ADRENERGIC RESPONSES IN SILENT AND PUTATIVE INHIBITORY PACEMAKER-LIKE NEURONS OF THE RAT ROSTRAL VENTROLATERAL MEDULLA IN VITRO.
Neuroscience 77 (1997) 199-217

Laboratoire de Physiologie Générale, URA CNRS 1446, Univ. Louis Pasteur, 21 rue R. Descartes, 67084 Strasbourg Cedex, France.

*To whom correspondence should be addressed: Abdallah HAYAR, Dept. Pharmacology / Univ. of Virginia School of Medicine / Box 448 HSC Charlottesville, VA 22908, U. S. A. Phone: 804-982-1040 / Fax: 804-982-3878

Abbreviations: APV, (±)-2-Amino-5-phosphonopentanoic acid; CNQX, 6-cyano-2,3-dihydroxy-7-nitroquinoxaline; GABA, gamma-aminobutyric acid; GAD, glutamic acid decarboxylase; KAc, potassium acetate; mACSF, modified artificial cerebrospinal fluid; NA, (-)-noradrenaline; PE, L-phenylephrine; PNMT, phenylethanolamine N-methyl transferase; PSPs, spontaneous postsynaptic potentials; Rin, cell input resistance; RVL, rostral ventrolateral medulla; SND, sympathetic nerve discharge; TTX, tetrodotoxin; UK-14,304, 5-bromo-N-(4,5-dihydro-1H-imidazol-2-yl)-6-quinoxalinamine.

RUNNING TITLE: Adrenergic responses in rat RVL in vitro

Abstract- Noradrenaline and adrenergic agonists were tested on pacemaker-like and silent neurons of the rat rostral ventrolateral medulla using intracellular recordings in coronal brainstem slices as well as in punches containing only the rostral ventrolateral medullary region. Noradrenaline (1-100 µM) depolarized or increased the frequency of discharge of all cells tested in a dose-dependent manner. The noradrenaline-induced depolarization was associated with an apparent increase in cell input resistance at low concentrations and a decrease or no significant change at higher concentrations. Moreover, it was voltage dependent and its amplitude decreased with membrane potential hyperpolarization. Noradrenaline caused a dose-related increase in the frequency and amplitude of spontaneous inhibitory postsynaptic potentials. The a1-adrenoceptor antagonist prazosin (0.5 µM) abolished the noradrenaline depolarizing response as well as the noradrenaline-evoked increase in synaptic activity and unmasked an underlying noradrenaline dose-dependent hyperpolarizing response associated with a decrease in cell input resistance and sensitive to the a2-adrenoceptor antagonist yohimbine (0.5 µM). The a1-adrenoceptor agonist phenylephrine (10 µM) mimicked the noradrenaline depolarizing response associated with an increase in membrane resistance as well as the noradrenaline-induced increase in synaptic activity. The a2-adrenoceptor agonists UK-14,304 (1-3 µM) and clonidine (10-30 µM) produced only a small hyperpolarizing response whereas the b-adrenoceptor agonist isoproterenol (10-30 µM) was without effect. Baseline spontaneous postsynaptic potentials were abolished by either strychnine (1 µM), bicuculline (30 µM) or both. However, only the strychnine sensitive postsynaptic potentials had their frequency increased by noradrenaline or phenylephrine and they usually occurred with a regular pattern. Tetrodotoxin (1 µM) eliminated 80-95 % of baseline spontaneous postsynaptic potentials and prevented the increase in synaptic activity evoked by noradrenaline and phenylephrine. Similar results were obtained in rostral ventrolateral medulla neurons impaled in both coronal slices and punches of the rostral ventrolateral medulla.
   It is concluded that noradrenaline could play an important inhibitory role in the rostral ventrolateral medulla via at least two mechanisms: an a2-adrenoceptor-mediated hyperpolarization and an enhancement of inhibitory synaptic transmission through activation of a1-adrenoceptors located on the somatic membrane of glycinergic interneurons. Some of these interneurons exhibit a regular discharge similar to the pacemaker-like neurons and might, at least in part, constitute a central inhibitory link in baroreceptor-vasomotor reflex pathway.

Key words: noradrenaline, rostral ventrolateral medulla, rat, pacemaker, interneuron, sympathoexcitatory, intracellular recording.

   The rostral ventrolateral medulla (RVL) is the medullary region with the densest population of adrenergic neurons. Moreover, it is a probable site of action of several types of antihypertensive agents which interfere with catecholaminergic transmission (for review, see Reis et al., 1994). It was initially suggested that the origin of vasoconstrictor and cardioaccelerator sympathetic tone is due in large part to the intrinsic pacemaker activity of a small group of reticulospinal excitatory neurons located at the extreme anterior tip of the RVL (for review, see Guyenet,1990).
   This hypothesis has been challenged by Barman and Gebber (1989) who reported that sympathetic nerve discharge (SND) was not desynchronized by intracisternal injection of kynurenate in baroreceptor-denervated rats indicating that RVL pacemaker neurons are not primarily responsible for the production of the rhythmicity in SND. Since many medullary sites were found to contain a 2- to 6-Hz oscillation correlated to that in SND, it was suggested that the oscillation in SND is an emergent brainstem network property rather than a property of the individual neurons that comprise the network (Zhong et al., 1993). On the other hand, it was found that an excitatory supraspinal drive to the spinal cord is essential for generating the 2- to 6-Hz oscillation in SND. However, the firing pattern of these neurons does not seem to be important for the production of this rhythmicity (Allen et al., 1993). Therefore, the regular firing pattern of some RVL neurons is not a sufficient criterion to classify pacemaker neurons as sympathoexcitatory. Interestingly, a recent study by Lipski et al. 28a showed that RVL presympathetic neurons recorded intracellularly in vivo displayed a largely irregular pattern of firing that resulted mainly from synaptic inputs, therefore supporting the ‘network’ hypothesis for the generation of vasomotor tone.
   Over 70% of the bulbospinal neuronal population of the RVL was found to be immunoreactive for the adrenaline-synthesizing enzyme, phenylethanolamine N-methyl transferase (PNMT). However, 38% of adrenergic neurons in this region were thought to be non-bulbospinal and might be local propriobulbar neurons (Ruggiero et al., 1994). The presence of barosensitive, adrenergic, non-bulbospinal RVL neurons was recently directly confirmed (Lipski et al., 1995).
   Immunoelectron microscopic studies on the localization of PNMT support the concept that adrenergic neurons might modulate the activity of neurons containing the same or other putative neurotransmitters in the RVL. However, only a few axons terminals containing immunoreactivity for PNMT have been observed and they were mainly found to form symmetric synapses with unlabelled dendrites (Milner et al., 1987). Moreover, there is a doubt whether adrenaline or another catecholamine acts as a transmitter in this region since it seems that adrenaline is not stored in vesicles but is rapidly degraded by intracellular monoamine oxidases (Sved, 1989; 1990).
   The mechanism of action of catecholamines and the type of adrenoceptor as well as the neuronal phenotype involved in their cardiovascular effects are still subject to controversy. a1 and a2-adrenoceptors have been shown to mediate respectively, the tachycardic and bradycardic responses to microinjection of adrenaline and clonidine in the intermediolateral column of the spinal cord (Malhotra et al., 1993). On the other hand, intracerebroventricular injection of clonidine has been reported to produce a2-adrenoceptor-mediated pressor and depressor responses in conscious and anaesthetized rats, respectively (Kawasaki and Takasaki, 1986). Clonidine (up to 1 µM) has been shown to produce no effect on RVL pacemaker neurons (Sun and Guyenet, 1990) and to produce an inhibitory action at a concentration of 10 - 30 µM through a bicuculline-sensitive mechanism (Sun and Reis, 1995). Clonidine has been found to stimulate the GABAergic system in spontaneously hypertensive rats (Czyzewska-Safran et al., 1991). Therefore, one possible mechanism of the sympatholytic effect of clonidine could be an indirect inhibitory action through stimulation of the inhibitory brainstem networks.
   We have recently recorded in most RVL neurons examined spontaneous postsynaptic potentials (PSPs) that often occurred in a regular pattern and were sensitive to bicuculline or strychnine (Hayar et al., 1996). As a consequence, we have hypothesized that at least some of the regular pacemaker-like RVL neurons are inhibitory interneurons and thus may not be sympathoexcitatory as it was assumed in previous in vitro studies (Sun et al., 1989; Sun and Guyenet, 1990). In order to provide more evidence supporting this hypothesis, we investigated the effects of NA and adrenergic agonists on the membrane properties of RVL neurons as well as on the spontaneous PSPs which probably reflect the activity of the intrinsic RVL inhibitory network. Moreover, we tested for the presence of functional a1 and a2-adrenoceptors and examined if activation of any of these receptors can affect the inhibitory synaptic neurotransmission. The present study is the first to use the isolated RVL punch preparation in order to test the hypothesis that the spontaneous regular PSPs originate from putative inhibitory pacemaker-like interneurons located within the RVL.

   Wistar rats (50 - 100 g, Etablissement Depre) were anesthetized with ether and decapitated. The brainstem along with the cerebellum were quickly removed and placed for 30 seconds in cold (2 - 4 °C) modified artificial cerebrospinal fluid (mACSF, see below) in which sucrose 248 mM was substituted for NaCl 124 mM (see also Refs 1, 40) and equilibrated with 95% O2 / 5% CO2. The tissue was trimmed with a razor blade and a block containing the medulla and a part of the cerebellum was glued with cyanoacrylate in front of an agar block on a petri dish and covered with the mACSF. Two to three transverse slices (400 µm thickness) containing the RVL were sectioned using an Oxford Vibratome and left to recover in mACSF at room temperature for 30 min. A slice was then transferred to the recording chamber and perfused at a rate of 1.5-2 ml / min with oxygenated ACSF (pH = 7.35) of the following composition (mM): NaCl 124, KCl 2 , CaCl₂ 2, MgSO₄ 2, NaHCO₃ 26, KH₂PO₄ 1.25, and D-Glucose 10. The temperature in the recording chamber was raised slowly from room temperature up to 31 - 32 °C and the slice was allowed to equilibrate for 1 hour before commencement of recording. Signals were recorded using a high-impedance bridge amplifier (Axoclamp-2B, Axon Instruments Inc.) filtered at 10 kHz bandwidth and displayed on an oscilloscope (Tektronix 5116). The output signal was also directed to a graphic thermal recorder (Gould TA240) and a digital tape recorder (DTR - 1201, Biologic) for storage and later analysis. Data analysis was performed using a personal computer equipped with Digidata 1200 analog/digital interface and pClamp software (Axon Inst., Foster City, CA). Electrodes were filled with KCl (3M) (30 - 70 MW) or in some experiments with potassium acetate (KAc, 2M) (80 - 130 MW) and pH was adjusted to 7.4 with KOH and hydrogen acetate HAc, respectively. Electrodes were guided towards the RVL with the assistance of a dissecting microscope and were advanced using a micromanipulator (Narishige, Japan). The region explored is the same as defined by Li and Guyenet28 and corresponds to that previously found to contain the highest concentration of adrenergic and other presympathetic cells (for review, see Ref 20). Some experiments were performed on punches of slices limited to the RVL region. For that purpose, the area that lies ventromedial to the compact rostral portion of the nucleus ambiguus and lateral to the inferior olive was excised from the entire coronal slice by dissection with a razor blade under visualization through a surgical microscope. The area cut out of the medulla was a rectangle, the length of which being the floor of the medulla (approximate dimensions 1500 X 800 µm). This operation was accomplished as soon as the slices were sectioned and in the same conditions described before for the slicing procedure. Drugs were dissolved in ACSF and applied via a three-way tap system, by changing the superfusion solution to one which only differed in drug content. The delay between turning the tap and the first arrival at the tissue of the exchanged solution was about 20 s. The time required to reach the concentration at equilibrium of a drug was 1 min.. The following compounds purchased from Sigma Co. were used: (-)-Noradrenaline (NA), L-phenylephrine, isoproterenol hydrochloride, clonidine hydrochloride, desipramine, prazosin, yohimbine hydrochloride, (-)-bicuculline methiodide, picrotoxin, strychnine, tetrodotoxin (TTX), and 6-cyano-2,3-dihydroxy-7-nitroquinoxaline (CNQX). (±)-2-Amino-5-phosphonopentanoic acid (APV) and 5-bromo-N-(4,5-dihydro-1H-imidazol-2-yl)-6-quinoxalinamine (UK-14,304) were purchased from Research Biochemicals International (RBI). Sodium metabisulphite (50 µM) obtained from RBI was added to NA solutions to prevent oxidation. The input resistance of RVL neurons was determined by measuring the asymptotic voltage drop caused by small hyperpolarizing currents of 100 ms duration. Changes in membrane potential due to drug application were offset by direct current injection to show effect on membrane input resistance. However, the accuracy in evaluating the drug-induced changes in membrane conductance cannot be guaranteed quantitatively, due to the space clamp limitations of the intracellular recording technique. Nevertheless, this should not affect any conclusion made regarding the effect of the drugs on cell input resistance. Spike amplitude and duration were calculated from the threshold to the peak and to the corresponding repolarization phase of the action potential, respectively. In silent neurons, these parameters were evaluated using the minimum current pulse (20 ms) needed to evoke an action potential. The resting membrane potentials were calculated upon withdrawal from cells. Statistics and curve-fitting were performed using Origin 3.0 program (Microcal software Inc., USA). Data are expressed as mean ± s.e.m unless otherwise stated.

Basic electrophysiological properties

This work is based on stable recordings from 120 RVL neurons. Ninety five neurons were impaled in the whole coronal brainstem slice preparation where the RVL region can clearly be identified by its anatomical location and 25 neurons were recorded in RVL punches (see methods). The data presented here refer to two groups of neurons which we previously designated as pacemaker-like and silent, respectively.21 Pacemaker-like neurons had a spontaneous pacemaker activity at rest with a mean frequency of 8.6 ± 0.5 Hz (range 5 to 14 Hz, n = 35, 5 of them in punches), or their membrane potential was oscillating near the threshold for spike generation and injection of a small depolarizing current was sufficient to render them active with repetitive discharge and no accomodation with a minimal frequency of 7-8 Hz. Silent neurons were quiescent and did not present membrane potential oscillations at resting potential. The membrane properties of the two groups of neurons investigated in the present study have been illustrated elsewhere.21
   The responses of these two groups of neurons to NA and adrenoceptor agonists were essentially identical at a given holding potential and no significant difference was found neither in their spike duration (< 1.2 ms) nor in their current-voltage relationship. Therefore, the data collected from these cells were considered together. The input resistance, spike duration and spike amplitude of 46 randomly chosen neurons were 114 ± 5 MW , 56 ± 0.8 mV and 0.91 ± 0.03 ms, respectively.
   It should be mentioned here that the concentration of K+ ions in the slice bathing solution was 3.25 mM. However, when the concentration of K+ ions was increased to 6.1 mM, all cells displaying membrane potential oscillations (and spiking at 8 Hz with injection of a positive current, n = 5) and a large part of the silent neurons (5 out of 9 tested) assumed a regular pacemaker activity. This certainly explains why a smaller proportion of the impaled neurons were pacemaker as compared to the one obtained by other investigators who used 6.1 mM K+ in their superfusion solution.48
   A third group of neurons previously termed by us irregularly firing21 included quiescent or slowly and irregularly firing neurons having a relatively longer action potential duration (> 1.2 ms). These neurons did not assume a regular pacemaker activity even with injection of positive currents. Since neurons of this group were less frequently encountered, no pharmacological investigation was performed on these cells in the present study.

Concentration and voltage-dependency of NA responses

NA (1-100 µM) was applied by superfusion for 90 sec in order to reach the equilibrium in concentration. In most cases, this time was sufficient for the membrane potential to reach a steady-state level in the presence of NA, otherwise the application was prolonged beyond this duration until the plateau of the response was obtained. Sometimes, a more pronounced depolarization was obtained upon washing NA especially at 10 or 30 µM. As the concentration of NA was increased, the time to the peak amplitude of the response was shorter and the recovery was longer (Fig. 1A).
   At a given concentration, the amplitude of the responses to NA was variable in different cells. To test whether reuptake processes could be at the origin of this variability, we examined the response to NA in control conditions and in the presence of desipramine (1 µM) which increased the amplitude and duration of the NA responses (n = 7, not illustrated). In general, when tested at the resting membrane potential and in the absence of desipramine, NA (10 µM) increased the firing rate of all pacemaker-like neurons tested (n = 19) and this was usually associated with an increase in the slope of the depolarizing ramp during the interspike intervals. In silent neurons, NA (10 µM) evoked a depolarizing response in almost all neurons tested (n = 40) at the resting membrane potential; however, in a few neurons, this response was obtained only at higher concentrations of NA.
   Concentration-dependent responses to NA (1 - 100 µM) in the absence of desipramine were compared in 6 neurons whose membrane potential was maintained by continuous current injection 20 mV below the threshold for spike generation. This condition permitted us to quantify the amplitude of NA responses in the absence of neuronal spiking in silent (n = 4) and pacemaker-like neurons (n = 2). In Figure 1B, the relative amplitude of the depolarizing response to NA in these 6 cells is plotted as a function of NA concentration yielding a half-maximal response (EC50) of 4.9 µM. The maximal depolarizing response obtained in the absence of desipramine was 15 mV. At the concentrations of 3 to 10 µM, an apparent increase in cell input resistance (Rin) was observed (132 ± 22% of control, n = 9). Rin was not affected or slightly decreased by NA at concentrations of 30 to 100 µM (95 ± 15% of control, n = 6).
   The NA-induced depolarizing responses either at 10 µM or 30 µM were compared in 4 neurons at different membrane potentials from a potential just above the threshold for spike generation to more negative potentials by steps of -10 mV (Fig. 2A).
   When the membrane potential was shifted to more negative levels, the amplitude of the depolarizing response to NA decreased. The amplitude of this response at a holding potential of -85 mV was less by a factor of 3.3 ± 0.3 (n = 8) than that at a holding potential of -65 mV, was further reduced at a holding potential of -90 mV (Fig. 2B), and was almost nullified at membrane potential beyond -100 mV.

Pharmacological characterization of spontaneous PSPs

   A significant proportion of RVL neurons (40%) (n = 48/120, silent and pacemaker-like) exhibited spontaneous postsynaptic potentials (PSPs) .There was a considerable variability in the frequency and amplitude of the potentials in different neurons. These PSPs were depolarizing when recorded with KCl-filled electrodes and hyperpolarizing when recorded with KAc-filled electrodes (for illustration, see Hayar et al.21), suggesting that they were Cl--mediated. In some cells, especially the pacemaker-like neurons, the PSPs were small in amplitude; in that case, a continuous negative current was injected intracellularly in order to enhance the magnitude of the synaptic events and increase the signal to noise ratio (KCl electrodes).
   In silent and pacemaker-like RVL neurons recorded both in the coronal slice and the punch, we often could identify evenly spaced PSPs with a mean frequency of 8.5 ± 0.6 Hz, (range 3 to 17 Hz, n = 38, 6 of them in punches). A similar regularity in the firing frequency was observed in pacemaker-like RVL neurons impaled in both preparations (nearly constant interspike intervals, Fig. 3, A1 and B1). The mean frequency of the PSPs in each neuron remained relatively constant over the duration of the recording (up to 4 hours in some cells), indicating that presynaptic cells at the origin of the PSPs had also a fairly stable frequency. The regular pattern of spontaneous firing and of PSPs is illustrated in Figure 3, A2 and B2, respectively. The amplitude of these rhythmic PSPs was quite variable and ranged from just above the baseline noise level (0.4 mV) to over 25 mV when measured at a membrane potential of -90 mV. They occasionally triggered a spike at resting nembrane potential. Moreover, it can be noted that there were occasionally some failures in the occurrence of PSPs leading to a double interval between two events (Fig. 3, B1 and B2).
   The PSPs were not abolished by silencing pacemaker cells with negative current injection indicating that this synaptic activity did not result from the release of neurotransmitter from the same neuron. In some neurons, we could identify two groups of rhythmic events having a different mean frequency suggesting that the spontaneously active neurons are not necessarily synchronous (not illustrated). Nevertheless, histograms of the frequency distribution of pacemaker-like neurons and that of regular events show a good correlation indicating that antecedent neurons at the origin of these synaptic events might correspond to the population of pacemaker-like RVL neurons (Fig. 3, C and D). The frequency of PSPs (n = 38) displayed a slightly wider distribution than that of the frequency of firing (n = 35); however, the two distributions were centered at almost identical values ( 8.5 - 8.6 Hz, P < 0.01).
   We tested the effect of bicuculline (30 µM), strychnine (1 µM), or both, on neurons displaying spontaneous PSPs either at resting membrane potential or at a holding potential sufficiently negative in order to reveal measurable PSPs. Out of 21 neurons tested, bicuculline abolished all the PSPs in 6 neurons and part of them in 3 neurons; in these 3 cells, the remaining PSPs were abolished by additional application of strychnine. Bicuculline increased the frequency and reduced the amplitude of the PSPs in 8 neurons and changed the pattern of PSPs from regular to bursting in 4 neurons. This bursting activity consisted of bursts of PSPs (duration 0.5 to 5 sec) spaced by silent period. The membrane potential was reversibly depolarized by bicuculline in 16 out of the 20 neurons tested by 6 ± 2.5 mV. This depolarizing effect was also induced by picrotoxin (100 µM, n = 3).
   Strychnine completely abolished the PSPs in 12 out of 20 neurons tested (Fig. 4A), and part of them in 3 neurons. The strychnine-insensitive PSPs were abolished by additional application of bicuculline (Fig. 4B). The strychnine-sensitive PSPs exhibited in most cases a regular pattern with a mean frequency of 6.8 ± 0.9 Hz (n = 10), whereas the bicuculline-sensitive PSPs were predominantly irregular. Unlike bicuculline, strychnine did not affect significantly the membrane potential of the neurons tested but progressively decreased the amplitude of the regular PSPs before they were abolished without altering their frequency or inducing a bursting pattern of PSPs.

Effects of NA and phenylephrine (PE) on spontaneous PSPs

   In almost all cells exhibiting PSPs, NA induced in addition to a depolarizing response an increase in baseline PSPs in a dose related manner in concentrations from 1 - 3 µM. The major effects of NA were an increase in the frequency of inhibitory PSPs as could be observed in neurons recorded with KAc-filled electrodes (Fig. 5).The regular pattern of these inhibitory PSPs was not altered during NA application. However, in the majority of the experiments we used KCl-filled electrodes to reverse the polarity and increase the amplitude of inhibitory PSPs as described above. Nevertheless, the effect of NA on baseline PSPs amplitude could not be assessed because additional PSPs of different amplitudes and compound PSPs appeared during NA superfusion. The analysis of PSPs amplitudes was further complicated by the high variability of PSPs amplitude even in control conditions when they were regularly occurring. Moreover, since the increase in Rin during NA application could by itself increase the amplitude of PSPs, it was difficult to establish whether NA altered neurotransmitter release or affected the sensitivity of the postsynaptic receptors to the neurotransmitters. At higher concentrations (30 - 100 µM), NA evoked a dramatic increase in the frequency of these PSPs (Fig. 6A) and in some cases their amplitude was depressed for a brief period probably because of an apparent decrease in Rin.

A similar increase in baseline PSPs was observed during PE (10 µM) application. In 12 out of 43 neurons which did not display PSPs in control, NA (10 µM) or PE (10 µM) elicited the appearance of PSPs (Fig. 6B).
   When bicuculline did not affect baseline PSPs, an increase in their frequency in its presence could still be induced by either NA (30 µM, n = 3) or PE (10 µM, n = 2). When strychnine abolished baseline PSPs, it prevented moreover the increase in synaptic activity induced by either NA (n = 5) or PE (n = 4, 2 in punches) (Fig. 7).

    On the other hand, PE application produced a slight increase in PSPs in only 2 out of 5 cells displaying bicuculline-sensitive PSPs. In all cases, NA and PE seemed to affect only those PSPs that occurred in a regular pattern.
   NA or PE might increase the frequency of PSPs through an action at the level of nerve terminals by enhancing neurotransmitter release or at the level of the soma of interneurons by increasing their firing frequency. To distinguish between these two possibilities, we examined the effects of NA and PE in slices treated with the fast sodium channel blocker TTX (1 µM). TTX blocked 80-95 % of the baseline PSPs (n = 9, 4 of them in punches). This effect was accompanied by an inhibition of spontaneous or evoked action potentials. The residual smaller amplitude miniature PSPs occurred infrequently, indicating a low rate of action potential-independent neurotransmitter release. TTX superfusion prevented the increase in frequency of synaptic activity evoked by either NA (10 µM) (n = 2) or PE (10 µM) (n = 4, 2 of them in punches); however, the depolarizing response evoked by NA or PE persisted in the presence of TTX in all cases (Fig. 8).

   On the other hand, in order to determine if there is a polysynaptic excitatory pathway intrinsic to the RVL and involved in the generation of the PSPs, we examined the effect of the excitatory amino acids receptors blockers on cells impaled in punches of RVL and exhibiting synaptic events. In 4 neurons tested, combined application of the N-methyl-D-aspartate (NMDA) receptors antagonist, APV (50 µM), and the non-NMDA receptors antagonist, CNQX (10 µM), did not affect the PSPs (Fig. 9).

   In 3 of these neurons, the PSPs occurred at regular intervals, their frequency was increased by PE and they were abolished by TTX (1 µM) and strychnine. The remaining neuron displayed irregular PSPs that were not affected by neither PE nor TTX (1µM) but abolished by bicuculline.

Effects of adrenoceptor agonists and pharmacological characterization of the NA responses

   PE (10 µM) was tested on 16 RVL neurons; 5 of them were impaled in RVL punches. All of them responded with a depolarization (6 to 12 mV, 8.7 ± 2.4 mV) from a holding potential maintained 20 mV below the threshold for spike generation. In all cells tested, there was an increase in the amplitude and frequency of PSPs. The response to PE was associated with an increase in cell input resistance (142 ± 20% of control, n = 8) when measured at the same holding potential as control.
   In all cells tested, incubation of the slice ( 8 to 15 min) with the a1 adrenoceptor antagonist prazosin (0.5 µM) abolished the depolarizing response of NA (10-30 µM, n = 16) or PE (10 µM, n = 2) as well as the NA evoked increase in synaptic activity (n = 4) and revealed an underlying hyperpolarizing response in 10 out of 16 cells tested with NA. This latter response was small in amplitude (5-10 mV) when measured at a membrane potential that was initially maintained 20 mV below threshold for spike generation but was NA dose dependent (30 - 300 µM) and was increased by 50 ± 15% in the presence of desipramine (0.5 µM, n = 3).
  Additional incubation (5 to 10 min) with the a2 adrenoceptor antagonist yohimbine (0.5 µM) reduced this residual hyperpolarizing response by 67 ± 12%, n = 5 (Fig. 10).
   UK-14,304 (1 -3 µM) suppressed the firing of all 3 pacemaker-like neurons tested and kept their membrane potential under high oscillation until enough time was allowed for recovery (10 to 20 min) (Fig. 11A). On the other hand, it did not affect the membrane properties of 4 out of 8 silent neurons tested at resting membrane potential; the membrane potential of the other four silent neurons was hyperpolarized by 2 to 5 mV and their Rin was slightly decreased (Fig. 11B). Clonidine (50-100 µM) suppressed the discharge and hyperpolarized the membrane potential of 3 out of 9 pacemaker neurons, and slightly hyperpolarized 3 out of 7 silent neurons tested. The amplitude of the hyperpolarizing response to clonidine never exceeded 4 mV. Clonidine slightly decreased the frequency of PSPs by 20 to 30 % in 4 neurons tested. Isoproterenol (10 - 30 µM) did not affect neither the resting membrane potential of 6 silent neurons tested (Fig. 11C) nor the spiking frequency of two pacemaker-like neurons. Only in one neuron, we observed an increase in PSPs after isoproterenol application.

   In this study, we have presented evidence that supports the existence of functional a1 and a2-adrenoceptors on silent and pacemaker neurons of the RVL. Our data show, moreover, that selective activation of a1-adrenoceptors facilitates inhibitory transmission via stimulating inhibitory interneurons. Some of these interneurons may have a regular discharge activity similar to pacemaker-like neurons suggesting that the latters could be responsible for the regular pattern of inhibitory PSPs. Finally, using RVL punches, we demonstrate that these regularly discharging interneurons have their soma located within the RVL.

a1- and a2 -adrenoceptor-mediated effects

   There were no apparent differences between silent and pacemaker-like neurons regarding their responsiveness to a1 and a2-adrenoceptors agonists. The two adrenoceptor subtypes were found to mediate opposing effects with possible coexistence of both of them on the same neuron. This phenomenon is not unique to the RVL since multiple effects of NA on different neurons and also on the same neuron have been reported for sympathetic preganglionic neurons as well as many other neurons in the brain like hippocampal pyramidal neurons and granule cells of the dendate gyrus (for reviews, see Refs 53, 9).
   On the other hand, we did not obtain any evidence for the existence of functional b-adrenoceptors in the RVL. We are unable to explain the discrepancy between our results and those of Sun and Guyenet46 who found that in vitro, RVL pacemaker neurons were activated by isoproterenol (10 µM) and were unaffected by phenylephrine (up to 100 µM). However, only few intracellular recordings were performed in their study; therefore, there may be some doubt about the type of neurons they investigated. Moreover, in the same laboratory later, Allen and Guyenet3 found that iontophoretic application of isoproterenol in the RVL in vivo produced no effect on any of the cells studied.
   In our study, the depolarization evoked by NA is probably due to activation of a1-adrenoceptors since it was blocked in all cases by the a1-adrenoceptor antagonist prazosin and was mimicked by the a1-adrenoceptor agonist PE and not by the b-adrenoceptor agonist isoproterenol. The associated increase in Rin, which could be due to a decrease in potassium conductance, further supports the involvement of a1-adrenoceptors in mediating the response as has been reported similarly for sympathetic preganglionic neurons53 and dorsal raphe neurons.37 This is supported by the fact that we determined an extrapolated reversal potential of the NA depolarizing response at -100 mV which is close to the equilibrium potential of K+ ions. The NA- and PE-induced depolarizations seem to be a postsynaptic mechanism since they persisted in the presence of TTX. The observation that TTX prevented the increase in PSPs in response to NA and PE also supports the hypothesis that a1-adrenoceptors are localized on the soma of interneurons rather than on the nerve terminals.
   To our knowledge, there has been no clear demonstration so far of a specific role of a1-adrenoceptors in the rat RVL itself. When injected in the RVL, NA and adrenaline have been found to lower arterial pressure and heart rate while the a1-adrenoceptor agonist PE which is also a weak a2-adrenoceptor agonist produced a small but significant pressor effect at low doses and elicited a slight fall in arterial pressure at higher doses.13 The potent a1-adrenoceptor agonists cirazoline and ST 587 produced dose-dependent hypotensive effects when injected in the nucleus reticularis lateralis of the cat.8 However, it is possible that the effect of theses substances resulted from the activation of distinct receptor types with different affinity when using the microinjection technique. Nevertheless, the attribution of a sympathoinhibitory role for the a1-adrenoceptors in the ventrolateral medulla is supported by a recent finding that NA injection in the ventrolateral depressor area (containing mainly GABAergic neurons projecting to the RVL), produced a depressor and bradycardic response that was blocked by prazosin.31
   In the presence of prazosin, a large proportion of RVL neurons was hyperpolarized by NA (30 -300 µM) and this effect was associated with a decrease in Rin. The latter observation could explain why no significant change in Rin was obtained with high concentrations of NA in control conditions and suggests that probably an a2-adrenoceptor-mediated decrease in Rin at relatively high concentrations of NA could counteract the increase in Rin induced by a1-adrenoceptors activation. The existence of an a2-adrenoceptor-mediated response would suggest that some of these neurons might belong to the C1 adrenergic group since immunoreactivity for a2A-adrenergic receptors has been detected in almost all the PNMT immunoreactive cells of the RVL.41 Consistent with this suggestion is a recent study demonstrating PNMT-immunoreactivity in a significant proportion of both silent and pacemaker RVL neurons.22 This would indicate that pacemaker or adrenergic neurons are not necessarily a functionally homogeneous population.
   The NA hyperpolarizing effect was only partially reduced by yohimbine and was small in amplitude. It is possible that this resulted from a high level of uptake of NA since the amplitude of the hyperpolarization was increased in the presence of desipramine. Moreover, the a2-adrenoceptor agonists UK-14,304 and clonidine had modest but significant effects on the membrane properties of RVL neurons. These substances generally suppressed the discharge of pacemaker neurons while only slightly hyperpolarizing the membrane potential of some pacemaker and silent neurons. Only high concentrations of clonidine (10-30 µM) were found to be able to silence RVL pacemaker neurons in a similar in vitro study.47 It is possible that in addition to activating a2-adrenoceptors, there might be a possible non-selective activation of dopamine D2 receptors by NA as has been reported in dopamine midbrain neurons.17 However, this possibility was not addressed in this study.

Pharmacological characterization of the receptors mediating the NA-induced enhancement of synaptic activity

   Monoamines-induced increases in inhibitory PSPs have been reported in many areas of the central nervous system. NA increased GABA-mediated IPSPs in the substantia gelatinosa of guinea-pig spinal trigeminal nucleus,18 in the pyramidal cells in piriform cortex,15 and in CA1 pyramidal neurons of the hippocampus.30 PE increased GABA-mediated IPSPs in midbrain dopamine neurons.17 Adrenaline elicited glycine-mediated IPSPs in sympathetic preganglionic neurons.35
   Our results suggest that at least a part of the neurons depolarized by NA or PE via a1-adrenoceptors are glycinergic interneurons - probably located within the RVL - whose excitation results in the increase in strychnine-sensitive PSPs. Furthermore, TTX blocked the NA and PE-induced increase in PSPs exhibited by RVL neurons in the coronal slice as well as in RVL punches, suggesting that the excited neurons are localed within the RVL. Additional contribution of glycine-projecting neurons to the RVL from adjacent nuclei in the slice like the nucleus tractus solitarius or the raphe nuclei cannot be excluded since the RVL was shown to have bilateral connections with many rostral ventral medullary nuclei.54 However, anatomical substrate supporting the presence of glycine RVL interneurons or glycinergic neurons projecting to the RVL is still lacking. Nevertheless, moderate levels of binding of 3H-strychnine have been demonstrated in the RVL and the paragigantocellular nucleus52 and the glycine transmitter transporter 2 mRNA is moderately expressed in the gigantocellular reticular nucleus.29
   The regular pacemaker neurons recorded in this study were probably responsible for the action of NA or PE on the regular glycine-mediated PSPs. This is supported by our finding that NA or PE depolarized the regular pacemaker neurons and increased their firing rate. Moreover, PSPs occurring at regular intervals and characterizing pacemaker activity were in most cases inhibited by strychnine at a concentration of 1 µM whereas bicuculline (30 µM) increased their frequency and reduced their amplitude only at higher concentrations (100-150 µM). On the other hand, bicuculline-sensitive PSPs were either slightly increased or not affected by NA or PE and they usually had an irregular pattern. One possible explanation is that most of them are miniature events caused by spontaneous release of GABA from nerve terminals and that only action potential-dependent events are affected by the a1-adrenoceptor agonists.
   The exact mechanism involved in the bicuculline-induced depolarization remains to be established. This depolarization could not be interpreted as a nonspecific effect of bicuculline since it was mimicked by picrotoxin but not by strychnine. Therefore, this observation supports the presence of a tonic GABAergic inhibition within the RVL. Interestingly, the present study provides evidence for a relatively higher contribution of glycine versus GABA receptors in mediating the inhibitory transmission at least in the neurons investigated. Diffusely scattered terminals and numerous perikarya containing the synthetic enzyme for GABA, glutamic acid decarboxylase (GAD), have been identified immunocytochemically within the RVL.42,33 In the ventral horn of the rat spinal cord, all GAD-positive terminals were found to be in direct apposition with glycine receptors.51
   On the other hand, GABA, glycine and hybrid GABA/glycine receptors appeared to mediate the responses to GABA and glycine applications in embryonic cultured rat medullary neurons with all three receptors sensitive to block by both glycinergic and GABAergic antagonists.25 It is tempting to suggest that GABA either alone or in combination with glycine may be endogenously released to activate strychnine-sensitive glycinergic receptors. It is also possible that GABAergic neurons innervating the RVL are predominently located in the caudal ventrolateral medulla (a region not included in our slice preparation), while the majority of RVL inhibitory interneurons are glycinergic. Alternatively, GABAA receptors synapses may be located electrically distal to the intrasomatic electrode, a location that could make bicuculline-sensitive PSPs undetected. On the other hand, the regular pattern of inhibitory glycinergic PSPs does not seem to result from a polysynaptic pathway involving pacemaker excitatory neurons. First, when using KAc-filled electrodes, depolarizing PSPs were almost never observed at resting membrane potential suggesting that there were few excitatory PSPs. Second, combined application of the excitatory amino acids receptors blockers, CNQX and APV, did not affect the regular TTX-sensitive PSPs in the RVL punches. This indicates the predominant inhibitory nature of the synaptic events and suggests, in addition, that they are unlikely due to a polysynaptic mechanism involving excitatory spontaneously active interneurons. This is supported by the observation that injection in the RVL of non-selective tracers labeled numerous neurons around the injection sites whereas [3H] D-aspartate injections labeled only few neurons in the RVL,44 suggesting that the majority of interneurons in the RVL are inhibitory.

Functional considerations

   Taken together, our data suggest that more than one mechanism may be operative in mediating the sympatholytic action of catecholamines when injected in the RVL. This could be accomplished by a direct inhibition via a2-adrenoceptors-mediated hyperpolarization or by facilitating inhibitory synaptic transmission via a1-adrenoceptors located on the soma of inhibitory interneurons. However, we cannot ascertain that these direct and indirect inhibitory actions produce sympatholytic actions in physiological conditions since the sympathoexcitatory nature of the responding neurons cannot be demonstrated in our in vitro experiments.
   Several lines of evidence suggest that the source of catecholamines released in the RVL is endogenous. Blessing et al.6 proposed that neurons in the area postrema are the only catecholamine-synthetizing cells in the medulla and pons with projections to the RVL (C1-area) in the rabbit. This means that neither the A5 neurons45 nor the A1 neurons of the caudal ventrolateral medulla16 project to the RVL (C1-region) as has been suggested by previous studies. On the other hand, Cunningham et al.10 found that the axons of the area postrema that traverse the region of the C1-cells group make few arborizations suggesting little functional contact. Therefore, it is most likely that the adrenergic receptors in the RVL are apposed to nerve terminals or dendrite of local adrenergic neurons.
   Furthermore, our results support the presence of glycinergic regularly firing interneurons within the RVL in the in vitro slice preparation. This hypothesis is supported by the presence of strychnine and TTX-sensitive PSPs with regular frequency in the majority of RVL neurons recorded in the coronal slice as well as in punches containing only the RVL region. It is also interesting to notice that the mean firing rate of the pacemaker neurons and the mean frequency of the PSPs were not significantly different and both were increased by PE and NA by activating a1-adrenoceptors.
   The RVL is the original ‘glycine sensitive area’ described by Guertzenstein and Silver.19 Glycine decreases the blood pressure when microinjected in the RVL of the rat; however, strychnine is without effect.4 Glycine has been shown to mediate baroreceptor inhibition of sympathetic activity at a spinal site.26 However, GABAergic but not glycinergic receptors were shown to mediate the depressor response evoked by activation of the caudal ventrolateral medulla.7 Nevertheless, detection of glycine release in the ventrolateral medulla by the microdialysis technique in vivo provides evidence for a physiologically relevant role of glycine in this region of the brain.23 The involvement of glycine in the cardiovascular regulation has recently been proposed by Moriguchi et al.,36 who found that the attenuation of the baroreflex by angiotensin II was associated with an augmented release of this amino acid in the RVL. Moreover, an increase of glycine release from ventral medullary surface occurred after baroreceptor stimulation.12 On the other hand, glycine-mediated inhibition was found to be vital for developmental maturation of the respiratory network within the ventrolateral medulla.38 However, glycine might not be as important as GABA in mediating the tonic sympathoinhibition within the RVL as shown by previous in vivo studies.4 Therefore, it is possible that the majority of the neurons investigated in our study, that receive a tonic glycinergic input, might be in fact propriobulbar neurons. This is supported by the finding that the reticulospinal projection is represented by a remarkably small number of RVL neurons, around 200 in the rat.43
   The silent and pacemaker-like RVL neurons could not be distinguished according to the membrane electrophysiological and pharmacological properties examined in this study. Therefore, it is possible that these neurons belong to a functionally homogeneous population but have different levels of membrane excitability ranging from completely silent, to oscillating, to spontaneously active. Since some neurons were capable of generating a rhythmic pattern of discharge when appropriately depolarized or when increasing the bathing extracellular K+ ions concentration to a higher level or in the presence of NA, it is difficult to determine whether spontaneously active neurons are true or ‘conditional pacemaker’ (for discussion, see Ref 20).
   On the other hand, we have previously characterized in the RVL a group of irregularly firing neurons with distinct electrophysiological properties, namely a relatively low discharge rate and longer spike duration and the presence of an inward rectifying current.21 Recently, only RVL neurons having a relatively slow and irregular firing were found to be inhibited significantly by clonidine in vivo2 and by the a2-agonist UK-14,304 in vitro.28 Unpublished data by us also indicate that the majority of irregular firing neurons were hyperpolarized by NA, suggesting that they have predominantly a2-adrenoceptors.
   Even though the neuronal circuit mediating the responses to catecholaminergic agonists might be very complex, we propose a simple hypothetical model for this circuit (Fig. 12). We suggest the existence of at least 3 types of RVL neurons: silent (50%), pacemaker-like (25%), and irregular firing (25%) based on our previous study. 21 Silent and pacemaker-like neurons could be in majority inhibitory interneurons containing essentially glycine whereas irregular firing neurons could be the putative C1 adrenergic neurons. However, glutamate might be the fast neurotransmitter colocalized with catecholamines and released in the spinal cord by the adrenergic neurons. Even though the excitatory spinal projection constitutes the greatest proportion of bulbospinal RVL neurons, it is not excluded that some local inhibitory neurons might also project spinally to inhibit sympathetic preganglionic neurons. a1- and a2-adrenoceptors are predominantly located on pacemaker-like and irregular firing neurons, respectively. The excitability of C1 adrenergic neurons could be limited by inhibitory somatic a2-adrenergic autoreceptors. NA might be also released by these neurons to stimulate a1-adrenoceptors located on glycinergic interneurons which, in turn, inhibit all types of RVL neurons. Finally, we would like to mention that this model is still highly speculative and should be considered carefully until direct evidence for receptor distribution and cell connectivity is provided in future studies.

   Further investigation is required to clarify whether the RVL pacemaker cells recorded in vitro are themselves responsible for the regular strychnine-sensitive PSPs. It seems unlikely that they correspond to the fast pacemaker bulbospinal cells recorded in vivo which are inhibited by baroreceptor activation since only a sympathoexcitatory function has been attributed to such neurons.3 Alternatively, it is possible that at least some of the RVL pacemaker neurons belong to as yet not clearly identified small population of neurons subserving a sympathoinhibitory function like the 12 % of bulbospinal RVL neurons found in the rabbit27 or the 25% of RVL neurons of the decerebrate rat14 that were excited by baroreceptor inputs. So far, there is more than one evidence supporting the existence of RVL neurons with pacemaker-like activity that act locally to provide tonic inhibition of sympathoexcitatory neurons or constitute an inhibitory link in the baroreceptor-vasomotor pathway. In conclusion, the present results underscore the strength and complexity of the inhibitory network in the RVL and suggest that it could be a target to the modulation by catecholaminergic agonists and centrally acting antihypertensive drugs.

1. Aghajanian G. K. and Rasmussen K. (1989) Intracellular studies in the facial nucleus illustrating a simple new method for obtaining viable motoneurons in adult rat brain slices. Synapse 3, 331-338.

2. Allen A. M., Adams J. M. and Guyenet P. G. (1993) Role of the spinal cord in generating the 2- to 6-Hz rhythm in rat sympathetic outflow. Am. J. Physiol. 264, R938-R945.

3. Allen A. M. and Guyenet P. G. (1993) *2-Adrenoceptor-mediated inhibition of bulbospinal barosensitive cells of rat rostral medulla. Am. J. Physiol. 265, R1065-R1075.

4. Amano M. and Kubo T. (1993) Involvement of both GABA-A and GABA-B receptors in tonic inhibitory control of blood-pressure at the rostral ventrolateral medulla of the rat. Naunyn-Schmiedeberg’s Arch. Pharmacol. 348, 146-153.

5. Barman S. M. and Gebber G. L. (1989) Basis for the naturally occurring activity of rostral ventrolateral medullary sympathoexcitatory neurons. Prog. Brain. Res. 81, 117-129.

6. Blessing W. W. and Willoughby J. O. (1987) Neurons in the area postrema are the only catecholamine-synthesizing neurons in medulla or pons with projections to the rostral ventrolateral medulla (C1 area) in the rabbit. Brain Res. 419, 336-340.

7. Blessing W. W. (1988) Depressor neurons in rabbit caudal medulla act via GABA receptors in rostral medulla. Am. J. Physiol. 254, H686-H692.

8. Bousquet P., Feldman J. and Schwartz, J. (1984) Central cardiovascular effects of alpha adrenergic drugs: Differences between catecholamines and imidazolines. J. Pharmacol. Exp. Ther. 230, 232-236.

9. Coote J. H. and Lewis D. I. (1995) Bulbospinal catecholamine neurones and sympathetic pattern generation. J. Physiol. Pharmacol. 43, 251-271.

10. Cunningham E. T., Miselis R. R. and Sawchenko P. E. (1994) The relationship of efferent projections from the area postrema to vagal motor and brain stem catecholamine-containing cell groups- an axonal transport and immunohistochemical study in the rat. Neuroscience 58, 635-548.

11. Czyzewska-Safran H., Jastrzebski Z., Remiszewska M. and Wutkiewicz M. (1991) Effect of clonidine on blood pressure and GABAergic mechanisms in spontaneously hypertensive rats. Eur. J. Pharmacol. 198, 115-120.

12. De Oliveira C. V. R., Assumpcao J. A., Confessor Y. Q., Covalan L. and Cravo S. L. (1995) Glycine release in the rat ventrolateral medulla during baroreceptor stimulation. Soc. Neurosc. Abstr. 21, 890.

13. Ernsberger P., Giuliano R., Willette R. N. and Reis D. J. (1990) Role of imidazole receptors in the vasodepressor response to clonidine analogs in the rostral ventrolateral medulla. J. Pharmacol. Exp. Ther. 253, 408-418.

14. Frugière A. and Barillot J.-C. (1994) The medullary rostral ventrolateral pressor region: an electrophysiological study in decerebrate rat. Neurosci. Lett. 68, 191-194.

15. Gellman R. L. and Aghajanian G. K. (1993) Pyramidal cells in piriform cortex receive a convergence of inputs from monoamine activated GABAergic interneurons. Brain Res. 600, 63-73.

16. Granata A. R., Numao Y., Kumada M. and Reis D. J. (1986) A1 noradrenergic neurons tonically inhibit sympathoexcitatory neurons of C1 area in rat brainstem. Brain Res. 377, 127-146.

17. Grenhoff J., North R. A. and Johnson S. W. (1995) Alpha1-adrenergic effects on dopamine neurons recorded intracellularly in the rat midbrain slice. Eur. J. Neurosci. 7, 1707-1713.

18. Grudt T. J., Williams J. T. and Travagli R. A. (1995) Inhibition by 5-hydroxytryptamine and noradrenaline in substantia gelatinosa of guinea-pig spinal trigeminal nucleus. J. Physiol. 485, 113-120.

19. Guertzenstein P. G. and Silver A. (1974) Fall in blood pressure produced from discrete regions of the ventral surface of the medulla by glycine and lesions. J. Physiol. 242, 489-503.

20. Guyenet P. G. (1990) Role of the ventral medulla oblongata in blood pressure regulation. In Central Regulation of Autonomic Functions. (eds Loewy A. D. and Spyer K. M.), pp. 145-167. Oxford University Press, New York.

21. Hayar A., Piguet P. and Feltz P. (1996) GABA-induced responses in electrophysiologically characterized neurons within the rostro-ventrolateral medulla in vitro. Brain Res. 709, 173-183.

22. Kangrga I. M. and Loewy A. D. (1995) Whole-cell recordings from visualized C1 adrenergic bulbospinal neurons: ionic mechanisms underlying vasomotor tone. Brain Res. 670, 215-232.

23. Kapoor V., Nakahara D., Blood R. J. and Chalmers J. P. (1990) Preferential release of neuroactive amino acids from the ventrolateral medulla of the rat in vivo as measured by microdialysis. Neuroscience 37, 187-191.

24. Kawasaki H. and Takasaki K. (1986) Central alpha-2 adrenoceptor mediated hypertensive response to clonidine in conscious, normotensive rats. J. Pharmacol. Exp. Ther. 236, 810-818.

25. Lewis C. A. and Faber D. S. (1993) GABA responses and their partial occlusion by glycine in cultured rat medullary neurons. Neuroscience 52, 83-96.

26. Lewis D. L. and Coote J. H. (1994) Mediation of baroreceptor inhibition of sympathetic nerve activity via both a brainstem and spinal site in rats. J. Physiol. 481, 197-208.

27. Li Y.-W., Gieroba Z. J., McAllen R. M. and Blessing W. W. (1991) Neurons in rabbit caudal ventrolateral medulla inhibit bulbospinal barosensitive neurons in rostral medulla. Am. J. Physiol. 261, R44-R51.

28. Li Y.-W. and Guyenet P. G. (1995) Neuronal excitation by angiotensin II in the rostral ventrolateral medulla of the rat in vitro. Am. J. Physiol. 268, R272-R277.

28a. Lipski J., Kanjhan R., Kruszewska B. and Rong W. (1996) Properties of presympathetic neurones in the rostral ventrolateral medulla in the rat: an intracellular study ‘in vivo’. J. Physiol. 490, 729-744.

28b. Lipski J., Kanjhan R., Kruszewska B. and Smith M. (1995) Barosensitive neurons in the rostral ventrolateral medulla of the rat in vivo: Morphological properties and relationship to C1 adrenergic neurons. Neuroscience 69, 601-618.

29. Luque J. M., Nelson N. and Richards J. G. (1995) Cellular expression of glycine transporter 2 messenger RNA exclusively in rat hindbrain and spinal cord. Neuroscience 64, 525-535.

30. Madison D. V. and Nicoll R. A. (1988) Norepinephrine decreases synaptic inhibition in the rat hippocampus. Brain Res. 442, 131-138.

31. Malhotra V. and Sapru H. N. (1994) Decrease in blood pressure and heart rate (HR) elicited by injections of glutamate (Glu) into the locus coeruleus (LC) are mediated by a1-adrenergic receptors in the nucleus ambiguus/ventrolateral medullary depressor area (NA/VLDA). Soc. Neurosc. Abstr. 20, 1181.

32. Malhotra V. K., Kachroo A. and Sapru H. N. (1993) Cardiac effects of injections of epinephrine into the spinal intermediolateral column. Am. J. Physiol. 265, H633-H641.

33. Meeley M. P., Ruggiero D. A., Ishitsuka T. and Reis D. J. (1985) Intrinsic *-aminobutyric acid in the nucleus of the solitary tract and the rostral ventrolateral medulla of the rat: an immunocytochemical and biochemical study. Neurosci. Lett. 58, 83-89.

34. Milner T. A., Pickel V. M., Park D. H., Joh T. H. and Reis D. J. (1987) Phenylethanolamine N-methyltransferase-containing neurons in the rostral ventrolateral medulla of the rat. I. Normal ultrastructure. Brain Res. 411, 28-45.

35. Miyazaki T., Coote J. H. and Dun N. J. (1989) Excitatory and inhibitory effects of epinephrine on neonatal rat sympathetic preganglionic neurons in vitro. Brain Res. 497, 108-116.

36. Moriguchi A., Mikami H., Otsuka A., Katahira K., Kohara K. and Ogihara T. (1995) Amino acids in the medulla oblongata contribute to baroreflex modulation by angiotensin II. Brain Res. Bull. 36, 85-89.

37. Pan Z. Z., Grudt T. J. and Williams J. T. (1994) a1-Adrenoceptors in rat dorsal raphe neurons: regulation of two potassium conductances. J. Physiol. 478, 437-447.

38. Paton J. F. R., Ramirez J.-M. and Richter D. W. (1994) Mechanisms of respiratory rhythm generation change profoundly during early life in mice and rats. Neurosci. Lett. 170, 167-170.

39. Reis D. J., Colanov E. V., Ruggiero D. A. and Sun M.-K. (1994) Sympatho-excitatory neurons of the rostral ventrolateral medulla are oxygen sensors and essential elements in the tonic and reflex control of the systemic and cerebral circulations. J. Hypertens. Suppl. 12, S159-S180.

40. Richerson G. B. and Messer C. (1995) Effect of composition of experimental solutions on neuronal survival during rat brain slicing. Expl. Neurol. 131, 133-143.

41. Rosin D. L., Zeng D., Stornetta R. L., Norton F. R., Riley T., Okusa M. D., Guyenet P. G. and Lynch K. R. (1993) Immunohistochemical localization of a2-adrenergic receptors in catecholaminergic and other brainstem neurons in the rat. Neuroscience 56, 149-155.

42. Ruggiero D. A., Meely M. P., Anwar M. and Reis D. J. (1985) Newly identified GABAergic neurons in regions of the ventrolateral medulla which regulate blood pressure. Brain Res. 339, 171-177.

43. Ruggiero D. A., Cravo S. L., Golanov E., Gomez R., Anwar M. and Reis D. J. (1994) Adrenergic and non-adrenergic spinal projections of a cardiovascular-active pressor area of medulla oblangata: quantitative topographic analysis. Brain Res. 663, 107-120.

44. Somogyi P., Minson J. B., Morilak D., Llewellyn-Smith I., McIlhinney J. R. A. and Chalmers J. (1989) Evidence for an excitatory amino acid pathway in the brainstem and for its involvement in cardiovascular control. Brain Res. 496, 401-407.

45. Sun M.-K. and Guyenet P. G. (1986) Effect of clonidine and *-aminobutyric acid on the discharges of medullo-spinal sympathoexcitatory neurons in the rat. Brain Res. 368, 1-17.

46. Sun M.-K. and Guyenet P. G. (1990) Excitation of rostral medullary pacemaker neurons with putative sympathoexcitatory function by cyclic AMP and *-adrenoceptor agonists 'in vitro'. Brain Res. 511, 30-40.

47. Sun M.-K. and Reis D. J. (1995) GABA-mediated inhibition of pacemaker neurons of rostral ventrolateral medulla by clonidine in vitro. Eur. J. Pharmacol. 276, 291-296.

48. Sun M.-K., Young B. S., Hackett J. T. and Guyenet P. G. (1988) Reticulospinal pacemaker neurons of the rat rostral ventrolateral medulla with putative sympathoexcitatory function: an intracellular study in vitro. Brain Res. 442, 229-239.

49. Sved A. F. (1989) PNMT-containing catecholaminergic neurons are not necessarily adrenergic. Brain Res. 481, 113-118.

50. Sved A. F. (1990) Effect of monoamine oxidase inhibition on catecholamine levels: evidence for synthesis but not storage of epinephrine in rat spinal cord. Brain Res. 512, 253-258.

51. Triller A., Cluzeau F. and Korn H. (1987) Gamma-aminobutyric acid-containing terminals can be apposed to glycine receptors at central synapses. J. Cell. Biol. 104, 947-956.

52. White W. F., O’gorman S. and Roe A. W. (1990) Three-dimensional autoradiographic localization of quench-corrected glycine receptor specific activity in the mouse brain using 3H-strychnine as the ligand. J. Neurosci. 10, 795-813.

53. Yoshimura M., Polosa C. and Nishi S. (1989) Multiple actions of noradrenaline on sympathetic preganglionic neurons of the cat studied in the spinal cord slice. Prog. Brain Res. 81, 181-190.

54. Zagon A. (1995) Internal connections in the rostral ventromedial medulla of the rat. J. auton. nerv. Syst. 53, 43-56.

55. Zhong S., Huang Z.-S., Gebber G. L. and Barman S. M. (1993) Role of the brain stem in generating the 2- to 6-Hz oscillation in sympathetic nerve discharge. Am. J. Physiol. 265, R1026-R1035.