Corresponding author B. N. Smith: Department of Anatomy and Neurobiology, Colorado State University, Fort Collins, CO 80523, USA. Email: firstname.lastname@example.org
1Whole-cell voltage-clamp recordings in an in vitro brainstem-cranial nerve explant preparation were used to assess the local circuitry activated by vagal input to nucleus tractus solitarii (NTS) neurones in immature rats.
2All neurones that responded to vagal stimulation displayed EPSCs of relatively constant latency. Approximately 50 % of these also demonstrated variable-latency IPSCs, and ∼31 % also displayed variable-latency EPSCs to vagal stimulation. All neurones also had spontaneous EPSCs and IPSCs.
3Evoked and spontaneous EPSCs reversed near 0 mV and were blocked by the glutamate AMPA/kainate receptor antagonists 6,7-nitroquinoxaline-2,3-dione (DNQX) or 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) at rest. Evoked EPSCs had rapid rise times (< 1 s) and decayed monoexponentially (τ= 2.04 ± 0.03 ms) at potentials near rest.
4At holding potentials positive to ∼−50 mV, a slow EPSC could be evoked in the presence of DNQX or CNQX. This current peaked at holding potentials near −25 mV and was blocked by the NMDA receptor antagonist dl-2-amino-5-phosphonovaleric acid (AP5). It was therefore probably due to activation of NMDA receptors by vagal afferent fibres.
5Fast IPSCs reversed near −70 mV and were blocked by the GABAA receptor antagonist bicuculline. In addition, bicuculline enhanced excitatory responses to vagal stimulation and increased spontaneous EPSC frequency. Antagonists to AMPA/kainate receptors reversibly blocked stimulus-associated IPSCs and also decreased the frequency of spontaneous IPSCs.
6These findings suggest that glutamate mediates synaptic transmission from the vagus nerve to neurones in the immature NTS by acting at non-NMDA and NMDA receptors. NTS neurones may also receive glutamatergic and GABAergic synaptic input from local neurones that can be activated by vagal input and/or regulated by amino acid inputs from other brainstem neurones.1. Whole-cell voltage-clamp recordings in an in vitro brainstem-cranial nerve explant preparation were used to assess the local circuitry activated by vagal input to nucleus tractus solitarii (NTS) neurones in immature rats.
The brainstem nucleus tractus solitarii (NTS) is the principal site of synaptic contact for visceral afferent fibres of the vagus nerve. Input to the NTS is arranged viscerotopically, with fibres carrying gustatory information terminating primarily in the rostral NTS and respiratory, cardiovascular and gastrointestinal afferents terminating primarily in the intermediate and caudal NTS regions. Central control of visceral function is generally thought to involve both reflex activation of local circuitry in the NTS and reciprocal connections with other brain areas implicated in visceral system regulation. The latter connections suggest a role for the NTS in relaying visceral information to other brain regions. The resulting integrated information is eventually transmitted by neurones in the dorsal motor nucleus of the vagus (DMNX) to their target organs.
In addition to local circuit reflexes and relay responses, initial processing of vagally mediated visceral afferent information may occur within the NTS prior to further integration via more rostral brain areas. Electrophysiological and morphological studies in preparations from adult mammals have suggested that neurones within the nucleus may be in communication with each other and with motor neurones in the DMNX (Champagnat et al. 1986; Rogers & McCann, 1993; Kawai & Senba, 1996). Application of amino acid receptor agonists and antagonists directly into the caudal NTS has been shown to rapidly alter visceral functions (Zhang & Mifflin, 1993; Ohta & Talman, 1994). In neonatal animals, where neuronal connections between the NTS and more rostral regions may not be fully developed (Khachaturian & Sladek, 1980; Sawchenko & Swanson, 1982; Rinaman et al. 1994), vagally mediated autonomic functions appear essentially intact (Phifer et al. 1986; Robinson et al. 1988). Therefore, neuronal circuits within the NTS probably serve to integrate and regulate vagally mediated visceral information. Despite its likely importance in the central processing of visceral afferent information, little is known about the local synaptic organization of this region, especially in immature animals, where such connections may be sufficient to co-ordinate autonomic function.
Brain slices are more amenable than intact animals for studying synaptic activity at high resolution with whole-cell patch-clamp recordings (Blanton et al. 1989). Although slices retain many of the local synaptic contacts present in intact animals, many connections are removed or cut during tissue preparation. We therefore used whole-cell patch-clamp recordings in a modified brainstem-cranial nerve explant preparation (Smith & Feldman, 1987; Barber et al. 1995) from immature rats in which the connections of the NTS with the proximal vagus nerve could be maintained in vitro. This preparation allowed stimulation of primary afferent fibres in isolation from possible simultaneous direct stimulation of local interneurones and also allowed us to assess synaptic activity in a relatively intact system. We tested the hypotheses that: (1) primary afferent information carried by the vagus nerve is glutamatergic, (2) vagally activated synaptic circuitry in the NTS is intact early in life, and (3) local circuits in the NTS regulate evoked and spontaneous synaptic activity within the nucleus. Our results suggest that the synaptic organization of the immature NTS is more complex than has been previously suggested, even for adult rats. This complexity may be due in part to the age of the animals, but may also be a reflection of the more intact circuitry of the three-dimensional preparation we used. Part of this study has appeared in abstract form (Smith et al. 1996).
Animals and tissue preparation
Whole-cell recordings were conducted using a brainstem preparation from immature (2–5 days old) Sprague-Dawley (Harlan, Indianapolis, IN, USA) rats. Animals were housed in a vivarium under a normal 12 h light-12 h dark cycle under the care of full-time veterinary staff. All procedures used in the study adhered to guidelines approved by the Colorado State University Animal Care and Use Committee. Rats were deeply anaesthetized by halothane inhalation and killed by decapitation while anaesthetized. The head was rapidly transected from behind the eye sockets to the lower cervical spinal column. The tissue was immediately transferred to a Sylgard-coated dissecting dish and immersed in ice-cold (0–4°C), oxygenated (95 % O2-5 % CO2) artificial cerebrospinal fluid (ACSF) containing (mm): 124 NaCl, 3 KCl, 26 NaHCO3, 1.4 NaH2PO4, 11 glucose, 1.3 CaCl2 and 8–10 MgCl2 (pH 7.3–7.4), with an osmolality of 290–315 mosmol kg−1. After a craniotomy, the bone tissue was pinned to the bottom of the Sylgard-coated Petri dish, and both cerebral hemispheres and cerebellum were removed under a dissection microscope. Iris scissors were then used to remove the remaining bone over the brainstem and proximal spinal cord. The jugular foramen was then carefully dissected away, exposing the tenth cranial nerve. The brainstem and vagus nerve proximal to the nodose ganglion were removed from the skull and transferred to cold ACSF containing Evans Blue for 30–60 s to improve the surface contrast of the tissue, and then returned to the cold ACSF. The meninges were carefully removed from the dorsal surface of the explant.
The explant was then transferred to a ramp-type recording chamber, modified to allow the perfusion of warmed (32–35°C) ACSF over the top of the tissue. The ACSF used for recordings was identical to that used in the dissection, except that 1.3 mm MgCl2 was used. Added to the bath solution for some specific experiments were the GABAA antagonist bicuculline methiodide (30 μm; Sigma), the glutamate NMDA receptor antagonist dl-2-amino-5-phosphonovaleric acid (AP5; 20–50 μm; Sigma), and the glutamate AMPA/ kainate receptor antagonists 6,7-nitroquinoxaline-2,3-dione (DNQX; 10–50 μm; Sigma) or 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX; 10 μm).
After an equilibration period of 1–2 h, whole-cell current recordings were obtained in the region of the dorsomedial NTS using patch pipettes with open resistances of 2–5 MΩ. Seal resistances were typically 1–4 GΩ and series resistances were 4–19 MΩ, uncompensated. Patch pipettes were filled with (mm): 130 potassium gluconate, 1 NaCl, 5 EGTA, 10 Hepes, 1 MgCl2, 1 CaCl2, 3 KOH, 4 ATP (pH 7.2). In some experiments, Cs+ was used instead of K+ as the primary cation in the electrode in order to block voltage-dependent K+ conductances (Hestrin et al. 1990) and thus allow better resolution of IPSCs at holding potentials positive to the expected Cl− equilibrium potential. Pipettes were pulled from borosilicate glass capillaries of 1.65 mm outer diameter and 0.45 mm wall thickness (Garner Glass Co., Claremont, CA, USA). Electrical stimulation of the vagus nerve was performed using a stimulating electrode made from a twisted pair of Teflon-coated platinum-iridium wires (0.75 μm diameter) inside a blunted glass micropipette placed over the nerve approximately 1–2 mm from the edge of the brainstem. Synaptic activity was recorded using an Axopatch-1D amplifier (Axon Instruments), low-pass filtered at 5 kHz, digitized at 44 kHz (Neuro-corder, Neurodata), stored on videotape, and analysed off-line on a 486/50 MHz computer with pCLAMP programs (Axon Instruments).
The criteria for detecting synaptic currents were fast rise times (< 1 ms) and exponential decays. A value of twice the baseline noise level for a given recording in control solutions was used as the detection limit for minimum PSC amplitude. EPSCs and IPSCs were separated pharmacologically or by the direction of the current (i.e. inward or outward) at a given holding potential. Response latencies were determined by measuring the time from stimulus onset to the onset of the synaptic event. More than five consecutive responses that varied by less than 0.5 ms within a given recording were considered to be of relatively constant latency. Measurements of 10–20 evoked or 100–200 consecutive spontaneous PSCs were used to obtain mean amplitudes. Changes in PSC frequency after a stimulus were measured by comparing the event frequencies in 100 ms bins after the stimulus to a period of pre-stimulus activity (minimally 1 s). At least five consecutive stimuli were averaged to determine the duration of stimulation-evoked changes in PSC frequency. Frequencies of spontaneous synaptic events were obtained over continuous 60–120 s periods. Once in the whole-cell configuration, cells were initially held near the resting membrane potential for 5–10 min to allow equilibration of the extracellular and recording electrode solutions. Synaptic currents were examined at rest or, for Cs+-loaded cells, near the mean resting potential (determined from recordings that used K+ as the primary cation), and at more positive (−60 to 10 mV) and negative (−100 to −70 mV) holding potentials. Student's unpaired, two-tailed t test was used for comparing data between recordings and to determine the duration of PSC bursts. Values are reported as means ± standard error of the mean (s.e.m.).
Electrodes contained 0.1 % biocytin (Horikawa & Armstrong, 1988) to label recorded neurones and verify their location. Following each recording, explants were fixed in 4 % paraformaldehyde and 0.05 % glutaraldehyde in 0.15 m NaPO4 buffer (pH 7.3) overnight at 4°C. Following fixation, explants were rinsed (3 × 5 min) in 0.01 m phosphate-buffered saline (PBS; pH 7.4), cryoprotected in PBS containing 30 % sucrose, and sectioned at 50–60 μm on a sliding microtome. After rinsing in PBS, endogenous peroxidase was removed (10 % methanol-3 % H2O2 in PBS; 60–70 min). The sections were again rinsed in PBS and the filled cell was visualized by incubating them overnight in an avidin-biotin-horseradish peroxidase complex (ABC kit, Vector Labs, Burlingame, CA, USA) in PBS (1 : 100; pH 7.3) containing 0.1 % Triton X-100 (Smith & Armstrong, 1990). The reaction product was visualized with diaminobenzidine at a concentration of 0.06 % with 0.003 % H2O2 in 0.1 m Tris-buffered saline (pH 7.4) to confirm the location of the recorded neurone within the NTS, and the tissue was subsequently dehydrated in alcohols and mounted in Permount. For the purposes of this study, recovered neurones were used only to verify their location. Quantitative aspects of neuronal morphology will be addressed in a future study.
Whole-cell patch-clamp recordings were obtained from forty-one NTS neurones in brainstem-cranial nerve preparations from immature (2–5 day) rats. Resting membrane potential for NTS neurones was −63 ± 2 mV (mean ±s.e.m.); input resistance ranged from 217 to over 1200 MΩ with a mean of 527 ± 58 MΩ. All neurones in this analysis were located in the dorsomedial NTS, determined by recording electrode placement, intracellular staining, or response to selective vagal stimulation. A diagram of the brainstem-cranial nerve explant preparation and the region of the NTS from which recordings were made is shown in Fig. 1.
Primary responses to vagal stimulation
To determine if vagal input to the NTS was glutamatergic, we monitored PSCs in NTS neurones while electrically stimulating the vagus nerve outside the brainstem (i.e. in isolation from possible direct activation of local neurones). Constant-latency (< 0.5 ms variability) responses to orthodromic stimulation of the vagus nerve consisted of fast PSCs, and were observed in twenty-nine of thirty-six NTS neurones tested. All of the vagally evoked constant-latency PSCs were inward at rest (Fig. 2). The mean latency for the primary response to vagal stimulation was 10.1 ± 0.6 ms. Using an estimated distance of ∼3 mm between recording and stimulating electrodes, this value indicated a conduction velocity for vagal central processes in the order of about 0.3 m s−1.
Vagally evoked responses were observed while voltage-clamping the neurone at several holding potentials between −100 and +20 mV. The amplitude of constant-latency PSCs was greater at more negative holding potentials and reversed at 8 ± 3.1 mV (Fig. 2), indicating that they were excitatory PSCs (EPSCs). Evoked EPSCs had 10–90 % rise times of less than 1 ms and decay time constants ranging from 1.3 to 2.8 ms at holding potentials near rest (2.04 ± 0.03 ms). All fast EPSCs were blocked by 10 or 50 μm DNQX or 10 μm CNQX at potentials more negative than −60 mV (Fig. 3). At more positive potentials, a DNQX- or CNQX-insensitive evoked EPSC was observed in seven of eleven neurones (Fig. 3). The I–V relationship of this current was non-linear, being largest at between −30 and −25 mV, and was reversibly blocked by 50 μm AP5. Together, these results suggest the involvement of NMDA receptors in the response to vagal stimulation in some neurones, as reported previously in data from slices (Aylwin et al. 1997). These data indicate that, as in the adult, primary vagal input to the NTS in immature rats is glutamatergic.
Secondary responses: inhibition
In addition to constant-latency glutamatergic EPSCs (Figs 2, 4, 5 and 8), fourteen of the twenty-nine responding neurones displayed mixed excitatory/inhibitory responses to vagus nerve stimulation (Fig. 4). Outward PSCs were of larger amplitude at more positive holding potentials and reversed polarity near −70 mV. They were therefore considered to be inhibitory PSCs (IPSCs). The mixed-inhibitory responses were composed of either an EPSC-IPSC complex or an EPSC followed by a barrage of IPSCs of variable latency and duration. The onset of the inhibitory phase of the response often overlapped the decay phase of the EPSC. At holding potentials between −20 and 0 mV, the IPSCs could be resolved in relative isolation from EPSCs. Although temporally related to the stimulus, evoked IPSCs did not have constant response latencies (Figs 4, 5, 6 and 8). They also did not reliably follow high-frequency stimulation (50–100 Hz; n= 10 neurones), even at stimulus intensities as much as 5 times those necessary to observe constant-latency EPSCs, which followed both stimuli with constant latency (n= 16; Fig. 4). In addition to inconsistent latency, vagally evoked bursts of IPSCs were of variable duration, lasting between 80 and 1200 ms. Evoked IPSCs were reversibly blocked by 30 μm bicuculline (n= 6; Fig. 5), suggesting that they were the result of activation of GABAA receptors.
None of the neurones in this study responded to vagal stimulation with constant-latency unitary IPSCs, indicating that GABAergic fibres do not play a role in the primary vagal input to NTS neurones. An additional seven NTS neurones that did not exhibit constant-latency responses to vagal stimulation were identified anatomically by the position of biocytin-filled neurones within the nucleus. Of these, three had variable-latency IPSCs that were associated with the stimulus. The primary response to vagal stimulation is therefore a glutamatergic EPSC, with secondary IPSCs also contributing to the response in many immature NTS neurones.
To test the extent to which inhibition in the NTS was controlled by glutamate receptor activation, we examined IPSCs in the presence of glutamate antagonists. Evoked IPSCs were reversibly blocked with 10–50 μm DNQX or 10 μm CNQX, indicating that they were generated secondary to glutamatergic activation of local inhibitory neurones (Fig. 6). Spontaneous IPSCs were observed in all neurones, regardless of whether evoked IPSCs could be detected. The mean reversal potential for spontaneous IPSCs in five neurones was −70 ± 3.9 mV. As with evoked IPSCs, spontaneous IPSCs could be blocked with 30 μm bicuculline (n= 6), indicating they were mediated by GABAA receptors. In addition to blocking EPSCs and evoked IPSCs, glutamate receptor antagonists also significantly reduced the frequency of spontaneous IPSCs in five neurones from a mean of 17.95 ± 6.4 to 5.35 ± 1.7 Hz (P < 0.05; Fig. 7).
Secondary responses: excitation
The bursts of IPSCs also suggested that local glutamatergic neurones were activated by vagal input, and that these neurones synapsed with other neurones in the NTS. Increasing orthodromic stimulus intensity often resulted in the addition of one or more relatively constant-latency EPSCs following vagal stimulation. This was probably due to recruitment of additional vagal fibres of slightly different length or conduction velocity. In addition to constant-latency EPSCs, nine neurones responded to vagal stimulation with a barrage of EPSCs, lasting from 50 to 600 ms (Fig. 8). Unlike the constant-latency EPSCs or the EPSCs that appeared with increased stimulus intensity, the evoked EPSC bursts were of variable latency and duration and did not require intense vagal stimulation. Therefore, although some EPSCs may have resulted from activation of additional vagal fibres, the barrages of variable-latency EPSCs were likely to have been due to orthodromic activation of local excitatory neurones in the brainstem. In neurones receiving EPSC barrages following vagal stimulation, stimulus-associated IPSCs (6 of 9 neurones) and/or spontaneous IPSCs (9 of 9) were also observed (Fig. 8). This suggested that some NTS neurones receive input from vagally activated local excitatory and inhibitory neurones in addition to vagal afferents.
To test the hypothesis that vagally evoked local inhibition acts to reduce the effects of excitatory connectivity in the NTS, we examined EPSCs in the presence of the GABAA receptor antagonist bicuculline (30 μm). In addition to blocking IPSCs, bicuculline increased the duration of evoked secondary EPSC bursts (Fig. 9). This suggested that a feedback inhibitory circuit can be activated by vagal stimulation.
Spontaneous EPSCs were observed in all neurones. Blocking IPSCs with bicuculline also significantly increased the frequency of spontaneous EPSCs, with the mean being increased from 5.8 ± 1.6 to 17.9 ± 6.7 Hz (P < 0.05) in six neurones tested (Fig. 10).
Several important features of synaptic connectivity in the immature NTS can be examined in vitro. We used a preparation that allowed examination of evoked synaptic input to individual NTS neurones in isolation from possible direct stimulation of locally projecting excitatory or inhibitory interneurones. In slices, electrical stimulation of the solitary tract can also cause action potentials in local neurones or axons of passage that project to the recorded cell. This has been used to advantage in studying inhibitory responses evoked within the NTS (Glaum & Miller, 1992; Grubauskas & Bradley, 1996), but also makes uncertain the makeup of putative vagal afferents. Stimulating the vagus nerve at a distance from the brainstem resulted in a response of relatively long (i.e. in the order of 10 ms) but invariant latency. The estimated 0.3 m s−1 conduction velocity, based on an approximate separation of ∼3 mm between stimulus and recording sites, is similar to that reported for C-type, unmyelinated fibres of vagal central processes in adult mammals (Ducreux et al. 1993; Fogel et al. 1996). These characteristics suggest that the responses we observed were the result of vagal fibre activation, and not due to input from local neurones or axons of passage that were activated by current spread at the tip of the stimulating electrode. Constant-latency excitatory responses also followed paired stimuli at high frequency. Based on previously established criteria for constant-latency responses in the NTS (Miles, 1986; Wang & Bradley, 1995; Aylwin et al. 1997), we interpret the constant-latency responses to be due to monosynaptic input from primary vagal afferents.
Most of the NTS neurones we recorded exhibited constant-latency EPSCs in response to stimulation of the vagus nerve. These EPSCs reversed near 0 mV and were blocked by glutamatergic non-NMDA receptor antagonists at holding potentials near the average resting potential. Although many studies in slices have suggested that the primary peripheral input to the NTS is glutamatergic (Andresen & Yang, 1990, 1995; Glaum & Miller, 1992; Fortin & Champagnat, 1993; Wang & Bradley, 1995; Grabauskas & Bradley, 1996; Kawai & Senba, 1996; Aylwin et al. 1997; Titz & Keller, 1997), the possibility that direct stimulation of local neurones in slices accounts for this response has not been tested directly. Similarly, few studies have excluded the possibility that GABAergic fibres are activated by solitary tract stimulation. We cannot exclude the possibility that additional neuroactive substances may also have participated in the response to vagal stimulation. However, all the fast PSCs we observed were blocked by amino acid receptor antagonists, suggesting that GABA and glutamate mediate fast synaptic transmission in the NTS. Our data support the hypothesis that glutamate mediates neurotransmission in primary vagal afferents to the NTS, and GABA does not.
Whether NMDA receptors also play a role in the response to vagal afferent activation has been controversial. Current-clamp recordings in slices have suggested that NMDA receptors were not significantly activated by primary vagal afferents, but instead play a mainly modulatory role in the response to vagal afferent stimulation (Andresen & Yang, 1990). However, others have suggested that NMDA receptors were involved in the response to solitary tract stimulation in the adult NTS (Aylwin et al. 1997; Titz & Keller, 1997). In most neurones we examined, a slow component of the monosynaptic EPSC was detected in the presence of DNQX or CNQX when the neurones were voltage clamped at membrane potentials positive to about −50 mV. The current was sensitive to AP5 and exhibited a non-linear current-voltage relationship. These data therefore support previous studies (Aylwin et al. 1997; Titz & Keller, 1997) which suggested that both NMDA and non-NMDA receptors on NTS neurones can be activated by vagal stimulation.
We were also able to assess the local circuitry activated by vagal input. Approximately half of the responses we observed had a variable-latency component to the response to vagal stimulation. Of the total responses recorded, 50 % included variable-latency IPSCs and 25 % had variable-latency EPSCs following the monosynaptic EPSC response. The stimulus-associated, variable-latency IPSCs were blocked by bicuculline, supporting the hypothesis that feedforward or feedback GABAergic circuits within the brainstem account for the inhibitory responses we observed. They were also sensitive to glutamate receptor antagonists, indicating that their activation was secondary to activation of primary vagal afferents. Finally, the variable duration of the bursts of EPSCs and IPSCs suggests that these inputs were due to neuronal activity in the brainstem. The variable-latency responses were therefore characteristically multisynaptic, being due to glutamatergic and GABAergic local circuitry that was activated either directly by vagal input or indirectly by local glutamate neurones that receive vagal input.
In transverse slices, electrical stimulation of the solitary tract region probably also activates nearby inhibitory neurones or their fibres (Glaum & Miller, 1992), making differentiation between vagally and locally activated inhibitory circuits difficult. Responses to solitary tract stimulation can also be contaminated by activation of local excitatory neurones or fibres in slices. Some attempts have been made to minimize this problem using horizontal slices, where the relatively larger distance between the stimulating and recording electrodes allows some delay between stimulus and response (Andresen & Yang, 1995; Grabauskas & Bradley, 1996). In those studies, as in this one, direct monosynaptic inhibition was not observed after primary visceral afferent stimulation. Some intracellular recordings in vivo have suggested that, although most responses are excitatory or mixed, some vagal inputs from the vagal or carotid sinus nerves may inhibit NTS neurones (Mifflin & Felder, 1988; Mifflin et al. 1988). Our use of voltage-clamp recordings effectively isolated temporally the two types of response, allowing us to detect excitatory inputs that may be strongly shunted by vagally activated inhibitory circuits in the brainstem.
A similar excitatory/inhibitory response pattern observed in the medial NTS has been described for neurones that express a prolonged period of reduced excitability following solitary tract stimulation (Champagnat et al. 1986; Fortin & Champagnat, 1993). These studies suggested that spontaneous inhibition was confined to this subset of NTS neurones, and that these neurones received very little spontaneous excitatory input. However, we detected spontaneous EPSCs in all neurones, including those that exhibited evoked IPSCs. Spontaneous IPSCs remained evident in the presence of DNQX or CNQX in our preparation, although they were reduced in frequency by about 70 %. This result suggests that a substantial portion of the inhibition in the nucleus is driven by activity in glutamate neurones with local connections. Our data suggest that excitatory input from the vagus nerve may activate a system of local circuit inhibitory neurones that can effectively decrease the primary effect of excitatory vagal input to many NTS neurones.
In addition to strong inhibition, some NTS neurones exhibited EPSC patterns consistent with the hypothesis that feedforward excitatory circuits exist in the brainstem. Analogous to the prolonged period of increased excitability observed after solitary tract stimulation in slices (Champagnat et al. 1986; Fortin & Champagnat, 1993), vagal stimulation triggered barrages of variable-latency EPSCs in 31 % of responding neurones. Unlike studies in slices from adult rats (Champagnat et al. 1986; Fortin & Champagnat, 1993; Kawai & Senba, 1996), most neurones exhibiting multisynaptic EPSCs (6/9) also displayed multisynaptic IPSCs associated with the stimulus. Further, spontaneous IPSCs were observed in all neurones that responded to vagal stimulation with multisynaptic EPSC barrages. The frequency of spontaneous EPSCs increased roughly threefold, and the duration of evoked multisynaptic EPSC bursts increased when GABAA receptors were blocked pharmacologically. These results suggest that local excitatory circuits contribute to the synaptic profile of the NTS in immature rats, and that these circuits are controlled by local inhibitory neurones in the brainstem. The results further emphasize the importance of local GABAergic inhibition in controlling and conditioning vagal input in the NTS of immature animals.
The relatively robust inhibitory tone we observed could have been due in part to developmental differences in GABAergic circuitry in the NTS. Although most intrinsic membrane properties of NTS neurones tend to be largely mature by the end of the first postnatal week (Kalia et al. 1993; Nabekura et al. 1994; Bao et al. 1995; Vincent et al. 1996), dendritic and axonal morphologies may not reach their full complexity until the end of the third postnatal week (Lasiter et al. 1989). Such findings suggest that differences in synaptic density or efficacy might explain the relatively robust synaptic responses we observed. However, the dendritic trees of NTS neurones tend to become more complex with age (Lasiter et al. 1989), implying that synaptic connectivity might be greater in more mature animals. Further, the response types we observed are qualitatively similar to those reported in the adult NTS (Champagnat et al. 1986; Fortin & Champagnat, 1993; Kawai & Senba, 1996), arguing against major age-related differences in connectivity as an explanation for the differences we observed. A repolarizing A-type K+ current that may not be as effective in younger animals as it is in adults has been described (Vincent & Tell, 1997). Inefficient spike repolarization could result in multiple action potentials being generated in some neurones after a single stimulus. This could help explain the long-duration PSC barrages we observed (∼1 s) if local afferent neurones fired bursts of action potentials in response to vagal stimulation. However, the PSC bursts were of similar duration to those observed in adults (Champagnat et al. 1986; Fortin & Champagnat, 1993), making an immature A-current an unlikely candidate to explain fully the synaptic responses we observed.
Evidence from anatomical studies (Khachaturian & Sladek, 1980; Sawchenko & Swanson, 1982) and functional evidence based on c-fos expression (Rinaman et al. 1994) suggests that connections between the NTS and hypothalamus mediating gastrointestinal function do not form completely until after the first postnatal week. Because of the relative maturity of visceral control in early postnatal animals, we used the immature NTS as a model for viscerosensory integration by a local synaptic network that is not dominated by extramedullary inputs, yet retains much of its local circuitry. Regardless of possible changes in intrinsic membrane properties, it is apparent from our data that GABAergic and glutamatergic circuitry within the NTS is intact and functional very early in life, implicating the NTS as a primary integrative centre for gastrointestinal and probably other visceral functions in very young animals.
All of our recordings were made in the caudal, dorsomedial region of the NTS, within ∼300 μm of the brainstem surface. Multisynaptic EPSCs are not widely reported in neurones from more rostral NTS regions in slices. Multisynaptic barrages of excitatory or inhibitory activity have been observed in the caudal NTS by only a few investigators (Fortin & Champagnat, 1993; Kawai & Senba, 1996), and both excitatory and inhibitory multisynaptic responses have been observed in the dorsomedial NTS (Champagnat et al. 1986). However, the two types of response have not been observed in the same neurone or neurone type in any of these studies. It is therefore possible that the feedforward excitation we observed, even in the presence of strong inhibition, is a unique feature of neurones in the dorsomedial NTS from which we recorded. This hypothesis has yet to be tested directly.
It was not possible for us to isolate specific components of the sensory input to the NTS. Labelling studies indicate that vagal afferents from the stomach preferentially innervate this region (Leslie et al. 1982; Shapiro & Miselis, 1985; Altschuler et al. 1989), and many baroreceptor afferents also terminate in this region of the nucleus (Mendelowitz et al. 1992). The dorsomedial NTS is reportedly rich in small, putatively GABAergic neurones (Blessing et al. 1984; Izzo et al. 1992). Previous recordings in slices have suggested that excitatory vagal inputs make synaptic contact with small interneurones in the NTS (Titz & Keller, 1997). If such interneurones are indeed GABAergic, this implies that excitatory vagal input to this region may result in net inhibition of NTS local circuitry. Recordings of unit activity in rat DMNX neurones indicate that gastric mechanoreceptor activation rapidly inhibits activity in most (80–90 %) of these motor neurones (McCann & Rogers, 1992; Fogel et al. 1996). Inhibitory neurone activity in the dorsomedial NTS may therefore play an important role in mediating feeding and satiety, especially in immature animals. Future studies designed to activate specific sensory inputs will be useful for determining if the responses we observed are associated with specific anatomical inputs or sensory modalities.
The greater local circuit interactions we observed might be due to the relatively intact nature of the local circuitry in the brainstem-cranial nerve explant. Neurones probably retained most or all of their local synaptic circuitry in this three-dimensional preparation. In slices, multisynaptic excitatory responses following solitary tract stimulation and an absence of spontaneous inhibition were observed in interneurones exclusively. Multisynaptic inhibitory responses with only weak spontaneous excitation were seen in projection neurones (Champagnat et al. 1986; Kawai & Senba, 1996). We observed spontaneous IPSCs and EPSCs in all neurones, and a subset of neurones responded to vagal stimulation with both multisynaptic EPSCs and IPSCs, suggesting that local GABAergic and glutamatergic connectivity in the NTS are probably not limited to single neurone types. At least some of the IPSCs we detected could have originated from GABAergic neurones whose somata were in NTS areas that are not present in slices, but remained intact in the brainstem-cranial nerve explant we employed. Alternatively, inhibitory neurones near the recorded neurone could be activated by feedforward excitation originating in other parts of the nucleus. Either possibility suggests that the various regions of the NTS may be in communication with one another.
Functionally, the strong inhibition observed in many neurones supports the hypothesis that stimulation of neurones in one region of the nucleus might inhibit activity in another. Such inhibition would be analogous to the ‘lateral’ inhibition described for other sensory systems (see Kandel & Schwartz, 1985). Unit recordings in intact animals suggest that stimulation of different NTS afferents can activate the same neurones in the nucleus. For example, individual NTS neurones can be activated by stimulating the renal and the carotid sinus nerves (Felder, 1986). Similarly, some NTS neurones receiving input from the carotid sinus nerve also receive contralateral carotid sinus (Mifflin & Felder, 1988), vagal, or superior laryngeal nerve input (Mifflin et al. 1988). Interactions between inputs from different structures tended to be inhibitory (Mifflin & Felder, 1988), but could also summate if the stimuli were applied simultaneously (Felder, 1986). Similar interactions were observed after stimulating different gastric vagal branches (Barber et al. 1990). Together, these observations suggest that local circuits in the NTS regulate and integrate inputs from several visceral systems and sensory modalities. Communication between various NTS regions could help to co-ordinate visceral inputs, thereby modifying output to an organ system in response to information from a different system. Further investigation of connectivity in the brainstem of adults will be necessary to determine if this hypothesis is correct.
Our data suggest that both excitatory and inhibitory synaptic interactions exist in the immature NTS. The relative importance of these interactions to NTS function may rely in part on the type of stimulus used to generate vagal afferent activity and the viscerotopic representation of those inputs in the nucleus. While it is possible that the various regions of the NTS receive different patterns of vagal input, similarities between our results and those observed in other NTS regions suggest that vagal input throughout the NTS is glutamatergic. Our data further indicate that the majority of evoked inhibitory connections are from interneurones located within the brainstem. These local inhibitory circuits may help in integrate vagally mediated inputs to various NTS regions. Further, synaptic interactions in the NTS probably play an important role in the control of visceral reflexes in very young animals. The final output of the NTS in response to vagally mediated afferent information will depend in part on the type of neurone (i.e. GABAergic or glutamatergic) receiving excitatory input from the vagus nerve, but will also be regulated by activation of this integrative circuitry in the brainstem. Important issues for future studies would be to determine the extent to which GABAergic neurones receive direct vagal input, whether intrinsic NTS neurones form a functional system of ‘lateral’ inhibition between NTS subnuclei, and how the multitude of neuroactive substances known to exist in the NTS affect this essential circuitry.
We thank J. Welton, M. Higgins and T. Sampson for technical assistance. This work was supported by a fellowship from the American Heart Association of Colorado, Inc. (B. N. S.), a grant from the AFOSR (F. E. D.), and by NIH grant NS27972 to W. D. B.