THE JOURNAL OF COMPAR4TlVE NEUROLOGY 301:65-79 (1990)

SimultaneousLabeling of Vagal hervation of the Gut andMerent Projections h m the Visceral Forebrain With Dil Injected Into the Dorsal Vagal Complex in the Rat HANS-RUDOLFBERTHOUD, AGNES JEDFUEJEWSKA, AND TERRY L. POWLEY Laboratory of Regulatory Psychobiology (H.R.B., T.L.P.), Department of Psychological Sciences,Purdue University, West Lafayette, Indiana 47907; Laboratory of Experimental Neuropathology (A.J.), Institute of Psychiatry and Neurology, Warsaw, Poland

ABSTRACT The vagal innervation of the different layers of the rat gastrointestinal wall was identified with the fluorescent carbocyanine dye Dil, injected into the dorsal motor nucleus of the vagus (dmnX). Multiple, bilateral injections were used to label all dmnX preganglionic motoneurons, and as a consequence, most of the vagal primary afferents that terminate in the adjacent nucleus of the solitary tract (nts) were also retrogradely and transganglionically labeled. With Fluorogold used to label the enteric nervous system completely and specifically,the Dil-labeled vagal profiles could be visualized and quantified in their anatomical relation to the neurons of the myenteric and submucous ganglia. In the myenteric plexus, vagal fibers and terminals were found throughout the gastrointestinal tract as far caudal as the descending colon, but there was a general decreasing proximodistal gradient in the density of vagal innervation. All parts of the gastric myenteric plexus (fundus, corpus, antrum), as well as the proximal duodenum, were extremely densely innervated, with vagal fibers and terminals in virtually every ganglion and connective. Further caudally, both the percentage of innervated myenteric ganglia and the average density of label within the ganglia rapidly decreased, with the exception of the cecum and proximal colon, where up to 65% of the ganglia were innervated. In the gastric and duodenal submucosa very few and in the mucosa no vagal fibers and terminals were found. With both normal epifluorescenceand laser scanning confocal microscopy, highly varicose or beaded terminal structures of various size and geometry could be identified. The Dil injections, which impregnated the dmnX as well as the adjacent nts, resulted in retrograde and anterograde labeling of all the previously reported forebrain connections with the dorsal vagal complex. We conclude that the myenteric plexus is the primary target of vagal innervation throughout the gastrointestinal tract, and that its innervation is more complete than previously assumed. In contrast, vagal afferent (and efferent) innervation of mucosa and submucosa seems conspicuously sparse or absent. Furthermore, the use of more focal injections of Dil offers the prospect to simultaneously identify specific subsets of vagal preganglionics and their central nervous inputs. Key words: entericnervous system, stomach, cecum, myenteric plexus, dorsal motor nucleus of vagus, nodose ganglion

Much progress has recently been made in understanding the functional anatomy of the central and peripheral neural structures that guarantee normal functioning of the gastrointestinal tract and other abdominal viscera. In particular, direct descending connections from various forebrain structures to the autonomic preganglionic motor nuclei in the dorsal vagal complex (Saper et al., '76; van der Kooy et

o 1990 WILEY-LISS, INC.

al., '84) and spinal cord (Kuypers and Maisky, '75) have now been characterized anatomically as well as neurochemically, and their physiological role has been shown in functional studies (Rogers and Hermann, '85, '87). At the peripheral end much has been learned about connectivity Accepted July 26,1990.

H.-R. BERTHOUD ET AL.

66 and neurochemistry (peptides) of the enteric nervous system and its interface with the spinal sympathetic innervation (e.g., Costa et al., '87). However, many anatomical and functional details of the interface between the vagal parasympathetic and the enteric nervous systems are still poorly understood. Only very recently, Kirchgessner and Gershon ('89) successfully demonstrated, using the PHA-L method, vagal efferent fibers and terminals in the gut wall and their topographical relationship to immunocytochemically identified enteric neurons. Earlier work, using silver staining for degeneration after vagotomy (Schofield, '62; Winogradova, '62; Mei, '78; Rodrigo et al., '75, '821, anterograde vagal transport of tritiated amino acids with subsequent autoradiography (Connors, '83; Sat0 and Koyano, '87), or HRP labeling of nodosal afferents (Clerc and Condamin, '87; Neuhuber, '87) was either limited by poor anatomical resolution, or by analysis of only select segments of the GI tract, and/or by an inability to co-visualize the enteric nervous system. We therefore used the recently introduced fluorescent cationic membrane probe "Dil" (Honig and Hume, '89) to label anterogradely vagal fibers and terminals in the gut, in combination with a new Fluorogold method that labels the entire enteric nervous system (Powley and Berthoud, '90). Because the emission spectra of the two fluorophores are completely separable with appropriate barrier filters, this double labeling strategy is particularly well suited to assess the topographical relationship between vagal fibers and enteric ganglia. Dil appeared to offer the additional features that it would: a) serve simultaneously as a strong retrograde label for central projections to the injection site, b) label all neurons within an injection site, and c) work concurrently with the retrograde labeling strategy we have extensively employed to identify specific pools of vagal preganglionic somata in the medulla (Powley et al., '87). In order to avoid the problem of false-negative findings in any part of the GI tract, the objective in this initial experiment was to label all neurons of the dorsal motor nucleus of the vagus (dmnX) which, in the rat, is the sole source of vagal preganglionics to the gut. This automatically meant that the adjacent nucleus of the solitary tract (nts) was also exposed to the tracer, resulting in retrograde transport to, and presumably past, the nodose ganglion in vagal primary afferents.

MATERZALSANDMETHODS Fifteen Male Sprague Dawley rats weighing between 160 and 300 g were housed under standard colony conditions, including a 12:12 hour LD schedule, constant temperature of 24"C, and ad libitum access to rat chow and water.

Injection

bocyanine perchlorate, Molecular Probes, in 100% EtOH) was then injected into each site by using a hamilton 1 ~1 syringe connected to the pipette by a length of PE 10 tubing. Typically, on each side of the brain, three to four sites were injected at different rostrocaudal levels. As needed the animals were manually ventilated to overcome the transitory respiratory depression occasionally observed. Muscle and skin were then separately sutured, and Nitrofurazone powder was applied to the wound. Three days before sacrifice some of the animals were given an i.p. injection of 5 mg of the retrograde tracer Fluorogold (Fluorochrome, Inc.) in 1.2 ml sterile saline, for purposes of labeling the enteric ganglia (Powley and Berthoud, '90) as well as the vagal preganglionic cell bodies in the dmnX (Powley et al., '87). Initially, survival periods of 15-60 days were allowed for Dil transport. Since no difference was detected in intensity and completeness in Dil labeling, even in the distal colon, survival times of 20-40 days were subsequently adopted.

Tissuepreparation Animals were anesthetized with an overdose of pentobarbital sodium (100 mg/kg) and, when fully unresponsive, perfused through the heart with phosphate-buffered (pH = 7.4) 10% formalin. The gastrointestinal tract was removed, slit near the mesentery attachment, rinsed, and stored in the buffered formalin solution. The cervical vagi, including the nodose ganglia, were also stored in formalin, and the entire brain in sucrose-formalin. One to several days later samples of 50 to 250 mm' were taken in a systematic fashion from all segments of the GI tract: from the ventral and dorsal sides of the nonglandular (fundus) and glandular (corpus and antrum) stomach, the duodenum at 1, 3, 5, and 8 cm from the pyloric sphincter (referred to as Dl-D8), the jejunum at 20,40,60,and 80 cm (J20480), the ileum at 5 cm from the ileocecaljunction, the ventral and dorsal sides of the apex and corpus of the cecum, the ascending, transverse, and descending colon, and finally from the proximal rectum. The samples were rinsed in phosphate-buffered saline (PBS, pH = 7.4) and dissected in a Sylgaard-coated petri dish containing PBS. Typically, with the serosal side down, the mucosa was first carefully scraped off by using a #10 scalpel blade, and the submucosa was then peeled off the smooth muscle layers under stereomicroscopic guidance. In some instances the external muscle layers were first peeled off, leaving a submucosa plus mucosa preparation that was subsequently sectioned in a cryomicrotome. The nodose ganglia and cervical vagi were stripped of their connective tissue capsules, but not sectioned. Finally, the brains were divided into three blocks (medulla, midbrain, and forebrain) and sectioned frontally at 56 pm in a cryomicrotome. All tissues were then dehydrated through a series of graded percentages (70%, 90%, and 2 x 100%) of glycerin, and finally mounted and coverslipped in 100% glycerin to which n-propyl gallate (2-5%, in Tris buffer, pH 7.8) had been added as an antifade agent. Before mounting, some of the specimens were also counterstained with 0.0001% bisbenzimide.

Each animal was anesthetized with pentobarbital sodium (60 mgikg i.p.1, treated with atropine (1mg/kg), and placed in a stereotaxic instrument with a head holder adapted to permit the neck to be sharply flexed. A dorsal incision was made over the neck muscles, and they were retracted to expose the dura overlying the fourth ventricle. The membrane was excised, exposing the cisterna magna and the Analysis surface features in the region of the obex. By using the obex to determine stereotaxic coordinates, a glass micropipette All tissue specimens were thoroughly examined with an (tip diameter 30-60 Fm) was then successively advanced to epifluorescence microscope equipped with filter cubes approseveral sites within the dmnX. Fifty nanoliters of a 3% priate for Dil (Leitz NJ, Fluorogold (A), and/or bisbenzimsolution of Dil(1,1'-dioctadecyl-3,3,3',3'-tetramethylindocar-ide (Ha).Myenteric plexus containing samples from all parts

VAGAL INNERVATION OF GI TRACT IN RATS

67

of the GI tract were systematically analyzed for Dil label in all animals. Submucosa samples from the duodenum were also analyzed in all animals, and from the glandular stomach and jejunum/ileum in selected animals only. Mucosa samples from stomach and small intestines were only spot checked in a few animals. In order to estimate the percentage of vagally innervated myenteric ganglia, sam~ in the ples from six rats were examined with a 2 5 objective following manner: in several scans (three to six) through the specimen, myenteric ganglia or ganglia rows were first identified with the appropriate filter cube for Fluorogold, and then checked for any vagal innervation by switching to the filter cube for Dil. Labeled fibers that merely passed through a ganglion without evidence for terminals were not

In overview, virtually all preganglionic somata in the dmnX were labeled, and Dil was anterogradely transported in the motor rootlets near the ventrolateral medulla, the fibers of the cervical vagus, and vagal branches penetrating the serosal surface of the gut wall. Concurrently, most of the vagal afferents including their somata in the nodose ganglia, as well as the descending CNS projections afferent to the dorsal vagal complex, were retrogradely labeled.

Fig. 1. A Photomicrograph of frontal section through dorsal medulla showing extent of diffusion of typical Dil injection (50 nl) into right dorsal motor nucleus. A’: Same brain section as in A viewed with different filter to show the retrogradely Fluorogold-labeled vagal preganglionics of the dmnX that remained viable following the Dil injection. B: Schematic diagram of the dorsal medulla showing the smallest and largest areas of Dil diffusion and labeling. ap = area postrema, subnuclei of the nucleus of the solitary tract: cen = central, com =

commissural, dl = dorsolateral, gel = gelatinosus, int = intermediate,v = ventral, vl = ventrolateral. dmnX = dorsal motor nucleus of the vagus, nXII = hypoglossal nucleus, ts = solitary tract. C : Photomicrograph of whole mount of nodose ganglion and cervical vagus showing the complete retrograde labeling of the vagal sensory neurons and the fibers of the cervical vagus with Dil. The inset (D) demonstrates that the peripheral neurites of nodosal neurons are also labeled. Scale bars: A,A’,D 100 ym, C 200 ym.

considered as innervation. A minimum of 100 ganglia at each jejunumiileum location and 50 ganglia elsewhere were examined for each rat.

RESULTS

H.-R. BERTHOUD ET AL.

68

Injection site and uptake of Dil b y v d efferents and derents The multiple bilateral injections successfully blanketed the extent of the dmnX as well as parts of adjacent nuclei with label. All perikarya as well as the neuropil were filled with granular accumulations of Dil (Fig. lA,B). Although Dil label was often so heavily concentrated in both the somata and the surrounding neuropil that it was difficult to resolve details of any individual soma, the concurrently employed Fluorogold label made it possible to visualize the preganglionics and then, by switching between appropriate filter cubes, to determine that the cells were labeled and that the entire dmnX fell within the effective sphere of the Dil injection (Fig. 1A‘).Additionally, the Fluorogold administered at the end of the long Dil survival period established that the heavily (Dil) labeled neurons were still viable and apparently uncompromised by the tracer. There was no evidence of a progressive disruption of cells with longer survival times. (Our observations on the general health, food consumption, and body weight of the animals suggested that Dil can be used as a supravital stain.) There was, however, a lack of Fluorogold label around the center of each injection site, indicating a sphere of necrosis of up to 300 Fm in diameter, that presumably was produced by the ethanol used as a vehicle. From this damage it was estimated that maximally (i.e., if the center of each injection was located within the dmnX) 20% of all preganglionics were destroyed. Finally, the efferent axons of the dmnX which course ventrolaterally from the lateral pole of the dmnX to the surface of the medulla were also conspicuously defined by the Dil label. Both nodose ganglia were virtually completely labeled (Fig. 10. No attempt to count the cells was made, but switching between the Fluorogold-counterstained ganglionic profile and the Dil-labeled cell distribution suggested that effectivelyevery cell and its proximal neurite contained granular accumulations of Dil. Individual “displaced” or ectopic somata were frequently seen in the cervical vagi peripheral to the nodose ganglia. Inspection of these neurons indicated that the label was also transported transganglionically into the peripheral neurites (Fig. 1D).We have to assume therefore, that most, if not all, vagal afferent fibers and terminals in the gut wall were labeled.

Vagal fibers and terminalsin the gut wall Dil-labeled vagal fibers and terminals were found in all sections of the gastrointestinal tract except the rectum (Table 1). Labeling of vagal fibers and terminals was abundant in the myenteric plexus, very sparse in the submucous plexus, and absent in the mucosa. The density of the vagal innervation of the myenteric plexus was highest in the stomach, particularly the glandular portion, and in the first part of the duodenum; intermediate in the rest of the small intestine, the cecum and ascending colon; and sparse in the distal colon (Table 1). The Dil label had generally a granular appearance, with the individual granules being very brightly fluorescent and thus easily distinguishable from the pale red of background autofluorescence sometimes present due to incomplete removal of the red blood cells. Stomach. In samples from the stomach wall close to the lesser curvature large nerve bundles containing many Dil-labeled fibers could be seen as they penetrated the

TABLE 1. Percentage’ of VagaUy InnervatedMyenteric Ganglia in Different Sedions of the GI Tract % Stomach Duodenum

Jejunum

Ileum Cecum Colon

Rectum

Fundus corpus Antrum D1 D3 D5 DR 520 540 J60 JRO Distal Apex corpus Ascending Transverse Descending P r o d

100 100 100 9625 R6+ 11 73 i7 5659 24 -t 5 40 k 7 40 + 4 35 2 5 35+ 4 66 i I 65 f 7 51 ? 4 20 i 6 1626 0

*

‘Means SEM of six rats. Labeled fibers that merely passed through a ganglion without evldence for terminals were not considered as innervation. For further details of analysis see Methods section. Note that this analysis does not take into account the number of individual enteric neurons contacted by vagal fibers in each ganglion. Generally this number also decreased at more distal gut locations,thus amplifylngthe decreasingproximudistal gradient of the overall density of veal innervation.

serosa and outer muscle layer and entered the myenteric plexus. Within the myenteric plexus of all parts of the stomach labeling was generally so dense and complete that the profile of the entire plexus (ganglia and connectives) was clearly delineated and each of the enteric neurons was defined in relief (Figs. 2B,C,C‘, 3A,A’).Because of the great number of intertwining axons that effectively obscured one another’s course, it was generally not possible to distinguish single axons from small bundles of more than one axon. This was particularly true for the glandular stomach (corpus and pyloric antrum) where the connectives contained the largest number of fibers (Fig. 2C,C’). In the nonglandular, fundic part of the stomach with less fibers, individual axons could be followed more easily (Fig. 3A‘-C). Also, because of the high density of Dil label in the stomach, it was usually not possible to determine the occurrence and degree of axon collateralization. The most easily recognizable terminals were highly varicose axons, often with large-diameter varicosities, giving the impression of a chain of beads (referred to as varicose or beaded terminals). They were short to very long, nonbranching or little branching, and either tightly hugging individual ganglion cells (Figs. 3A‘,C, 7A,F,G,J), or contacting many neurons on their meandering course through a ganglion (Fig. 2D,E;). They often ended abruptly in an enormous club or bouton of up to 10 km in diameter (Fig. 7D), or several such large varicosities were interspersed with smaller ones (Fig. 2D). In some instances part of the terminal seemed not to be close to myenteric ganglion cells, but rather to be in the connective or outside the plexus in the smooth muscle layers (Fig. 3B). Because of the granular or punctate Dil label of fibers of passage, it was more difficult or often impossible to distinguish them from terminals with very small varicosities. Again, this was particularly true for the heavily innervated ganglia in the glandular stomach, where the numerous fibers passing between the ganglion cells could potentially give the impression of small varicose terminals surrounding each cell (Fig. 2C’,E).Only in cases of less dense label (Fig. 7A), or single isolated myenteric ganglion cells (Figs. 3A‘,C, 7F,J), could the terminal character of such rings be clearly established.

VAGAL INNERVATION OF GI TRACT IN RATS

Fig. 2. Whole mounts of external smooth muscle layers of glandular stomach wall showing Dil-labeled vagal fibers and terminals and/or Fluorogold-labeled ganglia of the myenteric plexus. A: Low-power view of myenteric plexus from corpus of stomach. B: Low-power view of Dil-labeled vagal fibers and terminals in myenteric plexus of corpus. Note that virtually every ganglion and connective contains vagal profiles. C,C': Typical example of myenteric ganglion from corpus of stomach as double labeled by Fluorogold and Dil, respectively. Many fibers pass through the ganglion, while others apparently form varicose

69

ring-like terminals around individual ganglion cells. D Higher-power double exposure showing large-diameter varicose vagal terminal coursing through myenteric ganglion in gastric corpus. The ganglion cells are labeled by Fluorogold. E: Example of vagal innervation of myenteric ganglion in pyloric antrum of stomach. The double exposure shows Dil-labeled vagal fibers and terminals in white on background of Fluorogold-labeled ganglion cells (grey).Scale bars: A,B 100 pn,C,E 50 km, D 20 km.

70

Fig. 3. AA’: Typical example of myenteric ganglion from nonglandular (fundic) stomach double labeled by Fluorogold and Dil, respectively. B,C: Double exposures of fundic myenteric ganglia showing Dil-labeled vagal fibers and terminals (bright white) and Fluorogoldlabeled ganglion cells (greyish). Note the isolated enteric neurons surrounded by Dil-labeled fibers (arrowheads in A,A’, and C) D,E,E’:

H.-R. BERTHOUD ET AL.

Vagal innervation of submucous plexus in glandular stomach. D: Low-power overview of Fluorogold-labeled small groups or single enteric neurons scattered in the submucosa. Blood vessels (bv) are also outlined. E,E’: High-power view of same two-celled ganglion showing Fluorogold-labeled perikarya and Dil-labeled vagal varicose terminal coiling around them. Scale bars: A-C 50 pm, D 100 ( ~ mE, 20 km.

VAGAL INNERVATION OF GI TRACT IN RATS

Fig. 4. Vagal innervation of myenteric (A-D) and submucous (F,G) plexus in duodenum. A,A’: Low-power views of myenteric plexus from mid-duodenum (D5) showing only the Fluorogold-labeled ganglia (A) and double exposure with the additionally Dil-labeled vagal innervation (A’). Note that the myenteric plexus is organized in circular (vertical) rows of ganglia and that several vagal fibers or small fiber bundles (bright white) run longitudinally (horizontally) through the rows. B: Only the dense Dil-labeled vagal innervation of the proximal duodenum (D3) is shown in this photomicrograph. C,D: Higher-power double exposures showing vagal fibers and terminals innervating myenteric

71

ganglia at D3 ( C ) and D8 (D). E: Overview of duodenal submucous plexus showing Fluorogold-labeled ganglia that are not organized in rows. Blood vessels (bv) are also outlined. F: Dil-labeled vagal varicose terminal fiber in duodenal submucous plexus. G Low-power view of sectioned submucosa (sm) and attached mucosal villi (v). A Dil-labeled vagal terminal can be seen tightly hugging a Fluorogold-labeled submucosal ganglion cell. H Fluorogold-labeled cells in lamina propria of duodenal villus. No vagal innervation was seen. Scale bars: A,B,E,G 100 pm, C,D,F,H 50 pm.

72

In contrast to the relatively simple, clearly defined geometry of the terminals described above, more complex, less well-defined structures were also frequently observed, and tentatively classified as a distinctive type of terminal. They consisted of clusters of small varicosities or boutons connected by fine fibers, and are referred to as profusely arborizing terminals. Because there was a large brightness differential between the fibers and boutons, connectivity was difficult to establish, particularly in the stomach, where numerous other fibers obscured the view. (An example of a diffusive terminal is shown in Figure 7E from the myenteric plexus of the cecum.) The submucosa of the glandular stomach was very sparsely innervated by Dil-labeled vagal elements found only upon careful screening of the Fluorogold-labeled ganglion cells of the submucous plexus (Fig. 3D,E,E‘)at higher magnification. The submucous ganglia in the glandular stomach were much more scattered and contained fewer (typically one to three) neurons than those in the small intestines. A minority of these cell groups were approached by extremely thin vagal fibers producing varicose terminals that tightly hugged the neurons (Fig. 3E,E’, 7C). In the limited number of mucosa samples analyzed no Dil label was seen. Duodenum. In the proximal (2-4 cm) segment of the duodenum, Dil labeling in the myenteric plexus was quite complete, meaning that, as in the stomach, virtually every ganglion or circumferential row of ganglia received at least some vagal innervation (Table 1; Fig. 4A’,B). The average density of label within myenteric ganglia of the duodenum, however, was less than in the stomach. More distally in the duodenum, there was a rapid decrease in both the percentage of myenteric ganglia with label and the average density of label within the labeled ganglia (Table 1; Fig. 4C,D). Individual and small bundles of fibers were typically seen running longitudinally near the pancreatic attachment in the proximal duodenum. They often ran through numerous rows of the circularly oriented rows of myenteric ganglia, giving off collaterals to most of them (Fig. 4A’,B). In the submucous plexus of the duodenum, which consisted of a nonoriented network of small ganglia (5-10 neuronsiganglion, Fig. 4E), extremely few Dil-labeled vagal profiles were seen (Fig. 4F,G). Fluorogold-labeled cells occurred regularly in the lamina propria of the duodenal mucosa, but no Dil label was found around them (Fig. 4H). Jejunum and ileum. Both of these sections of the bowel were characterized by a more sparse label. Only 24-40% of the myenteric ganglia contained labeled vagal profiles, and then often only a few of the neurons within the labeled ganglion appeared to be contacted by vagal terminals (Table 1; Fig. 5 ) . In these sections of the GI tract (as well as the distal part of the duodenum), labeling had a “patchy” appearance apparently established by the terminal fields of separate vagal fibers. That is to say, individual fibers, or small bundles of fibers, could be observed entering the intestinal wall and then coursing through, making contacts, and/or sending collaterals into a number of neighboring ganglia (Fig. 5B,C,F, and schematic in 5G). This terminal zone might then be separated from the next innervated patch by an equally large or larger area without labeled vagal fibers and terminals. These terminal distributions or patches had another feature which may reflect the organization of the vagal projections to the gut. Labeled vagal fibers appeared to enter the gut wall from the mesentery attachment and

H.-R. BERTHOUD ET AL. typically distributed over one side or the other of the viscus towards the antimesenteric pole as well as longitudinally (Fig. 5G). As a result, the average percentage of innervated ganglia was higher in the side wall than at the antimesentery pole. Basically, the same types of terminal specializations as in the stomach could be seen in the small intestines (Figs. 5D,E, 7B), but large-diameter beaded or club-like terminals were rarely seen. No Dil-labeled vagal profiles were seen in the limited number of mucosal samples analyzed. Cecum and colon. In an exception to the gradual decrease in innervation density towards the distal alimentary canal, the cecum and ascending colon showed a remarkably high density of vagal innervation. More than half of the ganglia in the cecal and colonic myenteric plexus showed evidence of at least some vagal innervation (Table 1).This percentage decreased rapidly to only 16% in the descending colon. Individual Dil-labeled fibers were typically of very fine caliber, giving off collaterals to many ganglia, and could often be followed over large distances, if appropriately magnified. Within these ganglia the percentage of contacted neurons was highly variable (Figs. 6B,C,D, 7E), and basically the same types of terminal specializations were found as elsewhere (Figs. 6D-F, 7E).

CNS aerents to the dorsalvagal complex The paraventricular nucleus of the hypothalamus, particularly its medial and lateral parvicellular portions, was the most prominently labeled CNS area. At more caudal levels the most lateral portions of the paraventricular nucleus reached the most medial part of the internal capsule (Fig. 8E). Because the label was intense and filled the proximal dendrites, the architecture of many individual cells could be discerned at higher magnification (Fig. 8F). A few brightly labeled cells were scattered in the ventromedial hypothalamus, including parts of the retrochiasmatic area, the arcuate nucleus, and the ventromedial nucleus (Fig. 8E,G,H). At more caudal levels of the hypothalamus, clusters of well-labeled cells were also found in the dorsomedial nucleus and lateral hypothalamic area, as well as scattered between the two areas (Fig. 8G). A particularly dense cluster of brightly labeled cells, which rivalled the paraventricular nucleus in intensity, occupied the most posterior and lateral corner of the lateral hypothalamic area (Fig. 8H). A very small group of relatively brightly labeled cells was also routinely seen in the ventrolateral subnucleus of the bed nucleus of the stria terminalis, and immediately medial to this group, just ventral to the anterior commissure, was an area of brightly anterogradely labeled terminals in the ventral subnucleus (Fig. 8C,D). Scattered, more moderately labeled cells were also found dorsal to the anterior commissure in the dorsolateral subnucleus of the bed nucleus of the stria terminalis (Fig. 8C). The central nucleus of the amygdala, particularly its medial division, was labeled by a dense cluster of cells of intermediate brightness, without delineation of proximal dendrites (Fig. SE), and a smaller cluster of cells was seen in the substantia innominata, ventral to the internal capsule (Fig. 8E). Finally, relatively weakly labeled cells were found in the deeper layers of the agranular insular cortex, dorsal

VAGAL INNERVATION OF GI TRACT IN RATS

Fig. 5. Vagal innervation of myenteric plexus in small intestine. Single exposures show only Dil-labeled vagal fibers and terminals (A,F); double exposures additionally show Fluorogold-labeled rows of ganglia that are oriented in circular (vertical) direction. A: Vagal fibers in proximal jejunum (520). B: Single vagal fiber runs longitudinally (horizontally) through several rows of ganglia, supplying each with a collateral (540).C: Collateralizing vagal fiber in myenteric plexus of distal jejunum (J60).D,E: Higher-power views of two examples of vagal fibers and varicose terminals in far distal jejunum (580). F: Vagal

73

innervation of ileum. Several intertwining s o n s and their terminal trees are shown at low power. G Schematic view of small intestinal wall showing general pattern of vagal innervation of myenteric plexus (MP). Single vagal fibers or small bundles penetrate the serosa and longitudinal muscle layer (LM)at the mesentery attachment together with blood vessels (BV), and innervate several rows of ganglia in a patch-like pattern. Large areas around an “innervated patch” can be devoid of any vagal fibers. CM = circular muscle coat; SM = submucocsa; M = mucosa. Scale bars: A,B,C,F 100 pm, D,E 50 pm.

74

Fig. 6. Vagal innervation of myenteric plexus of large intestines. A: Low-power overview of Fluorogold-labeled myenteric plexus in cecum. B: Single exposure showing dense Dil-labeled vagal innervation of individual ganglion in cecum. C: Double exposure showing Dil-labeled vagal fibers and terminals (bright white) and Fluorogold-labeled myenteric ganglion (greyish) in cecum. D: Single Dil-labeled vagal fiber and

H.-R. BERTHOUD ET AL.

coiling terminal (bright white) in Fluorogold-labeled myenteric ganglion of distal colon, demonstrating the very sparse innervation at this level. E,F: Higher-power views of two examples of Dil-labeled vagal fibers and terminals in myenteric plexus of colon. Scale bars: A,D 100 pm, B,C 50 pm, E,F 20 km.

VAGAL INNERVATION OF GI TRACT IN RATS

Fig. 7. Dil-labeled, vagal terminal structures (bright white) shown in relation to Fluorogold-labeled enteric neurons (greyish). A-C: Examples of varicose or beaded terminals from the myenteric plexus of the nonglandular stomach (A), the distal jejunum (B), and the submucous plexus of the glandular stomach (C). D: Example of club-like terminal in the myenteric plexus of the glandular stomach with a very large terminal varicosity. E: Example of more complex, profusely arborizing terminal structure in the cecum. F:Vagal terminal forming

75

tight ring around isolated neuron of the gastric myenteric plexus. G-J: Gastric myenteric ganglia, optically sectioned with confocal, scanning laser microscopy, showing profiles of Dil-labeled vagal terminals in close contact with Fluorogold-labeled enteric neurons. In contrast to normal photomicrographs, Fluorogold appears in fine granules in the cytoplasm and can be easily distinguished from the larger Dil-labeled vagal profiles. Scale bars: 20 km.

H.-R. BERTHOUD ET AL.

76

Figure 8

VAGAL INNERVATION OF GI TRACT IN RATS

77

sory vagotomy, concluded that vagal sensory fibers terminate in duodenal villi in cats. Again, specific efferent vagal label has not been found in the rat mucosa (Kirchgessner and Gershon, ’89). From a functional point of view, it has DISCUSSION been argued that the vagal sensory endings that have been Vagal innervation of the gut wall demonstrated to respond to luminal metabolic substrates, The most striking finding in this study was the high pH, and osmolarity would best be situated close to the density of vagal fibers and terminals, particularly in the mucosal epithelium (e.g., Mei, ’85).We have not found any myenteric plexus of stomach and proximal duodenum. This Dil-labeled vagal fibers and terminals in gastric and duodeabundance of labeled vagal elements is in sharp contrast to nal mucosa even though we could identify Fluorogoldthe small number seen in any part of the gut by Kirchgess- labeled cells, which may be neurons, in the villous plexus. ner and Gershon (’89),who used PHA-L as an anterograde We do not know whether the discrepancy is due to a species tracer. One explanation for the discrepancy is that the differenceor to the inability of the Dil method to sufficiently additional label we observed is the vagal sensory compo- label these mucosal vagal afferents thought to be very thin nent, which was apparently not labeled in the PHA-L free nerve endings (Sat0and Koyano, ’87). In some specimens from the small intestine we found experiment. Since 70-80% of the fibers in the abdominal vagus are sensory (Prechtl and Powley, ’901, one would areas of up to 20 mm without any Dil-labeled vagal fibers expect to see considerably more labeling with a bidirection- and terminals, suggesting that the access route of vagal ally transported label such as Dil, unless the degree of innervation to the more distal parts of the small intestine is collaterlization for efferent fibers is much higher. A second via the mesentery. This assumption seems reasonable since explanation for the lighter labeling observed by Kirchgess- the vagal celiac branches, which have been shown to ner and Gershon (’89) may be that they labeled fewer vagal innervate the small intestine (Stavney et al., ’63; Berthoud motor neurons in the dmnX. This seems probable because et al., ’891, diverge from the gut axis at the esophagogastric PHA-L typically labels only a fraction of the exposed junction to take a more dorsal course, close to the celiac and neurons, thus accounting for the Golgi-likestaining pattern superior mesenteric ganglia (e.g., Boekelaar, ’85).However, at the injection site. Furthermore, in the PHA-L study, only by cutting the circumference of the intestinal wall at the mesentery attachment, we may have missed some vagal one injection (and sometimes only on one side) was made. A second important finding of our study was that the fibers descending there within the bowel, as suggested by entire gastrointestinal tract, including the nonglandular or Kirchgessner and Gershon (’89). Our combined DiVFluorogold protocol allowed us to fundic part of the stomach, the cecum, as well as the ascending, transverse, and descending colon, was vagally describe a t least one type of vagal terminal in terms of both innervated. This is consistent with earlier reports providing its architecture and its relationship to the enteric nervous functional evidence for such a complete vagal innervation of system. The main architectural characteristics were the the GI tract (Stavney et al., ’63; Rostad, ’73; Berthoud et small degree or complete absence of terminal arborization and the often large size of its constituent varicosities. The al., ’89). A third significant finding was the relatively weak vagal enormous variations in these and other geometrical paraminnervation of the submucous plexus and the apparent eters, like length, direction, and varicosity size and number, absence of vagal innervation of the mucosa. Sparse submu- would argue that there are in fact a number of different cosal labeling of vagal afferents following tritiated amino types of terminals hidden within this general description. acid or WGA-HRP injections into the nodose ganglion has This would also be supported by the observed differences in been reported in the rabbit stomach and duodenum (Sat0 these terminals’ appositions to enteric ganglia and neurons. and Koyano, ’87) and the rat esophagus (Clerc and Con- While some terminals seem to specifically “seek out” damin, ’87; Neuhuber, ’87).Vagal efferent innervation has individual enteric neurons in the middle of a ganglion, not been found in the submucosa of rat stomach and small others seem to contact a great number of neurons, and still intestine (Kirchgessner and Gershon, ’89).These observa- others seem to lie in connectives between the ganglia, or in tions would suggest that the submucosal label found in the the smooth muscle layers outside the myenteric plexus. It is present study was of vagal afferent origin. likely that such differences also reflect functional diversity. In the mucosa, Sat0 and Koyano (’87),using autoradiogIn the past, large, “club-like’’ terminals, similar to the raphy, found labeled fibers penetrating deep into the villi of ones seen in the present study (Fig. 7D), have been rabbit stomach and duodenum, and Mei (’781, on the basis observed throughout the autonomic nervous system in of a reduction of silver-stained fibers following vagal sen- many different species by using variations of the silver

to the rhinal fissure (Fig. 8A,B),and in some animals in the medial prefrontal cortex (not shown).

Fig. 8. Retrograde (and some anterograde) label in “visceral forebrain” structures following bilateral large Dil injections into the dorsal motor nucleus of the vagus. A,B: Low- and high-power views of Dil-labeled neurons in the deeper layers of the granular insular cortex, just dorsal to the rhinal fissure (RF). C,D: Dil label in different subnuclei of the bed nucleus of the stria terminalis from frontal sections at the level of the anterior commissure (ac). Weakly labeled cells in the dorsolateral subnucleus, stretching from the lateral ventricle (LV) to the lateral tip of the anterior commissure (ac). A few more brightly labeled cells lie within the ventrolateral subnucleus, lateral to a field of intensely anterogradely labeled terminals in the ventral subnucleus, just ventral to the anterior commissure. E: Montage of frontal section through anterior hypothalamus showing brightly labeled

perikarya in the paraventricular nucleus of the hypothalamus, including its most lateral portion, the central nucleus of the amygdala on the left, the substantia innominata just ventral to the internal capsule (ic), and the retrochiasmatic area at the bottom of the third ventricle (3V). F: Higher magnification of paraventricular nucleus showing heaviest label in medial and lateral parvicellular subnuclei. G . Montage from section through mid-hypothalamus at the level of the ventromedial nucleus (vmn), with labeled perikarya in the dorsomedial nucleus, the lateral hypothalamic area, and scattered between the two. A few brightly labeled cells are also in the most ventral parts of the ventromedial nucleus (vmn). H: Montage from sections through posterior-most hypothalamus showing a compact cluster of brightly labeled cells in the lateral hypothalamic area. Scale bars: A,C,E,G,H 100 pm, B,D,F 50 km.

78

H.-R. BERTHOUD ET AL.

impregnation method (Jabonero, '55; Kirsche, '55; Winogradova, '62; Rodrigo et al., '75). However, the beaded terminals with many large varicosities, so characteristic for the present study, were not reported in these earlier studies. Winogradova ('62) observed the disintegration of such "club-like" terminals within a few days following vagotomy and concluded that they must be of vagal origin. Schofield ('62) also reported the presence of darkly silver stained, spherical, or spindle-shaped structures, which he called argyrophilic swellings and which were peculiar to the rat. However, since their number increased following vagotomy (and sympathectomy) he concluded that they are degenerative products or remnants of retracting fibers. He further suggested that spontaneous degenerative changes in a portion of enteric fibers could account for the presence of such swellings in unoperated rats. Is it possible that Dil labeling could have produced pathological changes in vagal terminals in our study? The immediate destruction of some vagal preganglionics by the ethanol injection vehicle cannot account for it, since the degenerative changes, including fragmentation, take place rapidly and there is no time and conduit for the relatively slowly transported Dil to reach the terminals in the gut. The possibility that Dil exerts a slow toxic action over the relatively long survival period cannot be completely ruled out. But the facts that 1) the preganglionics were still able to retrogradely transport and accumulate Fluorogold in their somata at the end of the survival period and 2) that the animals showed no physiological signs indicative of subdiaphragmatic vagotomy speak against it. Our results are consistent with the existence of a second type of terminal, but we were not able to clearly characterize its more complex architecture. In the literature, more complex terminal structures described as profusely arborizing (Neuhuber, '87), laminar or reticular (Rodrigo et al., '75), or broom-like (Milochin, '63; Kolossow and Milochin, '63), were reported in the myenteric plexus of mainly the esophagus, and were always associated with relatively large-caliber myelinated fibers. While the earlier reports were based on silver impregnation methods and did therefore not prove vagal origin, Neuhuber ('87) apparently specifically labeled vagal afferents by injecting WGA-HRP into the nodose ganglion, and hypothesized that "such vagal myenteric sensory terminals represent a parsimonious device for tension perception, acting both on intrinsic neurons and superimposed brainstem centers." It is possible that the more complex, profusely arborizing terminal we found in the myenteric plexus throughout the GI tract serves the same purpose.

by damaged but not intact axom of passage. Following Dil injections into the nodose ganglia and later Fluorogold injections into the peritoneum, we found one population of dmnX neurons that was only Dil labeled and a distinct population that was only Fluorogold labeled, but no neurons that were double labeled. The presence of Fluorogold in the cell bodies was an indicator for intact axons that were not damaged by the Dil injection (unpublished observations). Anterograde terminal label was also observed in the parabrachial nucleus, the bed nucleus of the stria terminalis, and the ventral thalamus. By aiming the injections at the dmnX, it could be expected that those forebrain projections that terminate within or very close to the dmnX would be the most complete and strongest retrogradely labeled ones. The paraventricular nucleus of the hypothalamus has the most prominent projection directly into the dmnX (van der Kooy et al., '84) and showed the heaviest Dil label. The posterolateral hypothalamus, central nucleus of the amygdala, and bed nucleus of the stria terminalis, which all have been shown to preferentially project into the ventral nts immediately dorsal to the dmnX (van der Kooy et al., '841, were also strongly labeled. In contrast, the cortical areas, which have been shown to predominantly terminate in the more rostral, dorsal, and lateral portions of the nts, were more weakly or not regularly labeled by our Dil injections. The fact that even these large injections differentially labeled the forebrain inputs encourages us to use much more focal injections in the future, which will hopefully link a limited number, or a particular pattern of forebrain inputs, to smaller pools of function-specific, vagal preganglionic motoneurons. On the other hand, the virtually complete impregnation of all the dmnX preganglionics reliably and easily achieved with larger Dil injections will be useful to settle some of the controversies surrounding the vagal innervation of other viscera such as the pancreas, liver, and spleen.

CNS connectionswiththe dorsalvagal complex

Berthoud, H.-R., N. Carlson, and T.L. Powley (1989) Functional and anatomical organization of the abdominal vagal motor system in the rat. Dig. Dis. Sci. 34:970. Boekelaar, A.B. (1985) The Extrinsic lnnervation ofthe Stomach and Other Upper Abdominal Organs in the R a t Morphology and Morphogenesis. (PhD thesis). Amsterdam: Univ. of Amsterdam. Clerc, N., and M. Condamin (1987) Selective labeling of vagal sensory nerve fibers in the lower esophageal sphincter with anterogradely transported WG-HRP. Brain Res. 424t216224. Connors, N.A., J.M. Sullivan, and K.S. Kubbs (1983) An autoradiographic study of the distribution of fibers from the dorsal motor nucleus of the vagus to the digestive tube of the rat. Acta Anat. 115:266-271. Costa, M., J.B. Furness, and I.J. Lewellyn-Smith (1987) Histochemistry of the enteric nervous system. In L.R. Johnson (ed): Physiology of the Gastrointestinal Tract, Vol. 1.New York: Raven Press, pp. 1-40. Higgins, G.A., and J.S. Schwaber (1983) Somatostatinergic projections from the central nucleus of the amygdala to the vagal nuclei. Peptides 4:657-662.

The protocol we have used has proven to be highly sensitive and quite practical for labeling the descending CNS projections to the dorsal vagal complex. Virtually all of the well-established monosynaptic connections, including those from the paraventricular nucleus (e.g., Sofroniew and Schrell, '811, dorsomedial and lateral hypothalamus (e.g. van der Kooy et al., ,841, central nucleus of the amygdala (e.g., Higgins and Schwaber, '83), bed nucleus of the stria terminalis (e.g., Holstege, '871, as well as insular and medial prefrontal cortex (e.g., van der Kooy et al., '841, were readily observed in the present experiment. The possibility that Dil is also taken up and transported by axons of passage has to be considered. We have preliminary evidence for Dil uptake

ACKNOWLEDGMENTS We thank Elizabeth Baronowsky and Catherine Stanwyck for excellent technical assistance. The study was supported by National Institutes of Health grants DK27627, RR04802, and NS-26632. Some of the results were presented in an Abstract (no. 107.11) at the 18th Annual Meeting of the Society for Neuroscience, October 1989.

VAGAL INNERVATION OF GI TRACT IN RATS Holstege, G. i1987) Some anatomical observations on the projections from the hypothalamus to brainstem and spinal cord: An HRP and autoradiographic tracing study in the cat. J. Comp. Neurol. 260:98-126. Honig, M.G., and R.I. Hume (1989) Dil and DiO: versatile fluorescent dyes for neuronal labeling and pathway tracing. TINS 12333-341. Jabonero, V. (1955) Etudes sur les synapses du systeme neurovegetatif peripherique 11. Les synapses interneuronales dans les ganglions intramuraux du tube digestif. 2. Mikrosk.-Anat. Forsch. 61:549-570. Kirchgessner, A.L., and M.D. Gershon (1989) Identification of vagal efferent fibers and putative target neurons in the enteric nervous system of the rat. J. Comp. Neurol. 285:38-53. Kirsche, W. (1955) Synaptische Endigungen im Ganglion Coeliacum des Menschen. Z. Mikrosk.-Anat. Forsch. 61:624-640. Kolossow, N.G., and A.A. Milochin (1963) Die afferente Innervation der Ganglien des vegetativen Nervensystems. Z. Mikrosk.-Anat. Forsch. 70:426-464. Kuypers, H.G.J.M., and V.A. Maisky (1975) Retrograde axonal transport of horseradish peroxidase from spinal cord to brainstem cell groups in the cat. Neurosci. Lett. 1:9-14. Mei, N. (1978) Vagal glucoreceptors in the small intestine of the cat. J. Physiol. (Lond.)282:485-506. Mei, N. (1985)Intestinal chemosensitivity. Physiol. Rev. 65211-237. Milochin, A.A. (1962) Ueber eine besondere Form der Interorezeptoren des Verdauungstraktes. Z. Mikrosk.-Anat. Forsch. 69:145-152. 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. Powley, T.L., and H.-R. Berthoud (1990) A simple method for labeling enteric ganglia with Fluorogold. J. Neurosci. Methods (in press). 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 (Regul. Integr. Comp. Physiol. 22):R361-R370. Prechtl, J.C., and T.L. Powley (1990) The fiber composition of the abdominal vagus of the rat. Anat. Embryol. 181:lOl-115. Rodrigo. J., C.J. Hernandez, M.A. Vidal, and J.A. Pedrosa (1975) Vegetative

79 innervation of the esophagus 11. Intraganglionic laminar endings. Acta Anat. 92:79-100. Rodrigo, J . , J. de Felipe, E.M. Robles-Chillida, J.A. Perez Anton, L. Mayo, and A. Gomez (1982) Sensory vagal nature and anatomical access pathways to esophagus laminar nerve endings in myenteric ganglia. Determination by surgical degeneration methods. Acta Anat. 112.4747. Rogers, R.C., and G.E. Hermann (1985) Vagal afferent stimulation evoked gastric secretion suppressed by paraventricular nucleus lesion. J. Auton. Nerv. Syst. I3:191-199. Rogers, R.C., and G.E. Hermann (1987) Oxytocin, oxytocin antagonist, TRH, and hypothalamic paraventricular nucleus stimulation effects on gastric motility. Peptides 8:505-513. Rostad, H. (1973) Colonic motility in the cat 11. Extrinsic nervous control. Acta Physiol. Scand. 8 9 3 - 1 0 3 . Saper, C.B., A.D. Loewy, L.W. Swanson, and W.M. Cowan (1976) Direct hypothalamo-autonomic connections. Brain Res. I 1 7:305-312. Sato, M., and H. Koyano (1987) Autoradiographic study on the distribution of vagal afferent nerve fibers in the gastroduodenal wall of the rabbit. Brain Res. 4OUt101-109. Schofield, G.C. (1962) Experimental studies on the myenteric plexus in mammals. J. Comp. Neurol. 119:159-185. Sofroniew, M.V., and U. Schrell(1981) Evidence for a direct projection from oxytocin and vasopressin neurons in the hypothalamic paraventricular nucleus to the medulla oblongata: Immunohistochemical visualization of both the horseradish peroxidase transported and the peptide produced by the same neurons. Neurosci. Lett. 22:211-217. Stavney, L.S., T. Kato, C.A. Griffith, L.M. Nyhus, andH.N. Harkins (1963)A physiologic study of motility changes following selective gastric vagotomy. J. Surg. Res. 3,390-394. Van der Kooy, D., L.Y. Koda, J.F. McGinty, C.R. Gerfen, and F.E. Bloom (1984) The organization of projections from the cortex, amygdala, and hypothalamus to the nucleus of the solitary tract in the rat. J. Comp. Neurol. 224:l-24. Winogradova, O.N. (1962) Zur Frage der interneuronalen Verbindungen. Z. Mikrosk.-Anat. Forsch. 68:245-251.