Factors Regulating Vagal Sensory Development: Potential Role in Obesities of Developmental Origin

Abstract

Contributors to increased obesity in children may include perinatal under- or overnutrition. Humans and rodents raised under these conditions develop obesity, which like obesities of other etiologies has been associated with increased meal size. Since vagal sensory innervation of the gastrointestinal (GI) tract transmits satiation signals that regulate meal size, one mechanism through which abnormal perinatal nutrition could increase meal size is by altering vagal development, possibly by causing changes in the expression of factors that control it. Therefore, we have begun to characterize development of vagal innervation of the GI tract and the expression patterns and functions of the genes involved in this process. Important events in development of mouse vagal GI innervation occurred between midgestation and the second postnatal week, suggesting they could be vulnerable to effects of abnormal nutrition preor postnatally. One gene investigated was brain- derived neurotrophic factor (BDNF), which regulates survival of a subpopulation of vagal sensory neurons. BDNF was expressed in some developing stomach wall tissues innervated by vagal afferents. At birth, mice deficient in BDNF exhibited a 50% reduction of putative intraganglionic laminar ending mechanoreceptor precursors, and a 50% increase in axons that had exited fiber bundles. Additionally, BDNF was required for patterning of individual axons and fiber bundles in the antrum and differentiation of intramuscular array mechanoreceptors in the forestomach. It will be important to determine whether abnormal perinatal environments alter development of vagal sensory innervation of the GI tract, involving effects on expression of BDNF, or other factors regulating vagal development.

1. Introduction

1.1 Altered meal size is associated with obesity and eating disorders

The meal is the basic unit of food consumption, and alteration of some elements of the neural and hormonal systems that regulate meals are likely to be involved in obesity and eating disorders. Thus, understanding how meals are regulated could be key to understanding normal as well as disordered intake. As Greenberg and Smith [34]have pointed out, (1) meal size is increased in both animal [2, 67, 74] and human [9, 15, 61] obesity, (2) meal size is abnormal in anorexic and bulimic humans and (3) the cumulative effects of these alterations of meal size may result in disordered daily food intake and body weight regulation.

In the introduction to this symposium (cf. Adair, this issue) and in the following presentation (cf. Cottrell and Ozanne, this issue), the mounting evidence suggesting obesity has developmental origins was described. Both human and animal data have shown that malnutrition during pregnancy results in low birth weight, but if followed by nutritional sufficiency it can result in obesity. In contrast, continued exposure to nutritional insufficiency results in weight remaining low throughout life, but many symptoms of the metabolic syndrome develop. The opposite condition – overnutrition – results when mothers are hyperphagic and obese during pregnancy and lactation, or when litters are small, rendering newborns hyperphagic. Interestingly, overnutrition at these early ages can lead to a similar outcome as undernutrition: a predisposition to development of obesity and diabetes. Importantly, - similar to the observations suggesting that increased meal size is a feature associated with obesities of various etiologies - there is evidence that obesities with a developmental origin may be associated with increased meal size. For example, relative to animals raised in larger litters, adult rats predisposed to develop dietary obesity as a consequence of having been raised in small litters exhibited increased meal size when tested on chow and increased weight gain and adiposity when fed a high-fat high-carbohydrate diet [17]. Moreover, meal size predicted the magnitude of this dietary obesity, and it was the only feeding parameter examined that was strongly correlated with adiposity measures, including fat cell size and fat pad weight.

1.2 Controls of meal size

Given the potential significance of meal size to obesity and eating disorders, some of the key features of a model of the neural-hormonal controls of meal size are briefly reviewed here, followed by a review of the role of the vagus nerve in signaling satiation and the control of meal size.

One of the prominent models of neural-hormonal control of meal size (e.g., [71]) proposes that the timing of meal initiation is influenced by several internal and external factors including the time of day or the availability and palatability of food, whereas meal termination is more biologically controlled. Once eating has begun, food collects in the stomach and proximal small intestine where neural and hormonal satiation signals are generated. When these signals reach sufficient intensity they terminate feeding behavior. The bulk of these satiation signals are transmitted to the brain by sensory components of the vagus nerve that innervate the upper GI tract. Since eating usually must stop while a substantial proportion of the food consumed is still confined within the upper GI tract and thus not yet absorbed, vagal feedback signals play a significant role in determining when a meal ends and therefore in regulating the size of individual meals. This model further proposes that changes in levels of hunger or energy stores, for example, generate neural and hormonal signals that activate receptors in the hypothalamus and forebrain, which in turn lead to appropriate adjustments of food intake largely by modulating responses of neurons in the nucleus of solitary tract (NTS) to vagal satiation signals.

1.2.1 Role of vagal sensory innervation of the GI tract in controlling meal size

Vagal sensory pathways that innervate the gastrointestinal (GI) tract terminate centrally in the brainstem within the caudal half of the NTS. From the NTS, vagal sensory input is distributed either to several higher brain regions, or to vagal efferent neurons that regulate GI functions. The former route provides feedback about food present in the gut to feeding regulatory circuits in the brain, which contributes to meal termination, or satiation [14, 73]. In contrast, the latter route forms pathways mediating GI reflexes such as receptive relaxation and gastric emptying that can impact meal size indirectly (e.g., [1, 16, 68, 69, 76]). For example, changing the rate of gastric emptying can alter the rate at which the small intestinal mucosa is stimulated by nutrients, which in turn can influence the rate at which satiation develops. Similarly, changing the degree of receptive relaxation affects stomach capacity and thus the amount of food that can be consumed before satiation signals become sufficiently activated and terminate feeding.

The evidence for a vagal sensory role in satiation and determination of meal size derives mainly from studies that have attempted to manipulate some or all of the vagal sensory pathways while avoiding (or minimizing) effects on vagal motor ones. Several approaches have been employed for this purpose, including systemic application of the excitotoxin capsaicin [10, 42], or cutting the sensory rootlets unilaterally as they exit the medulla in combination with sectioning one or more of the contralateral subdiaphragmatic vagal branches [70, 72, 79]. Consistent with a vagal sensory role in satiation, both of these treatments led to increased meal size. In one instance this resulted from increased meal duration and increased maintained lick rates of sensory vagotomy animals compared to controls [70]. Also consistent with a satiation role, these sensory vagotomies blocked suppression of intake by preloads of some nutrients infused in the small intestine [78, 83], and reduced the satiation response to CCK [53, 72]. A third strategy for manipulating vagal GI afferents utilized mouse genetics or gene manipulations to perturb expression of genes that regulate development of vagal sensory nerves or tissue elements they associate with [31]. Several mutations have been associated with alterations restricted to specific subpopulations of vagal sensory receptors and specific changes in meal patterns consistent with disruption of either satiation (neurotrophin-4 knockout or knock-in [13, 29]), or vago-vagal digestive reflexes (c-Kit or steel factor mutations [12, 30]). Although each of these three approaches for manipulating vagal sensory pathways has interpretive issues [31], they have produced a consistent pattern of results, which taken together support an important role for vagal afferent innervation of the GI tract in satiation and determination of meal size.

1.3 Control of development of vagal sensory innervation of the GI tract

Under environmental circumstances such as malnutrition that lead to low birth weight, the growth of the brain is spared at the expense of damage to other organs. This damage includes reduced development of pancreatic b-cells and effects on muscle and adipose tissue that contribute to insulin resistance [26]. However, there is some evidence that development of neural pathways related to the control of feeding and body weight are affected. The offspring of female rats fed a low-sodium diet during the first half of gestation, or from the start of gestation through weaning exhibited dramatic changes in terminations of taste pathways in the NTS when examined in early adulthood, which were probably due to malnutrition [50]. Also, work described by Plagemann at the first IBRC symposium highlighted changes in hypothalamic structure and function that occurs in rats raised in small litters, which appear to result from hyperinsulinemia associated with their overnutrition, including increased expression of orexigenic neuropeptides such as galanin and NPY and decreased responsiveness to leptin and insulin [58]. Further, projections from the hypothalamic arcuate nucleus to other hypothalamic nuclei involved in regulation of food intake and body weight, including the paraventricular nucleus, develop after birth [7]. Interestingly, Richard Simerly reported at this symposium (cf. Simerly, this issue) the first demonstration in mice that perinatal under- or overnutrition both resulted in a predisposition to obesity and further, that each of these treatments was associated with a substantial reduction in the postnatal development of hypothalamic arcuate nucleus projections. Thus, the predisposition to develop obesity and the metabolic syndrome that occurs in response to both perinatal under- and overnutrition, has a parallel developmental neuroanatomical counterpart that could contribute.

These findings, taken together with the potential relevance of aberrant meal size to the etiology of obesities of developmental origin, and the importance of vagal sensory function in determining meal size raise the possibility that abnormal perinatal environments alter meal size in part through effects on vagal development. Abnormal nutritional or hormonal status could alter vagal development by modulating relevant gene expression, gene products, or sensitivity of tissues to gene products. For example, as discussed at this symposium by Susan Ozanne, she and her colleagues have identified selective alterations in the levels of insulin signaling proteins in adipose tissue and muscle of young men that had low birth weights [54, 55]. Further, these changes preceded the onset of type 2 diabetes and may reflect modifications of gene expression. Additionally, Susan Ozanne described how they have begun to explore the potential of epigenetic tagging, through mechanisms such as histone modification and DNA methylation, to mediate or maintain these changes in gene expression in adults as a memory of environmental effects that occurred during development [56].

To explore the potential for abnormal perinatal environments to alter the expression of genes involved in vagal development it would be valuable to have knowledge about the genetic program that regulates it. This information would also be useful for developing strategies for altering meal size as an adjunct to other treatments for obesity or eating disorders. Unfortunately however, little is known about regulation of vagal development, or surprisingly for that matter, about vagal development itself. Therefore, we have begun to investigate these issues with an initial focus on stomach innervation.

The first series of questions we addressed included: (1) what are the key events in development of vagal sensory innervation of the GI tract? (2) when do they occur? and (3) where do they occur (which stomach compartments and tissues are innervated)? Surprisingly little is known about the growth of vagal axons and terminals into the GI tract, and even less is known about the formation of sensory receptor structures. In particular, very few studies have directly characterized these developmental events, and to our knowledge only one of these studies employed a vagal-specific marker, but this study focused on a single time point fairly late in development. A serious consequence of this lack of knowledge is that it has often been assumed that vagal development is largely complete at birth. As a result, research on the neural/hormonal basis underlying the acquisition of controls of feeding behavior that occur after birth has focused on maturation of hypothalamic nuclei and their projections to the vagal nuclei in the brainstem [64], whereas little attention has been devoted to a potential vagal role. In addition to clarifying this issue, it will be valuable to identify the stomach tissues innervated by vagal sensory axons and terminals at each developmental stage because gene expression in or near these tissues at a given stage would be well positioned to influence vagal development. Therefore, to provide a more complete foundation for studies of factors controlling vagal development, we performed an initial survey of normal development of vagal stomach innervation in the mouse [51].

The emphasis of the initial studies reported here was on axons and two classes of vagal mechanoreceptors innervating the muscle wall of the GI tract - intraganglionic laminar endings (IGLEs), and intramuscular arrays (IMAs) - because their morphology and distribution have been characterized in the greatest detail as compared with other receptors in the rat and mouse [3, 27, 52, 66, 80]. This knowledge was particularly valuable for aiding identification of the precursors of these receptors that are present during development since their structures had not been previously described. Intraganglionic laminar endings consist of one or more aggregates or leaves of densely packed fine terminal puncta, which form a planar layer that is sandwiched between the neurons of a myenteric ganglion and its surrounding ganglion capsule. These putative mechanoreceptors are distributed throughout the myenteric plexus of the esophagus, stomach and intestines. They are thought to transduce muscle tension such as the shearing forces produced by the smooth muscle layers of these organs during peristaltic contractions [52, 80, 84].

The structure of IMAs is strikingly distinct from that of IGLEs. They are composed of groups of parallel rectilinear axonal telodendria oriented parallel to each other and interconnected by crossbridge fibers. Additionally, IMA telodendria run parallel and in close apposition with smooth muscle fibers and interstitial cells of Cajal (ICCs). These putative mechanoreceptors have a much more restricted distribution as compared with IGLEs. They are concentrated in both muscle layers of the forestomach and the circular muscle of the lower esophageal sphincter (LES) and pylorus. Intramuscular arrays are thought to detect muscle stretch that occurs, for example, as a food bolus passes through a sphincter, and when the forestomach fills with food during a meal [27, 80].

In parallel with the first series of questions posed above, we have begun to address a second series that includes: (1) what genes regulate development of vagal sensory receptor populations that supply the stomach? (2) where are these genes expressed? – i.e., which stomach compartments and tissues produce them? and (3) when are they expressed? Tissues that are both innervated by vagal afferents and express a particular gene are the tissues where this gene’s product has good opportunities to control vagal development, and also therefore, where vagal afferents are most likely to be affected by abnormal BDNF expression.

Several classes of genes are involved in neural development. Our initial investigations of the controls of vagal development have focused on one family of nerve growth factors, the neurotrophins, which regulate several facets of sensory neuron development including precursor proliferation, survival, differentiation, axon growth and axon guidance. Neurotrophin survival effects appear to be organ-specific as NT-4 is essential for the survival of vagal sensory innervation of the small intestine [29] and neurotrophin-3 (NT-3) for vagal afferents supplying the esophagus [60]. However, it is not known which neurotrophins regulate survival of vagal sensory innervation of the stomach, or whether they control other aspects of its development. We hypothesized that brain-derived neurotrophic factor (BDNF) may support survival of vagal sensory innervation of the stomach because it is the only other factor besides NT-3 and NT-4 supporting survival of a substantial population of vagal sensory neurons. Additionally, several criteria were used to assess the probability that BDNF regulates development of vagal mechanoreceptor innervation of the stomach. An analysis based on these criteria is summarized below.

Criterion 1. BDNF-deficient mice have a significant loss of vagal sensory neurons

Meeting this criterion suggests BDNF is essential for survival or maintenance of a subset of vagal afferents. Mice lacking BDNF have exhibited as much as 66% loss of vagal sensory neurons (e.g., [23]).

Criterion 2. Tissues supplied by vagal sensory receptors normally produce BDNF

This criterion and the following one are required for BDNF to act as a classic target-derived trophic factor and may also be crucial for it to impact the formation or structure of vagal sensory receptors. BDNF has been observed in GI tract regions of embryonic, postnatal and adult wild type mice that receive vagal sensory innervation [22, 35, 39, 46, 48, 49]. However, the localization and timing of BDNF expression within specific GI tissues during development has not been characterized in detail. Therefore, we have also investigated this in the present study, and as reported below we found that BDNF is expressed in specific tissues of the developing stomach wall that receive vagal sensory innervation. Therefore, this BDNF could support survival of these afferents once their axons reach the stomach wall, and it could also regulate other stages of vagal sensory receptor development.

Criterion 3. Vagal sensory neurons normally express BDNF receptors

Indeed these neurons express trkB, the high affinity BDNF receptor [22, 86], and BDNF is transported by vagal sensory axons in the retrograde direction from nerve terminals back to the cell bodies of origin [36]. Thus, two important conditions required for BDNF to act as a target-derived trophic factor have been met (this and the former criterion).

Criterion 4. BDNF mutants have (a) loss of mechanoreceptors, or (b) altered mechanoreceptor structures innervating tissues other than those of the GI tract

This criterion is based on the assumption that different types of mechanoreceptors share some essential properties with one another, but not with other classes of receptors and that these commonalities increase the likelihood that they are regulated in a similar manner by neurotrophins. The properties shared may be developmental, structural and functional ones, and may involve some molecular components of the transduction complex [18, 19, 24, 32, 59, 85]. To the extent that this assumption is correct, should a neurotrophin have effects on mechanoreceptors innervating tissues outside the GI tract, this increases the probability that a neurotrophin will have similar effects on vagal GI mechanoreceptors. BDNF levels in mice have in fact been found to affect the survival of mechanoreceptors innervating tissues other than those of the GI tract. Knockouts of the BDNF gene have typically been associated with decreased receptor numbers, whereas, BDNF overexpression has typically been correlated with increased numbers. For example, vagal baroreceptors were absent in BDNF-deficient mice [8], and low threshold cutaneous mechanoreceptors, longitudinal lanceolate endings and Ruffini endings were reduced in number [24, 33, 41]. Similarly, the morphology of mechanoreceptor complexes (i.e. sensory nerve terminals along with their accessory cells) is affected by BDNF levels. For example, a subpopulation of somatosensory neurons exhibited increased branching of axon terminals of in vitro in the presence of BDNF [45, 77], and BDNF overexpression in vivo has typically been associated with increased size and density of cutaneous mechanoreceptor nerve terminals [6, 44]. In contrast, these receptors are decreased in size and density in BDNF-deficient mice [62]. These alterations of nerve terminal structure can also affect interactions between the terminals and their accessory cells. For instance, in BDNF-deficient mice sensory fibers growing toward the vestibular sensory epithelia approached them as they normally would, but stopped short and never made contact with these epithelia [24]. Additionally, BDNF levels influenced the density of accessory cells associated with muscle spindles and Meissner corpuscles [25, 44].

Criterion 5. BDNF levels do not have significant effects on vagal efferent (preganglionic) neurons

Meeting this criterion is important because it reduces the likelihood that effects of BDNF deficiency on vagal sensory neurons are secondary to effects on vagal efferents. In fact, BDNF-deficient mice exhibit normal numbers of vagal preganglionic neurons, suggesting BDNF is not essential for their survival [41].

This analysis strongly favors the hypotheses that BDNF supports survival of vagalsensory innervation of the stomach and that it may also regulate other facets of this pathway’s development. Therefore, after characterizing development of vagal innervation and BDNF expression in the stomach wall, we investigated the roles of BDNF by examining vagal innervation in neonatal BDNF-deficient mice. The questions we addressed were (1) which vagal elements innervating the stomach were dependent on BDNF for normal development, (2) which aspects of these elements’ development depended on BDNF, and (3) in which stomach tissues were the affected vagal elements located.

2. Methods

2.1 Subjects

Mice were maintained at 22–23°C on a 12:12 hr (14:10 hr for breeding) light:dark schedule, lights on at 0500 with ad libitum access to tap water and Laboratory Rodent Diet 5001(PMI Nutrition International, St. Louis, MO). For Experiment 1, mice of the 129SvJae or C57BL/6 strains were used. For Experiment 2, BDNF^LacZ mice on a C57BL/6 background were obtained from Kevin Jones, University of Colorado, Boulder [41]. These mice express the have the LacZ reporter gene “knocked into” the BDNF locus and were utilized to map BDNF expression in the stomach. The rationale for using these mice for this purpose has been discussed [31]. Homozygous and heterozygous mutants and wild types were genotyped by PCR using DNA extracted from a tail sample for mice, or the yolk sack for embryos. For Experiment 3, BDNF heterozygous mutants (+/−), homozygous mutants (−/−) and wild-type mice(+/+) derived from BDNF heterozygous +/− breeder pairs on a C57BL/6 background obtained from Jackson Laboratories were used and were genotyped as described for Experiment 2. All procedures were conducted in accordance with Principles of Laboratory Animal Care (NIH publication No. 86-23, revised 1985) and American Association for Accreditation of Laboratory Animal Care guidelines and were approved by the Purdue University Animal Care and Use Committee.

2.2 Experiment 1. Development of DiI-labeled vagal axons and terminal specializations innervating the stomach wall

2.2.1 Embryos and postnatal mice

Timed matings were employed to acquire embryos at specific ages. Males were placed with females for 20 min between 2000 and 2100 hr and monitored to ensure copulation occurred. The following day at 1200 h was designated E0.5. Postnatal mice were obtained from a subset of these matings and the day of birth was defined as postnatal day (P)0. Development of axons, fiber bundles and nerve terminal specializations were examined in 129SvJae embryos (E12.5, n=8; E13.5, n=6; E14.5, n=3; E15.5, n=4; E16.5, n=7; E17.5, n=3), 129SvJae postnatal mice (P0, n=16; P1, n=6; P2, n=8; P3, n=13; P4, n=4; P5, n=4; P8, n=4; and P10, n=4), and C57Bl/6 P0 mice (n=6).

2.2.2 Harvesting and fixation of embryos and neonates

After euthanizing pregnant mice by cervical dislocation and then removing the uterus to chilled 0.1M sodium phosphate buffered saline (PBS) on ice, each embryo was dissected free and their abdominal organs exposed. Next, embryos were fixed for 3 d in 4% paraformaldehyde and 0.1% ethylene-diamine tetraacetate (EDTA) at 4°C. Postnatal mice were treated similarly except that they were first anesthetized with a lethal dose of methohexital sodium (Brevital Sodium, Eli Lilly, Indianapolis, Indiana; 100 mg/kg i.p.) and perfused transcardially with saline at 40°C for 5 min followed by the same fixative employed for embryos for 30 min.

2.2.3 Application of nerve tracer to the vagus nerve

The crystal form of DiI (perchlorate; D3911, Molecular Probes, Eugene, Oregon) was employed for postmortem nerve tracing to selectively label the developing vagal innervation of the stomach. When fixation was complete, the liver was carefully removed with the aid of a dissecting microscope to visualize the abdominal vagal trunks and branches associated with the esophagus and stomach. A single DiI crystal was inserted into the anterior vagal trunk immediately anterior to the bifurcation of the hepatic and gastric branches. This site was chosen because it was close enough to the stomach to keep the required DiI diffusion distance to a minimum yet far enough to significantly reduce the potential for DiI to redistribute from the insertion site to the stomach wall. Also, because this site is a readily identifiable anatomical landmark it was used to place DiI in a similar locus in each animal. After the DiI crystal wasplaced in the nerve a #5 straight tip fine forceps (Dumont, Switzerland) was used to crush it into the nerve fibers. Tissue paper sheets (1 inch sq) that had been twisted were used to keep tissues dry and to quickly remove any loose DiI crystal fragments, which aided in restricting spread of DiI from the insertion site. Specimens were then incubated in the same solution used for fixation at 37°C for 2.5–5 weeks. Next, wholemounts of the dorsal and ventral stomach walls were prepared and mounted directly in PBS, coverslipped and sealed with clear nail polish (AM Products, NJ, USA) and then DiI labeling was examined.

2.2.4 Imaging of DiI-labeled tissues

Since DiI emits red fluorescence at 565 nm, labeled vagal axons and nerve terminals were examined using a Nikon E1000 fluorescence microscope and standard rhodamine optical filters (TRITC filter cube; Tokyo, Japan), a multi-photon confocal microscope (Radiance 2100 MP Rainbow with green helium-neon laser, 543nm; Bio Rad, Hemel Hempstead, England) run with LaserSharp 2000 software (Bio-Rad), or an Olympus BX-DSU spinning disk confocal microscope. Image processing and three-dimensional reconstructions were done using Confocal Assistant (v4.02, Todd Brelje, University of Minnesota), or Slidebook (v.4.1, Intelligent Imaging Innovations, USA) software, respectively. Photoshop software (version 6.0 Adobe Systems, Mountain View, CA) was used for this and subsequent experiments to a) apply scale bars and text, b) adjust brightness and contrast, c) apply color correction, and d) organize figure layouts.

2.2.5 Criteria for including specimens in analysis

To qualify for analysis, specimens were screened using standard fluorescence microscopy to determine whether DiI labeling of vagal stomach innervation met several criteria:(1) In each compartment of the stomach wall a large proportion of vagal fibers and terminals were labeled with DiI, and these labeled elements extended from esophagus/cardia region to the greater curvature and pylorus. (2) DiI labeling in axons and terminals was stable throughout diffusion and imaging, i.e. DiI did not leak out of these labeled elements. (3) DiI spread from the insertion site in the vagal trunk to other tissues to produce false-positive labeling did not occur. Five percent of the specimens exhibited evidence of incomplete DiI-labeling, none had DiI leakage, and 20% showed DiI redistribution. None of these specimens were examined further.

2.2.6 Criteria for identification of afferent and efferent terminal specializations

Criteria that have been published for defining the mature forms of IGLEs, IMAs and efferent terminals were utilized here to identify their putative precursors as previously described [51]. Specifically, precursors were predicted to exhibit the defining features of the mature forms, but it was also expected that these features would be reduced in number, size and density as compared with adult forms. The criteria included morphology, associations with specific stomach tissue layers and distribution across stomach compartments as reviewed in the introduction section. Additionally, we verified the vagal origin of labeled axons and nerve terminals by demonstrating that this labeling was prevented when DiI application was immediately followed by vagotomy [51].

2.3 Experiment 2. Characterization of the temporal and spatial pattern of BDNF expression in the stomach wall

Timed matings were set up between heterozygous BDNF^LacZ mice, or between heterozygous BDNF^LacZ mice and wild-type mice. Embryos were harvested on E12.5, E13.5, E15.5 and E17.5, and GI tracts were dissected from P4 mice. Results were based on multiple embryos or neonates from at least 3 independent litters at each age studied.

2.3.1 Histochemical staining and imaging of β-galactosidase expression

Subjects were obtained as described for DiI experiments above. Embryos were fixed for 30 min on ice with 1% paraformaldehyde, 0.02% gluteraldehyde, 0.5 mM EGTA, and 2 mMMgCl₂ in 0.1M sodium phosphate buffer, pH 7.4. Neonatal mice were perfused transcardially at a flow rate of 3 ml/minute with 0.9% saline for 5 min at RT followed by the same fixative used for embryos for 30 min at 4°C. Immediately after fixation the embryos or abdominal organs were stained with X-gal as previously described (Fox, 2000). After staining was completed, tissue was postfixed 48 hr in 4% paraformaldehyde at 4°C, washed with PBS and then transferred to 10% buffered formalin at 4°C for a minimum of 5 days. Then the tissues were embedded in paraffin, sectioned at 8 µm thickness, air dried on gelatin-coated slides, alternate ribbons of sections were counterstained with 0.1% neutral red, and all sections were dehydrated in a series of graded alcohols (70%, 95%, 2 × 100%; 2 min each), cleared in xylene (3 × 2 min) and coverslipped with Cytoseal (Richard Alan Scientific, Kalamazoo, Michigan). X-gal-stained tissue was examined with standard bright-field illumination (Leica DM5000 microscope). Photomicrographs were acquired directly with a video camera (Spot RT Slider; Diagnostic Instruments, Inc., Sterling Heights, MI). Preliminary results of this experiment have been reported [31] and are reviewed here.

2.4 Experiment 3. Effects of BDNF deficiency on vagal sensory innervation of the stomach wall

The vagal innervation of the stomach wall in BDNF heterozygous mutants (+/−; n = 8),homozygous mutants (−/−; n = 3) and wild-type mice (+/+; n = 5) was labeled with DiI on P0 because BDNF knockout mice often die shortly after birth. All methods were the same as described above for Experiment 1, but additionally labeled neural elements were quantified as described in the next section. Two heterozygous mutants exhibited incomplete DiI label, and one heterozygous mutant and one of the wild-types demonstrated evidence of DiI redistribution. These specimens were not examined further, resulting in final group sizes of 5 for heterozygotes and 4 for wild types.

2.4.1 Quantification of axons, axon bundles and nerve terminal specializations

To obtain relative estimates of the number of vagal fibers innervating the stomach in wild types, individual DiI-labeled axons, fiber bundles and putative precursors of nerve terminal specializations were identified and quantified. Methods for sampling were similar to those used previously regarding the distribution and density of vagal mechanoreceptors in the stomach [27,80], and those for counting axons were similar to Cheng et al. [11] regarding vagal innervation of the aortic arch. In order to assess innervation throughout the stomach, a sampling grid was used to establish a proportional scale that normalized each wholemounted stomach so that the variability due to differences in stomach size or distention was minimized. For each stomach analyzed, a map of 80 sampling points covering all areas of the stomach was established in order to determine the microscope stage coordinates of each sampling point. At each sampling point a confocal microscope and a 60X water objective (total magnification 600X) were used to scan through the entire thickness of the stomach wall in a series of optical sections spaced 1 µm apart in the z-axis, which were stored as digital images. For each sampling location, a counting grid was used to assess the densities of single axons outside bundles (“separate axons”) and IMAs, as well as the numbers of axon bundles, IGLEs, and efferent terminals.

The counting grid consisted of equidistant lines (7 vertical and 9 horizontal) drawn on a transparency, which was attached to the computer screen that displayed the optical sections and covered a 108 µm × 144 µm area of the stomach wall. The position of each fiber on the counting grid was transferred in register onto grid paper by assigning successive rows on the grid paper to successive optical sections. In this manner, each fiber was tracked along their 3-dimensional path to avoid being counted more than once. Next, axons were counted starting with the first optical section scanned at a sampling site and continuing sequentially with each subsequent section. Using this technique, Cheng et al., [11] were successful in distinguishing adjacent fibers as close as 1 µm apart, a resolution also achieved with our confocal microscope, software and display configuration. Since vagal efferent terminals remained within a focal plane and the myenteric neurons they encircled were counted rather than their nerve processes, counts were made from two dimensional projections of each 3D series of optical sections. Separate axons were identified as individually labeled neurites with about the caliber of a single axon. Their density was quantified by counting the number of vertical or horizontal lines of the counting grid they crossed. Use of the vertical or horizontal set of grid lines for this counting was determined based on which set was closest to the axis perpendicular to the orientation of the nerve process being quantified. Vagal axon bundles were identified as a group of two or more labeled axons coursing in parallel through the tissue. They were quantified by counting the number of bundles present within the counting grid area. Intraganglionic laminar endings were quantified by counting the number of individual aggregates of terminal puncta or “leaves” present within the counting grid area. For IMAs, their density was quantified by adding the number of counting grid lines crossed by each of their telodendria in a similar manner to that described above for separate axons,

2.4.2 Statistical analysis and graphical display of data

Differences in numbers or densities of neural elements that were quantified between BDNF mutants and wild-type mice were tested using one-way ANOVAs, with genotype as the independent variable and each of the counts or densities of the neural elements examined individually as the dependent variables (Statistica, v6.0, StatSoft, Tulsa, OK). The main effect of genotype was used to determine significance and Fisher’s protected least-significant difference test was used to make post-hoc comparisons. Values reported are means ± SEM. For all statistical tests, p < 0.05 was required for statistical significance. Graphpad was used to construct all graphs (Graphpad Prism Version 4.0, Graphpad Software, Inc.).

3. Results

3.1 Experiment 1. Development of DiI-labeled vagal axons and terminal specializations innervating the stomach wall

Specimens ranging in age from E12.5 to P8 that were incubated for 2.5 – 3 weeks(embryos) or 4–5 weeks (neonates) after DiI application to the anterior vagal nerve trunk exhibited labeling of axons and terminal specializations innervating the ventral stomach wall that appeared to be largely complete. In particular, DiI labeled a large number of nerve fibers, which were distributed throughout the stomach wall, reaching as far as the greater curvature or pylorus. This labeling revealed four stages of vagal development elaborated below, including (1) growth of axons into the stomach wall, (2) fiber bundle formation and maturation within the myenteric plexus, (3) the appearance of the first precursors of sensory nerve terminal specializations, and (4) the continued appearance and maturation of these precursors. Significantly, this last stage extended into the second week after birth and it may be the most important stage with respect to acquisition of sensory function.

Stage 1. Initial vagal axon growth in the stomach wall (E12.5–E14.5)

DiI-labeled vagal innervation of the stomach wall was first examined at E12.5. At this age several axon bundles radiated out from the vagal anterior gastric branch as it traversed the LES to reach the surface of the cardia of the ventral stomach wall and these bundles extended only part of the distance to the greater curvature. By E13.5 the gastric branches approached the greater curvature and by E14.5 the gastric branches had reached it.

Stage 2. Maturation of vagal gastric fiber bundles and separate axons (E13.5–E16.5)

At E13.5 both small fiber bundles and individual axons running independent of these bundles formed a dense network throughout the stomach wall (Fig. 1A). As embryos developed beyond E13.5, the number of axons present continually increased, including those in bundles and separate axons. In parallel, the number of fiber bundles appeared to remain fairly stable, whereas stomach size progressively increased, which combined to result in increasing distances between fiber bundles and between separate axons at each successive age (Fig. 1A–C). At E16.5 large bundles of vagal fibers were present throughout all the major stomach compartments. Additionally, axon bundles throughout the stomach wall appeared to have completed their development and achieved a mature distribution (Fig. 1C). In parallel with development of vagal myenteric axons, at E13.5 individual axons had begun to exit the myenteric plexus to enter the adjacent smooth muscle, submucosal and mucosal layers. Between E14.5 and E15.5 these axons increased in number and by E16.5, in addition to these separate axons many small fiber bundles arose from the large myenteric nerve bundles and extended into the smooth muscle, submucosal and mucosal stomach wall layers.

Fig. 1.

Confocal images illustrating three stages of development of vagal axons and fiber bundles in the forestomach. A. At E13.5 a plexus-like pattern started to emerge consisting of separate axons and axon bundles of small diameter. B. At E14.5, axon bundle diameters and distances between separate axons and axon bundles increased. C. By E16.5, these diameters and distances increased further, and the pattern of organization of axon bundles and separate axons was similar to that observed at mature ages. Scale bars = 10 µm.

Stage 3. Initial formation of vagal gastric mechanoreceptors

A small number of putative IMA and IGLE precursors were first observed in most embryos examined at E16.5 and E17.5. These precursors each exhibited some of the unique characteristics of their fully differentiated forms, which permitted their identification. At their initial stage of development, putative IGLE precursors were often composed of planar clusters of terminal puncta covering all or a portion of a myenteric ganglion, similar to the adult form. The immature aspects of their structure included the smaller size of these terminal aggregates, the smaller number of terminals present in each cluster, and their looser packing as compared with mature IGLEs. In some instances however, putative IGLE precursors appeared to consist of aggregates of small numbers of growth cone-like structures that covered myenteric ganglia (e.g. at P0, Fig. 2A). In contrast, putative IMA precursors consisted of fine, short, individual processes – immature telodendria - innervating either the circular or longitudinal smooth muscle layer and oriented parallel to smooth muscle fibers (e.g. at P0, Fig. 3A). Also, crossbridge fibers that interconnect the telodendria of some IMAs were not observed at these ages.

Fig. 2.

Confocal images of DiI-labeled putative IGLE precursors shown in three stages of development in the forestomach. A. An early stage of putative IGLE precursor shown at P0, which consisted of DiI-labeled vagal fibers and axons in the myenteric plexus, terminating in growth cone-like puncta structures (arrow) present at the surface of myenteric ganglia. B. A second stage of putative IGLE precursor maturation observed here at P3 with increased numbers of putative terminal puncta, which in this instance consisted predominantly of small quasi-spherical structures (arrow). C. A third stage of putative IGLE precursor maturation shown here at P8 with continued development of putative terminals, consisting of more numerous terminal structures that were more densely packed in leaf-like patterns arranged in several groups. This and other IGLEs at this stage often exhibited some groups of terminal puncta lying below the plane of the myenteric plexus, and others above it. Scale bars = 10 µm.

Fig. 3.

Confocal images of DiI-labeled putative IMA precursors shown in four stages of development in the muscle layers of the forestomach. A. An early stage of putative IMA precursor development illustrated here at P0. An axon exited the myenteric plexus and entered the smooth muscle layer before branching into two processes that could have been growing axons, or precursors of IMA telodendria (arrows). A small neurite that extended from the termination of one of these processes (arrowhead) may have represented the initial formation of an additional putative telodendrion. Separate axons present were in the myenteric plexus. B. At the next stage of putative IMA precursor development also illustrated here at P0, single axons exited the myenteric plexus and entered the muscle layer, distributed additional axons and short rectilinear fibers forming precursors of IMA telodendria (e.g., arrows) that paralleled the muscle fibers, and also exhibited an interconnecting crossbridge fiber (arrowhead). C. Putative IMA precursor at P3 in the muscle layer at a later stage of maturation in which the telodendria (running diagonally from upper right to lower left) had lengthened and become more numerous - as had crossbridge fibers (arrowheads). D. Montage of confocal images of putative IMA precursors at P4 demonstrating the further lengthening of their maturing telodendria relative to earlier stages illustrated in panels A–C, and representing a portion of a field of multiple IMAs that was forming. Scale bars = 10 µm.

In addition to these structural features and associated target tissues, the distributions of IMA and IGLE precursors across stomach compartments aided their identification. At all ages where putative IMA and IGLE precursors were present they were distributed in a similar manner to their mature pattern. Putative IMA precursors were found mainly in the peri-esophageal region of the lesser curvature and in the forestomach. Within the forestomach IMAs that innervated the circular smooth muscle layer were concentrated near the lesser curvature and those that supplied the longitudinal smooth muscle layer near the greater curvature. Further, a region of overlapping circular and longitudinal IMAs in the forestomach was observed as early as P0 in some mice that may correspond to an immature form of the “fovea” observed in adult rats and mice, which has been suggested to be a high acuity region for stretch detection [27,80]. In contrast, IGLE precursors formed in the muscle wall throughout all major stomach compartments.

Stage 4. Maturation of vagal gastric mechanoreceptors

After E17.5, additional immature endings continued to arise as embryos and postnatal mice increased in age, and consequently putative vagal nerve terminal specializations continued to increase in number. In parallel, at a given age, putative vagal mechanoreceptor precursors that arose at earlier ages continued to mature, becoming more similar to their adult forms as development progressed through P8. Consequently, at each age examined, IMA and IGLE precursors at different stages of maturation were present.

At birth and shortly thereafter (P0 – P2), a subset of putative IMA and IGLE precursors already appeared to be more mature than their prenatal forms. Some putative IMA precursors exhibited telodendria of increased length, and occasionally they were interconnected by bridging collateral branches (Fig. 3B). Thus, some IMA precursors demonstrated all the features of the adult structure, but were still much smaller in size. Similarly, some of the putative IGLE precursors present had become larger in size and consisted of greater numbers of terminal puncta that were more tightly packed (Fig. 2B). Also, the terminal puncta of IGLE precursors exhibited one of, or a combination of, several morphologies, including growth cone-like, circular/spherical, or elongated structures. From P3 through P8, these putative mechanoreceptor precursors continued to mature along these same trajectories, and by P8 they had developed structures with characteristics approaching those of mature IMAs (e.g., Fig. 3C) and IGLEs (e.g., Fig. 2C).

3.2 Experiment 2. Characterization of the temporal and spatial pattern of BDNF expression in the stomach wall

At E12.5, the stomach wall consisted of a thick mesenchymal layer surrounding a thin epithelial layer. At this age BDNF was expressed in the mesenchyme of the caudal stomach region that develops into the corpus and antrum, whereas, there was no expression in the epithelial layer. A similar expression pattern was observed at E13.5, although in some specimens the tissue layers of the stomach wall had begun to differentiate and BDNF was expressed in the precursors of the lamina propria and muscle wall. By E15.5 the layers of the stomach wall had become distinct in all specimens. Strong BDNF expression at this age was restricted to the longitudinal and circular smooth muscle layers of the antrum and the lamina propria of the mucosa in the antrum and corpus. Additionally, the walls of blood vessels, which appeared to be restricted to arteries and arterioles that coursed along the outer surface of, and within the stomach wall - predominantly within the muscle wall and the adjacent portion of the submucosa - exhibited strong BDNF expression. This overall expression pattern was maintained at E17.5 (Fig. 4). At P4 this expression pattern remained the same, but was inconsistent ranging from none in some neonates to fairly strong in others. This suggests that expression levels may have been decreasing at different rates in different neonates. There was no expression in the submucosa or epithelium at any of the ages examined.

Fig. 4.

Brightfield photomicrographs of sections from BDNF^LacZ mice stained with X-gal illustrating BDNF expression in the stomach wall at E17.5. The images in panels A and C were counterstained with neutral red, illustrating all tissue layers of the stomach wall as well as X-gal staining, whereas B and D respectively are nearby sections that were stained only with X-gal to illustrate the extent of BDNF expression. A,B. BDNF expression occurred in tissues that developed into the lamina propria of the mucosa in the antrum and corpus. It was also present in arterial blood vessel walls (cross-sectioned in this sample) within the muscle wall and adjacent portion of the submucosa. Scale bar in A = 50 µm, also applies to B,C and D. C,D. BDNF expression was also present in some regions of tissues developing into both layers of the antrum smooth muscle wall. Additionally, near the center of the image in D there is an artery running on the surface of the stomach wall that exhibited BDNF expression and a branch of this vessel that entered the stomach wall and continued to express BDNF is present in the lower left of the image. Lamina propria, lp; smooth muscle, sm.

3.3 Experiment 3. Effects of BDNF deficiency on vagal sensory innervation of the stomach wall

Comparison of DiI labeled vagal innervation of the stomach in BDNF-deficient and wild-type mice did not support our prediction - based on the organ-specific hypothesis - that the survival of all vagal sensory receptor types innervating this organ was mediated by BDNF. Rather, the survival of only one receptor class, IGLEs, was affected. Quantitative analyses revealed reductions of approximately 50% in putative IGLE precursor numbers in heterozygous and homozygous mutants relative to wild types [F(2,9)=5.27, p<0.05; Fisher’s LSD for +/+ vs.+/− p<0.5, for +/+ vs. −/− p<0.5] (Fig. 5C). In contrast, there was an increase in vagal innervation of the stomach wall as reflected in a 50% increase in the density of separate axons in both heterozygous and homozygous mutants [F(2,9)=33.35, p<0.05; Fisher’s LSD for +/+ vs. +/− p<0.5, for +/+ vs. −/− p<0.5](Fig. 5B), whereas vagal axon bundle numbers [F(2,9)=0.59 ,p=0.58](Fig. 5A), IMA densities [F(2,9)=0.38, p=0.7](Fig. 5D), and efferent terminal numbers(wild type = 1970.5 ± 34.8; heterozygous mutants = 1574.4 ± 90.3; homozygous mutants = 1616.0 ± 221.2) [F(2,9)=2.41, p=0.15] were similar in wild types and mutants. Importantly, the lack of a detectable effect on efferent terminal numbers is consistent with the effects of BDNF deficiency on survival being restricted to sensory neurons.

Fig. 5.

Preliminary quantitative comparisons of the effects of BDNF +/+, +/− and −/− genotypes on vagal innervation of the stomach at P0. A. Mean total numbers of vagal bundles counted at all sampling sites. B. Mean total numbers of intersections of separate vagal axons with the lines of the counting grid added across all sampling sites. C. Mean total numbers of putative IGLE precursors counted at all sampling sites. D. Mean total numbers of intersections of telodendria of putative IMA precursors with counting grid lines added across all sampling sites. * Significantly different from wild type at p < 0.5 level.

An additional hypothesis proposed that BDNF may have roles in vagal sensory development other than survival, including axon growth or guidance and mechanoreceptor differentiation. Qualitative observations provided evidence consistent with this hypothesis. First, the morphology of DiI-labeled IMA precursors in some regions of the forestomach was abnormal in BDNF homozygous mutants: the elongation of their telodendria appeared to be stunted as these processes were short and exhibited larger-than-normal diameters, possibly representing prematurely stabilized growth cones (compare Fig. 3C,D and Fig. 6A with Fig. 6B). Second, the patterning of innervation in the antrum was abnormal as the vagal axons and fiber bundles exhibited a disorganized appearance, suggesting axon guidance or fasciculation was disrupted (compare Fig. 6C,D).

Fig. 6.

Confocal images of DiI-labeled vagal fibers and terminals compared in P0 BDNF knockout (−/−) and wild-type (+/+) mice. A. Putative IMA precursors present in the forestomach of BDNF +/+ mice with telodendria extending horizontally across the image. B. The growth of some of the telodendria (oriented diagonally across the image) of putative IMA precursors observed in the forestomach of BDNF −/− mice appeared stunted and they had larger-than-normal diameters. C. In the antrum of BDNF +/+ mice separate axons and axon bundles typically formed adult-like patterns of vagal stomach wall innervation as illustrated in this image. D. In contrast, in the antrum of BDNF −/− mice, separate vagal axons and fiber bundles often exhibited a disorganized pattern of innervation. The example shown here is an extreme instance, demonstrating both aberrant organization as well as a large increase in axon density. Scale bars = 10 µm.

4. Discussion

Development of vagal sensory innervation of the GI tract has been assumed to be largely complete at birth. Consequently, this development has not typically been considered to play a significant role in normal or disordered feeding behavior that develops postnatally. Similarly, it has not been raised as a possible target of abnormal perinatal environments that affect feeding behavior, body weight or metabolism. However, vagal development in the GI tract has not been directly observed, but has been indirectly inferred, for example on the basis of retrograde labeling of vagal sensory and motor neurons after tracer injections in the GI tract. In the present study anterograde nerve tracing with DiI was employed to label vagal projections to the mouse stomach wall during embryonic and early postnatal ages. Examination of labeled innervation from E12.5 - P8 revealed four stages of development involving axon growth into the stomach wall, formation of axon bundles, the initial appearance of mechanoreceptors, and finally mechanoreceptor differentiation, with the majority of this sensory receptor development occurring postnatally.

Given that neural development in the GI tract generally follows a rostrocaudal pattern, this raises the possibility that vagal development in not only the stomach, but in all GI tract organs may continue postnatally and thus be malleable in response to postnatal conditions. Therefore, we initiated examination of genetic control of vagal development, assuming that this is a major route through which abnormal environments could influence it. The first gene of interest we have begun to investigate is BDNF. Before examining the effects of BDNF deficiency on vagal development we characterized the expression pattern of BDNF in the stomach wall from E12.5 to P4, ages when key events in development of vagal stomach innervation occurred. At all ages examined, BDNF was produced in a subset of stomach tissues associated with developing vagal afferents, including the lamina propria of the mucosa in the corpus and antrum, the smooth muscle layers of the antrum, and the walls of arterial blood vessels supplying the stomach wall, raising the possibility that BDNF contributes to their regulation. BDNF deficiency reduced the number of developing IGLEs by half at birth, whereas surprisingly, axons separate from bundles increased in density by 50%. Additionally, IMAs in the forestomach exhibited abnormal morphology and patterning of axons and fiber bundles in the antrum appeared disordered.

Thus, BDNF contributes to the regulation of vagal development, including development of vagal mechanoreceptors at ages when abnormal environments can have effects that lead to obesities with a developmental origin (Fig. 7). Therefore, altering the expression of BDNF or other genes could be one possible mechanism these environments engage to ultimately change the course of development of vagal afferents important in signaling satiation and regulating food intake.

Fig. 7.

The temporal relationships between developmental age, BDNF expression in the stomach, and stages of vagal development are represented. These relationships illustrate the possibility for abnormal perinatal nutritional or hormonal environments to modify ongoing BDNF expression and thereby its regulation of vagal sensory development, involving vagal axon development at earlier ages and differentiation and growth of sensory receptors at later ages.

4.1 Development of DiI-labeled vagal axons and terminal specializations innervating the stomach wall

Characterization of the temporal and spatial pattern of developing vagal stomach innervation revealed four stages of normal development of vagal afferent innervation of the GI tract: (1) growth of axons into the stomach wall (E12.5–14.5), (2) fiber bundle formation and maturation within the myenteric plexus (E13.5–16.5), (3) the appearance of the first precursors of sensory nerve terminal specializations (E16.5 – 17.5), and (4) the continued appearance and maturation of these precursors (P0–8). Importantly, this finding of this last stage is contrary to the common assumption that vagal development is largely complete at birth, and suggests one of the most important developmental events with respect to acquisition of sensory function, namely, the differentiation of nerve terminal specializations occurs mainly after birth, rendering this process susceptible to postnatal environmental influences.

The pattern of vagal axon growth and fiber bundle formation within the stomach wall visualized in the present study using a method that labeled only vagal innervation was similar to the patterns observed utilizing techniques that additionally labeled non-vagal axons, including acetylcholinesterase in toto staining and P2X₃ immunostaining [5, 82]. This pattern is also consistent with results obtained using WGA-HRP to selectively label vagal sensory innervation at P10 [75]. Additionally, the timing of arrival of axons in the stomach wall agrees with a study that used DiI for retrograde labeling of vagal sensory neurons in the NTS and vagal preganglionic neurons in the DMV, which determined that vagal sensory axons first reach the gut by E12 and efferent ones by E13 or 14 [63].

To our knowledge only two other studies have examined vagal nerve fibers and terminals within the developing stomach wall, and for the most part our findings are consistent with their observations. In one of these studies, which employed anterograde nerve tracing with WGA-HRP in P10 rats, IGLEs appeared mature, they exhibited an adult-like distribution and their density was similar to adults [75]. In contrast, IMAs, which were distributed in a mature pattern – largely restricted to the forestomach, were much less dense than in adults. Our findings were consistent as the IGLE and IMA structures in the oldest mice we examined (P8) approached the forms previously described in adult rodents, and their distributions were adult-like(e.g., [3, 52, 65, 80]). In the other investigation of vagal terminal development in the stomach wall, which employed IHC detection of P2X₃ receptors in the rat, IGLEs were first observed at P1 and IMAs at P7 [82]. This is in contrast to our first observation of putative precursors of both of these receptor classes at E16.5. The factor that probably contributed most to this difference is that we included putative immature IMA and IGLE forms, whereas Xiang and Burnstock may have focused on terminals with a more mature appearance. Other factors that may have contributed to this difference include the delayed developmental timetable in the rat as compared with the mouse, the additional non-vagal enteric neural elements stained with P2X₃ may have obscured the simpler and more immature forms of IGLE and IMA precursors present early in development, and the initial immature forms of IGLE or IMA precursors may not express sufficient levels of the P2X₃ receptor for detection by IHC.

4.2 BDNF is expressed in specific stomach tissues during development

Neurotrophins may be produced at several sites along a sensory or sympathetic pathway, including the ganglion of origin, the axon trajectory and the target tissue (e.g., [57]), where they may be involved, respectively, in ganglion development, axon growth, and sensory receptor formation, survival or maintenance. It has long been known that BDNF is produced inthe region of the nodose-petrosal ganglion complex, that some neurons in these ganglia express the high affinity receptor for BDNF, trkB, and that a large proportion of these neurons are dependent on BDNF [22, 23, 81, 86]. However, the GI tract tissues innervated by these BDNF-dependent neurons are not known and although BDNF expression has been observed in the developing GI tract, the timing and localization of this expression has not been adequately characterized. Determining that BDNF is in fact present in the stomach wall during development would strengthen the possibility that this organ is innervated by BDNF-dependent vagal sensory neurons. Further, stomach wall tissues that contained both BDNF and developing vagal elements are among the most promising sites for BDNF regulation of vagal development, and ultimately for abnormal perinatal environments to influence this development through effects on BDNF expression.

In the present study we found that BDNF expression occurred in three distinct stomach tissues during mid-late embryonic and early postnatal ages, in parallel with important events in vagal development. These expression sites included: (1) the walls of blood vessels, mainly small arteries and arterioles on the stomach surface as well as within the muscle wall and adjacent portion of the submucosa, (2) the smooth muscle in the antrum, and (3) the lamina propria of the mucosa in the corpus and antrum. Each of these expression sites provides an opportunity for secreted BDNF to interact with developing vagal afferents. Since vagal fibers grow along blood vessels toward sites of innervation within the stomach wall, BDNF secreted from blood vessel walls would have access to these fibers and thus could contribute to their guidance or survival. BDNF expression in the muscle wall of the antrum is positioned to influence axons growing into this tissue and to affect the formation, morphology or survival of IGLEs, which are located between the muscle layers. Similarly, BDNF produced in the mucosa of the corpus and antrum could influence sensory innervation of these tissues.

4.3 BDNF deficiency involves altered development of vagal sensory innervation of the stomach

The effects of BDNF manipulation on development of vagal innervation of the stomachwas investigated to determine whether BDNF is a feasible candidate for mediating effects of abnormal perinatal environments on developing vagal GI afferents. Surprisingly, the mapping of vagal sensory innervation in BDNF-deficient mice at birth revealed that a substantial proportion of this innervation was present. This finding contrasts markedly with the 80–90% loss of IGLEs and axon bundles innervating the small intestine in NT-4 knockout mice [29] and the 50% loss of IGLEs from the esophagus in mice haploinsufficient for NT-3 or its high affinity receptor trkC[60]. It is also inconsistent with the organ-specific hypothesis of neurotrophin regulation of survival of vagal sensory innervation of the GI tract. However, it remains a possibility that at P0 developing fibers and nerve terminal specializations of neurons that have recently undergone programmed cell death, although still present were in the early stages of degradation and would have subsequently been lost. We are currently testing this hypothesis by examining vagal innervation in BDNF mutants that survive beyond P0. Alternatively, the present results suggest a significant proportion of BDNF-dependent vagal sensory neurons innervate organs other than the stomach, for example, the heart or lungs. The only targets of BDNF-dependent neurons identified to date are arterial baroreceptor and chemoreceptor afferents, which appear to be supported mainly by target-derived BDNF [8, 38].

Despite the substantial stomach wall innervation present in BDNF knockout mice at birth there was approximately a 50% loss of putative IGLE precursors, which suggests BDNF plays a role in survival of this class of vagal gastric sensory receptors, although we cannot rule out that their development was delayed. Moreover, this occurred in both heterozygous and homozygous mutants in all stomach compartments (not shown), implying that a partial reduction in BDNF levels is sufficient to produce this effect. Similar to this finding, heterozygous and homozygous BDNF mutants exhibited changes in sympathetic innervation of the pineal gland that were comparable in magnitude [43]. The widespread nature of this IGLE precursor loss suggests it was due to the absence of one or more BDNF sources capable of influencing a pool of neurons that supplies distributed throughout the stomach muscle wall. The candidate sources include BDNF expressed in the region of the nodose ganglion and in walls of blood vessels that were in close proximity to growing IGLE axons. Additionally, the loss of only half of the putative IGLE precursors in BDNF knockout mice raises the possibility that there are at least two different IGLE subpopulations innervating the stomach wall, one that is BDNF-dependent and one that is not. This finding extends the observation of differential dependence of IGLEs on neurotrophins observed in NT-4 knockout mice, which exhibited loss of IGLEs from the small intestine but not the stomach [29]. Consistent with IGLEs supplying these two organs having different properties, IGLEs in the stomach and intestine varied in some morphological parameters [40] and neurochemical properties (e.g., intestinal and some gastric IGLEs are differentially sensitive to capsaicin; [4]). However, it remains to be determined whether there are distinct morphological or neurochemical IGLE phenotypes present in the stomach that might coincide with the BDNF-dependent and –independent populations.

In contrast to the reduced numbers of putative IGLE precursors in BDNF-deficient mice, the density of putative IMA precursors did not appear to be decreased. Moreover, they were accurately targeted to the muscle layers where putative IMA precursor formation appeared to initiate normally, suggesting these processes do not depend on BDNF. However, putative IMA precursors in BDNF knockout mice exhibited abnormal morphology, characterized most notably by stunted telodendria, which suggests BDNF may be required for full outgrowth of these terminal processes and for development of normal IMA structure. Since this effect was most dramatic in the forestomach, which did not exhibit BDNF expression during development of its vagal innervation, it may involve loss of BDNF secreted in the region of the nodose ganglion or from blood vessel walls in close proximity to growing IMA axons as described above for putative IGLE precursors.

An interesting and unexpected consequence of BDNF deficiency was an increase in single fibers coursing outside axon bundles, which in normal adult rodents typically exit the bundles, traverse a short distance, and then distribute sensory receptors. One possible explanation for this increase is that the number of total axons (axons within fiber bundles plus axons outside bundles) had remained the same, but greater numbers than normal had exited the fiber bundles. However, given that the increase in separate fibers was large and qualitatively there was no apparent decrease in bundle sizes through out the stomach this is unlikely. Alternatively, there may have been a true increase in separate axons similar to the various forms of increased innervation reported in studies of other sensory and autonomic systems, which may have resulted from increased branching of developing axons (e.g., [47]). For example, BDNF knockout mice exhibited an increase in branching of sensory innervation of the tongue, as well as increases in Merkel innervation, most trkA-dependent innervation of the upper dermis, and sympathetic terminations supplying of hair follicle necks and the pineal gland [33, 43, 47, 62]. These findings suggest that counter to its more typical growth and survival promoting activities [24, 37], BDNF may function in some tissues to suppress axon growth and branching. Curiously, despite this increase in the density of separate axons, there was no obvious indication that increased receptor numbers were forming. It is possible they might have formed if mice survived beyond P0 as we observed the continued appearance of putative mechanoreceptor precursors in the week after birth in normal mice.

One additional effect associated with BDNF deficiency was aberrant patterning of vagal myenteric plexus bundles and separate axons, particularly in the antrum. This could have been a result of the loss of BDNF from the antrum muscle wall because BDNF secreted from this tissue would have access to the myenteric plexus where it could influence axon bundle formation, for example by acting as an axon guidance signal. Also, loss of BDNF from the lamina propria of the antrum may have contributed to this aberrant patterning of axons. For example, BDNF secreted by this tissue may guide axons from the myenteric plexus into the mucosa, so that in the absence of BDNF some of these axons may remain in the muscle layer and grow in somewhat random patterns searching for their missing guidance cue. A similar effect resulting in disorganized appearance of forestomach IMA innervation was observed in c-Kit and steel mutant mice that lack the intramuscular class of ICCs in the forestomach [28, 30]. Since these ICCs are cells that complex with IMA terminals, similar to the proposed mechanism in BDNF-deficient mice, the absence of ICCs may result in loss of an important chemoattractant signal for IMAs, resulting in their aberrant patterning.

One additional issue that should be raised is that examination of developing vagal sensory innervation of the stomach in BDNF-deficient mice may not reveal all BDNF developmental roles as a consequence of its interactions with other factors. For example, BDNF and NT-3 are known to cooperate in the survival of vagal sensory neurons [20], and therefore, it is also possible they interact in controlling other aspects of vagal sensory development. This is especially important for vagal innervation of the stomach as we have found NT-3 expression at various sites within the stomach wall, some that overlap with BDNF expression [31]. Therefore, it is possible that NT-3 present in the stomach wall could compensate for the lack of BDNF (e.g.,[21]).

4.4 Conclusions

The numerous effects of BDNF deficiency on vagal innervation of the stomach make BDNF a good candidate for mediating any effects that abnormal perinatal environments might have on development of vagal innervation of the stomach, which could in turn influence food intake. For example, if abnormal environments cause decreases in the expression of genes essential for development of vagal sensory neurons, including BDNF, this might lead to reduced numbers of surviving IGLEs and normally formed IMAs available to signal satiation. If this loss is not compensated for, the consequent reduction in negative feedback signaling would lead to increased meal size – as has been observed in NT-4 deficient mice [29]. Further, combined with other deficits induced by an abnormal environment – for example, the reduction of hypothalamic pathways originating in the arcuate nucleus that influence regulation of food intake (Simerly, current issue) – the increase in meal size could cumulatively result in an overall increase in food intake, and thus possibly lead to increased bodyweight. As illustrated in Fig. 7, environmental effects that influence subsequent feeding behavior and bodyweight occur during stages of development involving BDNF expression and vagal development in the stomach wall and thus at a time when these processes would be susceptible to those environmental influences.

Finally, irrespective of whether vagal development is disturbed by abnormal nutritional or hormonal environments, it is possible that vagal sensory pathways could be modulated to increase or decrease the satiating effect of food consumed as a means of aiding the treatment of eating disorders. Consistent with this possibility, a manipulation of vagal development that appeared to increase vagal sensory innervation of the small intestine was associated with enhanced satiation, increased sensitivity to the satiating effects of CCK, and reduced meal size[13].