting neurones, located in the lateral third of the DMV, are significantly larger than the stomach-projecting neurones located in the medial 2/3 of the DMV (see Fig. 5A and Table 2). Morphological diversity was reported both in rat and human DMV (Fox & Powley, 1992; Huang et al. 1993; Fogel et al. 1996; Jarvinen & Powley, 1999). Both Fogel et al. (1996) and Jarvinen & Powley (1999) reported four morphologically distinct groups of neurones in the rat; although it is difficult to compare absolute values between studies due to different planes of section, fixation techniques and measurement strategies, we have also observed distinctive morphological characteristics in the subgroups we analysed. Huang et al. (1993) identified six types of morphologically different neurones in the human DMV, subdivided in nine subgroups based on their relative location. Our data in the rat DMV seem to support the hypothesis formulated by Huang et al. that neuronal DMV subgroups may form functional units innervating specific organs.

Morphological and electrophysiological correlations

Whilst the observation of morphological differences in GI-projecting DMV neurones may not provide information with respect to function, we have shown that they have a counterpart in the diverse electrophysiological properties of the neuronal subgroups. For example, correlations can be made between the soma area (or soma diameter or volume), the frequency of firing and the AHP amplitude, as well as the kinetics of decay of I_A of both stomach- and intestine-projecting neurones (Fig. 6). These electrophysiological parameters were chosen because they are not exclusively dependent upon each other. Hence, our results suggest that such morphological and electrophysiological differences in neuronal groups may reflect underlying functional differences in the GI-projecting DMV neurones as well as an extension of the ‘size principle’ (Henneman et al. 1965).

The size principle, which is widely accepted for motor neurone activation and recruitment, states that neurones with the smallest cell bodies have the lowest threshold for synaptic activation and so can be activated by weaker synaptic inputs. As described above, stomach-projecting DMV neurones, particularly fundic neurones, are generally smaller and have a higher input resistance than intestine-projecting neurones. Thus, these neurones should respond to a greater range of synaptic inputs than the larger intestine-projecting neurones. Additionally, the larger frequency-responses of gastric neurones should permit the encoding and transmission of synaptic information over a much wider input range. The resting membrane potential of gastric smooth muscle is non-uniform with a gradient from approximately -48 mV in the proximal stomach to -70 mV in the distal stomach (Hunt, 1983). Since the threshold for contraction is around -50 mV (Szurszewski, 1987), the proximal stomach is in a constant state of partial tone or contraction, and hence relatively small changes in membrane potential can have dramatic effects on muscle tone. In fact, as seen by the effects of vagotomy on GI motor function, extrinsic innervation exerts a greater degree of influence over tonic gastric (particularly fundic) function than intestinal function (Hall et al. 1982; Meyer, 1987).

In conclusion, despite being at present unable to distinguish between DMV neurones involved in secretion/absorption or motility, the differences in morphology and firing frequency we have observed in the present study could reflect the different requirements of stomach-projecting neurones for a rapid yet finely tuned modulation of contractile activity.

In summary, we have shown that the rat DMV comprises heterogeneous populations of neurones which can be grouped according to their GI projections and classified based on their electrophysiological and morphological properties. Based on these results, we propose that the DMV is a nucleus capable of achieving a detailed modulation of parasympathetic preganglionic information via its differing properties rather than being a simple relay nucleus that transmits unmodulated motor information to the GI tract.