INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

GASTRIC DISTENSION-RELATED neurons of the nucleus of the solitary tract (NST) are strongly excited by subfemtomolar doses of the cytokine tumor necrosis factor- (TNF-; Ref. 9). Additionally, after exposure to TNF-, these neurons exhibit potentiated responsiveness to subsequent afferent stimulation. Given that vagal afferent connections with NST neurons are predominantly glutamatergic (27, 39, 41), it is possible that TNF- modulates responsiveness of NST neurons through a modification of glutamate neurotransmission.

Much evidence suggests that glutamate is the primary neurotransmitter released by vagal sensory neurons at the level of the NST (27, 39, 41). Microinjection of exogenous glutamate into this brain stem area evokes physiological changes such as cardiovascular responses, respiratory modulation, and esophageal contractions involved in swallowing (36, 40). Electrophysiological studies demonstrated that identified gastric distension-related NST neurons respond similarly to iontophoretic application of glutamate or natural stimulation. Responses to either stimulation could be blocked by the glutamate antagonist kynuretic acid (27). Axon terminals that contact the NST show glutamate immunoreactivity in all solitary subnuclei (23, 42). Immunohistochemical and pharmacological studies have shown that the N-methyl-D-aspartate (NMDA) and -amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) receptors are present in the various solitary subnuclei, and each carries a discrete functional role (2).

The neural circuitry of the dorsal vagal complex (DVC), made up of the sensory NST and the dorsal motor nucleus of the vagus (DMN), is involved in the control of physiological functions that may be affected during illness. Cytokines elicit a variety of physiological changes associated with illness, e.g., fever, nausea, and vomiting, through modulation of the central nervous system (CNS). The route by which these peptides evoke changes in CNS activity has been the subject of debate. Some investigators have proposed that vagal afferent fibers propagate information about peripheral cytokine generation centrally (12-14). Other studies have implicated direct entry of cytokines at the DVC, because it exhibits the characteristics of a circumventricular organ, i.e., fenestrated capillary network and absence of functional blood-brain barrier (8, 15, 16, 19, 29). Regardless of the route of activation of TNF- in the DVC, the majority of vagal afferents uses glutamate at the level of the NST (27, 39). Therefore, the possibility exists that cytokines, such as TNF-, may modulate the efficacy of glutamate-mediated activity of neurons in this area. Indeed, several studies have shown that cytokines can increase glutamate release in the NST (22, 25) as well as alter the firing rate of NST neurons that are coupled to vagal afferent stimulation, i.e., gastric balloon distension (9).

Our previous study used immunocytochemical identification of the immediate early gene product c-Fos as a marker for neuronal activation in the DVC (15). Peripheral administration (intravenous or intraperitoneal) of lipopolysaccharide (LPS; bacterial membrane component that induces the synthesis and release of cytokines from macrophages and glia) caused a significant increase in c-Fos in the DVC. The number of neurons within the DVC that expressed c-Fos activation after peripheral administration of LPS correlated with plasma levels of TNF-. This presumptive activation of DVC neurons did not require intact vagal pathways, suggesting that peripherally generated TNF- acts directly on these neurons (15). Our earlier gastric motility study (16) revealed that direct nanoinjection of TNF- into the DVC inhibited centrally stimulated motility in a dose-dependent manner. Given that those studies were performed in animals pretreated with dexamethasone, production of additional cytokines or the initiation of the cytokine cascade in response to TNF- was not a factor. Our electrophysiological studies (9) have shown that microinjection of TNF- onto identified neurons of the NST can rapidly and directly activate these cells as well as modify their responsiveness to afferent stimulation. Taken together, these studies strongly support the hypothesis that TNF- directly affects DVC neuronal circuitry.

The current study was designed to 1) definitively demonstrate that TNF-, and not some other cytokine, was responsible for c-Fos activation of NST neurons, 2) test the hypothesis that TNF- activation of NST cells is dependent on glutamate transmission, and 3) determine if glutamate antagonism retards the effectiveness of other agonists known to activate the NST (26) and cause gastric relaxation (34). It is possible that TNF- acts presynaptically (e.g., increased glutamate release) or that it modulates the postsynaptic action of glutamate. Either way, blockade of glutamate receptors should result in less c-Fos expression than with injections of TNF- alone. Therefore, we examined the induction of the c-Fos protein in the NST in response to nanoinjections of TNF- or in response to coinjection of TNF- with the AMPA receptor antagonist 1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-benzo quinoxaline-7-sulfonamide disodium (NBQX; Ref. 31) or the NMDA antagonist MK-801 (46).

Because practically all NST neurons receive glutamatergic inputs, it is possible that blocking AMPA or NMDA receptors may produce a nonspecific inhibition that affects the potency of any agonist operating on the NST. To evaluate this possibility, we took advantage of the fact that oxytocin (OT) directly activates NST neurons and also produces a potent gastroinhibition by operating on vago-vagal reflex circuitry (1, 26, 30, 34). If blockade of glutamate transmission with NBQX or MK-801 has a uniform effect to reduce NST excitability, then the numbers of c-Fos-expressing NST neurons will be reduced after either TNF- or OT.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals. Forty-six male Long-Evans rats (Simonsen Labs, Gilroy, CA) weighing 200-500 g were provided with food and water ad libitum and kept on an approximate 12:12-h light-dark cycle. All experimental protocols were performed according to guidelines set forth by the National Institutes of Health and were approved by the Ohio State University Institutional Laboratory Animal Care and Use Committee.

Chemicals. Animals were anesthetized with thiobutabarbitol (Inactin; Sigma-RBI, St. Louis, MO) dissolved to a concentration of 100 mg/ml in saline solution and administered at a dose of 200 mg/kg body wt ip. PBS solution (124 mM NaCl, 26 mM NaHCO₃, 2 mM KH₂PO₄; 304 mosmol/kgH₂O; pH 7.4) was used as a solvent diluent as well as a vehicle injection control. Recombinant rat TNF- (R&D Systems, Minneapolis, MN) was dissolved in PBS to a concentration of 3 × 10⁶ M, divided into 25-µl aliquots, and stored at 70^oC until use. OT (Bachem Bioscience, King of Prussia, PA) was dissolved in PBS to a concentration of 1 mM, divided into 25-µl aliquots, frozen, and stored. The concentration of OT used in this study was chosen because it was shown to evoke changes in DVC neuronal firing (26) and suppress gastric motility (34) in our previous work. The AMPA glutamate receptor antagonist NBQX (Sigma-RBI) was diluted to a concentration of 1 mM, aliquoted at 25 µl, and frozen until use. The NMDA glutamate receptor antagonist (5R,10S)-(+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclo-hepten-5,10- imine hydrogen maleate (MK-801; Sigma-RBI) was reconstituted in PBS to a concentration of 1 mM and was stored at 4^oC until use. Stored aliquots of TNF- were further diluted with PBS such that the injection pipette contained 10⁷ M, 10⁸ M, or 10⁹ M TNF-. These concentrations were chosen because they have been shown to excite single NST neurons and significantly suppress gastric motility in our previous studies (9, 16). NBQX was further diluted to a final concentration of 0.1 mM in the injection pipette (21). MK-801 was diluted such that the injection pipette contained a concentration of 0.67 mM (47).

Surgical preparation. The animal was deeply anesthetized with Inactin, and the trachea was cannulated to maintain an open airway. The animal was placed in a stereotaxic frame, the occipital skull plate was removed, and the dura and arachnoid meninges were resected exposing the dorsal surface of the brain stem.

Experimental design. Experiment 1 was designed to investigate whether neurons in the NST express c-Fos in response to nanoinjections of TNF- directly into the brain stem and whether that expression is dependent on the dose of TNF-. Experimental groups received unilateral nanoinjections of one of the following conditions: 1) PBS (n = 5), 2) 10⁷ M TNF- = 2 fmol = 34 pg (n = 5), 3) 10⁸ M TNF- = 0.2 fmol = 3.4 pg (n = 6), 4) 10⁹ M TNF- = 0.02 fmol = 0.34 pg (n = 5). These concentrations of TNF- were chosen as a result of our previous studies of the effects of the cytokine on DVC circuits controlling gastric motility (9).

8

Histological processing. Three hypotheses were tested using a total of nine experimental groups. Animals were assigned to the groups at random; therefore, the c-Fos processing across groups was also performed at random. Controls were run throughout the duration of all three experiments. Brain stems were sectioned on a freezing microtome at 40-µm thickness, and sections were collected in PBS. Sections were treated with 1% sodium borohydride to reduce remaining fixative in the tissue. Tissue sections were incubated in 10% normal sheep serum plus 0.3% Triton X in PBS to block nonspecific binding of the primary c-Fos antibody. The tissue was then incubated in primary c-Fos antibody (Oncogene AB-5; rabbit c-Fos, 1:20,000) plus Triton X in PBS for 17 h at room temperature with gentle agitation. The following day, the tissue was incubated in biotinylated goat, anti-rabbit IgG (Vector, 1:600) for 1 h. Sections were then reacted with Vector elite ABC (1:600 in PBS) for 1 h followed by Vector SG peroxidase detection reagents. Specificity of the c-Fos immunocytochemical reaction was verified by omitting the c-Fos antibody from randomly selected sections. Sections were rinsed, mounted on gelatin-coated glass slides, dried, cleared in Hemo-De (Fisher, Pittsburgh, PA), and placed under a coverslip with Entellan (Electron Microscopy Sciences, Fort Washington, PA).

Counting c-Fos-labeled cells. c-Fos-labeled nuclei in the NST were counted manually with the aid of an MD2 Microscope Digitizer (Minnesota Datametrics, St. Paul, MN) encoder attached to the stage of a Leitz Dialux Microscope. Inclusion of c-Fos-labeled neurons required that their stained nuclei be at least 6 µm in diameter and that they exhibit a nucleolus. These criteria guaranteed that staining artifacts and nuclear fragments would not be included in the count (15). c-Fos-stained nuclei were counted without knowledge of the experimental condition, and a second observer verified counts. The agreement between counts of the two observers was within 5%. Counts were made from one histological section that corresponded to site of the injection. The site of injection was taken to be the 40-µm-thick section that exhibited maximal c-Fos-labeled cells for a given animal.

Analysis. c-Fos counts among all nine experimental groups were analyzed using a one-way ANOVA. Statistical significance was determined at P < 0.05. Experiment 1 sought to establish a relationship between dose of TNF- microinjection and the resultant number of c-Fos-labeled neurons. Experiment 2 was designed to see if coinjection of the glutamate receptor antagonists (NBQX or MK-801) would affect the NST c-Fos expression evoked by TNF-. Experiment 3 was performed to examine the effect of NBQX and MK-801 on c-Fos induction evoked by OT. Therefore, subsequent post hoc comparisons were made by selected Bonferroni tests between relevant groups. All data are presented as means ± SE.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experiment 1: c-Fos labeling in the NST induced by TNF-. Compared with results obtained in PBS-injected rats, TNF- induced a dramatic increase in c-Fos labeling when injected into the NST. Indeed, PBS injection did not appear to produce any increase in NST c-Fos labeling over what we have obtained in rats receiving no NST injections (15). An overall ANOVA revealed a significant effect of treating experimental animals with the various protocols outlined in the current study (F = 5.996, degrees of freedom = 8, P < 0.0001). Microinjection of TNF- induced a significant rise in NST c-Fos labeling in rats subjected to a dose of 3.4 pg TNF- compared with vehicle-injection control (205.1 ± 29.5 vs. 62.0 ± 10.2; selected Bonferroni's multiple comparison test, t = 4.647, P < 0.001; Figs. 1 and 2). Increasing the dose to 34 pg TNF- also yielded a significant increase in c-Fos expression in the NST above PBS (213.6 ± 25.0; t = 4.522, P < 0.001; Fig. 1). Injection of 0.34 pg TNF- produced an increase in NST c-Fos expression that was not statistically significant (124.8 ± 25.7; P > 0.05; Fig. 1). This relationship parallels the results obtained in our electrophysiological studies (9). It is of interest to note that in all experimental groups, c-Fos expression was not confined to the side of the brain stem in which injections were made (right), although labeling was always most dense on this side.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   Experiment 1: effect of tumor necrosis factor (TNF)- on c-Fos expression in the nucleus of the solitary tract (NST). Graphic representation of c-Fos-labeled NST cell counts expressed as means ± SE. A dose of 3.4 pg TNF- induced c-Fos counts that were significantly different than those obtained from animals injected with PBS (*P < 0.001, selected Bonferroni's test). Decreasing the dose 10-fold (0.34 pg) did not produce statistically significant c-Fos labeling, whereas increasing the dose 10-fold (34 pg) resulted in significant c-Fos expression in the NST (*P < 0.001, selected Bonferroni's test).

View larger version (102K): [in this window] [in a new window]   Fig. 2.   Photomicrographs of coronal histological sections through the NST at the level of the area postrema (ap). c-Fos-labeled nuclei are characterized by their dark stain, a minimal 6 µm diameter, and the presence of nucleoli. A: in PBS-injected rats (20 nl), few c-Fos-positive cells were present (dmn, dorsal motor nucleus of the vagus; st, solitary tract). B: injection of TNF- (3.4 pg) induces a significant increase in c-Fos expression. C: coinjection of TNF- with the -3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) receptor antagonist 1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-benzo[f]quinoxaline-7-sulfonamide disodium (NBQX) results in little c-Fos production. D: coinjection of TNF- with the NMDA antagonist MK-801 causes scarce c-Fos labeling. Scale bar 700 µm for A-D.

Experiment 2: effects of glutamate antagonists on NST c-Fos labeling in response to coinjections with TNF-. When TNF- (3.4 pg) was coinjected with the AMPA glutamate receptor antagonist NBQX into the brain stem, NST c-Fos levels were significantly lower than when this dose of TNF- was injected alone (79.0 ± 10.8 vs. 213.6 ± 25.0; t = 3.891, P < 0.01; Figs. 2 and 3). Coinjection of TNF- with the NMDA glutamate receptor antagonist MK-801 also caused a significant reduction in c-Fos expression in the NST compared with TNF- alone (104.8 ± 13.3 vs. 213.6 ± 25.0; t = 3.256, P < 0.05; Figs. 2 and 3).

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Experiment 2: graphic representation of the effect of glutamate antagonists on c-Fos cell counts (means ± SE) induced by TNF-. Microinjection of TNF- (3.4 pg) results in significant c-Fos labeling in the NST compared with PBS (selected Bonferroni's test, *P < 0.001). Inclusion of the AMPA antagonist NBQX along with TNF- in the injection pipette inhibited significant c-Fos production. Similarly, when the N-methyl-D-aspartate (NMDA) antagonist MK-801 was injected with TNF-, there was no significant increase in NST c-Fos expression.

Experiment 3: effects of glutamate antagonists on NST c-Fos labeling in response to coinjections with OT. Microinjection of OT (20 ng) into the medulla induced a significant rise in c-Fos-labeled cells in the NST compared with PBS-injected rats (171.5 ± 16.7 vs. 62.0 ± 10.2; t = 3.064, P < 0.05; Figs. 4 and 5). In contrast to our results in experiment 2, coinjection of OT with either NBQX (241.3 ± 55.7) or MK-801 (163.8 ± 36.7) did not result in significantly different c-Fos counts than in rats treated with OT alone (171.5 ± 16.7; P > 0.05; Figs. 4 and 5). Again, c-Fos-labeled cells were not restricted to the side of the brain stem in which injections were performed in any experimental group.

View larger version (105K): [in this window] [in a new window]   Fig. 4.   Photomicrographs of c-Fos induction by microinjection of oxytocin (OT) in the presence and absence of glutamate receptor antagonists. A: PBS microinjection (20 nl) results in scarce c-Fos expression in the NST. B: OT injection induces significant c-Fos labeling in the NST. C: coinjection of OT with the AMPA antagonist NBQX does not affect c-Fos production induced by OT. D: microinjection of OT and the NMDA antagonist MK-801 generates c-Fos expression that is not different than that produced by OT alone. Scale bar 700 µm for A-D.

View larger version (25K): [in this window] [in a new window]   Fig. 5.   Experiment 3: effect of glutamate receptor antagonists on c-Fos expression in the NST induced by OT. Each bar represents c-Fos-positive cell counts expressed as means ± SE. Microinjection of 20 ng of OT into the NST induces a significant increase in c-Fos-labeled cells compared with control (*P < 0.05, selected Bonferroni's test). Injection of either NBQX or MK-801 with OT did not significantly inhibit c-Fos production in the NST.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

These studies demonstrated that nanoinjections of TNF- into the brain stem increased the expression of the immediate early gene product c-Fos in the NST. The results also suggest that NST activation (as estimated by c-Fos production) is dependent on glutamate neurotransmission, because inclusion of the AMPA receptor antagonist NBQX or the NMDA antagonist MK-801 along with TNF- in the injection pipette suppressed the increase in c-Fos production induced by TNF- alone. Finally, glutamate does not appear to set the general background excitability for NST activation. That is, neither NBQX nor MK-801 diminished c-Fos expression evoked by OT.

The effect of concentration of TNF- on NST neurons is consistent with the results of our electrophysiological studies (9). In this study, concentrations of 10⁸ M and 10⁷ M induced excitation of identified NST neurons, whereas a concentration of 10⁹ M TNF- failed to activate these cells (9). In the current study, the same concentrations of TNF- (10⁸ M and 10⁷ M) induced significant production of c-Fos in the NST, whereas the lowest dose (10⁹ M) produced a modest but nonstatistically significant increase in c-Fos expression. Although labeling was most dense on the side of the injection, c-Fos-positive cells were present on both sides of the brain stem. This is not surprising given the intrinsic projections between solitary nuclei (45) as well as the possibility of drug diffusion.

The data presented in this study reveal that both AMPA and NMDA receptors probably play a role in TNF- activation of NST neurons, because antagonists for both receptors block c-Fos activation in the NST in response to TNF-. An earlier study by Wan et al. (44) described how MK-801 inhibited c-Fos production in the NST in response to peripheral injections of LPS. However, it is still unknown how TNF- modulates glutamatergic transmission. It is possible that TNF- causes an increase in glutamate release presynaptically. Using dialysis techniques, Mascarucci et al. (25) demonstrated short-term enhancement (within 60 min) of glutamate release in the NST in response to intraperitoneal injections of LPS or interleukin-1. Glutamate release in the LPS preparation was simultaneous with elevated plasma levels of TNF- (25). Glutamate levels remain elevated in the NST for up to 3 h after intravenous LPS injection (22).

TNF- may also modulate the postsynaptic action of glutamate at the level of the NST. This type of relationship exists with corticotropin-releasing factor (CRF) in the cerebellum. Here, CRF amplifies both spontaneous and glutamate-induced single-unit activity of Purkinje cells (4, 5).

Other investigators have shown that NMDA and AMPA receptors in the DVC play a role in gastrointestinal function. Sivarao et al. (38) demonstrated that both receptors are involved in modulating intragastric pressure and motility. NMDA receptors play a role in satiety because direct injection of MK-801 into the NST (as well as fourth ventricular infusion) delays satiety (43, 47). One of these studies presented conflicting data on c-Fos expression in response to gastric distension. Although distension produced significant increases in c-Fos production in the medial NST, fourth ventricular infusion of MK-801 did not decrease this protein expression (47). In addition, injection of MK-801 increased c-Fos labeling in the NST in the absence of gastric distension (47). One may expect that MK-801 would have the opposite effect on c-Fos production in this experimental paradigm. Perhaps ventricular infusion of MK-801 causes inhibition of glutamatergic transmission elsewhere in the CNS, thus removing the NST from tonic inhibitory influences. This disinhibition would, in turn, lead to an excitation of NST neurons, thus an increase in c-Fos-positive cells. In the current study, MK-801 inhibited TNF- induced c-Fos production in the NST and not OT-induced c-Fos production. Perhaps this is because local injections of the antagonist do not allow for more widespread inhibition of glutamate neurotransmission.

The results presented here suggest that the relationship between TNF- and glutamate may be specific. That is, blocking of glutamatergic transmission in the NST blunted TNF- activation of NST neurons, while having no effect on OT activation of NST neurons. It has been demonstrated that OTergic neurons are present in the DVC and that there are receptors specific to OT in this region (1, 32). Other work has shown that OT directly excites NST neurons in vitro and in vivo (26, 30). The current study corroborates these findings by showing that injection of OT into the NST increases c-Fos expression. This OT-induced c-Fos activation was not dependent on glutamate receptors, because inclusion of either NBQX or MK-801 with OT did not decrease the number of c-Fos-positive cells in the NST.

Our previous study revealed that TNF- directly excites NST neurons that are responsive to gastric distension (9). Furthermore, these NST neurons that were preexposed to TNF- exhibited a potentiated response to afferent stimulation. The current study shows that NMDA (and AMPA) receptors are involved in TNF- activation of NST neurons. Perhaps the facilitation recorded in the neurophysiological study is similar to other known mechanisms of synaptic plasticity, e.g., long-term potentiation (LTP). NMDA receptors are required for LTP generation in the CA1 region of the hippocampus (the area in which LTP is most often studied), because MK-801 and other NMDA antagonists inhibit LTP (6). In this system, the AMPA receptor is initially responsible for depolarization of the cell. Once a threshold voltage is reached, the NMDA receptor is removed from inhibition and calcium is allowed to enter the cell, which is required for LTP induction (24). In our neurophysiological preparation, AMPA receptor activation was probably involved in the fast response to TNF-, whereas NMDA receptors were probably involved in sustained activation, and perhaps potentiation. Additional experiments are required to investigate these hypotheses further.

Perspectives

The NST sends long ascending connections with forebrain structures that are involved in the integration of autonomic, neuroendocrine, and behavioral responses to specific stimuli, such as the central nucleus of the amygdala (37). The amygdaloid nucleus, in turn, projects to the insular cortex, a temporal lobe area that has been implicated in conditioned taste aversion (CTA) behavior (3, 20). CTA is a model of learning and memory in which an animal acquires aversion to a novel taste when it is followed by digestive malaise (10). NMDA receptors are required for CTA production as well as LTP induction in the insular cortex (10). More recently, induction of LTP in amygdaloid projections to the insular cortex was directly linked to CTA behavior (11).

The NST has indeed been linked to CTA in several experimental paradigms. McCaughey et al. (28) showed that a burst of activity occurs in the NST in response to presentation of the conditioned stimulus to CTA conditioned rats. Houpt et al. (18) demonstrated c-Fos induction in the NST after CTA acquisition, and protein expression was not dependent on the integrity of the vagus nerve(s). Our data suggest a mechanism of enhanced responsiveness to afferent input after exposure to TNF-. Potentiation in the NST may be translated to the amygdala, which is then transmitted to the insular cortex for LTP induction and subsequent CTA acquisition. Such plasticity exhibited at multiple levels in the CNS would efficiently establish visceral aversion behavior in response to infection.