THE JOURNAL OF COMPARATIVE NEUROLOGY 416:451–460 (2000) Origins of Cholinergic Inputs to the Cell Bodies of Intestinofugal Neurons in the Guinea Pig Distal Colon ALAN E. LOMAX,* JONATHAN Y. ZHANG, AND JOHN B. FURNESS Department of Anatomy and Cell Biology, University of Melbourne, Parkville 3052, Victoria, Australia ABSTRACT Integration of function between gut regions is mediated by means of hormones and long neuronal reflex pathways. Intestinofugal neurons, which participate in one of these pathways, have cell bodies within the myenteric plexus and project their axons from the gut with the mesenteric nerves. They form excitatory synapses on neurons in prevertebral ganglia that in turn innervate other gut regions. The aim of the present study was to characterise immunohistochemically the synaptic input to intestinofugal neurons. The cell bodies of intestinofugal neurons that project from the distal colon were labelled with Fast Blue that was injected into the inferior mesenteric ganglia. Varicosities surrounding Fast Blue-labelled neurons were analysed for immunoreactivity for the vesicular acetylcholine transporter, vasoactive intestinal peptide, and bombesin. Most intestinofugal neurons were surrounded by nerve terminals immunoreactive for the vesicular acetylcholine transporter; many of these terminals also contained vasoactive intestinal peptide and bombesin immunoreactivity. This combination of markers occurs in axons of descending interneurons. Extrinsic denervation had no effect on the distribution of cholinergic terminals around intestinofugal neurons. A decrease in the number of vesicular acetylcholine transporter and vasoactive intestinal peptide immunoreactive terminals occurred around nerve cells immediately anal, but not oral, to myotomy operations. Consistent with previous physiological studies, it is concluded that intestinofugal neurons receive cholinergic synaptic input from other myenteric neurons, including cholinergic descending interneurons. Thus, intestinofugal neurons are second, or higher, order neurons in reflex pathways, although physiological data indicate that they also respond directly to distension of the gut wall. J. Comp. Neurol. 416:451–460, 2000. r 2000 Wiley-Liss, Inc. Indexing terms: prevertebral ganglia; enteric nervous system; immunohistochemistry; retrograde tracing Physiological and anatomic studies in the 1940s revealed unusual reflexes that bypass the central nervous system (Kuntz and van Buskirk, 1941; Kuntz and Saccomanno, 1944). These reflexes originate in the alimentary canal and pass back to the same or more proximal regions of the digestive tract by way of the abdominal prevertebral ganglia. The reflexes can be elicited by distension of the colon and by distension or chemical stimuli applied intraluminally in the small intestine (for review see Szurszewski and Miller, 1994). Early studies demonstrated that these intestinointestinal reflexes persist after decentralisation of the prevertebral ganglia, although the threshold for activating the reflexes is increased by decentralisation (Kuntz and Saccomanno, 1944; Semba, 1954; Shapiro and Woodward, 1959). r 2000 WILEY-LISS, INC. Since then, electrophysiological studies have shown that the cell bodies of sympathetic postganglionic neurons in prevertebral ganglia receive a prominent excitatory nicotinic synaptic input from axons arising in the viscera (Crowcroft et al., 1971; Szurszewski and Weems, 1976; Kreulen and Szurszewski, 1979). Retrograde tracing studies demonstrated that the peripheral inputs to the prever- Grant sponsor: National Health and Medical Research Council of Australia; Grant number: 963213. *Correspondence to: Alan Lomax, Department of Anatomy and Cell Biology, University of Melbourne, Parkville 3052, Victoria, Australia. E-mail: [email protected] Received 28 May 1999; Revised 29 September 1999; Accepted 29 September 1999 452 A.E. LOMAX ET AL. tebral ganglia originate from neurons with cell bodies in the myenteric plexus (intestinofugal neurons) of all regions of the gastrointestinal tract (Kuramoto and Furness, 1989; Furness et al., 1990; Messenger and Furness, 1991, 1992, 1993; Parr et al., 1993; Mann et al., 1995). The extent to which intestinofugal neurons are directly responsive to distension of the colon or are stimulated indirectly after activation of colonic intrinsic primary afferent neurons (IPANs) is uncertain. Many physiological studies have addressed this question without achieving a complete answer (Crowcroft et al., 1971; Weems and Szurszewski, 1978; Bywater, 1993; Parkman et al., 1993; Stebbing and Bornstein, 1993; Miller and Szurszewski, 1997). For these studies, preparations of intestine and connected prevertebral ganglia were set up in divided organ baths. By separately superfusing the segment of gut or the ganglia, comparisons were made of the responses of intestinofugal neurons to distension of the wall of the gut before and after synaptic transmission in the gut segment was blocked by nicotinic antagonists or low calciumcontaining saline. These studies suggested that (1) intestinofugal neurons receive synaptic input from other enteric neurons that are excited by distension, and (2) intestinofugal neurons also detect distension of the gut. Thus, it appears that some or all intestinofugal neurons receive synaptic input from other enteric neurons, and some or all intestinofugal neurons are also directly distension sensitive. It is possible that there are two populations of intestinofugal neurons, one that is directly sensitive to distension and one that is activated by inputs from other enteric neurons (Szurszewski and Miller, 1994; Sharkey et al., 1998). Sharkey et al. (1998) directly investigated whether intestinofugal neurons receive synaptic input from other enteric neurons by recording intracellularly from retrogradely labelled intestinofugal neurons that project from the distal colon to the inferior mesenteric ganglia (IMG) of the guinea pig. Fast excitatory postsynaptic potentials (fast EPSPs) were recorded in all intestinofugal neurons in response to stimulation of internodal strands and some intestinofugal neurons received on-going spontaneous fast EPSPs. The morphologies of the intestinofugal neurons were revealed by intracellular injection of biocytin by means of the recording electrode. None of the neurons from which recordings were made had Dogiel type II morphology, which is the morphology of the IPANs of the guinea pig small intestine, nor did intestinofugal neurons exhibit the electrophysiological behaviour characteristic of the IPANs that respond to stretch of the small intestine of the guinea pig (Kunze et al., 1998, 1999). In the present study, we have used a combination of retrograde neuronal tracing with Fast Blue, immunohistochemistry, and surgical interruption of nerve pathways to identify sources of synaptic input to intestinofugal neurons of the distal colon. MATERIALS AND METHODS Retrograde labelling of intestinofugal neurons All experimental procedures were in accordance with the guidelines for animal welfare set out by the National Health and Medical Research Council of Australia. Experiments were performed on 17 male guinea pigs, each weighing between 200 and 325g, from the University of Melbourne colony. The procedure for injection of the retro- grade neuronal tracer, Fast Blue, was identical for all guinea pigs and is summarised below. Each animal was anaesthetized by subcutaneous injection of sodium pentobarbitone (15 mg/kg; Boehringer-Ingelheim, New South Wales, Australia) 1 hour before the operation and an intramuscular injection of a mixture of fentanyl citrate (0.16 mg/kg) and droperidol (8.0 mg/kg) 15 minutes before the operation. The abdomen was opened by a midline incision and the IMG was exposed by exteriorising and retracting the small and large intestines. Fast Blue (Sigma Chemical Co., St Louis, MO) was dissolved in 10% dimethylsulphoxide (DMSO) in distilled water at 2% weight/ volume and injected into both lobes of the IMG by using a bevelled glass micropipette of 60–80 µm. An average of two injections into each lobe (rostral and caudal) was used (total volume, 1–3 µl). After injection of the IMG, and return of the intestines to the body cavity, the peritoneum and abdominal muscles were sutured and covered with an antibiotic powder, cicatrin (Wellcome, New South Wales, Australia) and the skin was stapled together. Each animal received a 0.1 ml intramuscular injection of terramycin (50 mg/ml; Pfizer Agricare, New South Wales, Australia), a broad spectrum antibiotic. Surgical interruption of nerve pathways Seven to 12 days after Fast Blue was injected into the IMG of 10 animals, operations were carried out to surgically ablate specific nerve pathways. In four animals, two myotomies, separated by 15–70 mm, were performed. In this procedure, the animals were anaesthetised and opened in the same manner as before and a pair of parallel circumferential cuts were made through the external musculature of the distal colon, thus severing the nerve pathways in the myenteric plexus (Messenger and Furness, 1990). A loose loop of surgical thread was tied around the nearest mesenteric blood vessel to mark each operation site. Extrinsic denervations were performed on four animals. The lumbar colonic nerves were located and severed as they emerged from the IMG, denervating the distal colon of sympathetic input and of those sensory fibers that follow the sympathetic nerves. The surrounding mesentery was also cut and, to ensure that no fibers in close proximity to the inferior mesenteric artery remained intact, the artery was lightly swabbed with cotton wool soaked in 80% phenol. The site of the operation was marked by a small piece of surgical thread to enable identification in dissection. The two remaining animals underwent combined double myotomy plus extrinsic denervation operations. After these operations, the animals were left for a further 6–10 days to allow degeneration of the axons that were severed. Once enough time was allowed for the Fast Blue to fill intestinofugal neurons as completely as possible (7–12 days) and the axonal ablations to take effect (a further 6–10 days), animals were stunned by a blow to the back of the head, then killed by severing the carotid arteries and the spinal cord. In the animals without myotomies, a 5- to 8-cm-long segment of the distal colon adjacent to the IMG was removed as a control. Segments of colon were immersed in phosphate-buffered saline (PBS; 0.9% NaCl in 0.01 M sodium phosphate buffer, pH 7.2) containing the muscle relaxant nicardipine (3 µM, Sigma). The segment of colon was then cut along the antimesenteric border, as most of the intestinofugal neurons lie close to the mesen- SYNAPTIC INPUT TO INTESTINOFUGAL NEURONS TABLE 1. Characteristics of Primary Antibodies1 453 TABLE 2. Secondary Antibodies or Streptavidin Complexes Used1 Tissue antigen Host Dilution Code and reference Antibody or streptavidin label Dilution Source Calbindin Calretinin VAChT Bombesin 5-HT (serotonin) Neural NOS NPY Somatostatin TK (substance P) TH VIP Rabbit Rabbit Goat Mouse Rat Rabbit Rabbit Mouse Rat Mouse Rabbit 1:1600 1:1000 1:1000 1:100 1:500 1:200 1:1600 1:400 1:200 1:400 1:200 DEMLR 1, Furness et al., 1988 7696, Mann et al., 1997 1624, Li and Furness, 1998 2AII, Costa et al., 1984 3M55, Wardell et al., 1994 N74, Anderson et al., 1995 JM263, Maccarrone and Jarrot, 1985 S895, Buchan et al., 1985 NCI/34HL, Cuello et al., 1979 T341, Boehringer Mannhein Furness et al., 1981 Biotinylated donkey anti-rabbit IgG Biotinylated donkey anti-sheep IgG Biotinylated horse anti-mouse IgG Biotinylated horse anti-rabbit IgG Donkey anti-rabbit IgG FITC Donkey anti-sheep IgG FITC Donkey anti-rat IgG FITC Streptavidin-Texas Red 1:100 1:100 1:100 1:100 1:50 1:50 1:100 1:100 Jackson Jackson Vector Jackson Amersham Jackson Jackson Amersham 1Supply companies: Amersham Pty, Ltd., Melbourne, Australia; Jackson Immunosearch Laboratories, West Grove, PA; Vector Lab., Burlingame, CA. IgG, immunoglobulin G; FITC, fluorescein isothiocyanate. 1VAChT, vesicular acetylcholine transporter; NOS, nitric oxide synthase; NPY, neuropeptide Y; TH, tyrosine hydroxylase; VIP, vasoactive intestinal peptide. teric attachment (Kuramoto and Furness, 1989; Messenger and Furness, 1993), and fecal pellets were removed. To visualise intestinofugal neurons better, the mesentery was carefully removed and the tissue pinned tautly on balsa board. The tissue was then fixed by immersion in 2% formaldehyde plus 0.2% picric acid in 0.1 M sodium phosphate buffer (pH 7.0) at 4°C overnight and subsequently cleared in DMSO (3 ⫻ 10 minute changes) followed by PBS (3 ⫻ 10 minute changes). The mucosa, submucosa, and circular muscle were removed, leaving myenteric plexus and longitudinal muscle whole-mount preparations. Immunohistochemistry Whole-mount preparations of longitudinal muscle with attached myenteric plexus were incubated for 30 minutes in 10% normal horse serum in 1% Triton X-100 in PBS, to limit background staining, before exposure to primary antisera. Excess serum was then removed, and the preparations were incubated in mixtures of primary antibodies for 2 nights at room temperature in a humidified chamber (Table 1). Cell body staining was also carried out on whole-mount preparations of longitudinal muscle plus myenteric plexus of guinea pig distal colon that had been incubated in culture medium containing colchicine under sterile conditions for 24 hours before fixation (Messenger and Furness, 1990). After incubation at room temperature in combinations of primary antisera, tissue was washed in PBS and then incubated for 2 hours in a mixture of secondary antibodies (see Table 2) comprising one antibody linked to biotin and one directly labelled with fluorescein isothiocyanate (FITC). The tissue was then washed for 30 minutes in PBS and incubated with streptavidin-Texas Red for 90 minutes. A final wash in PBS was made before tissue was mounted in glycerol buffered with 0.5 M sodium carbonate buffer (pH 8.6). Preparations were examined on a Zeiss Axioplan fluorescence microscope (Carl Zeiss, Oberkochen, Germany) equipped with the appropriate filter cubes for discriminating between Fast Blue, FITC, and Texas Red fluorescence. Preparations were scanned carefully by using the filter cube to reveal Fast Blue fluorescence. Each successively located Fast Blue-labelled neuron was examined by switching the filter cube to reveal fluorescence for one of the immunohistochemical markers of synaptic terminals within the myenteric plexus. The fluorescence observed for these immunohistochemical markers, along with the Fast Blue fluorescence of the intestinofugal neurons, was captured by using a Sensys 12-bit cooled CCD camera (Photometrics, Ltd., Tucson AZ) and V for Windows imaging software (Digital Optics, Ltd., Auckland, New Zealand). Contrast and sensitivity adjustment were carried out by using Corel Photopaint imaging software (Corel Corporation, Dublin, Ireland). Some preparations were examined by using a confocal scanning laser microscope system (BioRad MRC 1024 attached to a Zeiss Axioplan fluorescence microscope). The laser was an argon/krypton mixed gas laser with excitation and emission wavelengths 448 and 522/535 nm, respectively, for FITC and 568 and 605/632 nm for Texas Red. Samples were scanned sequentially to collect light emitted as red or green fluorescence by using a 100⫻ oil immersion objective and ⫻1.4 zoom. Optical sections of 0.5 µm thickness were taken in each case. Images of 768 ⫻ 512 pixels were obtained and processed by using Confocal Assistant and Corel Photopaint. Retrogradely labelled neurons were evaluated under the microscope and in the stored images, to determine whether they were surrounded by pericellular baskets of immunohistochemically labelled nerve terminals. To do this, we followed the criteria of Pompolo and Furness (1995) for the identification of baskets. Pericellular baskets consist of closely spaced axonal varicosities that are juxtaposed to cell bodies. They are at a greater density around innervated cells than in the neuropil, or in relation to uninnervated cells, and ultrastructural examination shows that varicosities in baskets form synapses on the cell bodies that they surround (Pompolo and Furness 1995; Mann et al., 1997). RESULTS Injection of Fast Blue into the IMG resulted in labelling of neuronal perikarya in the myenteric plexus of the distal colon. The staining of these neurons was usually restricted to the cytoplasm, although occasional nuclear staining was also observed. The shapes of neurons were not as well defined as in previous studies, in which tissue was generally taken after 4–10 days. In the present work, 13–22 days elapsed, so that there was time for retrograde accumulation before nerves were severed, and then time for degeneration of the severed fibers. In this time, much of the Fast Blue relocated to lysosomes away from the cell surface and dendrites (e.g., Fig. 1). Retrogradely labelled nerve cells were distributed around the circumference of the colon, although they were more numerous at the mesenteric aspect of the gut, as previously reported (Furness et al., 1990). Neurons whose shapes were revealed fell into two reasonably distinct morphologic groups. The first group had many short lamellar dendrites, whereas the second group had one or more lamellar dendrites but also had up to five longer dendritic processes that had distinct irregularities on their surfaces. Both cell types correspond 454 Fig. 1. Fluorescence photomicrographs showing the immunoreactivities of nerve terminals that surround the cell bodies of intestinofugal neurons that were retrogradely labelled with Fast Blue. A,A8: Images of the same field illuminated to reveal Fast Blue (A) and vesicular acetylcholine transporter (VAChT) immunoreactive (-IR) (A8). The image in A is the cell body of an intestinofugal neuron that is surrounded by a basket-like arrangement of VAChT-IR nerve terminals (image in A8). Most of the Fast Blue is in lysosomes in this neuron. B,B8,B9: Another field illuminated to reveal Fast Blue (B), VAChT-IR A.E. LOMAX ET AL. (B8), and vasoactive intestinal peptide (VIP) -IR (B9). The intestinofugal neuron (B) is surrounded by nerve terminals that are immunoreactive for VAChT (B8) and VIP (B9). Arrows indicate where VAChT-IR and VIP-IR appear to colocalise within varicosities. C: An intestinofugal neuron that is surrounded by VIP-IR (C8) and bombesin-IR (BN; C9) terminals. Arrows indicate colocalisation of VIP-IR and bombesin-IR within varicosities. Scale bars ⫽ 10 µm in A,B (apply to A,A8,B,B8,B9), 20 µm in C (applies to C,C8,C9). SYNAPTIC INPUT TO INTESTINOFUGAL NEURONS to the class of neurons designated by Dogiel (1899) as type I (see Sharkey et al., 1998). Preparations of longitudinal muscle plus myenteric plexus containing Fast Blue labelled intestinofugal neurons were stained with a variety of immunohistochemical markers of nerve terminals of populations of myenteric neurons. Fast Blue-labelled nerve cell bodies were commonly surrounded by immunoreactive (IR) varicosities which outlined cell bodies, sometimes forming several layers of varicosities (Fig. 1A,A8). Such relationships between varicosities and cell bodies, described as pericellular baskets, have been examined by electron microscopy, and it had been found that varicosities within these baskets form synapses with the cell bodies (Pompolo and Furness, 1995). One hundred ninety-nine of 215 nerve cell bodies (93%) were surrounded by basket-like arrangements of axon varicosities that were immunoreactive for the vesicular acetylcholine transporter (VAChT), which is a marker of cholinergic nerve terminals (Fig. 1A,A8,B,B8). The cell bodies of intestinofugal neurons were also examined for their relation to vasoactive intestinal peptide (VIP)-IR and bombesin-IR terminals in preparations stained for VAChT plus VIP or bombesin plus VIP. One hundred sixty-nine of 212 intestinofugal neurons (80%) were surrounded by VIP-IR varicosities, whereas 77 of 88 intestinofugal neurons (88%) examined were surrounded by bombesin-IR terminals (Fig. 1B,B88,C,C88). These doublestaining experiments showed that terminals that were immunoreactive for each of these markers formed baskets around intestinofugal neurons and that these markers were commonly colocalised within individual varicosities (Fig. 1B8,B88,C8,C8). Varicose terminals with bombesin-IR were found around other nerve cells, but some nerve cells were not surrounded by bombesin-IR varicosities. Bombesin-IR varicosities surrounded most bombesin-IR nerve cells (Fig. 3A,B). Descending inputs Myotomy and extrinsic denervation operations were performed to identify the projection patterns of cholinergic neurons that innervate intestinofugal neurons. In preparations from the animals in which myotomy operations were performed, there was a decrease in the innervation of intestinofugal neurons in the 10 mm on the anal sides of the lesions and quantitative data presented below on the effects of myotomy operations was obtained from the first 10 mm anal or oral to a myotomy. There was a dramatic reduction in the number of intestinofugal neurons that were surrounded by VAChT-IR and VIP-IR varicosities just anal to a myotomy operation (i.e., where the axons of neurons with descending (oral to anal) projections had been severed and allowed to degenerate). In untreated animals, 93% of intestinofugal neurons examined were surrounded by VAChT-IR terminals (see above), whereas after a myotomy operation performed oral to the area examined 37 of 141 intestinofugal neurons (26%) had VAChT innervation that was similar to controls; a further 40 of 141 intestinofugal neurons (28%) had few VAChT-IR terminals close to them, insufficient to be recognised as a pericellular basket (Fig. 2B,B8), and the remaining 64 neurons (46%) had no VAChT-IR terminals close to the cell bodies. In areas just anal to a myotomy, 10 of 87 (12%) intestinofugal neurons had normal innervation by VIP-IR terminals; a further 18 of 87 (20%) had reduced innervation and 59 of 87 (68%) had no close VIP-IR terminals (Fig. 455 2C,C8). The innervation of intestinofugal neurons by VIP-IR varicosities returned to normal about 10 mm anal to the myotomy operation site. This finding is consistent with the findings of Messenger and Furness (1990), who reported that VIP fibers in myenteric ganglia in the guinea pig colon were reduced by oral myotomy operations but that normal VIP innervation of myenteric ganglia was observed within approximately 8 mm anal to the site of the myotomy. Ascending inputs The effects of myotomy operations that interrupt ascending fibers within the myenteric plexus on the innervation of intestinofugal neurons by VAChT-IR terminals were examined. In the 10 mm oral to myotomies, 51 of 60 intestinofugal neurons (85%) had normal VAChT innervation, suggesting that ascending pathways do not contribute significantly to the innervation of intestinofugal neurons by cholinergic terminals (Fig. 2A,A8). Extrinsic inputs Extrinsic denervation operations were performed to determine the contribution of extrinsic neurons to the innervation of intestinofugal neurons, and immunohistochemistry for tyrosine hydroxylase (TH) was used to monitor the effectiveness of the denervation. In preparations from unoperated animals, TH-IR fibers formed a dense network within myenteric ganglia, whereas after extrinsic denervation, only rare TH-IR fibers could be seen, consistent with previous findings (Furness, 1970). In segments of colon that had been extrinsically denervated, 36 of 43 intestinofugal neurons (84%) received normal VAChT innervation. In the two animals that had undergone both extrinsic denervation and myotomy operations, there was a reduction in VAChT innervation anal to the myotomies that recovered at about 1 cm anal as described above, but further anal between the myotomy operations no reduction was observed. These results, along with the finding of only infrequent innervation of intestinofugal neurons by TH-IR fibers in untreated preparations (5 of 36; 14%, Table 3) suggest that the innervation of intestinofugal neurons originates predominantly within the gut wall. Because VAChT, VIP, and bombesin were localised in many terminals that surrounded intestinofugal neurons, we investigated whether there were myenteric nerve cell bodies with the appropriate neurochemistry to be a source of the terminals. Staining of myenteric neurons of the guinea pig distal colon that had been pretreated with colchicine to enhance cell body immunoreactivity for peptides revealed that choline acetyltransferase (ChAT), VIP, bombesin, and nitric oxide synthase (NOS) immunoreactivities are colocalised in a subset of nerve cells (Fig. 3). Many of these nerve cells were surrounded by bombesin-IR terminals. Sixty-six of 100 bombesin-IR myenteric nerve cell bodies were ChAT-IR, 50 of 56 bombesin-IR were VIP-IR, and 78 of 79 were NOS-IR. We also examined whether intestinofugal neurons had terminals surrounding them that were immunoreactive for neurochemical markers of subclasses of myenteric neurons other than VAChT, VIP, and bombesin (Table 3). No other markers stained terminals that surrounded such a large proportion of intestinofugal neurons. A complication that made accurate counting of the number of intestinofugal neurons that were surrounded by varicosities immunoreactive for a particular substance was that the 456 A.E. LOMAX ET AL. Fig. 2. Paired photomicrographs that show the effects of myotomy operations on the distribution of labelled nerve terminals around intestinofugal neurons. A: An intestinofugal neuron in a region within 5 mm oral to a myotomy that caused degeneration of ascending fibers is surrounded by vesicular acetylcholine transporter (VAChT) –immunoreactive (-IR) terminals (A8). Note that the labelled varicosities appear to be closely aligned with the surface of the cell body and with one of its dendrites (arrow). B: An intestinofugal neuron from a region within 5 mm anal to a myotomy that severed descending axons has a marked reduction in the number of VAChT-IR terminals that approach the cell body (B8). No labelled terminals can be seen to contact the cell body or its processes. C: Another intestinofugal neuron from a region where descending axons have been severed. In this case, there is an obvious reduction in the amount of vasoactive intestinal peptide (VIP) -IR terminals in the ganglion (C8, compare with Fig. 1B9 and Fig. 1C8) and no immunoreactive terminals can be seen close to the cell body. Scale bars ⫽ 15 µm in A–C (apply to A–C8). cell bodies of intestinofugal neurons are immunoreactive for NOS, bombesin, VIP, calretinin, calbindin, and ChAT (Furness et al., 1990; Mann et al., 1995). This was not a problem with bombesin and VIP, because cell body staining of myenteric neurons with these markers is weak without colchicine treatment, whereas cell body staining of intestinofugal neurons with calbindin, calretinin, and NOS is intense and precludes accurate assessment of whether cell bodies of intestinofugal neurons are surrounded by terminals immunoreactive for these markers. To determine the proportion of intestinofugal neurons with inputs from calbindin-, calretinin-, or NOS-IR terminals, only retrogradely labelled nerve cell bodies that did not have the same immunohistochemistry were examined. Of SYNAPTIC INPUT TO INTESTINOFUGAL NEURONS TABLE 3. Incidence of Pericellular Baskets Markers used1 Number of intestinofugal neurons examined Number of intestinofugal neurons surrounded by terminals (%) 88 30 47 69 17 36 77 (88) 7 (23) 6 (13) 15 (23) 12 (70) 5 (14) 215 141 60 43 199 (93) 37 (26) 51 (85) 36 (84) 212 87 169 (80) 10 (12) Bombesin Calbindin Calretinin NOS Substance P TH VAChT Normal colon Anal to myotomy Oral to myotomy After extrinsic denervation VIP Normal colon Anal to myotomy 1For abbreviations, see footnote to Table 1. the intestinofugal neurons whose cell bodies were not immunoreactive for calretinin, calbindin or NOS, 6 of 47 (13%) were surrounded by calretinin-IR terminals, 7 of 30 (23%) by calbindin-IR terminals, and 15 of 69 (23%) by NOS-IR terminals respectively (Table 3). DISCUSSION Sources of innervation We have found that the cell bodies of intestinofugal neurons that project to the inferior mesenteric ganglia from the distal colon are surrounded by cholinergic nerve terminals. The possible sources of these terminals are intrinsic neurons, extrinsic neurons whose axons reach the colon by means of the mesenteric nerves, vagal axons that take an intramural course to reach the colon and branches of the pelvic nerves that run orally within the gut wall (intramural pelvic nerves; see Furness and Costa, 1987). None of the extrinsic sources seem to contribute significantly. When the mesenteric nerves were severed and their endings allowed to degenerate, there was no detectable loss of innervation of the cell bodies of intestinofugal neurons. Moreover, in the region between two circumferential myotomies, which would sever both vagal fibers descending within the gut wall and ascending fibers from the pelvic nerves, normal innervation persisted in ganglia toward the anal part of the neurally isolated segment. The appearance was the same if the two myotomies and the mesenteric nerve section were performed together. Thus, the pericellular fibers at the distal part of the segment between myotomies must arise from neurons with cell bodies within the neurally isolated segment, and, because the loss of nerve terminals occurred at the oral part of this segment only, the neurons must be anally projecting. However, not all fibers were lost from the oral region. The remaining fibers are, therefore, of local origin, either from cell bodies in myenteric ganglia or from cell bodies in submucosal ganglia. The lack of detectable loss of innervation oral to circumferential lesions of intramural nerve pathways indicates that ascending fibers do not contribute significantly to the innervation of the cell bodies of intestinofugal neurons. We have identified the class of cholinergic descending interneuron that makes a large contribution to the cholinergic innervation of intestinofugal neurons. VAChT, VIP, and bombesin all labelled nerve terminals that surround the majority of intestinofugal neurons. Moreover, by exam- 457 ining the cell bodies of myenteric neurons in preparations of distal colon that had been pretreated with colchicine, a class of descending interneurons was identified that was immunoreactive for ChAT, VIP, and bombesin. In studies on the myenteric plexus of guinea pig distal colon that used single label immunohistochemistry, descending interneurons with immunoreactivity for VIP, NOS, and bombesin have been identified (Messenger and Furness 1990; McConalogue and Furness, 1993; present results). Bombesin/VIP-IR terminals also surrounded many other nerve cells, including other bombesin-IR neurons, but there were cells without this innervation; thus, these descending interneurons evidently make specific connections with intestinofugal neurons, descending interneurons, and other classes of neurons, which, by analogy with the small intestine, probably include inhibitory motor neurons (Kunze and Furness, 1999). Relations of labelled nerve terminals to nerve cell bodies The inputs to the intestinofugal neurons have been examined by fluorescence microscopy. Dense baskets of cholinergic (VAChT-IR) varicosities were closely apposed to the nerve cells. Electrophysiological studies support the conclusion that the baskets of VAChT-IR varicosities represent true synapses, because fast EPSPs were recorded from the intestinofugal neurons, and, although this was tested in only a few cases, the fast EPSPs were blocked by an antagonist of cholinergic transmission, hexamethonium (Sharkey et al., 1998). Similar associations of baskets of terminals have been examined by electron microscopy, and in all cases, many of the varicosities were found to form true synapses (showing presynaptic accumulations of vesicles and synaptic densities), whereas some varicosities formed close contacts (closely apposed membranes, presynaptic accumulations of vesicles but no discernible synaptic densities). The presence of synapses within baskets of immunoreactive terminals in enteric ganglia has been confirmed by electron microscopy and includes calretinin-IR inputs to calretinin-IR interneurons (Pompolo and Furness, 1993), somatostatin-IR inputs to somatostatin-IR interneurons (Portbury et al., 1995; Pompolo and Furness, 1998), NOS-IR inputs to VIP-IR secretomotor neurons (Li et al., 1995), bombesin-IR inputs to calretinin-IR motor neurons (Pompolo and Furness, 1995), NOS-IR inputs to NOS-IR motor neurons (Young et al., 1995), 5-HT-IR inputs to 5-HT-IR interneurons (Young and Furness, 1995), and somatostatin-IR inputs to NOS-IR cell bodies (Mann et al., 1997). The numbers of varicosities seen in apposition to nerve cells with fluorescence immunohistochemical methods is likely to be an overestimate of the number of true synaptic inputs (Mann et al., 1997). This is because some of the varicosities are separated from the surfaces of nerve cells by glial cell processes. Physiological significance Bywater (1993) made the interesting observation that when synaptic transmission within the colon was blocked, the initial response of intestinofugal neurons to distension was not significantly reduced, whereas the response after 10–15 seconds of sustained distension was almost abolished. This finding suggests that reflexes that pass from the gut by means of intestinofugal neurons to the prevertebral ganglia have two components: one that is a direct 458 A.E. LOMAX ET AL. Fig. 3. Paired photomicrographs illustrating the cell body immunoreactivities of the class of descending interneuron that supplies terminals that surround the cell bodies of intestinofugal neurons within the myenteric plexus of guinea pig distal colon. A,A8,B,B8: Confocal microscope images, optical thickness 0.7 µm. C,C8: Photomicrographs acquired by using conventional fluorescence illumination. These descending interneurons are immunoreactive for bombesin (BN), vasoactive intestinal peptide (VIP), choline acetyltransferase (ChAT), and nitric oxide synthase (NOS). Bombesin immunoreactivity (A,B,C), colocalised with VIP-immunoreactive (-IR) (A8), ChAT-IR (B8), and NOS-IR (C8). These neurons receive bombesin-IR varicose terminals, as can be seen in A and B. Arrows indicate cell bodies where colocalisation occurs. Scale bars ⫽ 15 µm in A–C (apply to A–C8). response of intestinofugal neurons to distension and the other that is a response to synaptic input from other enteric neurons to intestinofugal neurons (i.e., intestinofugal neurons are indirectly activated by the distension stimulus). Szurszewski and Miller (1994) have suggested that intestinofugal neurons may form a homogenous population of neurons that are both directly and indirectly activated. Evidence in support of this theory was provided by Sharkey et al. (1998), who found that all intestinofugal neurons from which intracellular membrane potential recordings were taken had S-type electrophysiological characteristics and had Dogiel type I morphology; in other words, they appear to form a homogenous population. Experimental evidence also suggests that all intestinofugal neurons have the same primary transmitter, acetylcholine (Szurszweski and Miller, 1994; Mann et al., 1995; Sharkey et al., 1998). The results of the present study support the idea that intestinofugal neurons are a population that is homogenous in its most significant characteristics. Ninety-three percent of intestinofugal neurons were surrounded by basket-like arrangements of VAChT-IR varicosities. It is not possible to say that the other 7% are not innervated by cholinergic nerve terminals because the VAChT-IR terminals may not always form a distinct pericellular basket and may also contact dendrites of intestinofugal neurons that are not revealed by Fast Blue fluorescence. On the other hand, we found that about 25% of the intestinofugal neurons were surrounded by calbindin terminals. Calbindin is a marker of Dogiel type II neurons, which are presumed to be IPANs in the colon (Lomax et al., 1999) but is also contained in other types of neuron. Thus, it is possible that a minority of intestinofugal neurons receive direct inputs from IPANs. There is strong experimental evidence that intestinofugal neurons are length sensors (Szurszewski and Miller, 1994), whereas there is good evidence that the IPANs of the ileum respond to tension generated in the external muscle in response to stretch rather than the stretch per se, i.e., they are tension detectors (Kunze et al., 1998, 1999). These IPANs have Dogiel type II morphology and AH electrophysiological characteristics (Furness et al., 1998). A recent study has revealed the neurons in the myenteric plexus of the guinea pig distal colon that have similar morphologic, immunohistochemical, and electrophysiological characteristics to the IPANs of the ileum (Lomax et al., 1999). The observations of Bywater (1993) might be accounted for if intestinofugal neurons respond to changes in length directly, which would account for the SYNAPTIC INPUT TO INTESTINOFUGAL NEURONS initial response to distension that persists after blockade of synaptic transmission within the gut. The slowly developing increase in tension in response to this distension might excite the IPANs, but not the intestinofugal neurons directly. These IPANs then excite each other and provide input to the descending interneurons, which in turn provide fast synaptic input to the intestinofugal neurons. Convergence of fast excitatory postsynaptic potentials onto the intestinofugal neurons would cause action potential firing in the intestinofugal neurons in response to distension (by means of the increase in tension) that would be sensitive to synaptic blockade. Neurons with similar chemical coding (ChAT, VIP, bombesin, NOS) to the class of descending interneuron that provides cholinergic input to intestinofugal neurons in the distal colon have been examined in the guinea pig small intestine. Anatomic and pharmacologic studies indicate that this class of interneuron forms descending chains and that these neurons innervate the inhibitory motor neurons that relax the circular muscle (Young et al., 1995; Yuan et al., 1995; Kunze and Furness, 1999), i.e., they are conduits of local descending reflex pathways in the small intestine. It is likely that similar descending interneurons in the distal colon that innervate the intestinofugal neurons are also involved in local (intrinsic) reflex pathways in the colon because, as discussed above, they innervate other nerve cells in the myenteric ganglia. This raises the interesting possibility that the sustained response to distension that is observed in intestinofugal neurons is controlled by the same class of descending interneuron that is involved in local descending inhibitory reflexes in the gut. ACKNOWLEDGMENTS We thank Dr. Colin Anderson and Dr. Heather Young for expert advice on the manuscript and Heather Woodman and Clare Delaney for excellent technical assistance. LITERATURE CITED Anderson CR, Furness JB, Woodman HL, Edwards SL, Crack PJ, Smith AI. 1995. Characterisation of neurons with nitric oxide synthase immunoreactivity that project to prevertebral ganglia. J Auton Nerv Syst 52:107–116. Buchan AMJ, Sikora LKJ, Levy JG, McIntosh CHS, Dyck I, Brown JC. 1985. An immunocytochemical investigation with monoclonal antibodies to somatostatin. Histochemistry 83:175–180. Bywater RAR. 1993. Activity following colonic distension in enteric sensory fibers projecting to the inferior mesenteric ganglion in the guinea-pig. J Auton Nerv Syst 46:19–26. Costa M, Furness JB, Yanaihara N, Yanaihara C, Moody TW. 1984. Distribution and projections of neurons with immunoreactivity for both gastrin-releasing peptide and bombesin in the guinea-pig small intestine. Cell Tissue Res 235:285–293. Crowcroft PJ, Holman ME, Szurszewski JH. 1971. Excitatory input from the distal colon to the inferior mesenteric ganglion in the guinea-pig. J Physiol (Lond) 219:443–461. Cuello AC, Galfre G, Milstein C. 1979. Detection of substance P in the central nervous system by a monoclonal antibody. Proc Natl Acad Sci USA 76:3532–3536. Dogiel AS. 1899. Über den Bau der Ganglien in den Geflechten des Darmes und der Gallenblase des Menschen und der Säugetiere. Arch Anat Physiol (Leipzig) Anat Abt Jg 1899:130–158. Furness JB. 1970. The origin and distribution of adrenergic nerve fibers in the guinea-pig colon. Histochemie 21:295–306. Furness JB, Costa M. 1987. The enteric nervous system. Edinburgh: Churchill Livingstone. 459 Furness JB, Costa M, Walsh JH. 1981. Evidence for and significance of the projection of VIP neurons from the myenteric plexus to the Taenia coli in the guinea-pig. Gastroenterology 80:1557–1561. Furness JB, Keast JR, Pompolo S, Bornstein JC, Costa M, Emson PC, Lawson DEM. 1988. Immunohistochemical evidence for the presence of calcium binding proteins in enteric neurons. Cell Tissue Res 252:79–87. Furness JB, Kuramoto H, Messenger JP. 1990. Morphological and chemical identification of neurons that project from the colon to the inferior mesenteric ganglia in the guinea-pig. J Auton Nerv Syst 31:203–210. Furness JB, Kunze WAA, Bertrand PP, Clerc N, Bornstein JC. 1998. Intrinsic primary afferent neurons of the intestine. Prog Neurobiol 54:1–18. Kreulen DL, Szurszewski JH. 1979. Reflex pathways in the abdominal prevertebral ganglia: evidence for a colo-colonic inhibitory reflex. J Physiol (Lond) 295:21–32. Kuntz A, Saccomanno G. 1944. Reflex inhibition of intestinal motility mediated through decentralized prevertebral ganglia. J Neurophysiol 7:163–170. Kuntz A, van Buskirk C. 1941. Reflex inhibition of bile flow and intestinal motility mediated through decentralized celiac plexus. Proc Soc Exp Biol 46:519–523. Kunze WAA, Furness JB. 1999. The enteric nervous system and regulation of intestinal motility. Annu Rev Physiol 61:117–142. Kunze WAA, Furness JB, Bertrand PP, Bornstein JC. 1998. Intracellular recording from myenteric neurons of the guinea-pig ileum that respond to stretch. J Physiol (Lond) 506:827–842. Kunze WAA, Clerc N, Bertrand PP, Furness JB. 1999. Contractile activity in intestinal muscle evokes action potential discharge in guinea-pig myenteric neurons. J Physiol (Lond) 517:547–561. Kuramoto H, Furness JB. 1989. Distribution of nerve cells that project from the small intestine to the coeliac ganglion in the guinea-pig. J Auton Nerv Syst 27:241–248. Li ZS, Furness JB. 1998. Immunohistochemical localization of cholinergic markers in putative intrinsic primary afferent neurons of the guineapig small intestine. Cell Tissue Res 294:35–43. Li ZS, Young HM, Furness JB. 1995. Do VIP- and nitric oxide synthaseimmunoreactive terminals synapse exclusively with VIP cell bodies in the submucous plexus of the guinea-pig ileum? Cell Tissue Res 281:485– 491. Lomax AEG, Sharkey KA, Bertrand PP, Low AM, Bornstein JC, Furness JB. 1999. Correlation of morphology, electrophysiology and chemistry of neurons in the myenteric plexus of the guinea-pig distal colon. J Auton Nerv Syst 45: 45–61. Maccarrone C, Jarrott B. 1985. Differences in regional brain concentrations of neuropeptide Y in spontaneously hypertensive (SH) and Wistar Kyoto (WKY) rats. Brain Res 345:165–169. Mann PT, Furness JB, Pompolo S, Mäder M. 1995. Chemical coding of neurons that project from different regions of intestine to the coeliac ganglion of the guinea pig. J Auton Nerv Syst 56:15–25. Mann PT, Southwell BR, Young HM, Furness JB. 1997. Appositions made by axons of descending interneurons in the guinea-pig small intestine, investigated by confocal microscopy. J Chem Neuroanat 12:151–164. McConalogue K, Furness JB. 1993. Projections of nitric oxide synthesizing neurons in the guinea-pig colon. Cell Tissue Res 271:545–553. Messenger JP, Furness JB. 1990. Projections of chemically specified neurons in the guinea-pig colon. Arch Histol Cytol 53:467–495. Messenger JP, Furness JB. 1991. Calbindin-immunoreactive nerve terminals in the guinea pig coeliac ganglion originate from colonic nerve cells. J Auton Nerv Syst 35:133–142. Messenger JP, Furness JB. 1992. Distribution of enteric nerve cells that project to the coeliac ganglion of the guinea-pig. Cell Tissue Res 269:119–132. Messenger JP, Furness JB. 1993. Distribution of enteric nerve cells projecting to the superior and inferior mesenteric ganglia of the guinea-pig. Cell Tissue Res 271:333–339. Miller SM, Szurszewski JH. 1997. Colonic mechanosensory afferent input to neurons in the mouse superior mesenteric ganglion. Am J Physiol 272:G357–G366. Parkman HP, Ma WH, Stapelfeldt WH, Szurszewski JH. 1993. Direct and indirect mechanosensory pathways from the colon to the inferior mesenteric ganglion. Am J Physiol 265:G499–G505. Parr EJ, Davison SN, Davison JS, Sharkey KA. 1993. The origin and distribution of neurons with projections passing through the inferior mesenteric ganglion of the guinea-pig. J Auton Nerv Syst 44:91–99. 460 Pompolo S, Furness JB. 1993. Origins of synaptic inputs to calretinin immunoreactive neurons in the guinea-pig small intestine. J Neurocytol 22:531–546. Pompolo S, Furness JB. 1995. Sources of inputs to longitudinal muscle motor neurons and ascending interneurons in the guinea-pig small intestine. Cell Tissue Res 280:549–560. Pompolo S, Furness JB. 1998. Quantitative analysis of inputs to somatostatin immunoreactive descending interneurons in the myenteric plexus of the guinea-pig small intestine. Cell Tissue Res 294:219–226. Portbury AL, Pompolo S, Furness JB, Stebbing MJ, Kunze WAA, Bornstein JC, Hughes S. 1995. Cholinergic, somatostatin-immunoreactive interneurons in the guinea pig intestine: morphology, ultrastructure, connections and projections. J Anat 187:303–321. Semba T. 1954. Intestino-intestinal inhibitory reflexes. Jpn J Physiol 4:241–245. Shapiro H, Woodward ER. 1959. Pathway of enterogastric reflex. Proc Soc Exp Biol 101:407–409. Sharkey KA, Lomax AEG, Bertrand PP, Furness JB. 1998. Electrophysiology, shape and chemistry of intestinofugal neurons projecting from guinea pig distal colon to inferior mesenteric ganglia. Gastroenterology 115:909–918. Stebbing MJ, Bornstein JC. 1993. Electrophysiological analysis of the A.E. LOMAX ET AL. convergence of peripheral inputs onto neurons of the coeliac ganglion in the guinea-pig. J Auton Nerv Syst 46:93–105. Szurszewski JH, Miller SM. 1994. Physiology of prevertebral ganglia. In: Johnson LR, editor. Physiology of the gastrointestinal tract. New York: Raven Press. p 795–877. Szurszewski JH, Weems WA. 1976. A study of peripheral input to and its control by postganglionic neurones of the inferior mesenteric ganglion. J Physiol (Lond) 256:541–556. Wardell CF, Bornstein JC, Furness JB. 1994. Projections of 5-hydroxytryptamine-immunoreactive neurons in guinea-pig distal colon. Cell Tissue Res 278:379–387. Weems WA, Szurszewski JH. 1978. An intracellular analysis of some intrinsic factors controlling neural output from inferior mesenteric ganglion of guinea pigs. J Neurophysiol 41:305–321. Young HM, Furness JB. 1995. An ultrastructural examination of the targets of serotonin-immunoreactive descending interneurons in the guinea-pig small intestine. J Comp Neurol 356:101–114. Young HM, Furness JB, Povey JM. 1995. Analysis of connections between nitric oxide synthase neurons in the myenteric plexus of the guinea-pig small intestine. J Neurocytol 24:257–263. Yuan SY, Bornstein JC, Furness JB. 1995. Pharmacological evidence that nitric oxide may be a retrograde messenger in the enteric nervous system. Br J Pharmacol 114:428–432.