Vagal preganglionic axons arborize in the myenteric plexus into two types: nitrergic and non-nitrergic postganglionic motor pools?

Abstract

Vagal preganglionic neurons innervate myenteric ganglia. These autonomic efferents are distributed so densely within the ganglia that it has been impractical to track individual vagal axons through the myenteric plexus with tracer labeling. To evaluate whether vagal efferent axons evidence selectivity, particularly for nitrergic or non-nitrergic myenteric neurons within the plexus, we limited the numbers and volumes of brainstem dextran biotin tracer injections per animal. Reduced labeling and the use of immunohistochemistry generated cases in which some individual axons could be distinguished and traced in three dimensions (Neurolucida) within and among successive (up to 46) myenteric ganglia. In the myenteric plexus of all stomach regions, the majority (∼86%) of vagal efferents were organized into two distinct subtypes. One subtype (∼24% of dextran-labeled efferents, designated “primarily nitrergic”) selectively contacted and linked—both within and between ganglia—nitric oxide synthase positive (nNOS+) neurons into presumptive motor modules. A second subtype (∼62% of efferents, designated “primarily non-nitrergic”) appeared to selectively contact and link—both within and between ganglia—non-nitrergic enteric neurons into a second type of effector ensemble. A third candidate type (∼14% of labeled preganglionics), appeared to lack “nitrergic selectivity” and to contact both nNOS+ and nNOS− enteric neurons. In addition to the quantitative assessment of the efferent axons in stomach, qualitative observations of the proximal duodenum indicated similar selective vagal efferent projections, in proportions comparable with those evaluated in the stomach. Limited injections of tracer, three-dimensional (3-D) tracing of individual axons, and histochemistry of myenteric neurons might distinguish additional efferent phenotypes.

NEW & NOTEWORTHY The present study highlights the following: 1) one type of vagal efferent axon selectively innervates nitrergic upper gastrointestinal myenteric neurons; 2) a second type of vagal efferent selectively innervates non-nitrergic gastrointestinal myenteric neurons; and 3) the two types of vagal efferents might modulate peristalsis reciprocally and cooperatively.

INTRODUCTION

Physiological experiments have established that the vagus nerve projects extensively to the myenteric plexus in the upper gastrointestinal (GI) tract (e.g., see Ref. 1). The relevant vagal preganglionic axons originate in the dorsal motor nucleus of the vagus (dmnX) in the brainstem (e.g., see Ref. 2). These efferents project to many upper intestinal enteric neurons, including the two most ubiquitous neurochemical phenotypes of enteric neurons, the mutually exclusive nitrergic or neuronal nitric oxide synthetase positive (nNOS+) versus the non-nitrergic or nNOS negative (nNOS−) postganglionic neurons (e.g., see Refs. 3–6). Many of these intrinsic “postganglionic” neurons then, in turn, modulate the two phases of smooth muscle peristalsis, relaxation or inhibition (a nNOS+-mediated postganglionic effect), and contraction or excitation (a nNOS− postganglionic effect) (4, 5, 7–10).

Physiological, pharmacological, and genetic experiments directly in the wall of the GI tract (7, 11, 12) or within the brainstem dmnX (8, 13–15) suggest that excitation and inhibition of GI motility might be modulated by separate vagal neurons, originating in different rostrocaudal regions of the longitudinally organized dmnX gastric column (16, 17), that coordinate different types of postganglionic enteric neurons.

Consistent with local physiological observations suggesting extensive projections from the brainstem to the stomach, early anterograde neural tracer injections into the dmnX (e.g., see Ref. 3) identified an extensive network of preganglionic terminals surrounding nearly all enteric neurons in all myenteric ganglia in the proximal GI tract [distal esophagus (18), stomach and duodenum (6, 9, 19)].

This labeling is so inclusive that vagal preganglionic axons must, on statistical grounds, form appositions with both nNOS+ and nNOS− enteric neurons in the myenteric plexus. It is unclear and has not been established, however, how individual vagal preganglionic fibers are organized with respect to the two particularly ubiquitous myenteric neurochemical phenotypes. No unambiguous evidence of subtypes of preganglionic axonal projections has emerged from tracer mapping.

This experiment introduced two methodological adjustments to facilitate the identification of individual complete tracer-labeled axons. First, tracer injection sites were reduced in size and number. Second, axons (though not their terminal arbors) were traced digitally in three dimensions within the myenteric plexus, directly from the whole mounts, as opposed to the conventional two dimensions of photomicrographs. The changes were adopted to limit the number of vagal motor axons labeled and thus to make it practical to follow single efferents through multiple ganglia. Such sampling was then used to determine whether individual efferent axons are specialized as subtypes defined by their terminal arbors and target-specific neurochemistries within the myenteric ganglia. (See Refs. 17 and 20 for preliminary discussions of the basic conclusions.)

MATERIALS AND METHODS

Animals

Two- to four-month-old male Sprague–Dawley rats (number of animals = 38; RRID:RGD_737903; Envigo, Indianapolis, IN) with an average weight of 253 g (SD: 42 g) at the time of tracer injection were housed individually in shoebox cages. All of the animal cages were maintained in an Association for Assessment and Accreditation of Laboratory Animal Care-approved temperature (22–24°C)-and-humidity (40%–60%)-controlled colony room. The room was maintained on a 12-h light-dark schedule. A standard rat chow (Teklad Global Diet 2018; Envigo, Indianapolis, IN) and filtered tap water were provided ad libitum, except for the night before tracer injection, when food but not water was removed. All husbandry practices conformed to the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals (8th edition) and were reviewed and approved by the Purdue University Animal Care and Use Committee. All efforts were made to minimize any suffering as well as the number of animals used.

Some comparable GI tract whole mounts from male Sprague–Dawley rats from earlier complementary experiments (19, 21) were also examined to compare the effect of injected volume. For extensions of the stomach observations to the duodenum, the proximal duodena of another 10 animals that had been similarly obtained, housed, fed, and injected, but which were counterstained according to a previous protocol (21) were also qualitatively examined.

Tracer Injection

Overnight-fasted animals were anesthetized with isoflurane (Akorn Animal Heath, Lake Forest, IL). After anesthesia induction, glycopyrrolate (0.2 mg/mL sc; Akorn Animal Health, Lake Forest, IL) was injected to minimize secretions.

The cisterna magna of the medulla was exposed and the rostral-most obex was used as the reference point for coordinates for injection sites into the dorsal motor nucleus of the vagus nerve (dmnX) (locations spanning the accessible dmnX: LHS: from −0.2, −0.8, −0.5 to −0.6, 1.2, −0.7; RHS: from 0.2, −0.8, −0.5 to 0.6, 1.2, −0.7; measurements in mm; x: left/right −/+; y: caudal/rostral −/+; z: deep/superficial −/+. The neuron-sparse rostral-most and caudal-most extremes of the longitudinally spindle-shaped dmnX were not injected.). Each animal received a single injection into the left, and another into the right, side of the dmnX, but the injection location within the dmnX was varied across the population of rats studied. A 7.5% solution of lysine fixable, 10,000 MW dextran-biotin conjugate (Thermo Fisher Scientific, Waltham, MA, Cat. No. D1956, RRID:AB_2307337) in PBS was pressure injected through a glass micropipette (ID 25 μm) with a Picospritzer III Intracellular Microinjection Dispense System (Parker Precision Fluidics, Hollis, NH) into the dmnX at 40 psi. The left and right side 4–6 ms injections (30–35 nL volume corresponding to a diameter of ∼400 µm) of the tracer solution were placed into homologous locations on either side. The pipette was left in each site for 2–3 min to allow the solution to disperse and prevent excess leakage from the site of penetration.

Following the final retraction of the glass pipette, the muscle and then the skin incisions above the cisterna magna were closed with interrupted sutures. The animal was then transferred to a circulating water heating pad until its righting reflexes had returned and subsequently to its home cage. Buprenorphine (0.01 mg/kg) was given subcutaneously before suturing as analgesia. Further analgesia was provided the day following surgery (Rimadyl, 5 mg/kg sc).

Tissue Fixation and Stomach Dissection

Following dmnX injections, a survival period of 19 days was used for the anterograde tracer transport into the vagal terminals in the GI tract. Rats were then weighed and euthanized with a lethal dose of a combination of ketamine and xylazine (275 mg/kg ketamine and 27.5 mg/kg xylazine). The animals had food available ad libitum until they were administered the overdose so that the stomach was typically full and relaxed. Once the animal was unresponsive to paw pinch, the abdomen and chest cavity were opened and heparin (0.5 mL; 1,000 U/mL) was injected into the heart, followed by transcardial perfusion with physiological saline. In addition, as needed, the stomach was distended with 10 to 15 mL of PBS to provide uniformity in organ distension. Tissue fixation was then achieved by transcardial perfusion of 500 mL of 4% paraformaldehyde (PF) in 0.1 M PBS (pH 7.4; 4°C).

After perfusion, the upper GI tract was freed from the other abdominal viscera with transections of both the distal esophagus and the proximal duodenum and then removed. The stomach was then cut open longitudinally along the greater curvature, and the stomach contents were gently rinsed away with tap water. The specimen was then trimmed to include the distal lower esophageal sphincter and the proximal pylorus, and the ventral and dorsal stomach walls were separated by cutting along the lesser curvature, to yield two whole mounts per animal. The whole mounts were transferred into fixative (4% PF in PBS) for at least 18 h or up to about 6 mo. In some cases, 4-cm long proximal duodenal samples were also removed and opened longitudinally with an incision just lateral and parallel to the mesenteric attachment.

To facilitate penetration of histological solutions, the gastric mucosa and submucosa were removed with forceps while the sample was submersed in PBS. For optimal nNOS immunohistochemistry, tissue separations were completed within 24 h of perfusion and processing was immediately begun.

Staining

Whole mounts (stomach and duodenum) were processed free-floating for all tracer processing and immunohistochemistry. Immediately following removal of the gastric mucosa and submucosa, the whole mounts were rinsed in PBS and treated with hydrogen peroxide-methanol block (1:4) for 30 min to quench endogenous peroxidase activity, rinsed again in PBS, and then soaked 3–5 days in PBS containing 0.5% Triton X-100 (and 0.08% Na azide) to facilitate solution penetration. Following a PBS rinse, whole mounts were incubated for 60 min in avidin-biotin-horseradish complex (Vectastain Elite ABC HRP kit, RRID:AB_2336819, Vector Laboratories, Burlingame, CA). After the avidin-biotin complex was established, the specimens were rinsed in PBS and then soaked in a diaminobenzene (DAB)/H₂O₂ solution for 3–5 min to yield a permanent gold-brown stain of all labeled efferent fibers.

Following completion of dextran-biotin histochemistry, all whole mounts were processed immunohistochemically for nNOS+ neurons (22). The tissue was washed in cold ultrapure DI water, rinsed in PBS, and then incubated overnight at room temperature in serum block. The following day, the tissue was washed in PBS, and sequentially incubated in an avidin-blocking solution and in a biotin blocking solution, washed in PBS, and then incubated for 48 h in nNOS polyclonal antibody (Invitrogen 61-7000, RRID:AB_2313734, Thermo Fisher Scientific, Waltham, MA), dilution 1:2,000, at room temperature. Whole mounts were turned over after 24 h. Then, the tissue was washed in PBST (0.3% Triton X-100, PBS), incubated for 2 h in Biotin-SP AffiniPure Goat Anti-Rabbit IgG (H L) (111-065-003, RRID:AB_2337959, Jackson Immunoresearch, West Grove, PA), dilution 1:500, washed in PBS, incubated for 60 min in Vectastain Elite ABC HRP kit, washed in PBS, and stained for 5 min with steel gray chromogen (VECTOR SG Peroxidase (HRP) Substrate Kit, RRID:AB_2314425, Vector Laboratories, Burlingame, CA). Owing to interruptions in antibody availability (and also to provide validation observations), in some earlier specimens in the laboratory (and in some preliminary analysis of more recent duodenal tissue), nNOS+ neurons had been stained histochemically for NADPH, as detailed in Ref. 21.

The validity and specificity of the primary antibodies were established with converging checks: 1) the pattern of labeling of myenteric nitrergic neurons was compared with our and others’ mapping (e.g., see Refs. 2, 9, and 21–24). 2) Earlier published extensive characterizations, including dilution information, for the primary antibodies were used as cross checks to determine baselines and to guide our dilution trials. 3) We varied our dilutions in initial development of the protocols. 4) Our observations confirmed common staining patterns and axonal-myenteric neuron organizations for the marker protocols we employed [Invitrogen antibody; NADPH histochemistry].

Finally, the tissue was washed in cold ultrapure H₂O and mounted circular-muscle-side up on gelatin-coated slides. After flattening, the samples were left to air dry overnight, and then were dehydrated in an ascending series of alcohols and xylene and coverslipped with Cytoseal.

Neural Tracing, Image Analysis, Photography, and Efferent Terminal Inventories

All whole mounts were first systematically surveyed microscopically at ×200 to identify potentially isolated and traceable vagal efferents. In the case of these axon candidates, three separate sets of criteria and categorizations were then used successively (different columns of Table 1) to determine which fibers could be characterized and to what extent. For the first set of criteria, as listed in Table 1, 1st column, an axon needed to be judged “traceable” according to five characteristics: 1) well-labeled, 2) complete, 3) sufficiently isolated from other fibers to enable unequivocal identification of the complete fiber, and with 4) relatively few artifacts such as folds, tears, debris etc. obscuring the fiber, and with 5) adequate counterstaining quality.

Table 1.

Criteria for axon selection, ganglia exclusion, and assignment of axon-myenteric neuron interaction type

Axon Selection Criteria	Excluded Ganglion Criteria	Categorization of Axon-Neuron Interactions
		Type	Criteria
Well labeled Complete Sufficiently isolated from other axons to enable unequivocal identification of the complete axon With relatively few artifacts such as folds, tears, debris, etc. obscuring the axon With adequate counterstaining quality	Multiple efferent axons were present, and interactions could not be definitely assigned to the traced axon Local counterstaining was too faint to make any determination of the interaction type as nNOS+ or nNOS− Examination indicated that the efferent axon passed through the ganglion without interaction Efferent axon terminals were present, but it was unclear which neuron type they interacted with Local background issues in the image prevented a clear determination of the interaction pattern	nNOS+ type	nNOS+: definite interaction with nNOS+ type neurons of the myenteric plexus, such as contact with, or encircling of, nNOS+ myenteric neurons nNOS+?: likely appositions with nNOS+ type neurons of the myenteric plexus (neurons present but faint or possessing less clear cut contact)
		Mixed?	M: possible evidence of apparent appositions with both nNOS+ and nNOS− type neurons of the myenteric plexus
		nNOS− type interactions	nNOS−?: clear encircling of unstained “cell” with axons of the myenteric plexus clearly visible in the vicinity of the efferent axon rings but no stained neurons of the myenteric plexus visible in the area of view nNOS−: clear encircling of nNOS-unstained “cell” shadow, with other well-stained nNOS+ neurons and axons of the myenteric plexus clearly visible in the close vicinity of the efferent axon encircling the neuronal shadow

The table lists the criteria for axon selection (column 1); exclusion of ganglia images from analysis (column 2); and assignment of individual ganglia images according to the type of axon-myenteric neuron interactions observed (column 3). nNOS, neuronal nitric oxide synthetase.

For axons that satisfied the first set of five criteria, Neurolucida 360 software by MBF BioScience (RRID:SCR_001775) controlling the motorized stage of a Zeiss (Oberkochen, Germany) Axio Imager Z2 microscope equipped with DIC optics and long-working-distance objectives (×40 and ×63 oil) was used to trace fibers. All branches of an arbor were digitized in three dimensions as the parent fiber branched, arborized, and finally terminated. Fibers were only included in the final, fully analyzed population if tracing of the fiber could be completed with confidence, despite any folds, tears, or intermingling of branches of a neighboring fiber.

Photomicrographs for offline visual assessment of all the ganglia associated with each individual fiber identified as traceable were acquired using a Leica DMRE widefield microscope (Leitz, Wetzlar, Germany), equipped with a Spot Flex camera controlled through the Spot software Advanced Plus V4.7 (Diagnostic Instruments, Sterling Heights MI; www.spotimaging.com), using ×20 and ×40 objectives as needed to encompass entire ganglia. Some large-field photographs (all-in-focus image mosaics) for this publication were acquired using a Leica DM5500 microscope equipped with a ×100 oil long-working-distance objective and Surveyor software (Objective Imaging, Cambridge, UK). DIC transmitted light optics on both the Leica microscopes and the Zeiss MBF microscope digitizing station facilitated critical resolution for both final two-dimensional (2-D) imaging and photomicrographs as well as three-dimensional (3-D) analyses.

Individual myenteric ganglia apparently traversed by a vagal efferent axon were then assessed with the second set of criteria. Candidates were excluded from further analysis if they failed on any of five features: 1) multiple efferent fibers were present, and interactions could not be definitely assigned to the traced efferent fiber; 2) local counterstaining was too faint to make any determination of the cells as nNOS+ or nNOS−; 3) examination indicated that the efferent fiber passed through the ganglion without interaction; 4) efferent fiber terminals were present, but it was unclear which cells they interact with; or 5) background issues in the image prevented a clear determination of the interaction pattern (see column 2 in Table 1).

For all fibers and associated ganglia satisfying the various assessments earlier, the vagal varicosities and appositions on ganglion neurons were then categorized by ganglion in terms of the observed nNOS+ and nNOS− appositions. The overall analysis depends on the assumption that all nNOS+ type neurons in a local ganglionic area are comparably well stained, so a ringed unstained “shadow” neuron in an area with adjacent stained cells can be assumed to correspond to an nNOS− type neuron but not an unstained nNOS+ type neuron. Each photograph of each associated ganglion was examined and the type of interaction between the efferent fiber and the contacted neurons within the ganglion was assigned to one of five semiquantitative categories (see Table 1, column 3, as well as Supplemental Material; see https://doi.org/10.6084/m9.figshare.19953074.v1 for illustrative images of each category).

Finally, Photoshop was used to adjust brightness, contrast, and sharpness of the images included in this report, as well as to organize layouts, add text, and scale bars.

Data associated with this study (25–27) were collected as part of the Stimulating Peripheral Activity to Relieve Conditions (SPARC) program and are available through the SPARC Portal (RRID: SCR_017041) under a CC-BY 4.0 license. A detailed protocol is available through Protocols.io: https://dx.doi.org/10.17504/protocols.io.kxygxmqqwl8j/v2.

RESULTS

As noted previously, earlier experiments concentrated on developing a method for establishing the complete vagal innervation of the myenteric plexus and produced results such as those illustrated in Fig. 1 (e.g., see Refs. 19 and 21). Such samples were typically generated by injecting larger amounts of a tracer into the longitudinally elongated dorsal motor nucleus (dmnX), using multiple injection locations and repeated pipette insertions to maximize labeling. The photomicrographs in Fig. 1 are taken from samples of stomach tissue that were double-labeled, with the DAB staining of the vagal efferents combined with counterstaining of the nNOS+ neuronal cells of the myenteric plexus by either nNOS antibodies or by NADPH (21). The images illustrate both the dense innervation of the myenteric plexus of the stomach by vagal efferent fibers and the complex tangles that the fibers can make with the enteric neurons of the myenteric plexus. Individual enteric neurons are in some cases tightly encircled, and in other cases, they evidence more limited but discrete appositions from tracer-labeled vagal axons. However, the combination of multiple labeled efferent fibers in the same area of tissue frequently makes it impossible to tease out the details of the interactions between particular efferent fibers and the individual neuronal cells of the myenteric plexus.

Figure 1.

Myenteric ganglia are heavily innervated by preganglionic efferent axons. The three panels show examples of heavily innervated ganglia, largely from previous work (21) resulting from multiple large injections of neuronal tracer into the dorsal motor nucleus of the vagus (dmnX). In A the nitrergic neurons of the myenteric plexus were labeled using neuronal nitric oxide synthetase (nNOS) staining while in B and C the nitrergic neurons of the myenteric plexus were labeled using NADPH staining. Although the more extensive dextran biotin injection demonstrates the heavy innervation of the myenteric plexus by preganglionic efferent axons, it does not enable the disentangling of the details of interactions between individual axons and individual myenteric neurons. Though single axons cannot be distinguished unequivocally, B and C suggest a tendency for vagal efferents to contact a specific type of myenteric neurons. Scale bars: 250 µm (A); 50 µm (B and C). A: a new version of an image associated with Powley (17). Images from two males.

Therefore, in the results we present here, the amount of dextran-biotin conjugate (tracer) injected into the dmnX was significantly reduced from that common in previous studies, in an effort to increase the likelihood that individual isolated efferent fibers could be identified and fully traced. Instead of the multiple injection sites used in previous work, e.g., see Refs. 16 and 28, only a single injection site was used on each of the left and right sides of the dmnX and the location of the injection sites was varied from animal to animal to sample the entire volume of the dmnX. This technique allowed us to generate a set of stomach samples with some traceable efferent fibers. In different animals, different injection locations were used (see materials and methods) in an attempt to explore any regional organization within the dmnX. The limited number of well-labeled cases and the variability in injection site locations, however, precluded meaningful conclusions.

Intraganglionic Analysis

The photomicrographs in Fig. 2 show a selection of myenteric ganglia through which a previously identified and fully traced single efferent fiber was followed. Although the innervation of the individual ganglia might appear complex in some of these 2-D images with multiple fiber branches present and as captured within the focal dimension of the objective, the additional capability of Neurolucida software being able to trace and focus in the z-direction established that in all analyzed cases a single efferent fiber was distinguishable and traceable. All the axons in our final analysis were fully traced and were either clearly isolated fibers even in 2-D or could be successfully and completely disentangled from adjacent fibers when traced by focusing through the z-direction in 3-D tracing.

Figure 2.

Myenteric ganglia innervated by single preganglionic efferent axons. Each panel shows an example of a myenteric ganglion innervated by a single preganglionic efferent axon, determined by complete tracing of individual axons. Efferent axons were stained with dextran biotin injected into the dorsal motor nucleus of the vagus (dmnX), whereas the nitrergic neurons of the myenteric plexus were labeled with neuronal nitric oxide synthetase (nNOS) staining. A and B each show an efferent axon interacting specifically with nNOS+ type neurons, whereas C–E each show an efferent axon interacting specifically with nNOS− type neurons. Scale bars: 25 µm (A); 50 µm (B–E). Images from four males.

Examination of the images in Fig. 2 suggests a pattern: it appears that within an individual ganglion, each identified and defined efferent fiber interacts predominantly with either nNOS+ type neurons or nNOS− type neurons. To examine this phenomenon quantitatively, we photographed and analyzed images of all the myenteric ganglia associated with the span of each of 54 vagal fibers that had been fully traced using Neurolucida 360 (Of an initial population of 69 fully traced fibers, 15 could not be used for this analysis due to inadequate counterstaining).

From the population of 54 vagal fibers, a total of 814 ganglia were photographed. Based on the visible neurons of the myenteric plexus and the behavior of the efferent fiber, the nature of the interaction between the efferent fiber and the local neurons of the myenteric plexus was categorized as described in materials and methods (see Table 1, column 3, and see also the Supplemental Material). As indicated in Table 1, we used two levels of discrimination between interactions of different types. A simplified description of the distribution that groups together nNOS+/nNOS+? categories (and equivalently nNOS−/nNOS−? categories) is as follows: 21% of ganglia images (176) showed nNOS+ type interactions, whereas 53% of ganglia images (424) showed nNOS− type interactions; 5% (42) showed possible evidence of mixed interactions and 21% of images (172) were excluded according to the criteria outlined in Table 1, column 2. If we break out the nNOS+ type and nNOS− type interactions to include the finer detail of Table 1 column 3, the results are as follows: nNOS+ 140 (or 17%); nNOS+? 36 (or 4%); Mixed 42 (or 5%); nNOS−? 151 (or 19%); nNOS− 273 (or 34%), again with 172 (or 21%) ganglia images excluded.

Therefore, of the ganglia images that were analyzable (see materials and methods for criteria, and Supplemental Material for illustrative examples), in 93% of the cases analyzed vagal efferent axons appeared to have “selective” to “exclusive” interactions with either nNOS+ type neurons or nNOS− type neurons and only 7% of images showed evidence of possible mixed postganglionic interactions.

We extended our analysis of vagal efferent fiber interaction with nNOS enteric cells to the proximal duodenum, and we have discovered the same pattern of interaction (qualitative assessment only due to available sample size), namely, vagal preganglionic axons in the duodenum appear to selectively innervate either nNOS+ type neurons or nNOS− type neurons, as illustrated in Fig. 3.

Figure 3.

Preliminary examination of duodenal tissue reveals a similar pattern of specific interactions. Each panel shows an example of interactions by a single efferent axon with cells of the myenteric plexus in the duodenum. Efferent axons were stained with dextran biotin injected into the dorsal motor nucleus of the vagus (dmnX), whereas the nitrergic neurons of the myenteric plexus were labeled with neuronal nitric oxide synthetase (nNOS) staining. A and B each show an efferent axon interacting specifically with nNOS− type neurons, whereas C shows an efferent axon interacting specifically with nNOS+ type neurons. Scale bars: 50 µm (A–C). Images from three females.

Interganglionic Analysis

The results of the analysis of individual ganglia images beg a second question—do single axons through their pathway show the same pattern of consistent interaction with either nNOS+ type neurons or nNOS− type neurons but not both? In other words, do interactions of the vagal preganglionic fiber with postganglionic neurons expressing one nitrergic profile in a particular ganglion predict the pattern of that axon’s contact with neurons in neighboring myenteric ganglia it traverses? In many cases, a fiber may interact with only a single myenteric neuron within each of a series of ganglia that it traverses (see Figs. 1, 2, and 3 for an illustration of the variety in the number of myenteric neurons per ganglion). This opens the possibility that within a greater length of a fiber, we could see apparent selectivity at the level of the individual ganglion but see a mixed pattern of interaction for the fiber as a whole.

To address this question of consistency across multiple traversed ganglia, the results were analyzed by fiber, and Fig. 4 summarizes the results for that subset of 28 vagal fibers in the traceable group with greater than 10 ganglia analyzed per fiber (i.e., with more than 10 ganglia traversed per fiber where the interactions between fiber and cells could be satisfactorily interpreted). The number of ganglia assessed for fiber-myenteric neuron interactions was often less than the total number of ganglia traversed by the fiber owing to ganglia images excluded because of legibility—see Table 1 column 2 in the materials and methods for exclusion criteria; see columns to the right of bar chart in Fig. 4 for a count of excluded ganglia). In Fig. 4, each row summarizes the distribution of observed fiber-myenteric neuron interactions for a particular fiber, as either nNOS+ type, nNOS− type, or mixed. If fibers interacted randomly with nNOS+ and nNOS− neurons in different ganglia, then we would anticipate that all the fibers would have a mix of interactions, representing the balance of nNOS+ and nNOS− type myenteric neurons within the myenteric plexus. However, this is not what we saw. For 24 out of 28 (i.e., 86% of the) fibers, more than 80% of ganglia analyzed along the analyzable segment of each axon show the fiber interacting with either nNOS+ or nNOS− neurons consistently from ganglion to ganglion. The four exceptions in the 28 cases are outlined in the center of Fig. 4. In two of the four “mixed” cases, no type of fiber-neuron interaction predominates, suggesting the possibility of a fiber that interacts with both nNOS+ type neurons and nNOS− type neurons along its length. In the other two of the four cases, interactions with nNOS− type neurons predominate, but potentially mixed interactions are observed in more than 20% of ganglia images—thus this may reflect the limits of the method or be as much a reflection of the arbitrary criteria we used.

Figure 4.

Analysis by axon—distribution of axon-ganglia interactions that are assigned as neuronal nitric oxide synthetase (nNOS)−, nNOS+, or mixed. The chart shows the observed distribution of assigned axon-ganglia interactions as either nNOS+ type, nNOS− type, or mixed for each axon examined. The data correspond to the set of 28 axons traversing more than 10 ganglia for which assignment of axon-ganglia interactions was possible and each row in the chart represents the distribution for a single axon. The three columns of data on the right show: 1) the total number of ganglia traversed by the axon; 2) the number of ganglia analyzed; and 3) the approximate location of the axon within the stomach: A, antrum; AC, antrum/corpus; C, corpus; CF, corpus/forestomach; F, forestomach. For 24 of the 28 axons more than 80% of the analyzed ganglia have interactions assigned as either all nNOS+ type (axons 1–6) or all nNOS− type (11–28 axon). The four exceptions are outlined in the center: two axons show no dominant type of interaction, whereas two axons show predominantly nNOS− interactions but with more than 20% of the ganglia showing possibly mixed interactions. The 28 axons came from 17 males.

The analysis in Fig. 4 included only vagal efferent fibers that could be unequivocally Neurolucida 360-traced as traversing more than 10 ganglia with unambiguous interactions. This 10-ganglia cutoff provided both enough sites of interaction per fiber as well as a large enough population of fibers to analyze. Table 2 provides a summary of results using various cutoff criteria, whereas the graph in Fig. 5 shows the percentage of fibers that consistently expressed appositions with either nNOS+ type neurons or nNOS− neurons but not both (setting the criterion as no more than 20% of ganglia images analyzed showing evidence of mixed interactions or interactions with the other cell type) as a function of the minimum number of ganglia per fiber analyzed, together with the number of fibers at each point.

Table 2.

Distribution of efferent axon type based on analysis of axon-myenteric neuron interactions

Minimum Number of Analyzed Ganglia/Axon	Number of Traced Efferent Axons	Number of Axons With >80% Interactions With nNOS+ Type Neurons Only, No. (%)	Number of Axons With >80% Interactions With nNOS− Type Neurons Only, No. (%)	Number of Axons With Possible Mixed nNOS+/nNOS− Interactions, No. (%)
2	54	15 (28)	29 (54)	10 (19)
5	40	11 (28)	22 (55)	7 (18)
11	28	6 (24)	18 (62)	4 (14)
16	15	3 (19)	11 (75)	1 (6)

The table shows the distribution of efferent axons between the three categories of neuronal nitric oxide synthetase (nNOS)+ interacting, nNOS− interacting, or mixed, depending on the criterion of the number of traversed ganglia analyzed. The first column shows the criterion for the minimum number of axon-ganglia images per axon used in the analysis, the second column shows the number of fully traced efferent axons meeting that criterion, and the final three columns show the number (and percentage) of axons identified as nNOS+ interacting, nNOS− interacting, or potentially mixed (no dominating nNOS+/nNOS− pattern), based on patterns of axon-myenteric neuron interaction. In all cases, an axon is identified as nNOS+ interacting (or nNOS− interacting) if more than 80% of ganglia show specific nNOS+ (or nNOS−) interactions. Data come from a population of 54 axons from 31 males.

Figure 5.

Axon specificity as a function of the number of ganglia per axon that were analyzed. The solid data points show the number of efferent axons in the population of 54 axons as a function of the minimum number of analyzed ganglia per axon. The open data points show the percentage of axons with at least 80% of axon-ganglia interactions showing axon contacts with either only neuronal nitric oxide synthetase (nNOS)− type neurons or with only nNOS+ type neurons. At the left all 54 axons are included; proceeding to the right progressively fewer axons are included as the required number of analyzed ganglia increases. The analysis demonstrates that, independent of how many ganglia per axon could be analyzed, the efferent axons are almost all specifically nNOS+ interacting or specifically nNOS− interacting, rather than exhibiting interactions with a mixture of nNOS+/nNOS− type myenteric neurons across the span of the axon. The 54 axons came from 31 males.

The percentage of fibers that meet the cutoff criterion of less than 20% of the traversed ganglia showing some interaction with nNOS− neurons in the case of predominantly nNOS+ interacting fibers, or with nNOS+ type neurons in the case of predominantly nNOS− interacting fibers, rises from ∼80% when the entire population of fibers is included to 100% when only the longer fibers and/or more completely counterstained tissue are included. The fact that none of the longer/more completely counterstained fibers are categorized as mixed may reflect the fact that more optimally stained tissue allows for more unequivocal assignment of nNOS+ versus nNOS− interaction, or it could indicate that fibers showing mixed interactions are in general shorter, or it could simply reflect the small population of fibers traversing 19 or more ganglia allowing unambiguous analysis.

Our tracer-injection method to limit labeling vagal preganglionic efferent fibers certainly showed variability in results—some whole mount samples still evidenced so many multiple, interwoven efferent fibers that they could not be disentangled from each other, whereas other whole mounts showed no labeled fibers at all. Furthermore, even after optimization of the counterstaining process some variability in the extent of counterstaining across the whole mount was sometimes seen. Hence, fibers for analysis were chosen to ensure 1) that the fibers themselves were strongly labeled and complete, 2) that an individual fiber could be identified and followed throughout (i.e., the fiber was completely traced using Neurolucida 360), and 3) that an adequate number of ganglia within the span of the fiber were consistently and well counterstained. In all, 69 efferent fibers from 38 animals were inventoried and completely traced. However, of those 69 fibers, only 54 were analyzed for this paper, the rest being rejected for inadequate counterstaining of the myenteric neurons in the area of the fiber. Our goal in this analysis was to characterize axons from all regions of the stomach (antrum, corpus, and forestomach—Accomplished. See column on the right of Fig. 4 for the location of each fiber). However, the higher density of preganglionic efferent fibers in the antrum meant that relatively few axons from the antrum could be isolated and fully traced; hence the distribution of analyzed fibers across the stomach in Fig. 4 underestimates the incidence of preganglionic efferents in the antrum.

In our assessment of fiber-ganglia interactions, we initially distinguished between images where interactions were identified as unequivocally nNOS+ (or nNOS−) and images where we concluded that interactions were more ambiguously nNOS+ (or nNOS−) contacts—see Table 1; and also the Supplemental Material for sample images. The initial assignment into separate nNOS+ and nNOS+? categories (and, equivalently, nNOS− and nNOS−? categories) represented a conservative approach to the analysis allowing for the possibility that nNOS+? and nNOS−? assignments were potentially incorrect assignments, but the results show a strong grouping between nNOS+ and nNOS+? (and equivalently between nNOS− and nNOS−?) assignments. Table 3 shows the incidence per axon of traversed ganglia of different assignments for the 28 axons from Fig. 4. The data shows that traversed ganglia where the interactions are assigned as nNOS+? are almost exclusively found along axons where the rest of the traversed ganglia show interactions assigned as nNOS+ (i.e., nNOS+ type axons) and are found rarely where the rest of the traversed ganglia show interactions assigned as nNOS− (i.e., nNOS− type axons), and equivalent statements can be made for ganglia where the interactions are assigned as nNOS−?. If the assignment nNOS+? or nNOS−? represented a significant uncertainty in whether the axon was actually interacting with nNOS+ or nNOS− type neurons, then we would expect both assignments to show up frequently with axons assigned to the opposite type. This does not occur and, therefore, in our quantitative analysis of axon-ganglia interactions we grouped together these pairs of categories (nNOS+? with nNOS+ and nNOS−? with nNOS−).

Table 3.

Number of traversed ganglia per axon with the designated type of axon-myenteric neuron interaction, by assigned axon type

Axon-Neuron Interaction Assignment	Axon Type (Number)
	nNOS+ Type (6)	Mixed type? (4)	nNOS− Type (18)
nNOS−	0.2	5.0	11.0
nNOS−?	0.5	3.5	6.7
Mixed	0.3	4.3	0.7
nNOS+?	3.2	0.3	0.1
nNOS+	13.8	1.5	0.2
Mean number of analyzed ganglia/axon	18.0	14.5	18.8

The table shows the number of traversed ganglia per axon of each type of axon-myenteric neuron interaction, for the three categories of assigned axon type. The 28 axons considered are those shown in Fig. 4. Assignment of axons as neuronal nitric oxide synthetase (nNOS)+ type (or nNOS− type) is based on greater than 80% of traversed ganglia showing specifically nNOS+ (or nNOS−) type interactions; mixed? axons have less than 80% of traversed ganglia showing a single type of interaction. Negligible numbers of ganglia where the interactions are assigned as nNOS+? (or nNOS−?) are found associated with nNOS− (or nNOS+) type axons. Data come from a population of 28 axons from 17 males.

Note that, of course, we are only visualizing a subset of the axon-ganglia interactions—multiple unlabeled axons pass through each ganglion—so our observations relate only to the traced axon—we are in no way implying that each ganglion has only a particular type of interaction with preganglionic vagal efferents. We are simply using the visualized neurons to infer the properties of the labeled axon.

DISCUSSION

Myenteric ganglion neurons have long been typed according to neurochemical, morphological, and/or electrophysiological characteristics (10, 23, 24, 29). Vagal afferents have also been repeatedly recognized in two different phenotypes, and the distinctions are based on both their arbor morphologies and their target tissues (20, 30). In contrast, though, vagal efferents, the focus of the present report, have been difficult to divide into subtypes or different phenotypes. In the present experiment, we report that vagal efferents can be categorized into two distinct types (as well as perhaps a third, less numerous type of more ambiguous functional correspondence) on the basis of their terminals forming appositions around nNOS-stained neurons. One phenotype of vagal efferents to the myenteric plexus of the upper GI tract selectively and predominantly forms appositions around nNOS+ enteric neurons organizing them into potential apparent nitrergic motor neuron pools; and a second phenotype of vagal efferents selectively and predominantly produce appositions around nNOS− neurons, organizing them into non-nitrergic ensembles or modules. A third and smallest group of fibers, potentially a third phenotype, appears to form appositions with both nNOS− and nNOS+ enteric neurons.

Functionally, the easiest subgroup of efferent axons to interpret is the 24% of the axon segments that selectively contacted the nNOS+ neurons—both within and between ganglia–of the myenteric plexus. Such nNOS+ postganglionic neurons in the myenteric plexus have long been implicated in the relaxation responses of the smooth muscle layers making up the external muscle wall, and it is recognized that the vagus nerve can modulate the relaxation responses in the smooth muscle motility (31). It is hypothesized that this “nNOS+ or nitrergic phenotype” of vagal efferents is responsible for coordinating these modulating influences on smooth muscle relaxation. And further, it seems reasonable to hypothesize that the vagal axon linkages of similar ganglionic effectors may very well organize motor neuron pools and coordinate the responses into smoother, more synchronous patterns of modulation. nNOS+ myenteric neurons are distributed throughout the stomach and proximal small intestines (22), and, correspondingly, we observed this nNOS+ phenotype of vagal efferents throughout the stomach and proximal duodenum.

The second prominent subgroup of vagal efferents extensively contacts nNOS− ganglion neurons. Interestingly, this is apparently the most prevalent phenotype. A larger percentage of vagal efferent segments was made up of the “nNOS− phenotype” than of the “nNOS+ phenotype.” This imbalance might reflect a variety of different factors. “nNOS− myenteric postganglionic efferents” include the population usually assumed to modulate the muscarinic excitatory or contraction phase of gastric motility. In this role, the “nNOS− phenotype” may be more prevalent as a reflection of an imbalance in the vagal influences over, respectively, contraction and relaxation.

Alternatively, of course, the “nNOS−”pattern of innervation might also reflect the facts that there is considerable unclarified heterogeneity in nNOS− neurons and that the myenteric ganglia coordinate more intrinsic responses than just contraction and relaxation of smooth muscle layers, and, hence, the larger number of vagal efferents might project predominantly to “nNOS−”enteric neurons that do not, in turn, project to the smooth muscle coat. A subset of the nNOS− linked myenteric neurons may well coordinate functions other than those of the smooth muscle wall contractions. An obvious alternative response is secretion. Cholinergic stimulation certainly elicits gastric secretion, and the active secretion of gastric acid, enzymes, and fluid seems not to have a strong NOS+ restraint.

And yet alternatively again, the “nNOS−”projection pattern we observe may reflect a neuronal staining—or lack of staining—bias that generated more false positives.

As noted in results, the two phenotypical patterns of efferent contacts with myenteric neurons are relative, not absolute. One type of efferent predominantly contacted nNOS+ neurons. The other group of efferents predominantly contacted nNOS− neurons. But neither phenotype exhibited complete neurochemical selectivity of all efferents. Some of the nNOS+ phenotype of vagal efferents apparently made limited and occasional contacts with nNOS− neurons, and, conversely, some of the nNOS− phenotype made occasional contacts with nNOS+ neurons. This apparent absence of exclusivity might reflect rater errors because of staining limitations and/or difficulties assessing appositions at the resolution of light microscopy. Alternatively, though, the lack of absolute selectivity might be a reflection of the tight coordination of phase relationships and secretory coordination that must occur in peristalsis.

Finally, there is a third subgroup, the smallest of the three efferent axon types, that consists of axon segments that appears to contact both the “nNOS+” ganglionic neurons and the “nNOS−” ganglionics. This subgroup pattern could reflect one or more of the sampling biases mentioned immediately earlier. Alternatively, some, or all of this mixed phenotype could reflect a basic pattern of linkage between the relaxation and the contraction responses that are important in coordinating or phasing the modulation of the two antithetical response patterns.

An observation with implications for neuromodulation stimulation should be mentioned. Most vagal efferent axons reaching the rat upper GI tract are fine, unmyelinated axons (32–34). In keeping with these reports, the vagal fibers observed contacting nNOS+ and nNOS− enteric neurons in the upper GI myenteric plexus were consistently fine (apparently <0.8 μm), seemingly unmyelinated fibers. On the one hand, this caliber of axonal process has significant implications for the choice of parameters that should be selected when the vagus is stimulated to mobilize the gastric efferent responses. On the other hand, the fact that the majority of vagal efferents contacting nNOS+ neurons or nNOS− neurons are so similar in caliber will complicate attempts to selectively stimulate a response class without simultaneously eliciting a counter response that might cancel out the effect sought.

Significantly, the electrophysiology of vagal efferents corresponds to the phenotypic pattern observed with the present neural tracing and immunohistochemistry. In their analysis of the vagal efferent projections to the stomach, Davison and Grundy (35), in their classic analysis, characterized vagal efferents in teased-nerve single-unit preparations, and identified two basic, reciprocal phenotypes; in effect each Davison–Grundy phenotype was divided into two subtypes. One basic phenotype of vagal efferents increased its firing as gastric distension was increased [“Type I,” which presumably would activate nNOS+ linked postganglionic neuron pools and stimulate relaxation of smooth muscle]. The second basic type increased its activity when gastric distension decreased [“Type II,” which presumably would normally activate cholinergic neuron sets and stimulate contraction of smooth muscle]. The investigators also described biphasic variants (Type III and Type IV, respectively) of the two phenotypes. The fundamental patterns and the biphasic pattern thresholds could have been determined, of course, by afferents. And conceivably the patterns might reflect the possibly incomplete predominances we have noted. In addition, as we have reported elsewhere (36), changes in stomach volume are a dominant stimulus of vagal afferent signaling controlling stomach responses. It appears that vagovagal relaxation and contraction reflexes might constitute a mechanism modulating peristalsis and affecting nutrient handling by the stomach.

LIMITATIONS

One issue that must be noted is that contact, i.e., synaptic contact, between an axon terminal and a “post-synaptic” neuron can only be fully gauged at the electron microscopic level. We have exclusively used light microscopy and the criteria of such microscopy, and thus we did not attempt to evaluate the efferents for synaptic contents. The striking appositions, enlarged varicosities, and calyxes that the efferents produce around the postsynaptic somata seem compelling [and have been confirmed to form vagal synaptic contacts on enteric neurons ultrastructurally (6) as well as in numerous other neural systems], and thus we have tentatively assumed that the various close appositions were contacts. Technically, of course, the nature of the putative contacts cannot be determined without an ultrastructural assessment.

The survey protocol we employed certainly shaped the sample proportions we observed. Different volumes or concentrations of tracer injections into different regions of the dmnX might change the proportions of the different phenotypes observed. The multiple criteria that we used and that had to be satisfied to indicate the need to trace a fiber were intentionally stringent, and other criteria might also affect the final set. Certainly, one case in point is the possibility that another phenotype of efferent terminal is so complex and convoluted that it remained cryptic by forming tangles of terminals around a neuron, and thus complexity excluded our tracing this phenotype(s). Any other sampling protocol and/or staining protocol might also have uncovered other specialized phenotypes that have yet to be identified.

Estimates of percentages of vagal axons that project selectively to either nNOS+ or nNOS− postganglionic myenteric neurons are approximations. As previously mentioned, the criteria used certainly affect the exact percentages observed. No evidence in the present sample clearly suggested a more fine-grained typology of vagal efferents to the upper GI tract, but alterations of the sampling, counting, and staining criteria would presumably reveal somewhat different, perhaps multimodal, estimates of vagal motor phenotypes. To some extent, we probed this issue through our more fine-grained assignment of fiber-myenteric neuron interactions into five rather than three types (see the discussion around Table 3, toward the end of results). Still, the interaction categories deemed less-clear cut (nNOS+? and nNOS−?) were strongly correlated with the more definitive assignments of nNOS+ and nNOS−.

Certainly, myenteric ganglia contain different types of enteric neurons defined by neurochemistry, morphology, and/or electrophysiology (e.g., see Ref. 23). For example, in a different animal model (guinea pig; proximal intestines myenteric plexus), Furness (24) has calculated that there may be 14 functionally distinct neuron types, and it is noteworthy that in the mouse dmnX, Tao et al. (13) have demonstrated seven types of preganglionic neurons defined genetically. Hybrid cases in which vagal efferent axons seem to contact a heterogeneity of enteric neuron types, as well as nNOS+ or nNOS− type neurons, might well reflect important specializations or subtypes of the two broad phenotypes we report here.

PERSPECTIVES AND SIGNIFICANCE

Vagal axons project to the stomach, and they influence both gastric relaxation (presumptive nNOS+ effects) and contraction (presumptive nNOS− effects). However, there has been no clear resolution of how the two counteracting motor responses are differentiated by vagal preganglionic projections. In this anterograde tracing experiment, we observe for the first time (to our knowledge) that one type of vagal efferent axon selectively establishes appositions with nNOS+ myenteric neurons, whereas a second type of vagal efferent axon selectively establishes appositions with nNOS− myenteric neurons. These specializations could help explain vagal modulation of the phases of peristalsis.

DATA AVAILABILITY

Data associated with this study are available through the SPARC Portal (RRID:SCR_017041) under a CC-BY 4.0 license (see Refs. 25–27).

SUPPLEMENTAL DATA

Supplemental Material: https://dx.doi.org/10.17504/protocols.io.kxygxmqqwl8j/v2.

GRANTS

As per the National Institutes of Health reporting guidelines, the research reported in this publication was supported by the National Institutes of Health Grants DK27627 and OD023847.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

D.M.J. and T.L.P. conceived and designed research; J.L.M., E.A.B., and D.B. performed experiments; D.M.J. analyzed data; D.M.J. interpreted results of experiments; D.M.J. prepared figures; D.M.J. and T.L.P. drafted manuscript; D.M.J., J.L.M., and T.L.P. edited and revised manuscript; D.M.J., J.L.M., E.A.B., D.B., and T.L.P. approved final version of manuscript.