Afferent innervation of gastrointestinal tract smooth muscle by the hepatic branch of the vagusкод для вставкиСкачать
THE JOURNAL OF COMPARATIVE NEUROLOGY 384:248–270 (1997) Afferent Innervation of Gastrointestinal Tract Smooth Muscle by the Hepatic Branch of the Vagus ROBERT J. PHILLIPS, ELIZABETH A. BARONOWSKY, AND TERRY L. POWLEY* Purdue University, West Lafayette, Indiana 47907 ABSTRACT To survey the vagal hepatic branch afferent projections to and the terminal specializations in the gastrointestinal tract, male Sprague-Dawley rats were given subdiaphragmatic vagotomies, sparing only the common hepatic branch, and were injected with 3 µl of 8% wheat germ agglutinin-horseradish peroxidase in the left nodose ganglion. The nodose ganglia, the stomach, the first 8 cm of duodenum, and the cecum were prepared as wholemounts and were processed with tetramethyl benzidine. Hepatic afferent innervation of the ventral stomach consisted of one or more bundles entering at the lower esophageal sphincter and coursing to the forestomach, where they branched into distinct terminal fields. The only fibers on the dorsal forestomach were distal branches and terminals that wrapped around the greater curvature from the ventral side. Hepatic afferents supplied the forestomach with both intraganglionic laminar endings (IGLEs; putative mechanosensors that coordinate peristalsis) and intramuscular arrays (IMAs; considered tension receptors). IGLEs were located primarily on the ventral wall of the stomach, whereas IMAs were distributed symmetrically. Afferents were also supplied to the distal antrum and the pylorus, with pyloric innervation consisting almost exclusively of IMAs. Innervation of the proximal duodenum was denser in the first 3 cm and decreased progressively caudally, with only meager innervation after 6 cm. Cecal innervation consisted of a few fibers at the ileocecal junction. Duodenal and cecal endings were predominately IGLEs. These results indicate that the hepatic branch carries sensory information from the forestomach, antrum, pylorus, duodenum, and cecum. Furthermore, the different terminals it supplies suggest that the branch mediates a multiplicity of gastrointestinal functions. J. Comp. Neurol. 384:248–270, 1997. r 1997 Wiley-Liss, Inc. Indexing terms: duodenum; liver; stomach; vagotomy; visceral afferents The vagus nerve contains both sensory and motor axons that have been implicated in the control of gastrointestinal physiology and ingestive behavior. Peripherally, the vagi enter the abdomen as two trunks coursing along the esophagus. After passing through the diaphragm, these mixed trunks, the dorsal and ventral, divide into five distinct branches: paired gastric branches, paired celiac branches, and a single hepatic branch that originates from the ventral trunk (Boekelaar, 1985; Prechtl and Powley, 1985, 1987, 1990). Traditionally, the hepatic branch of the rat vagus was assumed to be an uncomplicated and coherent bundle projecting simply, as its name suggests, to the liver. Indeed, the hepatic branch has been shown to innervate the liver, but it has only recently been shown to have projections to the gastrointestinal (GI) tract (Berthoud et al., 1991, 1992; Prechtl and Powley, 1987). The hepatic branch ramifies from the ventral vagal trunk in one to three bundles of fascicles that travel close r 1997 WILEY-LISS, INC. together. The branch then runs between the esophagus and the hepatic artery proper in loose association with the hepatoesophageal artery. On this path, the bundles that comprise the branch split into a plexus of fascicles, a few of which course into the more distal lesser omentum (Boekelaar, 1985; Prechtl and Powley, 1987). Near the junction of the hepatoesophageal artery and the hepatic artery proper, the hepatic branch plexus forms into five to seven fascicles, which organize into two secondary divisions, one that turns toward the liver hilus (hepatic branch proper) and a second that turns to course with the common hepatic and Grant sponsor: NIH; Grant number: DK27627; Grant sponsor: NIMH; Grant number: MH01023. *Correspondence to: T.L. Powley, Department of Psychological Sciences, 165 Peirce Hall, Purdue University, West Lafayette, IN 47907. E-mail:firstname.lastname@example.org Received 12 December 1996; Revised 6 March 1997; Accepted 6 March 1997 VAGAL HEPATIC BRANCH AFFERENTS gastroduodenal arteries (gastroduodenal branch; Berthoud et al., 1992; Prechtl and Powley, 1987). Recognizing that two divisions separate from the classic hepatic branch and course to different targets, the classical ‘‘hepatic branch,’’ as it separates from the ventral vagal trunk and courses parallel to the hepatoesophageal artery, is now often more explicitly designated as the common hepatic branch1. On average, approximately two-thirds of the afferent fibers contained in the common hepatic branch bypass the liver hilus on their way to the gastroduodenal and common hepatic arterial plexuses (Berthoud et al., 1992). It has also been demonstrated that a portion of the vagal efferents contained in the common hepatic branch bypasses the liver hilus and innervates the gastric antrum, pylorus, duodenum, and pancreas (Berthoud et al., 1991; Berthoud and Powley, 1991). Although the findings of Berthoud et al. (1991) provide evidence for hepatic motor innervation of the GI tract, they may still underestimate the projection field of the common hepatic branch due to their sampling of efferents, which are only 7% of the fibers in the common hepatic branch (for fiber counts, see Prechtl and Powley, 1987). Furthermore, the findings of Berthoud and coworkers may also have underestimated the projection field, because the somata of the common hepatic branch efferents are widely distributed within the dorsal motor nucleus (Fox and Powley, 1985), and it is difficult to obtain complete injections in this highly elongated nucleus. Reinforcing the concern that the Berthoud et al. efferent study underestimated the extent, and perhaps the breadth, of the common hepatic branch(es) GI projection field, a recent survey of the spatial location of vagal afferent innervation of the GI tract (Powley and Wang, 1995; Wang, 1995; Wang and Powley, 1994) found that the left nodose ganglion (the source of common hepatic branch afferents) provided significantly more afferents to the duodenum than did the right nodose ganglion. This pattern led the authors to speculate that the difference was due to a heavy contribution from the common hepatic branch. In addition to the issue of distribution, another question concerns the types of afferent terminal specializations or endings the common hepatic branch of the vagus supplies to its different targets. Although terminal specializations or endings provided by hepatic branch afferents to the liver have been characterized (Berthoud et al., 1992, 1995; Prechtl and Powley, 1987), such information has not been available for the corresponding projections to the GI tract. The characterization of the hepatic branch distribution based on efferents does not speak to the question, of course, and previous experiments characterizing vagal afferent terminals in the GI tract (see, e.g., Berthoud and Powley, 1992; Berthoud et al., 1995; Kressel et al., 1994; Powley et al., 1994; Wang, 1995) have not distinguished hepatic branch afferents from other vagal afferents. These experiments, however, have distinguished two terminal specializations [intraganglionic laminar endings (IGLEs) and intramuscular arrays (IMAs)] in the smooth muscle wall of the GI tract, either or both of which could conceiv- 1In this report, the classic term hepatic branch and the more explicit expression common hepatic branch are used synonymously to refer to the initial segment of the branch that leaves the ventral trunk and courses with the hepatoesophageal artery. The more distal secondary divisions of this branch are consistently referred to as the hepatic branch proper and the gastroduodenal branch. 249 ably be found among hepatic branch afferents. The goals of the present study were 1) to describe both qualitatively and quantitatively the innervation pattern of the common hepatic branch of the vagus along the GI tract by using a protocol that would delineate this pattern in detail and 2) to provide a survey of the types of afferent endings within the projection field. MATERIALS AND METHODS Animals Male Sprague-Dawley rats (Harlan Industries, Indianapolis, IN; n 5 25) weighing 300–350 g (ranging from 13 to 15 weeks old) at the time of first surgery were housed individually and maintained on a 12:12 hour light:dark schedule at 23°C. Animals had ad libitum access to tap water and 17-hours-per-day (1700–1000 hours) access to Isocal (Mead Johnson, Evansville, IN), a nutritionally complete liquid diet (1.05 kcal/ml). All procedures were conducted in accordance with American Association for Accreditation of Laboratory Animal Care guidelines and were approved by the Purdue University Animal Care and Use Committee. Surgeries and tracers Vagotomy. Rats were anesthetized by using sodium pentobarbital (60 mg/kg, i.p.) and then treated with atropine (1.0 mg/kg, s.c.). Each animal was then laparotomized and selectively vagotomized, as previously described (Fox and Powley, 1985; Powley et al., 1987). To eliminate the vagal innervation of the GI tract, except for that supplied by the common hepatic branch of the vagus, all animals were given partial subdiaphragmatic vagotomies, consisting of a transection of 1) the posterior trunk above the point where it bifurcates into the gastric and celiac branches as well as both 2) the accessory celiac and 3) the anterior gastric branches below the point where the common hepatic branch separates from the ventral trunk. The appropriate branch was isolated by dissection, supported by a microdissecting hook, and cauterized with a hightemperature ophthalmic cautery; the proximal stump was then recauterized, effectively sealing the exposed nerve end. Fluoro-Gold injection. To verify the sparing of the common hepatic branch, we used a fluorescent tracer strategy that effectively retrogradely labels autonomic as well as other peripheral axon terminals that lack complete blood-nerve barriers (Fox and Powley, 1985; Powley et al., 1987). Eight days postvagotomy, each animal received an i.p. injection of 1 mg/1 ml saline of Fluoro-Gold (Fluorochrome, Inc., Englewood, CO) to label the cell bodies of intact efferent fibers. Wheat germ agglutinin-horseradish peroxidase injection. To label the afferent innervation of the GI tract supplied by the common hepatic branch, an adaptation of the wheat germ agglutinin-horseradish peroxidase (WGAHRP) protocol introduced by Aldskogius and coworkers (1986), specifically for its success in labeling visceral afferents, was used. In addition, Robertson et al. (1992) have shown that all nodose ganglion neurons appear to incorporate WGA-HRP. Three days prior to perfusion, each rat was reanesthetized and placed in a supine position, and a ventral midline incision was then made in the neck. The left vagus, including the nodose ganglion, was exposed by blunt dissection, and WGA-HRP (3.0 µl; 8% in phosphate-buffered saline; Sigma, St. Louis, MO) was pressure injected (PicoSpritzer II, General Valve Corporation, Fair- 250 R.J. PHILLIPS ET AL. field, NJ) through a glass micropipette (I.D. 25 µm) into the ganglion. The incision was then closed and treated with nitrofurazone, and the animal was allowed to recover on a heating pad and then returned to its cage. Tissue preparation Two weeks after the vagotomy and 3 days after the WGA-HRP injection, each animal was deeply anesthetized with a lethal dose of pentobarbital. When it was completely unresponsive to nociceptive stimuli, the animal was injected in the left ventricle with 0.1 ml Heparin (Elkins-Sinn, Inc., Cherry Hill, NJ) (1,000 units/ml) to prevent coagulation and 0.1 ml propranolol (Ayerst Laboratories, Inc., Philadelphia, PA) to produce vasodilation and was then perfused transcardially with 500 ml of 0.9% saline at 40°C. To produce matched stomach and duodenal sizes across animals, the jejunum was clamped during this saline perfusion, and the stomach and area proximal to the clamp site were filled with 0.9% saline (15 6 5 ml; volume varied depending on quantity of stomach contents) at 40°C. Once the organs were filled with saline, the tissues were fixed by transcardial perfusion of 3% paraformaldehyde and 0.4% glutaraldehyde in 0.1 M sodium phosphate buffer, pH 7.4, at 4°C, and, finally, 10% sucrose in the same buffer at 4°C. The brainstem was removed and cryoprotected in 15% sucrose in phosphate buffer. The medulla was then blocked, sectioned coronally at 56 µm with a cryotome, and prepared to verify both the vagotomy (Fluoro-Gold: threefourths of the sections) and the nodose injection (WGAHRP: one-fourth of the sections). The nodose ganglia, the entire stomach, the first 8 cm of the duodenum, and the cecum were prepared as wholemounts, as described previously by this laboratory (Powley et al., 1994). Briefly, the whole GI tract was rinsed with cold tap water and then separated into the areas of interest. The stomach and cecum were divided into dorsal and ventral halves by cutting along the greater and lesser curvatures. The duodenum was divided into two 4-cm lengths and was opened with a longitudinal cut along the mesenteric attachment. Wholemounts were then processed with tetramethyl benzidine (TMB) according to the protocol of Mesulam (1978). Sections for the Fluoro-Gold analysis were thaw-mounted on slides, air dried, dehydrated in alcohol, cleared in xylene, and coverslipped with DPX (Aldrich, Milwaukee, WI). Sections and wholemounts for WGA-HRP analysis were cleared in xylene and coverslipped with DPX. Medulla sections, nodose ganglia, and GI tract wholemounts processed for WGA-HRP were examined with light- and darkfield illumination, and all animals were screened for the presence of labeled fibers in the nucleus of the solitary tract and left nodose ganglion. Counts of the different elements were conducted as described below. Specific stereological techniques were not employed, because 1) only relative comparisons were used, 2) the observations were to be compared with earlier relative estimates that did not employ a director strategy, and 3) most of the tissue was examined in wholemounts, not in serial sections. Verification of hepatic-spared vagotomies To verify the vagal surgery, the Fluoro-Gold test of selective vagotomies (Powley et al., 1987) based on the characteristic distributions of efferent somata within the dorsal motor nucleus of the vagus (dmnX) was employed (no comparably sensitive protocol is available specifically for afferents). In this procedure, Fluoro-Gold-labeled neurons in the dmnX were counted, and their distributions were analyzed by an observer, who was blind to each animal’s identity and WGA-HRP profiles. Because only three-fourths of the coronal sections of the brainstem were used for Fluoro-Gold analyses, the total cell counts were corrected (34/3). Because any ‘‘split-cell’’ counting bias would have inflated the resultant counts toward an overestimate of the number of dmnX neurons spared and a correction was problematic owing to the sequence of histological sections counted, no split-cell correction was used. A priori, brainstem Fluoro-Gold patterns had to satisfy four criteria for an animal to be classified as a common hepatic branch-spared case with no extraneous label: 1) Numerically, in terms of left vs. right dmnX distributions, labeled cells had to be found almost exclusively on the left side (i.e., the side occupied by the hepatic branch column). To be considered successful, verified cases had to have less than 20 labeled cells on the right and, then, only in the caudal part of the nucleus, where the left and right half-nuclei are fused on the midline. 2) Also numerically, the total of labeled preganglionic somata on the left side had to be less than or equal to the total number of hepatic motor neurons estimated in the quantitative description of the hepatic branch column by Fox and Powley (1985). 3) Spatially, the distribution of the labeled neurons in the left dmnX had to agree with the earlier characterizations of the column (Fox and Powley, 1985; Norgren and Smith, 1988; Powley et al., 1987). 4) Conversely, in contrast to the last criterion, the selective vagotomy was considered inappropriate if the distribution of Fluoro-Gold-labeled cells was organized in a pattern similar to the distinctive columns associated with either of the other two branches, i.e., the accessory celiac and ventral gastric branches, also organized in the left dmnX. Evaluation and screening of wholemounts A need for intact (required to preserve spatial information), low-artifact (required to permit counts of fine fibers and terminals) wholemounts made it necessary to screen specimens and eliminate those that could not be reliably quantified. In particular, it was found that the physical integrity of some wholemounts was compromised by dissection damage resulting from surgically induced tissue adhesions, and others were compromised during the removal of the mucosa from the muscle wall. In terms of the staining protocol, strong reactions of mast cells and perhaps other TMB-reactive nonneuronal cells, such as granulocytes and intravasal granular leukocytes associated with the surgery and related scarring, produced excessive artifact, which masked axon staining in some specimens. Before quantitative analyses were undertaken, each wholemount of each organ was evaluated. Those specimens that were judged excessively damaged or potentially masked by artifact were excluded from formal quantitative analyses. All judgments were made blind, prior to the Fluoro-Gold verification assessments. Quantification of fibers and terminals Counting of all intact, low-artifact wholemounts processed with WGA-HRP was then done without knowledge of the results of the verification procedure for the selective vagotomy. Criteria for counting. Based on the findings of Wang (1995), four main vagal afferent elements are found in the muscle layers of the rat GI tract and needed to be counted VAGAL HEPATIC BRANCH AFFERENTS to provide a description. The first two are bundles and fibers. Bundles were defined as two or more axons running together (see Fig. 1A), and fibers were defined as individual axons (see Fig. 1A). The third and fourth elements are terminal specializations, which are found on vagal afferents innervating the GI tract (Berthoud and Powley, 1992; Powley and Wang, 1995; Rodrigo et al., 1982; Wang, 1995; Wang and Powley, 1994). These consist of IGLEs and IMAs. An ending was determined to be, and was counted as, an IGLE if it consisted of fine terminal puncta within the neuropil of a myenteric ganglion, effectively encapsulating all or part of the ganglion (see Fig. 1B; see also Neuhuber, 1987; Neuhuber et al., 1995; Rodrigo et al., 1975, 1982). Finally, an ending was counted as an IMA (Berthoud and Powley, 1992; Wang, 1995) if it consisted of an array of parallel axonal telodendria in close proximity, interconnected by bridging axonal elements, and was located in either the circular or the longitudinal muscle layer (see Fig. 1C). Counting grid. To facilitate a quantitative reconstruction of the innervation of the GI tract by the common hepatic branch of the vagus (as well as to facilitate comparison across experiments), the sampling procedure of Wang (1995) was used. Two different reticles, a counting grid and a sampling grid, were employed. Briefly, at each sampling point (see below), a camera lucida drawing arm was used to merge an image of a 1.7 3 1.7 mm counting grid, consisting of 9 3 9 equidistant lines (forming 64 squares), with the tissue image in the microscope. Counting was done at a final magnification of 350 (eyepiece 38, objective 36.3). Before counts were taken at a sampling point, the grid was rotated, so that one set of lines paralleled the circular muscle, and the second set paralleled the longitudinal muscle of the organ wall. Each of the four elements (bundles, fibers, IGLEs, and IMAs) were then counted in a similar manner by using the counting grid. This grid was visually scanned from left to right/top to bottom, and the innervation pattern was then recorded. For bundles and fibers, the total number of intersections within the grid was the characteristic counted. For example, a single fiber (or bundle) traveling the width of the grid in a longitudinal direction would intersect with nine of the lines that make up the counting grid, so the fiber (or bundle) would be counted as nine longitudinal intersections. Similarly, a case in which two fibers (or bundles) were traveling in a similar fashion would yield a count of 18 longitudinal intersections, and so on. Due to the fact that distinguishing between an IMA and several IMAs is problematic, we used a modification of the counting criteria for IMAs. The same sampling grid was used, and direction of innervation was determined; but, instead of total number of intersections being counted, total number of squares in a counting grid that contained IMAs was recorded. So, a sampled area could not have more than 64 squares innervated. In the case of IGLEs, total number of squares containing IGLEs was counted. If an IGLE was contained entirely within one square, then it was counted as one; but, if the IGLE overlapped into an adjacent square, then that square was also counted. However, even if a single square contained two or more separate IGLEs, it was still only recorded as one. In composite, the counts of the four elements (bundles, fibers, IGLEs, and IMAs) as they intersected the grid lines oriented parallel to the two different muscle layers pro- 251 vided useful profiles of innervation patterns. For example, a region with a number of individual fibers as well as coherent bundles would presumably be an area in which axons were dispersing from connectives to form terminal fields, whereas a region with bundles but no fibers would represent an area that the vagus was merely traversing. For another example, a region in which most axons (or IMAs) intersected grid lines parallel to the circular muscle and few fibers (or terminals) intersected the grid lines parallel to the longitudinal muscle would be an area in which fibers (or endings) were traveling longitudinally. Sampling algorithms: Stomach and pylorus. Because of the morphological differences among the organs surveyed, the counting grid was positioned according to different sampling algorithms in the different organs. In the case of the stomach, the wholemounts of the dorsal and ventral halves were each sampled at 45 similar areas by the use of a sampling grid. To minimize variability resulting from differences in stomach size or distention, the tissue was first normalized with this sliding-scale sampling grid. For this purpose, each half of the stomach was measured and fitted with grid dimensions that provided ten equidistant intervals in the cranial-to-anal direction and six similar intervals from the lesser to the greater curvature. The survey generated by then positioning the counting grid at each of the 45 intersections within the sampling grid included 10% of the area of each wholemount. In addition to quantification of the gastric innervation with the counting and sampling grid strategy, the distribution pattern of the vagal bundles coursing into the stomach was also mapped graphically to facilitate describing the innervation of the stomach in terms of the different regions (e.g., forestomach, corpus, antrum). For these latter maps, each stomach wholemount was placed in an enlarger, and the specimen was projected onto a sheet of graph paper. The contour of the stomach was traced, and WGA-HRPlabeled bundles were then drawn onto the organ outline by using a camera lucida strategy (see Figs. 3A,B, 6A,B). The radial or circumferential extent of the pylorus that was innervated by common hepatic axons, the types of endings, and the location of the innervation were also determined through the use of the same counting grid. The entire pylorus was successively blanketed by the counting grid, and the total number of squares containing innervation and location of the innervation was recorded. Types of endings and location of the endings were also recorded. Sampling algorithm: Duodenum. The first 8 cm of the duodenum were sampled in a manner similar to the strategy for the stomach, allowing for the differences in the nature of the tissues. Specifically, the duodenal wholemount was counted in transverse sampling bands situated every 3.2 mm along its entire length. Distance was measured aborally from the pylorus. The circumferential midpoint of the transverse band (i.e., the antimesentery pole, given the dissection procedure) was located by measuring from the edges of the longitudinal tissue cut, and the counting grid was centered at that point, with the grid lines oriented parallel to the fibers of the two muscle sheets. Bundles, fibers, IGLEs, and IMAs within the counting grid were then recorded exactly as for the stomach. After the antimesenteric pole was counted in this manner, the entire circumferential band of duodenum was sampled by moving the counting grid in both radial directions in successive 1.7-mm steps. After the entire 252 Fig. 1. Darkfield photomicrographs illustrating vagal elements that were quantified. A: Bundles and fibers. A bundle of axons (solid arrow) with several individual fibers (open arrows) separating from the main bundle. B: Intraganglionic laminar ending (IGLE). A single afferent vagal fiber ending in an IGLE. Notice the distinct punctate pattern of the ending as it encapsulates portions of a myenteric ganglion. C: Intramuscular array (IMA). A bundle of afferent axons R.J. PHILLIPS ET AL. wrapping around from the ventral forestomach and terminating as several IMAs (arrows), which end in the longitudinal muscle. The diffusely distributed white spots are labeled mast cells that are not part of the ending. Mast cells and perhaps other granulocytes are labeled due to the reaction of tetramethyl benzidine (TMB) with the endogenous peroxidase. Scale bars 5 200 µm. VAGAL HEPATIC BRANCH AFFERENTS 253 transverse band had been counted, the counting grid was then relocated at the antimesenteric point 3.2 mm aboral to the band just counted, and the procedure was repeated. By using this procedure, approximately 53% of the tissue area was sampled. Sampling algorithm: Cecum. Due to the sparse innervation of the cecum by hepatic vagal fibers, the entire organ was systematically scanned, and the total numbers of fibers and types of endings were counted and described without the use of the counting grid. Graphical display of data. Graphpad Prism (version 2.0; Graphpad Software, Inc., San Diego, CA) software was used for line graphs; three-dimensional topographic and surface plots were created with Surfer Version 6.0 (Golden Software, Inc., Golden, CO). Control for possible regeneration of vagal branches Recently, it has been reported that vagal branches exhibit plasticity after vagotomy (Phillips et al., 1996; Powley et al., 1996). This plasticity involves regeneration of cauterized axons and possibly reorganization of any spared axons. Changes in the pattern of vagal fibers are found from 6 to 18 weeks after selective vagotomy. To address the possibility that the patterns of innervation observed in the present experiment might reflect some axonal reorganization, three animals were treated as described in Materials and Methods, except that, instead of a common hepatic-spared vagotomy, they received a complete subdiaphragmatic vagotomy, including the common hepatic branch. These animals were then examined for possible indications of vagal plasticity, such as the presence of bundles, fibers, and growth cone profiles (Phillips et al., 1996; Powley et al., 1996). RESULTS Verification of hepatic-spared vagotomies with retrograde transport of Fluoro-Gold Twenty-five rats satisfied the criteria used to establish that the selective vagotomy procedure had eliminated all vagal abdominal branches except the common hepatic branch. In terms of the four specific criteria employed: 1) Labeled hepatic preganglionics were essentially confined to the left dmnX, with a mean (6S.E.M.) of only 8 (60.8) Fluoro-Gold cells per animal observed in the right dmnX. 2) Compared with animals with all vagal branches and their corresponding dmnX columns intact (cf. Fig. 2A, control section), hepatic-spared animals had only a few Fluoro-Gold-labeled motor neurons distributed sparsely through the left dmnX (mean 6 S.E.M. 5 112 6 8; cf. Fig. 2B; for comparison, the other two branches of the ventral trunk are represented by approximately 3,000 somata in the left dmnX; Fox and Powley, 1985). 3) In all animals judged to have successful hepatic-spared vagotomies, these labeled preganglionic neurons were distributed throughout the left dmnX in the sparse columnar pattern characteristic of the hepatic column (cf. Fox and Powley, 1985; Norgren and Smith, 1988; Powley et al., 1987). This pattern included a distinct and consistent ‘‘arch’’ or ‘‘horseshoe’’ of preganglionic somata at the level of the area postrema (Fig. 2C). The arch gets its appearance because preganglionic neurons of the common hepatic branch column form a thin dorsal plate of cells spanning through Fig. 2. Photomicrographs of frontal sections of dorsal medulla at level of area postrema showing Fluoro-Gold label in the dorsal motor nucleus of the vagus (dmnX). With this protocol, neurons of the dmnX with intact projections to viscera in the peritoneal space label strongly with Fluoro-Gold (‘‘primary labeling’’), whereas neurons and other tissue associated with heavily vascularized sites without fully effective blood-nerve barriers (cf. the hypoglossal neurons and area postrema) sequester some Fluoro-Gold that is circulated systemically after i.p. injections (‘‘secondary label’’). The dmnX neurons axotomized by cauterization contain only secondary Fluoro-Gold labeling and appear no brighter than neurons in the hypoglossal nucleus. Also note that tissue autofluorescence at background was high in the present tissue, because the glutaraldehyde used in the wheat germ agglutininhorseradish peroxidase (WGA-HRP) protocol induces autofluorescence. A: Control animal with no previous surgery. The left and right dmnX can be seen clearly with the gastric (solid arrows) and celiac (open arrows) poles intact. B: Two hepatic motor cells with primary Fluoro-Gold labeling (arrows) can be seen in the region of the left gastric pole. The number of labeled motor cells is unequivocally less in the hepatic-spared case compared with the control case. C: A distinct arch pattern of cells with primary Fluoro-Gold labeling (arrows) can be seen in the left dmnX. This pattern was found to occur when scanning through the sections starting at the area postrema and moving distally through the medulla. Almost all animals with hepatic-spared vagotomies had this distinct pattern. Scale bar 5 200 µm. the central part of the nucleus (viewed in transverse sections), with the more medially and laterally situated cells dispersed from dorsal to ventral. 4) None of the animals judged to have successful hepatic-spared surgeries had condensations of preganglionic neurons in either the distinctive ventral gastric column or the compact celiac column of the left dmnX. WGA-HRP-labeled fibers and terminals in the stomach and pylorus General pattern. Wholemounts of 11 complete (dorsal and ventral halves) stomachs from the sample of verified 254 animals met the criteria for analysis (i.e., the specimens were intact and low in artifact) and were mapped, quantified, and photographed. These 11 specimens were representative of the larger set of 25 animals, because their mean dmnX verification profiles (left dmnX 5 114, right dmnX 5 7) were similar to those of the other 14, which were not suitable for counting (left dmnX 5 110, right dmnX 5 8) as well as the profile for the entire set of 25 animals (left dmnX 5 112, right dmnX 5 8). Stomach bundles and fibers. When the normalized sampling grid was fitted to the individual stomachs and used to determine regions for counting, each animal exhibited a similar pattern (for discussion of exceptions, see below) of innervation across the different points sampled. The hepatic afferents were predominately distributed within the ventral wall of the stomach (Figs. 3A,C,E, 4A,C), with the dorsal gastric wall receiving only 27% of the total innervation (Figs. 3B,D,F, 4B,D). Innervation of the ventral stomach consisted of one or more large bundles that entered at the lower esophageal sphincter and traveled through the upper portion of the forestomach (close to the lesser curvature) toward the greater curvature, where it wrapped around to the dorsal side and terminated (Figs. 3A,B, 6C). As the bundles coursed through the ventral forestomach, they branched into smaller bundles and individual fibers, some of which also wrapped around to the dorsal forestomach (Fig. 3C–F). Seventy three percent of the bundles (Fig. 3C,E) and 73% of the fibers (Fig. 4A,C) were found to innervate the ventral stomach. Neither bundles nor fibers evidenced a preferential direction of innervation (i.e., paralleled one of the muscle fiber axes) on either side of the stomach. On the ventral side, 47% of the bundles (Fig. 3C) and 44% of the fibers (Fig. 4A) traveled in the longitudinal direction. A similar pattern was determined for the dorsal side, with 52% of the bundles (Fig. 3D) and 52% of the fibers (Fig. 4B) coursing in the longitudinal direction. Bundles and fibers were also found to innervate both sides of the stomach in the region of the antrum (almost always circular in direction), but these bundles and fibers usually terminated quickly upon entering the stomach wall from the lesser curvature. Stomach terminals. Both types of afferent terminals that have been observed in the gastric wall (i.e., IGLEs and IMAs) were formed in the forestomach by common hepatic branch afferents. IGLEs were diffusely distributed throughout the rostral portion of the ventral and dorsal forestomach, with 70% of the terminals located on the ventral side (Fig. 4E,F). Relatively speaking, IMAs were symmetrically distributed on both sides of the stomach, with 59% of the terminals on the ventral side and 41% on the dorsal. Whereas IGLEs were distributed diffusely throughout the upper forestomach, IMAs were concentrated in a particular location. This locus was on the greater curvature of the ventral and dorsal forestomach (Fig. 5A,B), with IMAs forming large arrays of axons running parallel to each other (Figs. 1C, 5C). In addition to being concentrated at one location, IMAs were also found to innervate principally longitudinal muscle, with 93% of the endings running in the longitudinal direction (Figs. 1C, 5C). Range of stomach innervation patterns. The pattern of innervation described above was the typical pattern observed in most of the stomachs counted (n 5 8) and was validated as the characteristic pattern when the average R.J. PHILLIPS ET AL. for the eight cases were compared with averages for all 11 stomachs. However, this description does not completely describe the range of patterns produced by the hepatic innervation of the stomach in some animals. In one of the stomachs studied, the projection consisted of bundles and fibers that traveled along the limiting ridge of the ventral forestomach to wrap around the greater curvature of the dorsal forestomach at a site somewhat more distal than the typical location. Finally, the remaining two stomachs were found to be densely innervated compared with the typical stomach, with bundles and fibers encompassing the whole ventral forestomach, including extensive wraparounds to the dorsal side (Fig. 6A,B). Pylorus. The innervation of the pylorus by the common hepatic branch was limited to a few fibers ending exclusively in IMAs. In 10 of 12 stomach wholemounts counted (five ventral and dorsal pairs), the IMAs were localized within a small region of the pylorus (33% of the total area of the pylorus) on the greater curvature side of the stomach (Fig. 7). In the remaining two stomach specimens counted (one ventral and dorsal pair), the IMAs were not localized to one discrete area but were distributed more diffusely throughout the entire pylorus. This less typical pyloric pattern was not correlated with the less typical gastric innervation patterns. Duodenum and cecum General pattern. A total of 16 duodena were analyzed. These 16 specimens were representative of the larger set of 25 animals, because their average dmnX verification profiles (left dmnX 5 111, right dmnX 5 7) were similar to those of the other nine, which were not suitable for counting (left dmnX 5 112, right dmnX 5 8), as well as the profile for the entire set of 25 animals (left dmnX 5 112, right dmnX 5 8). The number of transverse sampling bands that could be counted per specimen ranged from 3 to 13, with a mean of 11 bands for the first wholemount of the duodenum and 7 bands for the remaining wholemount. Duodenal bundles and fibers. Afferent bundles entered the duodenum from the mesentery at about 3.2 mm distal to the pylorus and densely innervated the tissue for the next 3.0 cm, falling off to almost no innervation by 5.0 cm (Figs. 8A, 10A, 11A). Bundles were also seen to innervate the first 3.2 mm (see 0 mm, Figs. 8A, 10A, 11A), but the innervation consisted primarily of fascicles that branched from larger bundles at more distal sites and traveled toward the pylorus. In addition, a few bundles were found to enter the most proximal duodenum from the mesenteric attachment, but hepatic afferent bundles were never seen crossing the pylorus from the stomach. Bundles ran preferentially in the longitudinal direction parallel to the mesentery for approximately the first 3.5 cm (Figs. 8A, 11). Rarely were bundles at the antimesentery found to be traveling in the longitudinal direction (Fig. 11B,C). Bundles traversing the tissue in the circular direction (Figs. 8A, 9, 10) were also denser along the mesenteric attachment and were progressively sparser moving toward the antimesentery (Fig. 10B,C). At all points, however, fewer bundles ran in the circular orientation than in the longitudinal direction. The quantitative survey of fibers revealed patterns similar to those of the bundles. The fibers entered the duodenum from the mesentery at about 3.2 mm and densely innervated the tissue for the first 3 cm, then progressively declined in density more distally, until there VAGAL HEPATIC BRANCH AFFERENTS Fig. 3. Map and plots of the hepatic branch innervation of the ventral and dorsal stomach by bundles. For ease in comparing symmetrical sites from the two half stomachs, the ventral gastric wholemount has been ‘‘flipped over.’’ In effect, the viewer is examining the ventral gastric wholemount from the mucosal side and the dorsal wholemount from the conventional serosal side. This same convention is used throughout this report for all maps, plots, and drawings. A,B: Camera lucida tracing of the ventral (A) and dorsal (B) stomach from the same rat. This map illustrates the typical pattern of innervation by bundles in the fundus, although the one bundle on the ventral 255 stomach that branches from the main trunk and crosses over the corpus to terminate in the antrum is atypical. C–F: Surface and contour plots of counts and direction of innervation by hepatic bundles. A silhouette of a stomach wholemount (see A for orientation and regions) is superimposed on the contour plots to highlight the location of the innervation. The x-axis is the sampling grid column sampled from, the y-axis is the sampling grid row sampled from, and the z-axis is the total number of intersections by bundles per counting grid averaged across the 11 stomachs sampled. Scale bars 5 2 mm. 256 Fig. 4. A–D: Plots of the same type shown Figure 3C–F, except the plots are of the fibers counted. E,F: Plots of the same type shown in Figure 3C–F, except the plots illustrate the location of IGLEs in the stomach. Notice the concentration of IGLEs in the region of the R.J. PHILLIPS ET AL. antrum. Antral fibers ended primarily in IGLEs. The x- and y-axes are the same as previous plots, but the z-axis now represents the density of IGLEs per counting grid averaged across stomachs sampled. VAGAL HEPATIC BRANCH AFFERENTS Fig. 5. A,B: Plots of IMA terminal fields. The x- and y-axes are the same as the plots in Figures 3C–F and 4A–D, but the z-axis now represents the number of squares containing IMAs per counting grid averaged across stomachs sampled. The IMAs are unequivocally and 257 solely focused on the greater curvature of the forestomach. C: Darkfield image of IMAs (arrows) running in the longitudinal muscle. Scale bar 5 200 µm. 258 Fig. 6. A,B: Camera lucida tracing of hepatic bundles in the ventral (A) and dorsal (B) stomach from the same rat. These maps illustrate the extensiveness of the innervation by the hepatic branch of some animals (n 5 2). Bundles and their branches are contained in the region of the ventral forestomach, with some wrapping around to the dorsal forestomach where they terminate. None is seen to cross the R.J. PHILLIPS ET AL. limiting ridge into the corpus of the stomach, and only one bundle was found in the antrum. C: Darkfield image of a bundle (solid arrows) in the ventral forestomach branching into a smaller bundle and several individual fibers (open arrows) as it courses toward the greater curvature. Scale bars 5 2 mm in A,B, 200 µm in C. VAGAL HEPATIC BRANCH AFFERENTS 259 Fig. 8. Graphs of the bundles (A), fibers (B), and IGLEs (C) counted in the duodenum. Error bars are 6S.E.M. There are two main findings conveyed by the three graphs. First, for all three graphs, the innervation is denser in the second 3.2-mm circular band sampled compared with the first circular band, which was sampled from the region adjacent to the pylorus. Finally, the error bars become relatively small to nonexistent at about 50 mm, indicating a consistent diminution of hepatic branch afferent innervation at this distance. Fig. 7. Darkfield photomicrograph demonstrating the typical innervation seen in the rat pylorus by vagal hepatic afferent fibers. The dominate pattern was for bundles, fibers, and IMAs to be concentrated in the greater curvature side of the pylorus. The long arrow indicates a bundle that projects to the pylorus, and the short arrows delineate the region of the pylorus that is innervated by IMAs. Scale bar 5 200 µm. 260 Fig. 9. Darkfield image taken from the 1st cm of the duodenum. A bundle (solid arrow) can be seen entering from the mesentery (not shown) and traveling in a circular direction (vertically in the photograph) toward the antimesentery. Numerous fibers separate from the R.J. PHILLIPS ET AL. bundle, forming a meshwork of fine fibers, which travel in both circular and longitudinal directions. A multitude of IGLEs are distributed across the duodenal tissue. A typical fiber (open arrow) and IGLE (open arrowhead) are indicated. Scale bar 5 200 µm. VAGAL HEPATIC BRANCH AFFERENTS Fig. 10. A: Surface and contour maps of bundles running in a circular direction of the duodenum. The x-axis is the distance anal from the pylorus that was sampled, the y-axis is the width of the tissue sampled (with 0 mm indicating the center of the antimesentery), and the z-axis is the total number of intersections per counting grid averaged across the duodena sampled. B,C: Head-on views of the 261 surface map from the proximal end (B) and the distal end (C). Notice the paucity of innervation of the antimesentery. D: Darkfield image taken from the 2nd cm of the duodenum. Bundles (arrows) can clearly be seen entering from the mesenteric border (top) and traveling toward the antimesentery. Scale bar 5 200 µm. 262 Fig. 11. A: Surface and contour maps (for conventions, see Fig. 10) of duodenal bundles running in a longitudinal direction. B,C: Head-on views of the surface map from the proximal end (B) and the distal end (C). Notice how the bundles are densest along the sides where the duodenum is attached to the mesentery, and note the sparseness of R.J. PHILLIPS ET AL. innervation of the antimesentery. D: Darkfield image taken from the 3rd cm of the duodenum. Large longitudinal bundles (arrows) run parallel to the mesenteric border (bottom of picture; not shown). Scale bar 5 200 µm. VAGAL HEPATIC BRANCH AFFERENTS was little innervation at about 6 cm (Figs. 8B, 12A, 13A). In the same vein as innervation by bundles, fibers were also found in the first 3.2 mm (see 0 mm, Figs. 8B, 12A, 13A), but not to the same extent as more distal sites. These fibers branched from more distal bundles at 3.2 mm and traveled toward the pylorus. In the case of fibers, antrumto-duodenum transpyloric crossings did occur in a small percentage of the duodena surveyed. This similarity between the patterns of bundles and fibers partially reflects the fact that fibers separated from the bundles (Figs. 9, 12D, 13D), so that they would naturally parallel, at least to some extent, the innervation pattern of the bundles. Where fibers differed from bundles was in the distribution radially throughout the tissue. Fibers had no predominant orientation and distributed around the full circumference of the tissue (Figs. 8B, 12B,C, 13B,C), creating a meshwork (or web-like) pattern across the entire width of the tissue from the mesentery to the antimesentery (Figs. 9, 12D, 13D). Duodenal terminals. Hepatic afferent duodenal terminal specializations consisted almost exclusively of IGLEs, with IMAs comprising only about 0.5% of the terminals counted. IGLEs followed a distribution pattern similar to that of duodenal fibers. The densest area of terminals occurred in the first 3 cm of the duodenum and progressively declined in density moving distal to the pylorus (Figs. 8C, 14A). IGLEs were scattered evenly and transversely around the tissue, although a slightly greater concentration was seen at the mesenteric border (Fig. 14A–C). IGLEs were generally similar in appearance from region to region, although some differences were seen based upon the density of innervation along the length of the duodenum. In the first 3 cm, where the terminals were densest, several fibers entered or continued from a single IGLE (Figs. 9, 14F), and an individual IGLE would frequently be situated in close proximity to neighboring IGLEs (Figs. 9, 14D). In contrast, from 4 cm distal to the pylorus on, IGLEs were associated with fewer (typically one) fibers (Fig. 14E) and were sparsely distributed in the tissue. At this distance from the pylorus, IGLEs were also usually located in close proximity to the mesenteric border (Fig. 14D). The small subset of hepatic duodenal afferents ending in IMAs (Fig. 15A) always consisted of very fine axons running exclusively in the circular direction (Fig. 15A) adjacent to the mesenteric border. Cecum. Seven cecal wholemounts met the criteria to be counted and photographed. The cecum was very sparsely innervated by common hepatic branch afferents. The projection usually consisted of 2 to 15 fibers (mean 5 8) entering at the ileocecal junction and terminating close to that site. Terminals consisted exclusively of IGLEs (Fig. 15B), with an average of four IGLEs per cecum. Control for regeneration No indication of regeneration was found to occur by the 2-week sampling point in the three control animals that received complete subdiaphragmatic vagotomies, thus verifying that the innervation seen in the hepatic-spared animals was the normal or ‘‘intact’’ hepatic branch pattern and was not contaminated by regeneration of vagal axons transected by selective vagotomy. Two weeks after the complete vagotomy procedure, as verified by the lack of Fluoro-Gold-labeled cell bodies in the dmnX (mean 6 S.E.M.; left dmnX 5 6 6 2, right dmnX 5 3 6 1 labeled cell bodies), not a single axon was found in the stomachs and 263 ceca of the three control animals. On the other hand, a mean of one axon was found to innervate the first 8 cm of the small intestine. Because only a mean of one axon per animal (with one of the three cases having no axons) was found, the pattern presumably reflects either the existence of a stray minor fascicle that was missed when the vagal bundles were cauterized at surgery or else the earliest (still inconsequential) stage of vagal plasticity occurring at 2 weeks postvagotomy. DISCUSSION The present results show that hepatic afferents not only innervate the duodenum, pylorus, and antrum, as implied by previous studies of efferents (Berthoud et al., 1991), but that afferents from the branch also project to the forestomach and cecum. In addition to describing this regional specificity, we also demonstrate that the hepatic branch has distinct patterns of terminal specializations, varying by region or organ (e.g., the combination of both IGLEs and IMAs in the forestomach or the preponderance of IGLEs in the duodenum). Furthermore, the patterns and densities of hepatic terminals represent discrete, organized, and localized concentrations that cannot be explained as random and proportional subsets of the overall vagal truncal afferent pattern or of the density of myenteric ganglia (Gabella, 1987; Powley and Wang, 1995; Wang, 1995; Wang and Powley, 1994). Before considering the results for the different organs in detail, however, two critical elements of the protocol should be considered. Effectiveness of selective vagotomy verification and nodose injections The interpretations of the present results hinge on the vagotomy verification and the afferent labeling strategy. Two critical questions arise. Were we able to accurately determine a selective hepatic-spared vagotomy with the Fluoro-Gold test, and were only vagal afferents, and not efferent fibers of passage, labeled by nodose injection of WGA-HRP? Three relevant observations on the use of the FluoroGold verification test indicate that hepatic-spared cases were effectively distinguished. First, the vast majority (96%) of the preganglionic somata in the dorsal motor nucleus of the vagus are organized into two symmetrical pairs of coherent subnuclei or columns, which issue the efferent axons of the two gastric (ventral and dorsal) and the two celiac (ventral or ‘‘accessory’’ and dorsal) branches of the vagus, respectively (Fox and Powley, 1985). If a fraction of one of these gastric or celiac branches was inadvertently spared, then the nonaxotomized neurons would have been detected both as a substantial increase in the counts of labeled cells in the dmnX and as a remaining fragment of one of the subnuclei (for an example of incomplete branch cuts, see Powley et al., 1987). Second, the common hepatic branch also contains the axons of a distinctive column of preganglionic neurons within the left dmnX (Fox and Powley, 1985; Powley et al., 1987). This hepatic column pattern includes a distinguishing ‘‘arch’’ of cells at the level of the area postrema that is noticeable in the figures of previous reports (see Fox and Powley, 1985; in particular, for a good example, also see Fig. 2D in Berthoud et al., 1990a). Furthermore, the distinctive distribution of this column at different longitudinal levels constitutes a series of signature features that would be all 264 Fig. 12. A: Surface and contour maps (for conventions, see Fig. 10) of circular fibers distributed evenly across the width of the duodenum. B,C: Head-on views of the surface map from the proximal end (B) and the distal end (C). Note how the fibers are located in the antimesentery R.J. PHILLIPS ET AL. as well as in the mesentery. D: Darkfield photograph taken from the first 4 cm of the duodenum. The fibers (open arrows) form a characteristic meshwork pattern that consists of fibers running in the longitudinal and circular directions. Scale bar 5 200 µm. VAGAL HEPATIC BRANCH AFFERENTS Fig. 13. A: Surface and contour maps (for conventions, see Fig. 10) of longitudinal fibers uniformly distributed across the width of the duodenum. B,C: Head-on views of the surface map from the proximal end (B) and the distal end (C). Note how, unlike vagal bundles, the 265 fibers are located in both the antimesentery and the mesentery. D: Darkfield photograph of longitudinal fibers (arrows) ending in IGLEs. Scale bar 5 200 µm. 266 Fig. 14. IGLEs were evenly distributed across the width of the duodenum and progressively diminished in density, moving distal to the pylorus. A: Surface and contour maps (for conventions, see Fig. 10) of IGLEs in the duodenum. The x- and y-axes are the same as in previous plots, but the z-axis now represents the density of IGLEs per counting grid averaged across duodena sampled. B,C: Head-on views of the surface map from the proximal end (B) and the distal end (C). D: R.J. PHILLIPS ET AL. A single afferent bundle (solid arrow) can be seen entering from the mesentery (top) and terminating in several IGLEs (open arrowheads). This was standard for the innervation of the duodenum at about 5–8 cm distal from the pylorus. E: A single fiber can be seen in the antimesentery running in the longitudinal direction and ending in an IGLE. F: Several fibers are seen innervating a single IGLE. Scale bars 5 200 µm. VAGAL HEPATIC BRANCH AFFERENTS Fig. 15. A: Darkfield image taken from the duodenum, 5 cm distal from the pylorus. Several fibers (open arrow) can be seen separating from a bundle (long solid arrow) and ending in several IGLEs (arrowhead) and IMAs (short solid arrow). IMAs were rarely found to occur in the duodenum in this study. B: A single afferent vagal fiber (arrows) entering the cecum at the ileocecal junction and ending in two IGLEs (arrowheads). IGLEs were the only type of ending found on hepatic afferent fibers innervating the cecum. Scale bars 5 200 µm. but impossible to duplicate by randomly sparing minor elements of the other branches of the abdominal vagus. Third, the dmnX cell counts of the present experiment made it possible to compare the present hepatic branch preganglionic population with that previously estimated (Fox and Powley, 1985). The somewhat lower counts in the present experiment (112 6 8 vs. 215 for the left dmnX) probably represent different counting criteria—particularly a higher threshold for cell recognition produced by the higher background autofluorescence yielded by the fixation protocol of the present experiment. Alternatively, if one were to argue that the lower cell counts in the present experiment resulted from partial surgical damage to the hepatic branch, then it would follow that the present survey of the distribution of hepatic branch projections, despite its finding more extensive patterns than were previously recognized, would be a conservative underestimate of the full projections. Concerning the second methodological point above, namely, the issue of selectivity of afferent labeling, it is important that 1) the method of labeling used (WGA-HRP into the nodose ganglion) produced at least relatively complete labeling of vagal afferent projections to the GI 267 tract and 2) did not also label efferent fibers of passage. In terms of the labeling of vagal afferents, it should be noted that all afferent neurons in the nodose ganglion appear to bind and transport WGA-HRP (Robertson et al., 1992). In addition, the volume of the injections employed in the present study was determined, on the basis of pilot work, to infiltrate the entire ganglion with tracer and to produce asymptotic labeling. It should also be noted that any incompleteness of labeling would have caused us to underestimate the full extent of hepatic afferent labeling. Because the present results identified all of the characteristic projections to the intestines and antrum suggested by the earlier work on hepatic efferents as well as described additional discrete fields of projections in the gut, a major (and systematic across-animals) underestimation seems unlikely. Several observations also suggest that the innervation patterns reported here did not result from inadvertently labeling efferent fibers of passage. Perhaps most tellingly, the types of hepatic branch endings observed in the present analysis corresponded to the afferent endings described in previous work that employed selective motor and sensory vagotomies to distinguish unequivocally the distinctively different types of endings in the gut (Berthoud and Powley, 1992; Berthoud et al., 1991; Powley et al., 1994; Wang, 1995). In addition, the mechanism of WGA-HRP labeling assures that, as demonstrated previously (Neuhuber, 1987; Pugh and Kalia, 1982), nodose injections of the conjugate preferentially label vagal afferents. This relative selectivity for afferents results from the combined facts that WGA-HRP binds to membrane sites that are internalized by endocytosis and that endocytosis typically occurs at the cell body, not at the axon. Similarly, nodose ganglion injections of other tracers [e.g., 1,18dioctadecyl-3,3,38,38-tetramethylindo-carbocyanine perchlorate (DiI) injected in vivo; Berthoud and Powley, 1992; Berthoud et al., 1992; Kressel et al., 1994] relying on endocytosis also do not generally label fibers of passage. Finally, these various tracer experiments suggest that only damaged fibers of passage would take up and possibly transport the marker. With such damage, retrograde, not anterograde, labeling has typically been observed. These arguments are consistent with the fact that we did not observe classical efferent profiles, i.e., varicose processes terminating in myenteric ganglia (Berthoud et al., 1990b; Holst et al., 1997), in the present series. In summary, then, by a variety of criteria, the WGAHRP labeling in the present experiment was both relatively complete and selective. Furthermore, sampling biases would have led to some underestimation of what is a dense, although regionally restricted, set of projections to the GI tract. These projections are considered below. Stomach and pylorus Previous studies (Powley and Wang, 1995; Sugitani et al., 1993; Wang, 1995; Wang and Powley, 1994) have shown that the vagal innervation of the esophagus and stomach is predominately lateralized, with the left trunk innervating the ventral side of the stomach and the right trunk innervating the dorsal side of the stomach (although, for an experiment claiming the projections are bilateral, see Hudson, 1989). The present observations refine and extend this idea. We found that the elements of the hepatic branch (bundles, fibers, IGLEs) of the left vagus predominately innervated the ventral side of the stomach. In 268 particular, it is instructive that the hepatic afferents that originate from the left nodose express this innervation selectivity for the ventral (or ‘‘ipsilateral’’) side of the stomach, even though the afferent axons of the hepatic branch reach the stomach by a very different trajectory than the axons that stay in the ventral vagal trunk on the esophagus and that form the ventral gastric branch. The distribution of hepatic branch IMAs in the stomach seems to be in apparent contrast to the pattern just described. Unlike the distributions for bundles, individual axons, and IGLEs, the IMA terminal specializations were essentially evenly distributed on both sides. This discrepancy, however, may reflect the fact that the IMAs of hepatic branch origin were located predominately on the greater curvature, where they innervated longitudinal muscle. In effect, these IMAs were situated on the midline rather than throughout either side of the stomach. Consistent with the general pattern of a predominant distribution of the hepatic afferents on the ventral side, these IMAs typically originated from axons traveling in bundles in the ventral wall of the stomach. The findings that hepatic IMAs are located primarily in one gastric locus (greater curvature of the forestomach) and that they innervate almost exclusively the longitudinal muscle layer raise the possibility that this branch is important for the detection of a particular type of stimulus localized to a restricted area of the organ. The type of stimulus (possibly distention; Berthoud and Powley, 1992) and the role of the hepatic branch in the detection of this stimulus have yet to be elucidated. However, the fact that overall hepatic gastric innervation was contained primarily in the forestomach indicates that this branch may play an important role in the reservoir function (Gabella, 1987) that is associated with this region. In addition, by supplying the forestomach with two morphologically distinct terminal specializations, hepatic afferents may detect more than one gastric cue. Consistent with these ideas, it is worth noting that animals with the hepatic branch spared but with all other abdominal vagal branches cut are still capable of detecting gastric distension (Gonzalez and Deutsch, 1981; Phillips and Powley, unpublished observations). Kressel et al. (1994) found that DiI-labeled vagal afferents innervating the circular muscle of the pylorus ended exclusively in IMAs. Due to the structural similarity of these endings to previously described in-series tension receptors, the authors proposed a possible functional role for these endings in the detection of passive distention and sphincter constriction. In agreement, we found that hepatic afferents innervated the circular muscle of the pylorus and ended exclusively in IMAs. Although the Kressel and coworkers experiment describes the afferent innervation of the pylorus by either the ventral or the dorsal trunk of the vagus, our findings take the next step by determining the innervation of the pylorus by an individual branch of the vagus. The fact that the hepatic branch is only one of five vagal subdiaphragmatic branches and yet it innervates a discrete region (about one-third) of the pyloric torus may indicate a prominent role for this branch in pyloric function. It should also be taken into account that the present study only looked at the innervation of the pylorus in a superficial manner, so additional studies looking at the innervation of the pylorus by the hepatic branch are needed before any strong claims about hepatic primacy can be made. R.J. PHILLIPS ET AL. Duodenum and cecum The first 8 cm of the duodenum were chosen for study, because this proximal region, including the duodenal bulb, appears to be particularly important in the feedback control of gastric activity as well as feeding behavior. Anatomically, this is the region of the small intestine that receives bile duct secretory products, is attached to the ligament of Treitz, and receives overlapping efferent projections from the gastric, celiac, and hepatic branches of the vagus (Berthoud et al., 1991). Examination of the region also made it possible to compare the present afferent results with the earlier efferent projection results. Physiologically, the proximal duodenum receives the initial rush of chyme from the stomach and has been shown to play an important role in the detection of nutrients (Walls et al., 1995a,b), distension, (Hölzer and Raybould, 1992) and hormones (Smith et al., 1995). Wang (1995) found that the left nodose ganglion provided significantly more vagal afferent innervation to the duodenum compared with the right nodose ganglion. This pattern led him to speculate that this difference was due to the additional contribution of the hepatic branch. Although it is not possible to determine directly from the present findings the percentage of the total innervation of the duodenum contributed by the hepatic branch, through comparisons between the present study and the Wang (1995) study, we can make some inferences about the contribution of the common hepatic branch to the duodenal field. It appears that the hepatic branch provides the majority of the vagal innervation to the first 3 cm of the small intestine. Due to the importance of this region in GI physiology, we can only speculate that the hepatic branch, as the prominent source of sensory feedback in this region, must play a key role in ingestive behavior and GI physiology. Even though the innervation of the cecum by hepatic afferents was sparse, the present results still provide direct evidence of hepatic vagal involvement in this organ. In addition, one may be able to extrapolate from the finding of hepatic afferents in the cecum as well as the duodenum to the innervation of the intestines, which we did not directly examine. It would be surprising to find that the hepatic branch only innervated the small length of intestine that we examined. A more parsimonious picture of its innervation pattern would be that the hepatic branch innervates the duodenum heavily at its initial site of contact and then travels through the mesentery toward the cecum, innervating the small intestine periodically along its length. However, we cannot rule out the possibility that there may be other areas of the small intestine (i.e., ileum and jejunum) that are heavily innervated by hepatic afferents. Implications of IMA and IGLE distributions The present results establish that the IMAs and IGLEs supplied by the vagal hepatic branch to the GI tract have quite different distributions, as has been reported more generally by Wang in his survey of all subdiaphragmatic vagal afferents projections (Wang, 1995; Wang and Powley, 1994). Indeed, some regions innervated by the hepatic branch (e.g., the pylorus) contain only IMAs, whereas other regions (e.g., the duodenum and cecum) have almost exclusively IGLEs. This spatial segregation of putative receptors offers a useful perspective on the earlier observa- VAGAL HEPATIC BRANCH AFFERENTS tions of Berthoud and Powley (1992) indicating that single afferent axons of the vagus in the forestomach could terminate polymorphically, with some collaterals forming IMAs and others forming IGLEs. The present results illustrate that the overall distributions of IMAs and IGLEs are not highly correlated and that, therefore, the polymorphic character of some fibers in the forestomach is by no means a universal pattern of vagal afferent termination within the muscle wall of the GI tract. Overall, the more extensive distributions of IGLEs and the relatively restricted regional concentrations of IMAs reinforces the speculation that the two endings are different types of receptors. The widespread distribution of IGLEs is consistent with the hypothesis that they serve as mechanoreceptors that detect the shearing action of muscle layers during movement of the wall of the GI tract (Neuhuber, 1987; Neuhuber et al., 1995) and, even more specifically, as receptors (as well as perhaps axon reflex elements) involved in the coordination of peristalsis and other intrinsic motility patterns of the gut (Wang, 1995; Wang and Powley, 1994). In contrast the more localized concentrations of IMAs, particularly in sphincters and in the forestomach, suggest that these endings may operate at strategic GI sites to detect distension provided by the passage of ingesta and the presence of intraluminal contents (Wang, 1995). CONCLUSIONS In conclusion, the results of this study as well as others on the hepatic branch (Berthoud et al., 1992; Prechtl and Powley, 1987) depict a complex nerve that serves as a pathway for fibers with numerous destinations. Not only are these destinations spread extensively across a single organ, they are also distributed across several abdominal organs from the liver to the cecum. These projections suggest that a single vagal afferent branch has the capability to detect stimuli from multiple areas that play key roles in the maintenance of GI function and feeding behavior. The unique anatomic innervation pattern displayed by the hepatic branch makes it a likely candidate for orchestrating a finely tuned system of interorgan communication and cross-organ coordination to facilitate the maintenance of body homeostasis. ACKNOWLEDGMENTS We thank M.-C. Holst and F.B. Wang for graciously demonstrating the WGA-HRP technique as well as J. Kelly (photography) and J. Mitchell (art work) for their expert technical assistance. In addition, we thank Drs. Robert Meisel, Susan Swithers, Zixi Chang, and Elwood Walls as well as Mike Jarvanin, Fred Martinson, and Wyatt Wollmann for their comments on earlier drafts of this paper. A preliminary report of the present findings was given in abstract at the Annual Meeting of the Society for Neuroscience (1995). The research was performed in partial fulfillment of the Master’s degree to R.J.P. LITERATURE CITED Aldskogius, H., L.-G. Elfvin, and C.A. Forsman (1986) Primary sensory afferents in the inferior mesenteric ganglion and related nerves of the guinea pig. J. Auton. Nerv. Syst. 15:179–190. 269 Berthoud, H.R., and T.L. Powley (1991) Morphology and distribution of efferent vagal innervation of rat pancreas as revealed with anterograde transport of DiI. Brain Res. 553:336–341. Berthoud, H.R., and T.L. Powley (1992) Vagal afferent innervation of the rat fundic stomach: Morphological characterization of the gastric tension receptor. J. Comp. Neurol. 319:261–276. Berthoud, H.R., E.A. Fox, and T.L. Powley (1990a) Localization of vagal preganglionics that stimulate insulin and glucagon secretion. Am. J. Physiol. 258:R160–R168. Berthoud, H.R., A. Jedrzejewska, and T.L. Powley (1990b) Simultaneous labeling of vagal innervation of the gut and afferent projections from the visceral forebrain with DiI injected into the dorsal vagal complex in the rat. J. Comp. Neurol. 301:65–79. Berthoud, H.R., N.R. Carlson, and T.L. Powley (1991) Topography of efferent vagal innervation of the rat gastrointestinal tract. Am. J. Physiol. 260:R200–R207. Berthoud, H.R., M. Kressel, and W.L. Neuhuber (1992) An anterograde tracing study of the vagal innervation of rat liver, portal vein and biliary system. Anat. Embryol. 186:431–442. Berthoud, H.R., M. Kressel, and W.L. Neuhuber (1995) Vagal afferent innervation of rat abdominal paraganglia as revealed by anterograde DiI-tracing and confocal microscopy. Acta Anat. 152:127–132. Boekelaar, A.B. (1985) The extrinsic innervation of the stomach and other upper abdominal organs in the rat. Faculteit der Wiskunde en Natuurwetenschapen, Universiteit van Amsterdam: Doctoral Thesis, pp. 1–94. Fox, E.A., and T.L. Powley (1985) Longitudinal columnar organization within the dorsal motor nucleus represents separate branches of the abdominal vagus. Brain Res. 341:269–282. Gabella, G. (1987) Structure of muscles and nerves in the gastrointestinal tract. In L.R. Johnson (ed): Physiology of the Gastrointestinal Tract, 2nd Ed. New York: Raven Press, p. 335. Gonzalez, M.F., and J.A. Deutsch (1981) Vagotomy abolishes cues of satiety produced by gastric distension. Science 212:1283–1284. Holst, M.-C., J.B. Kelly, and T.L. Powley (1997) Vagal preganglionic projections to the enteric nervous system characterized with PHA-L. J. Comp. Neurol. 381:81–100. Hölzer, H.H., and H.E. Raybould (1992) Vagal and splanchnic sensory pathways mediate inhibition of gastric motility induced by duodenal distension. Am. J. Physiol. 262:G603–608. Hudson, L.C. (1989) The location of extrinsic efferent and afferent nerve cell bodies of the normal canine stomach. J. Auton. Nerv. Syst. 28:1–14. Kressel, M., H.-R. Berthoud, and W.L. Neuhuber (1994) Vagal innervation of the rat pylorus: An anterograde tracing study using carbocyanine dyes and laser scanning confocal microscopy. Cell Tissue Res. 275:109– 123. Mesulam, M.-M. (1978) Tetramethyl benzidine for horseradish peroxidase neurohistochemistry: A noncarcinogenic blue reaction-product for visualizing neural afferents and efferents. J. Histochem. Cytochem. 26:106– 117. Neuhuber, W.L. (1987) Sensory vagal innervation of the rat esophagus and cardia: A light and electron microscopic anterograde tracing study. J. Auton. Nerv. Syst. 20:243–255. Neuhuber, W.L., M. Kressel, M. Dütsch, J. Wörl, and H.-R. Berthoud (1995) Relationships of IGLEs to enteric glia and neurons in the rat esophagus: Further indications of a mechanosensor-local effector role. Soc. Neurosci. Abstr. 21:1633. Norgren, R., and G.P. Smith (1988) Central distribution of subdiaphragmatic vagal branches in the rat. J. Comp. Neurol. 273:207–223. Phillips, R.J., E.A. Baronowsky, and T.L. Powley (1996) Reinnervation of the stomach by vagal afferents after selective vagotomy: Terminals. Soc. Neurosci. Abstr. 22:395. Powley, T.L., and F.B. Wang (1995) Mapping regional distributions of vagal projections to the gastrointestinal tract (abstract). Gastroenterology 108:A670. Powley, T.L., E.A. Fox, and H.-R. Berthoud (1987) Retrograde tracer technique for assessment of selective and total subdiaphragmatic vagotomies. Am. J. Physiol. 253:R361–R370. Powley, T.L., M.-C. Holst, D.B. Boyd, and J.B. Kelly (1994) Threedimensional reconstructions of autonomic projections to the gastrointestinal tract. Microsc. Res. Tech. 29:297–309. Powley, T.L., E.A. Baronowsky, and R.J. Phillips (1996) Reinnervation of the stomach by vagal afferents after selective vagotomy: Axons and bundles. Soc. Neurosci. Abstr. 22:395. 270 Prechtl, J.C., and T.L. Powley (1985) Organization and distribution of the rat subdiaphragmatic vagus and associated paraganglia. J. Comp. Neurol. 235:182–195. Prechtl, J.C., and T.L. Powley (1987) A light and electron microscopic examination of the vagal hepatic branch of the rat. Anat. Embryol. 176:115–126. Prechtl, J.C., and T.L. Powley (1990) The fiber composition of the abdominal vagus of the rat. Anat. Embryol. 181:101–115. Pugh, W.W., and M. Kalia (1982) Differential uptake of peroxidase (HRP) and peroxidase-lectin (HRP-WGA) conjugate injected in the nodose ganglion of the cat. J. Histochem. Cytochem. 30:887–894. Robertson, B., B. Lindh, and H. Aldskogius (1992) WGA-HRP and choleragenoid-HRP as anterogradely transported tracers in vagal visceral afferents and binding of WGA and choleragenoid to nodose ganglion neurons in rodents. Brain Res. 590:207–212. Rodrigo, J., C.J. Hernandez, M.A. Vidal, and J.A. Pedrosa (1975) Vegetative innervation of the esophagus. II. Intraganglionic laminar endings. Acta. Anat. 92:79–100. Rodrigo, J., J. De Felipe, E.M. Robles-Chillida, J.A. Pèrez Antòn, I. Mayo, and A. Gòmez (1982) Sensory vagal nature and anatomical access paths R.J. PHILLIPS ET AL. to esophagus laminar nerve endings in myenteric ganglia. Determination by surgical degeneration methods. Acta Anat. 112:47–57. Smith, G.P., E.A. Rauhofer, and J. Gibbs (1995) CCK-8 inhibits meal size in rats that have only the hepatic-duodenal vagal branch intact. 1994 Benjamin Franklin/Lafayette Symphaguim (abstract). Appetite 24:93. Sugitani, A., P.E. Donahue, M.D. Doyle, K. Anan, and L.M. Nyhus (1993) The ipsilateral organization of the afferent nerves to the stomach. J. Surg. Res. 54:212–221. Walls, E.K., R.J. Phillips, F.B. Wang, M.-C. Holst, and T.L. Powley (1995a) Suppression of meal size by intestinal nutrients is eliminated by celiac vagal deafferentation. Am. J. Physiol. 269:R1410–1419. Walls, E.K., F.B. Wang, M.-C. Holst, R.J. Phillips, J.S. Voreis, A.R. Perkins, L.E. Pollard, and T.P. Powley (1995b) Selective vagal rhizotomies: A new dorsal surgical approach used for intestinal deafferentations. Am. J. Physiol. 269:R1279–1288. Wang, F.B. (1995) Inventory and distribution of vagal afferent projections in the gastrointestinal tract of rat. Purdue University: unpublished dissertation. Wang, F.B., and T.L. Powley (1994) Taxonomy of vagal afferent projections to the gastrointestinal tract. Soc. Neurosci. Abstr. 20:1375.