A direct hepatic osmoreceptive afferent projection from nucleus tractus solitarius to dorsal hypothalamus

Thirty units that responded antidromically to electrical stimulation of the zona incerta (ZI) or the dorsal portion of hypothalamus were recorded in the nucleus tractus solitarii (NTS). These units were analyzed in relation to hepatoportal afferent inputs. Electrical stimulation of the hepatic branch of the vagus nerve facilitated six units (facilitatory units) and suppressed 10 units (suppressed units). Effect of the portal infusion of hypertonic saline was examined on six facilitatory and eight suppressed units. One facilitatory unit and one suppressed unit increased their discharge rates in response to portal infusion. Four facilitatory units and one suppressed unit decreased their discharge rates in response to the same stimulation. Increased or decreased discharge rates in response to portal infusion of hypertonic saline were observed in units that responded antidromically to electrical stimulation of the paraventricular hypothalamic nucleus, the lateral hypothalamic area, or the ZI. It is concluded that hepatoportal osmoreceptive signals are conveyed to the hypothalamus or the ZI directly from the NTS.     

Membrane and synaptic properties of nucleus tractus solitarius neurons projecting to the caudal ventrolateral medulla

Projections from the nucleus of tractus solitarius (NTS) to the caudal ventrolateral medulla (CVLM) are important in mediating autonomic reflexes. However, little is known about the cellular properties of the CVLM-projecting NTS neurons. In this study, the CVLM-projecting NTS neurons were retrogradely labeled by fluorescent microspheres injected into the CVLM. Whole cell voltage- and current-clamp recordings were performed on labeled NTS neurons in coronal brainstem slices. Compared with unlabeled neurons, the labeled NTS neurons had more depolarized resting membrane potentials, larger input resistance, and higher firing activity in response to depolarizing currents. Bath application of an ionotropic glutamate receptor antagonist kynurenic acid and a non-NMDA receptor antagonist CNQX significantly decreased the firing activity in the majority of labeled NTS neurons. In contrast, an NMDA receptor antagonist AP5 failed to alter the firing activity in labeled neurons tested. While the glycine receptor antagonist strychnine had no effect on the firing activity, blockade of GABA(A)receptors with bicuculline significantly increased the firing rate in the majority of labeled NTS neurons. Furthermore, CNQX blocked the majority of spontaneous excitatory postsynaptic currents (EPSCs) and evoked EPSCs elicited by stimulation of the tractus solitarius. The residual spontaneous and evoked EPSCs were abolished by the nicotinic receptor antagonist mecamylamine and the purinergic P2X receptor antagonist iso-PPADS. Finally, while bicuculline completely blocked the miniature inhibitory postsynaptic currents (IPSCs), the spontaneous and evoked IPSCs were abolished by a combination of bicuculline and strychnine in labeled NTS neurons. Collectively, these data suggest that the CVLM-projecting neurons are a population of neurons with distinctive membrane properties.     

Effect of nucleus tractus solitarius lesions on cardiovascular responses elicited from the caudal ventrolateral medulla.

Neurons in the caudal ventrolateral medulla (CVLM) which inhibit sympathetic vasomotor tone may have reciprocal connections with the nucleus tractus solitarius (nTS). This study determined whether changes in arterial pressure elicited by chemical excitation or inhibition of neurons in the vasodepressor region of the rabbit CVLM depend on the integrity of the nTS. Unilateral injections of L-glutamate (10 pmol to 100 nmol), or bilateral injections of GABA (1 nmol to 125 nmol), were made into the CVLM, and dose-response effects on arterial pressure determined. The nTS was then bilaterally cauterized, or inhibited by local injections of muscimol, and the dose-response curves were repeated. Neither cauterization nor injection of muscimol significantly altered the slope of the log dose-response curves for L-glutamate, but nTS muscimol increased the fall in arterial pressure for each dose of L-glutamate (P less than 0.01). Cauterization of the nTS significantly (P less than 0.01) increased the slope of the curve relating dose of GABA to rise in arterial pressure observed, after injection of GABA into the CVLM. This increase in slope was similar to the increase observed when GABA is injected into the CVLM in baroreceptor-denervated rabbits. We conclude that neither the depressor nor the pressor response evoked by stimulation or inhibition of the CVLM is dependent on the integrity of the nTS. Inactivation of the nTS tends to increase the magnitude of the CVLM responses, possibly by removal of baroreceptor-mediated buffering of the responses.     

Projection of nucleus tractus solitarius units influenced by hepatoportal afferent signal to parabrachial nucleus

In urethan-chloralose anesthetized rats, units in the nucleus tractus solitarius (NTS) which antidromically responded to electrical stimulation of the parabrachial nucleus (PB) were investigated for their responses to hepatic-vagal afferent signals. Among 63 NTS units examined, 25 (40%) were excited, 17 (27%) inhibited and the remaining 21 (33%) unaffected by single shock electrical stimulation of the hepatic branch of the vagus nerve. Topical application of Na+ produced an increase in discharge rate of 9 units and a decrease in 5 units. Portal infusion of hypertonic saline produced an increase in discharge rate of 3 units and a decrease in 5 units. Furthermore, 3 units responded to both topical application of Na+ and portal infusion of hypertonic saline.     

Stimulation of the ventrolateral medulla inhibits the baroreceptor input to the nucleus tractus solitarius

Extracellular recording experiments were done in urethane- and chloralose-anaesthetized, paralyzed and artificially ventilated rabbits. Thirty-one baroreceptor-sensitive neurones were identified in the region of the nucleus tractus solitarius (NTS) by their excitatory responses to stimulation of the ipsilateral aortic nerve. A conditioning stimulus delivered to the ipsilateral or contralateral rostral ventrolateral medulla (rVLM) inhibited the excitatory responses of 80.0% (19 out of 24) neurones to an aortic test stimulus as early as 3 ms and extending as long as 400 ms after conditioning. The same inhibitory effect was also observed by application of excitant amino acid DL-homocysteic acid (0.2 M, 100 nl) into the ipsi- or contralateral rVLM area in 6 units. In 5 units inhibited by rVLM stimulation, evoked discharges were inhibited by prolonged electrical stimulation of the superficial peroneal nerve (SP) with low intensity and low frequency (0.1-0.3 mA and 5-10 Hz). These results provide the electrophysiological evidence for the suppressing effect of the rVLM on the excitatory responses of NTS neurones to baroreceptor afferent stimulation.     

Hypoxia-induced cellular and vascular changes in the nucleus tractus solitarius and ventrolateral medulla

Major changes in arterial pressure, autonomic, and respiratory activity occur in response to hypoxia. We analyzed structural damage and increased vascular permeability in the ventrolateral medulla and nucleus tractus solitarius, which control autonomic, respiratory, and cardiovascular functions in adult Wistar rats subjected to 2 hours of hypoxia (7% oxygen + 93% nitrogen) for up to 14 days after hypoxicexposure. Brainstem tissue levels of vascular endothelial growth factor (VEGF), nitric oxide (NO), and glutamate were significantly increased over control levels after hypoxic injury. By electron microscopy, swollen neurons and dendrites, degenerating axons, disrupted myelin sheaths, and swollen astrocyte processes were observed in the nucleus tractus solitarius and ventrolateral medulla. Leakage of intravenously administered horseradish peroxidase was observed through vascular walls in hypoxic rats. These results suggest that increased VEGF and NO production in hypoxia resulted in increased vascular permeability, which, along with increased levels of glutamate, may have induced structural alterations of the neurons, dendrites, and axons. Administration of the antioxidant neurohormone melatonin (10mg/kg) before and after the hypoxia reduced VEGF, NO, and glutamate levels and improved ultrastructural abnormalities induced by hypoxia exposure, suggesting that it may have a therapeutic potential in reducing hypoxia-associated brainstem damage.     

Projections from the nucleus tractus solitarii to the rostral ventrolateral medulla

Projections from the nucleus tractus solitarii (NTS) to autonomic control regions of the ventrolateral medulla, particularly the nucleus reticularis rostroventrolateralis (RVL), which serves as a tonic vasomotor center, were analyzed in rat by anterograde, retrograde, and combined axonal transport techniques. Autonomic portions of the NTS, including its commissural, dorsal, intermediate, interstitial, ventral, and ventrolateral subnuclei directly project to RVL as well as to other regions of the ventrolateral medulla. The projections are organized topographically. Rostrally, a small cluster of neurons in the intermediate third of NTS, the subnucleus centralis, and neurons in proximity to the solitary tract selectively innervate neurons in the retrofacial nucleus and nucleus ambiguus. Neurons generally located in more caudal and lateral sites in the NTS innervate the caudal ventrolateral medulla (CVL). The RVL, CVL, and nucleus retroambiguus are interconnected. A combined retrograde and anterograde transport technique was developed so as to prove that projections from the NTS to the ventrolateral medulla specifically innervate the region of RVL containing neurons projecting to the thoracic spinal cord or the region of the nucleus containing vagal preganglionic neurons. When the retrograde tracer, fast blue, was injected into the thoracic spinal cord, and wheat germ agglutinin-conjugate horseradish peroxidase (HRP) was injected into the NTS, anterogradely labeled terminals from the NTS surrounded the retrogradely labeled neurons in the RVL and in the nucleus retroambiguus in the caudal medulla. Among the bulbospinal neurons in the RVL innervated by the NTS were adrenaline-synthesizing neurons of the C1 group. When fast blue was applied to the cervical vagus, and HRP was injected into the NTS, anterogradely labeled terminals from the NTS surrounded retrogradely labeled neurons in the rostral dorsal motor nucleus of the vagus, the region of the nucleus ambiguus, the retrofacial nucleus, and the dorsal portion of the RVL, a region previously shown to contain cardiac vagal preganglionic neurons. This combined anterograde and retrograde transport technique provides a useful method for tracing disynaptic connections in the brain. These data suggest that the RVL is part of a complex of visceral output regions in the ventrolateral medulla, all of which receive afferent projections from autonomic portions of the NTS. Bulbospinal neurons in the RVL, in particular the C1 adrenaline neurons, may provide a portion of the anatomic substrate of the baroreceptor and other visceral reflexes.     

Monosynaptic projections from the nucleus tractus solitarii to C1 adrenergic neurons in the rostral ventrolateral medulla: Comparison with input from the caudal ventrolateral medulla

The rostral ventrolateral medulla (RVL) contains reticulospinal adrenergic (C1) neurons that are thought to be sympathoexcitatory and that form the medullary efferent limb of the baroreceptor reflex pathway. The RVL receives direct projections from two important autonomic regions, the caudal ventrolateral medulla (CVL) and the nucleus tractus solitarii (NTS). In the present study, we used anterograde tracing from the CVL or the NTS combined with immunocytochemical identification of C1 adrenergic neurons in the RVL to compare the morphology of afferent input from these two autonomic regions into the RVL. NTS (n = 203) and CVL (n = 380) efferent terminals had similar morphology and vesicular content, but CVL efferent terminals were slightly larger than NTS efferent terminals. Overall, efferent terminals from either region were equally likely to contact adrenergic neurons in the RVL (21% for NTS, 25% for CVL). Although efferents from both regions formed both symmetric and asymmetric synapses, NTS efferent terminals were statistically more likely to form asymmetric synapses than CVL efferent terminals. CVL efferent terminals were more likely to contact adrenergic somata than were NTS efferents, which usually contacted dendrites. These findings 1) support the hypothesis that a portion of NTS efferents to the RVL may be involved in sympathoexcitatory, e.g., chemoreceptor, reflexes (via asymmetric synapses), whereas those from the CVL mediate sympathoinhibition (via symmetric synapses); and 2) provide an anatomical substrate for differential postsynaptic modulation of C1 neurons by projections from the NTS and CVL. With their more frequent somatic localization, CVL inhibitory inputs may be more influential than excitatory NTS inputs in determining the discharge of RVL neurons. © 1996 Wiley-Liss, Inc.     

Synaptic contacts of substance P-immunoreactive axon terminals in the nucleus tractus solitarius onto neurons projecting to the caudal ventrolateral medulla oblongata in the rat

The possibility that substance P (SP)-immunoreactive axon terminals in the nucleus tractus solitarius (NTS) make synaptic contacts onto NTS neurons projecting to the catecholaminergic cell region in the caudal ventrolateral medulla oblongata (CVLM) was examined in the rat using a retrograde tract-tracing method combined with immunohistochemistry. After injection of a retrograde tracer, wheat germ agglutinin-conjugated horseradish peroxidase-colloidal gold complex (WGA-HRP-gold), into the CVLM region where tyrosine hydroxylase-immunoreactive neurons were situated, many retrogradely labeled neurons were detected in the dorsal parts of the NTS, especially at levels between 1.0 mm caudal and 0.5 mm rostral to the obex. Immunoelectron microscopy revealed synaptic contacts between SP-immunoreactive axon terminals and WGA-HRP-gold-labeled neurons in the NTS. These findings indicated that SP regulates NTS neurons which project to the catecholaminergic cell region of the CVLM. ©1997 Elsevier Science B.V. All rights reserved.     

Sonic hedgehog signalling in neurons of adult ventrolateral nucleus tractus solitarius

The transmembrane receptor Patched (Ptc) mediates the action of the diffusing factor Sonic hedgehog (Shh), which is implicated in establishing morphogenetic gradients during embryonic development. Whereas alteration of Ptc function is associated with developmental abnormalities and brain tumors, its functional activity and roles in the adult brain have yet to be elucidated. Here we describe the complementary pattern of Shh and Ptc expression in the rat dorsal vagal motor nucleus and the ventrolateral nucleus tractus solitarius (vNTS), respectively. Those two interconnected structures regulate the cardiorespiratory function during hypoxia. Bath application of a subnanomolar concentration of aminoterminal Shh protein (ShhN) to a slice preparation of the vNTS induces a rapid decrease in neuronal firing followed by a bursting activity that propagates in the neuronal network. Intracellular current injections show that bursts result from an action on the neuronal membrane electro-responsiveness. Both inhibiting and bursting effects are blocked by the monoclonal Shh antibody 5E1 and may require the Ptc binding site of ShhN. Thus, ShhN acting on specific neuronal sites controls electrophysiological properties of differentiated neurons of the vNTS. We speculate on a retrocontrol of cardiorespiratory signals in the vNTS, by Shh generated in dorsal vagal motoneurons.     

Reciprocal connection between nucleus ambiguus and caudal ventrolateral medulla

To investigate interactions between parasympathetic preganglionic neurons in the nucleus ambiguus (NA) and sympathoinhibitory neurons in the caudal ventrolateral medulla (CVLM) the activity of CVLM neurons was recorded during glutamate stimulation of cardioinhibitory sites in the NA of urethane-anesthetized rats. Neurons in the CVLM were identified as cardiovascular neurons if they increased their activity after i.v. phenylephrine and displayed cardiac cycle-related rhythmicity. Of 23 cardiovascular neurons studied, 10 decreased activity after glutamate (GLU) microinjection in the NA, five neurons were excited and eight did not respond. In another series of experiments, the nature of the influence of the CVLM on unit activity in the NA was investigated. Sites in the CVLM from which decreases in arterial pressure and heart rate were elicited after GLU microinjection were identified and the activity of cardiovascular neurons in the NA was recorded. Of 22 NA cardiovascular neurons studied, eight decreased activity after microinjection of GLU in the CVLM and 14 did not change firing frequency. These results demonstrate the existence of a reciprocal pathway between the NA and CVLM and provide evidence for functional interactions between medullary sites implicated in the control of the parasympathetic and sympathetic nervous systems.     

Reciprocal connections between nucleus tractus solitarii and rostral ventrolateral medulla

Experiments were done to test the hypothesis that there are reciprocal connections between the nucleus of the tractus solitarius (NTS) and the rostral ventrolateral medulla (RVLM). Spontaneous activity was recorded from units in the right RVLM or NTS of urethan-anesthetized and artificially ventilated rats. Twenty-four of 42 RVLM and 12 of 21 NTS units were classified as cardiovascular because they were inhibited by baroreceptor activation and displayed a cardiac rhythm. Electrical stimulation of depressor sites in the NTS inhibited 14 and excited 10 RVLM units. Stimulation of pressor sites in the RVLM excited 10 and inhibited 2 NTS units. None of the units in the NTS or in the EVLM could be activated antidromically. These results provide evidence that there are reciprocal excitatory and inhibitory connections between NTS and RVLM and that these connections are not monosynaptic.     

Respiratory modulation of afferent terminal excitability in the nucleus tractus solitarius

The excitability of vagal and carotid sinus nerve afferent terminals within the nucleus of the tractus solitarius has been studied in cats and rabbits using the technique of antidromic activation. Although the excitability of superior laryngeal nerve afferent terminals was observed to fluctuate in phase with the central respiratory cycle, no equivalent variations in sinus nerve terminal excitability were observed. Whilst recording from single identified vagal afferents, no respiratory-related changes in the excitability of baroreceptor afferent terminals could be demonstrated, though such modifications in terminal excitability were readily seen in lung stretch afferent terminals.     

Immunocytochemical localization of glutamate containing neurons in the ventrolateral medulla oblongata and the nucleus tractus solitarius of the rat.

Using antisera raised against excitatory neurotransmitter glutamate fixed to carrier proteins and purified by affinity chromatography, localization of glutamate in the ventrolateral medulla oblongata and the nucleus tractus solitarius of the rat was studied. Within the rostral portion of the ventrolateral medulla, glutamate-like immunoreactive neurons were found in the lateral wing of the raphe magnus and in the region of the paragigantocellular reticular nucleus. In the caudal portion of the ventrolateral medulla, glutamate-positive neurons were clustered within the nucleus reticularis lateralis and a lesser number of stained neurons were scattered in the region around this nucleus. Glutamate-immunoreactive neurons were found in the nucleus tractus solitarius at the level of the area postrema. They were densely distributed especially in the medial subdivision of the nucleus whereas other subdivisions contained considerable glutamate positive cells. These results provide immunohistochemical evidence for the presence of glutamate-containing neurons in the ventrolateral medulla oblongata and the nucleus tractus solitarius of the rat.     

The ventrolateral medulla as a source of synaptic drive to rhythmically firing neurons in the cardiovascular nucleus tractus solitarius of the rat

We sought to determine whether the caudal ventrolateral medulla (cVLM), at the level of area postrema, influences the rhythmically beating neurons found within the dorsomedial NTS in rat brainstem slices. Intra- or extracellular recordings of neurons firing rhythmically at around 5 Hz were characterized as either auto-active (i.e. pacemaker; AA) or synaptically driven (SD) by pharmacological interventions. The nature of inputs evoked from the ipsilateral cVLM were orthodromic and the majority were excitatory (latency 3-20 ms). Further, this excitatory influence was found to be tonically active in 25/47 cells studied since inactivating the ipsilateral cVLM by localized cooling reduced the firing rate by 0.5-3.0 Hz (23% on average). Neuronal characterization showed that the most consistent and pronounced effect occurred on SD rather than AA cells. Control experiments that cooled other areas of the slice closer to the recording site proved ineffective. Additional studies showed that most rhythmically firing cells in the NTS received an excitatory synaptic input from the solitary tract (ts; latency 3-30 ms). This input was reduced or blocked by inactivating the cVLM in neurons in which the ts latency of activation was greater than 8 ms in half of the neurons tested. Subsequent pharmacological tests revealed that these neurons were predominantly SD. Identified AA neurons received an input from the ts at a shorter latency, typically less than 8 ms, and this was unperturbed by cooling the cVLM in all cases. Further, there was no obvious difference in the baseline discharge rates between cells in the hemi-slice and those recorded in an intact slice. In a hemi-coronal slice cooling the cVLM also produced a 20% decrease in firing rate in identified SD neurons but no consistent change in AA cells. We conclude that (1) the ipsilateral cVLM contributes principally tonic excitatory drive to rhythmically active neurons in the dorsomedial NTS in vitro and this preferentially effects SD neurons; (2) other excitatory drives other than those from the ipsilateral cVLM impinge upon SD cells, the origin of which are relatively local and likely to be in the NTS; (3) in the slice the projection from the cVLM to the NTS appears to be present but the reciprocal connection is absent.     

Catecholamine inputs to the nucleus tractus solitarius

The projections of brainstem catecholaminergic (CA) cell groups to the rat nucleus tractus solitarius (NTS) were examined using 6-hydroxydopamine (6-OHDA) injections and glass microknife cuts. 6-OHDA (4 micrograms) was injected into the intermediate NTS, and this resulted in depletion of CA fluorescent varicosities from the NTS at this rostrocaudal level, except for varicosities along the periventricular edge of the NTS. In addition, a band of swollen fluorescent axons extended between the CA A1 cell group of the ventrolateral medulla and the lateral NTS. Microknife cuts were used to interrupt the projections of the CA A1 and A2 cell groups (located in the caudal NTS) and tissues were examined for changes in CA varicosity density within the intermediate NTS. Following transverse knife cuts of the intermediate NTS, rostral to the A2 cell group, fluorescent varicosities rostral to the cut virtually disappeared, and the fluorescence intensity of the ipsilateral A2 neurons caudal to the cut was increased. These cuts also eliminated the 6-OHDA-resistant varicosities along the periventricular NTS. After microknife cuts lateral to the intermediate NTS, the fluorescent varicosity density in the NTS was unchanged. These results indicate that the major CA projection to the NTS arises from the ipsilateral A2 cell group. The 6-OHDA-resistant varicosities arising from neurons caudal to the knife cut probably arise from the adrenergic C2 cell group.     

Electrophysiological and morphological characteristics of nucleus tractus solitarii neurons projecting to the ventrolateral medulla

Electrophysiological and morphological properties of a direct projection from the nucleus of the tractus solitarius (NTS) to the ventrolateral medulla (VLM) were investigated. NTS neurons projecting to the VLM exhibit a monosynaptic excitatory response followed by an inhibitory one after the tractus solitarius stimulation. These neurons show spontaneous inhibitory postsynaptic currents, and have medium to large soma (14-26 microm in diameter). It is concluded that the projection from the NTS to the VLM is mediated mostly by medium to large neurons that are inhibited locally by GABAergic interneurons within the NTS.     

Nitric oxide inhibits excitatory vagal afferent input to nucleus tractus solitarius neurons in anaesthetized rats

Objective

Endogenous nitric oxide (NO) has been implicated in the regulation of neuronal activity which mediates cardiovascular reflexes. However, there is controversy concerning the role of NO in the nucleus tractus solitarius (NTS). The present study aims to elucidate the possible physiological role of endogenous NO in modulating the excitatory vagal afferent input to NTS neurons.

Methods

All the experiments in the rat were conducted under anaesthetic conditions. Ionophoresis method was used for the application of NO donor or nitric oxide synthase (NOS) inhibitor, and single unit recording method was employed to detect the effects of these applications on vagal afferent- or cardio-pulmonary C-fibre reflex-evoked neuronal excitation in NTS.

Results

Ionophoresis applications of L-arginine (L-Arg), a substrate of NOS, and sodium nitroprusside (SNP), a NO donor, both attenuated the vagal afferent-evoked discharge by (51.5±7.6)% (n = 17) and (68.3±7.1)% (n = 9), respectively. In contrast, application of D-Arg at the same current exerted no overall effect on this input. Also, both L-Arg and SNP inhibited spontaneous firing of most of the recorded neurons. In contrast, ionophoresis application of NG-nitro-Larginine methyl ester (L-NAME) enhanced vagal afferent-evoked excitation by (66.3±11.4)% (n = 7). In addition, ionophoresis application of L-Arg and SNP significantly attenuated cardio-pulmonary C-fibre reflex-induced excitation in the tested NTS neurons.

Conclusion

Activation of local NO pathway in the NTS could suppress vagal afferent-evoked excitation, suggesting that NO is an important neuromodulator of visceral sensory input in the NTS.     

Direct catecholaminergic projection from nucleus tractus solitarii to supraoptic nucleus

To determine whether the supraoptic nucleus (SON) receives a direct projection from catecholamine cells of the nucleus tractus solitarii (NTS), retrograde transport of rhodamine-tagged latex microspheres was combined with a procedure for the fluorescence histochemical visualization of catecholamines. SON tracer injections, made via transpharyngeal approach, retrogradely labelled cells at all levels of NTS, although the majority were located caudal to obex with an ipsilateral predominance. Approximately half of these cells were also identified as catecholaminergic; the relatively caudal level in the dorsomedial medulla of most of these cells suggests that they probably correspond to the A2 catecholamine cell group.     

Adrenergic projection from the caudal part of the nucleus of the tractus solitarius to the parabrachial nucleus in the rat: immunocytochemical study combined with a retrograde tracing method

The presence of an adrenergic projection from the nucleus of the tractus solitarius (NTS) to the parabrachial nucleus (PB) was demonstrated by the immunocytochemistry combined with a retrograde tracing method. Numerous neurons containing both phenylethanolamine N-methyltransferase, a marker for adrenaline, and wheat germ agglutinin-conjugated horseradish peroxidase, a retrograde tracer, were detected in the dorsolateral part of the NTS at the level of the area postrema after injection of the tracer into the dorsal PB.