INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

THE RETICULAR FORMATION of the caudal ventrolateral pons receives most of its input from brain stem areas that are involved in autonomic regulation (2). Stimulation of the ventrolateral pons in anesthetized rats alters cardiovascular and nociceptive reflexes (1, 7, 20, 26). Some of these effects have been attributed to the activation of a large group of noradrenergic neurons known as the A5 cells present in this region. A5 cells project to brain stem autonomic centers (2), and they are among a very small number of brain stem neurons that establish monosynaptic connections with vasomotor sympathetic preganglionic neurons (9, 10, 28). In addition, unit recording data indicate that the activity of the spinally projecting neurons of the A5 region is strongly influenced by afferent inputs from the cardiopulmonary region, including arterial baroreceptors and peripheral chemoreceptors (6, 8).

The cellular properties of A5 neurons have not been examined yet. The present study was designed to start filling this gap. The first objective was to determine whether A5 neurons possess characteristic electrophysiological properties that would permit their identification in vitro without having to resort to post hoc histology. The second main objective was to determine whether A5 neurons are autoactive like several other cell groups involved in sympathetic tone generation and, if so, to try and identify some of the underlying mechanisms.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Whole cell recordings in the neonate slice. This work was carried out in thin slices from neonate rat brain, material that presents the advantage that individual cells can be visualized before whole cell recording (4). All recordings were obtained from cells retrogradely labeled with a fluorescent tracer injected into the spinal cord. This procedure was used to tag a cell population highly enriched in A5 noradrenergic neurons. The methods have been described in detail in previous papers dedicated to the study of C1 adrenergic neurons (16, 17). Briefly, Sprague-Dawley rat pups (3 days old) were anesthetized by deep hypothermia (24), and a suspension of fluorescein isothiocyanate (FITC)-labeled microspheres (0.3-0.5 µl; Molecular Probes) was injected bilaterally into the upper thoracic spinal cord. Two to seven days later, the pups (5-10 days old) were deeply anesthetized by hypothermia and decapitated. The brain stem was blocked and immersed in a sucrose-artificial cerebrospinal fluid (aCSF) mixture composed of (in mM) 26 NaHCO₃, 1 NaH₂PO₄, 5 KCl, 5 MgSO₄, 0.5 CaCl₂, 10 glucose, and 248 sucrose and equilibrated with 95% O₂-5% CO₂ (pH 7.3). Coronal slices (130 µm thick) were cut with a microslicer (Dosaka), preincubated at 35°C for 30 min, and then brought to room temperature (22°C) in a lactic acid-aCSF mixture composed of (in mM) 124 NaCl, 26 NaHCO₃, 5 KCl, 1 NaH₂PO₄, 2 MgSO₄, 2 CaCl₂, 10 glucose, and 4.5 lactic acid and equilibrated with 95% O₂-5% CO₂ (pH 7.3-7.4). For recording, a single pontine slice was placed in a recording chamber on an upright, epifluorescence microscope (Olympus BH-2). Slices located in immediate proximity to the exit point of the facial nerves were selected for recording. Slices containing the A5 region were identified under a ×10 objective by their characteristic pattern of retrograde labeling (for details, see Pattern of retrograde labeling in the ventrolateral pons of the neonatal rat). In this chamber, the slice was continuously superfused at the rate of 2-3 ml/min with normal aCSF equilibrated with 95% O₂-5% CO₂ (pH 7.4; composition identical to lactic acid-aCSF without the lactate). The reduced-Ca²⁺ aCSF used in some experiments contained 0.1 mM CaCl₂ and 5 mM MgSO₄. In other experiments, NaCl (124 mM) was replaced by an equimolar concentration of N-methyl-D-glucamine (NMDG) titrated with HCl. All experiments were performed at room temperature. Individual retrogradely labeled neurons were visualized with a water-immersion ×40 objective via epifluorescence and Hoffman modulation optics. Patch pipettes were pulled from borosilicate glass capillaries (1.5 mm OD; Clark) on a pipette puller (Sutter P87), coated with Sigmacote (Sigma Chemical, St. Louis, MO), and filled with a solution of the following composition (in mM): 114 K gluconate, 17.5 KCl, 4 NaCl, 4 MgCl₂, 10 N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid, 0.2 ethylene glycol-bis(-aminoethyl ether)-N,N,N',N'-tetraacetic acid, 3 Mg₂ATP, and 0.3 Na₂GTP and 0.02% lucifer yellow (Molecular Probes). Osmolarity was adjusted to 270 mosmol, and the pH was adjusted to 7.3. Electrode resistance was 5-7 M. Whole cell current- and voltage-clamp recordings were made with an Axoclamp-2A amplifier. The liquid junction potential was measured (8-12 mV), and all reported voltage measurements have been corrected for this potential. Series resistance compensation was not employed because the recorded currents were small enough (<100 pA) that voltage errors due to series resistance should have been negligible. Current and voltage data were collected through a DigiData-1200 interface with pCLAMP software version 6.0 (Axon Instruments) and were stored on videotape for off-line analysis. Power spectrum density analysis was used to analyze membrane potential oscillations. In this case, the membrane potential was sampled every 10 ms, and the power spectra represented averages of ten 20-s segments. The SD of the membrane potential during a 100-s segment (sampling every 10 ms) was also used to quantify membrane potential variability. In this case, the signal was filtered from 0.1 to 50 Hz because power spectrum density analysis revealed that the main density was distributed within this range.

Drugs and solutions of different ionic content were applied to the slice by switching the perfusion solution via a three-way electronic valve system. Time to onset of drug action was ~30 s.

After the recording was made, every slice was fixed in freshly prepared 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.3). Immunostaining for tyrosine hydroxylase (TH) was done with an avidin-biotin-based reaction (mouse anti-TH monoclonal antiserum from Chemicon, dilution 1:750; biotinylated goat anti-mouse antiserum from Vector, 1:150 dilution; avidin-conjugated Texas red from Molecular Probes, 1:200 dilution). Computer-assisted mapping of the location of retrogradely labeled, TH-immunoreactive (TH-ir), and/or lucifer yellow-stained neurons was done with a Ludl motor-driven stage and Neurolucida Software (MicroBrightfield, Colchester, VT). The atlas of Paxinos and Watson (23) was used for reference and nomenclature.

Drugs and chemicals. Tetrodotoxin (TTX), kynurenic acid, bicuculline methiodide, strychnine HCl, and NMDG were obtained from Sigma Chemical (St. Louis, MO).

Statistics. Results are expressed as means ± SE. Data were analyzed by either paired t-tests or analysis of variance. Significance was set at P < 0.05.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Pattern of retrograde labeling in the ventrolateral pons of the neonatal rat. These anatomic experiments were designed to determine the pattern of retrograde labeling of the ventrolateral pontine reticular formation in the neonate after injection of FITC-labeled microbeads into the thoracic spinal cord. The tracer was injected bilaterally into the upper thoracic spinal cord of three neonatal rats (3 days old). Two days later, serial 40-µm-thick coronal sections were cut with a vibratome throughout the rostral medulla and pons. A one-in-three series was processed for the immunofluorescent detection of TH, and adjacent sections were Nissl stained to add anatomic accuracy. As shown in Fig. 1 (left), the caudal pons is well differentiated at this early age. The distribution of spinally projecting neurons (with and without TH-ir) was mapped with the assistance of a computer. The number of each type of cell was then counted within a fixed-size rectangular window (0.58 × 0.53 mm) that was placed medial to the exiting root of the facial nerve over the area that contained the highest concentration of retrogradely labeled cells (Fig. 1). In this window, we counted 159 ± 18 TH-ir cells · animal¹ · side¹, 63.5% of which (101 ± 10 cells) were retrogradely labeled with microbeads (examples of single and dual labeling are shown in Fig. 2). In the same area, 66.9% of all retrogradely labeled cells counted (100 ± 12 of 151 ± 8 cells · side¹ · rat¹) were TH-ir. Note (Fig. 1, right, insets) that the proportion of TH-ir cells among retrogradely labeled neurons was highest laterally, in the immediate proximity of the ventral surface.

View larger version (66K): [in this window] [in a new window]   Fig. 1.   Pattern of retrograde labeling in caudal pons of neonate rats. Left: Nissl-stained coronal sections at 3 representative levels (A to C, caudal to rostral). Right: pattern of retrograde labeling after injection of fluorescent microbeads into upper thoracic spinal cord (same 3 levels as on left). Insets: standardized rectangular windows used to obtain cell counts. , Microbeads only; , tyrosine hydroxylase (TH)-immunoreactive (TH-ir) cells devoid of microbeads; , retrogradely labeled TH-ir cells. In locus ceruleus (lc; B), only retrogradely labeled TH-ir cells have been represented. cgp, Central gray matter of pons; co, cochlear nucleus; fn, facial nerve; ic, inferior colliculus; ml, median lemniscus; Mo5, trigeminal motor nucleus; pb, parabrachial nucleus; Pr5, principal sensory trigeminal nucleus; rpo, rostral periolivary region; so, superior olive; tz, trapezoid nucleus. Bar, 0.5 mm.

View larger version (133K): [in this window] [in a new window]   Fig. 2.   Retrograde labeling in coronal section (thickness 40 µm) of ventrolateral pons. A: TH-ir. B: fluorescent microbeads. Arrows, retrogradely labeled TH-ir neurons. * Retrogradely labeled cells devoid of TH-ir. Bar, 100 µm.

Location and phenotype of the recorded neurons. Whole cell recordings were obtained from a total of 85 spinally projecting neurons recorded in the region of the ventrolateral pons outlined in Fig. 1, insets. Sixty-three lucifer yellow-stained neurons were recovered after histology. All contained FITC-labeled microbeads. Sixty percent of the recovered cells (38 of 63) were TH-ir, i.e., A5 cells. An example of one recorded cell identified as an A5 neuron is shown in Fig. 3. The presence of FITC-labeled microbeads did not interfere with the detection of lucifer yellow (the two A5 cells shown in Fig. 3 had an equivalent amount of microbead labeling). Also, as illustrated in Fig. 3, the presence of the two latter markers did not interfere noticeably with the detection of TH-ir, which is especially intense in A5 neurons. The location of all recovered cells was mapped by computer and replotted on three standardized coronal sections and is represented in Fig. 4.

View larger version (70K): [in this window] [in a new window]   Fig. 3.   Recorded A5 neuron. A: lucifer yellow-labeled neuron (ultraviolet incident light, fluorescein isothiocyanate filter). B: same field under green light illumination to reveal TH-ir (Texas red filter). Note presence of a 2nd spinally projecting TH-ir cell that was not recorded. C: higher power photograph of recorded cell under the same illumination condition as in A to reveal microbeads. Bars: 25 µm in A and B; 10 µm in C.

View larger version (26K): [in this window] [in a new window]   Fig. 4.   Computer-assisted map of location of recorded cells. Location of 59 of the histologically recovered neurons (34 A5 cells on right, 25 non-TH-ir cells on left) is plotted at 3 representative levels. Sp50, spinal trigeminal nucleus, oral; 7n, facial nerve; VCA, ventral cochlear nucleus; LSO, lateral superior olive; Me5, mesencephalic trigeminal nucleus.

General electrophysiological properties of spinally projecting neurons in the ventrolateral pons. Monoaminergic neurons often exhibit an electrophysiological signature that is sufficiently distinct from that of other types of surrounding neurons to render post hoc histological identification unnecessary. Accordingly, the first objective of the present study was to determine whether A5 cells can be distinguished from the other spinally projecting neurons of the ventrolateral pons. However, as summarized in Table 1, both types of spinally projecting cells had similar general characteristics (discharge rate, mean interspike membrane potentials, conductance, spike amplitude measured from threshold, and width measured halfway between threshold and apex). The action potential threshold was close to 47 mV. All cells included in this study had action potential overshoots of at least +10 mV (+21 mV on average). The majority of the recorded cells (69 of 85) had a low level of spontaneous activity. This was true regardless of phenotype (79% of A5 cells, 88% of non-A5 cells; an example of an A5 neuron is shown in Fig. 5A). Slower cells fired irregularly and exhibited slow membrane oscillations. Faster cells (>2 Hz) had a more regular discharge pattern (Fig. 5A). Injection of a hyperpolarizing current converted a regular firing pattern into an irregular one with slow oscillations (Fig. 5A). At rest, each pattern could be found in both TH-ir and nonnoradrenergic spinally projecting cells of the ventrolateral pons. Spontaneous action potentials were followed by 6-12 mV afterhyperpolarizations lasting for 100-250 ms and were best observed in slowly firing or silent cells. Afterhyperpolarizations merged into slow depolarizations in cells with firing rates of >2 spikes/s (Fig. 5A). Bursts of action potentials produced by injection of depolarizing current (1 s duration) were followed by a long-lasting hyperpolarization of relatively modest amplitude (<5 mV; Fig. 5A). In current clamp, inward rectification was typically observed regardless of cell type (e.g., A5 cell in Fig. 5B), and in voltage clamp, a slowly developing inward relaxation [hyperpolarization-activated current (I_h)-like current] was produced when the membrane potential was clamped below 80 mV (e.g., A5 cell in Fig. 5, C and D). A variable and generally low frequency of postsynaptic potentials (PSPs) was present, which, in some cases, contributed to the generation of action potentials. Figure 5E represents a case where PSP frequency was above average. One action potential was triggered by a PSP (Fig. 5Ea), and another action potential was triggered from a slow depolarization (Fig. 5Eb). Perfusion with a mixture of synaptic blockers including kynurenic acid (a nonselective glutamate-receptor antagonist; 0.5 mM), bicuculline (a -aminobutyric acid_A-receptor antagonist; 20 µM), and strychnine (a glycine-receptor antagonist; 1 µM) eliminated observable PSPs (assessed during periods when membrane potential was hyperpolarized to about 70 mV; data not illustrated; note that the Cl equilibrium potential equals 35 mV under our experimental conditions). This treatment did not significantly change the average resting discharge rate of nine spontaneously active cells tested (sample includes seven histologically identified A5 cells; 2.5 ± 0.7 spikes/s before drug and 2.4 ± 0.8 spikes/s after 10-min perfusion with the drug mixture), although slight increases or decreases were found in a few cases. Minimal accommodation of firing was found when neuronal discharges up to 8-10 Hz were induced by injection of up to 30 pA of depolarizing current (n = 22 cells, including 18 identified A5 cells; an example of one A5 cell is shown in Fig. 6A). Accordingly, frequency-current curves were almost linear in the range of current tested.

View larger version (25K): [in this window] [in a new window]   Fig. 5.   General properties of A5 cells. A: typical slow discharge pattern of an A5 cell. Note slow irregular discharge pattern and slow membrane oscillations that develop when 5 pA of hyperpolarizing current is injected (middle segment). Excerpt at right, small afterhyperpolarization that follows an induced volley of action potentials. B: depolarizing sag in response to square-wave pulses of hyperpolarizing current (up to 60 pA in 10-pA increments) in histologically identified A5 cell with no tetrodotoxin (TTX) present. C: slowly developing inward current in response to voltage steps (holding potential 50 mV, 10-mV steps) in histologically identified A5 cell with 1 µM TTX. D: current-voltage plot of experiment shown in C. , Instantaneous current; , current measured at end of voltage step. E: A5 cells exhibiting an unusually high frequency of postsynaptic potentials (PSPs). Spike a was triggered from a PSP and spike b from a slow depolarization.

View larger version (12K): [in this window] [in a new window]   Fig. 6.   Accommodation properties of A5 cells. A: response of 1 cell to 3 levels of current injection. B: mean response of 18 histologically identified A5 cells (average firing frequency during depolarizing step). Solid line, linear regression (firing frequency = 0.21 × intensity + 1.76; r2 = 0.986).

Membrane potential oscillations. All slow-firing cells exhibited spontaneous membrane oscillations between action potentials (typical example in Figs. 5A and 7A, top trace), and the same pattern could be produced by injection of a small amount of hyperpolarizing current into more active cells (Fig. 5A). In all cases, when enough negative current was injected to eliminate the action potentials (mean membrane potential below 58 mV), all A5 cells (n = 38) and most others (18 of 22 non-A5 and 14 of 16 unrecovered) showed spontaneous membrane potential oscillations (example of one A5 cell in Fig. 7A, middle trace). These irregular oscillations were characterized by slow rising and falling phases that lasted 100-500 ms with an amplitude of 2-7 mV. These membrane oscillations were voltage dependent and disappeared when enough hyperpolarizing current was injected to bring the membrane potential down to 65 mV or lower (Fig. 7A, bottom trace). The voltage dependency of the oscillations was quantified by measuring the SD of the membrane potential at two different voltage levels (means of 58 and 71 mV) in a group of 11 A5 cells. As indicated in Table 2, the SD was significantly reduced. The voltage dependence of the oscillations was another feature that distinguished these events from PSPs besides kinetic considerations; i.e., PSPs had a much faster rising time and tended to become larger rather than smaller when the membrane was hyperpolarized (Fig. 5E). However, to test more rigorously the possibility that the membrane oscillations might be due to some kind of synaptic activity, we determined the effect produced by perfusing the slice for 10-15 min with a medium containing reduced levels of Ca²⁺ (0.1 mM) and increased Mg²⁺ (5 mM). This treatment failed to reduce the membrane oscillations (Table 2, Fig. 7B). The mean membrane potential at which the oscillations were recorded was the same in the control medium and the low-Ca²⁺ medium (Table 2).

View larger version (21K): [in this window] [in a new window]   Fig. 7.   Slow oscillations are voltage dependent and not due to synaptic activity. A: membrane trajectory of 1 A5 cell at rest (top trace; spikes truncated) and during injection of 2 levels of hyperpolarizing current (middle and bottom traces). B: membrane potential of an A5 cell in control medium (top trace) and during perfusion with low-Ca2+-high-Mg2+ medium (middle and bottom traces; bottom trace during injection of hyperpolarizing current).

Contribution of Na⁺ currents to the membrane oscillations. Three types of experiments were performed to test whether the membrane oscillations were due to a TTX-sensitive (TTX_s) persistent Na⁺ current (I_Na). The first experiment tested the effect of NMDG (extracellular Na substitution) on the magnitude of the oscillations. In the second experiment, we searched for the presence of a persistent voltage-activated TTX_s inward current using a classic voltage-clamp paradigm (19). Finally, we determined whether the membrane oscillations were sensitive to TTX.

View larger version (24K): [in this window] [in a new window]   Fig. 8.   Effect of lowering extracellular Na+ on the membrane oscillations. A: low-Na+ medium [124 mM NaCl substituted by N-methyl-D-glucamine (NMDG) HCl] hyperpolarizes cell (histologically identified A5 neuron) and reduces amplitude of membrane fluctuations even after membrane potential has been restored to control level by current injection (period indicated by b). B: magnified view of membrane fluctuations before, during, and after perfusion with low-Na+ medium. C: power spectral analysis of membrane potential in control state (top line) and in presence of low-Na+ medium (bottom line).

View larger version (17K): [in this window] [in a new window]   Fig. 9.   TTX-sensitive persistent current. A: a, current produced by voltage steps in inset (200 ms; from 65 to 60, 55, and 50 mV); b, current elicited by the same steps after perfusion with 0.1 µM TTX; c, difference current (a minus b); d, effect of Na+ substitution with NMDG (difference current, control minus NMDG). B: typical effect of TTX on membrane oscillations of 1 A5 cell.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

This study is the first to describe the cellular properties of A5 neurons in vitro. Although spinally projecting neurons were selectively recorded, this population is probably a fair representation of the entire A5 cell group given that the majority of A5 neurons (up to 75% in one rat) have an axonal projection to the spinal cord. The present study was done in the neonate because the optical properties of neonatal brain slices permit individual neuronal cell bodies to be seen and patched with relative ease. This is not the case in adult tissue due to the development of myelin. The results of the present study may not be strictly applicable to adult A5 neurons.

General properties of A5 and non-TH-ir spinally projecting neurons. Monoaminergic neurons often exhibit an electrophysiological signature that is sufficiently distinct from that of other types of surrounding neurons to render post hoc histological identification unnecessary. For example, the dopaminergic neurons of the midbrain can be identified by a combination of features that include broad action potentials, spontaneous activity, and a large time- and voltage-dependent inward current [I_h; (5)]. These characteristics have diagnostic value even in thin slices of neonate brain (27). In the present case, we found no obvious electrophysiological difference between A5 neurons and the rest of the surrounding spinally projecting cells except for the fact that A5 cells were selectively responsive to ₂-adrenergic stimulation [see companion paper (6a)].

A5 and non-TH-ir spinally projecting cells had similar discharge rates at rest, a similar input resistance, and equally broad action potentials. Their I_h-like current was unremarkable, and their accommodation properties were the same. It is unlikely that many of the neurons classified as noncatecholaminergic could have been A5 cells exhibiting lower than normal levels of TH-ir. One reason is that the TH-ir of A5 cells is very intense and easy to detect (Fig. 3). The second is that, even under optimal staining conditions (thin sections prepared exclusively for histology), the A5 region contains spinally projecting cells that are clearly not TH-ir (see Figs. 1 and 2). Finally, the cells that were not TH-ir seemed as healthy as the A5 cells (Table 1). On the basis of the results of recent viral retrograde tracing data, it is possible that some of the nonnoradrenergic cells might also innervate sympathetic preganglionic neurons (10).

The general properties of A5 cells were similar to those of the C1 adrenergic neurons previously recorded under similar conditions (16). The resemblance includes all of the basic parameters summarized in Table 1: the minimal accommodation of firing, the generally modest inward rectification, and the presence of slow and irregular membrane oscillations from which the action potentials of the slower firing cells appear to be triggered. Autoactivity in vitro is by no means a unique characteristic of the catecholaminergic neurons and is found in many other brain stem neurons (e.g., Refs. 11, 22).

Autoactivity of A5 neurons. The generally slow and irregular discharges of A5 cells were most likely due to intrinsic cell properties for the following reasons: PSPs were infrequent, and the discharge rate of the cells was not changed significantly by adding a mixture of synaptic blockers that eliminated detectable synaptic activity (kynurenate, strychnine, and bicuculline). In addition, in the slower cells, action potentials appeared to be triggered from slow membrane depolarizations. These membrane depolarizations were probably not due to synaptic activity because they were unaffected by synaptic blockade with reduced extracellular Ca²⁺ (Table 2).

Possible mechanisms underlying the membrane potential oscillations of A5 neurons. The power spectrum of the membrane potential of A5 neurons displayed one or more peaks between 0.5 and 2 Hz (Fig. 8), suggesting that the membrane potential fluctuations were not random. However, the power was rather broadly distributed in the 1- to 8-Hz range, consistent with the fact that the oscillatory behavior of A5 cells is not as regular or as large as in many other types of neurons (e.g., Refs. 12, 18, 21). In other systems, the frequency of neuronal oscillations increases steeply, with the membrane potential in the range of 60 to 50 mV (12, 18). The low frequency found in A5 cells may reflect the fact that the membrane oscillations were examined while the membrane potential was maintained close to 60 mV to prevent action potential generation.

The most common mechanisms of membrane potential oscillations involve some form of voltage-activated I_Na (e.g. Ref. 18) and or a low-threshold Ca²⁺ current (e.g., Refs. 12, 14). Because 1 µM TTX reduced the amplitude of the oscillations significantly (Table 2), a TTX_s voltage-activated I_Na contributes to the phenomenon (30). Voltage-clamp experiments such as the one shown in Fig. 9 suggest that A5 neurons may have a TTX_s persistent inward current that activates close to the mean resting potential of 60 to 55 mV (Fig. 9). This interpretation is based on the assumption that the persistent component of the TTX_s inward current was not an artifact due to a poor space clamp in dendrites (29). The TTX_s persistent I_Na is thought to play a role in the oscillatory behavior of many central nervous system neurons (e.g. Refs. 14, 19). Its origin is attributed to one of three possible mechanisms: a window current due to the overlap between the activation and inactivation properties of the transient Na⁺ channels, a current generated by channels distinct from the latter, and a modal change in the inactivation properties of the transient Na⁺channels (for a review, see Ref. 3).

Because a large part of the subthreshold oscillations of A5 neurons typically persisted in the presence of 1 µM TTX (Fig. 9, Table 2), TTX_s I_Na appears to amplify membrane oscillations caused by other types of conductances. These conductances are also presumably activated by depolarization because the oscillations were severely attenuated by holding the cell soma at or below 65 mV (Fig. 7, Table 2). Two possible candidates are a low-threshold Ca²⁺ conductance or a TTX-insensitive (TTX_r) I_Na (for a review, see Ref. 30). Because all the inward current recorded during a step depolarization to 50 mV was sensitive to 0.1 µM TTX (Fig. 9), this voltage-clamp protocol did not provide evidence for either of the two possibilities. Perhaps this negative result indicates that the voltage-activated conductances responsible for the membrane oscillations reside primarily in distal dendrites. The TTX_r I_Na hypothesis is consistent with the fact that extracellular Na⁺ substitution with NMDG reduced the oscillations to about the same degree as hyperpolarization (Fig. 8, Table 2). Also, the disappearance of the oscillations in low-Na⁺ medium was apparently not due to the membrane hyperpolarization per se because the oscillations were not restored by depolarizing the cells to the original membrane potential (about 59 mV) with a depolarizing current injection (Fig. 8). However, lowering extracellular Na⁺ causes intracellular acidification and a rise in intracellular Ca²⁺ (15, 25). Either of these effects could conceivably alter a variety of conductances that may decrease the ability to observe membrane oscillations.

In summary, the present experiments suggest that the membrane oscillations of A5 neurons are amplified but not caused by TTX_s I_Na. This amplification could be important functionally by causing the oscillations to reach threshold for action potential generation. TTX_s I_Na could be even more important for action potential generation in the more active cells that tend to be slightly more depolarized. A noninactivating form of TTX_s I_Na may also contribute to the autoactivity of C1 cells in the neonate (13). The exact mechanism underlying the TTX_r oscillations of A5 neurons remains uncertain. The role of a TTX_r I_Na is consistent with the data, e.g., attenuation of oscillations both by mild hyperpolarization and by Na⁺ substitution with NMDG (Table 2). However, lowering extracellular Na⁺ with NMDG produces multiple effects unrelated to I_Na reduction (15, 25). Accordingly, the contribution of several other conductances to the generation of the membrane fluctuations, including a low-threshold Ca²⁺ conductance, is not excluded.

Because neonatal A5 cells in vitro have a high input resistance, the opening of a small number of channels may be sufficient to produce the subthreshold membrane potential oscillations observed in the present study. In addition, these membrane oscillations would be less apt to trigger action potentials if the cells had more negative resting potentials. Whether adult A5 neurons have the proper mix of persistent current, input resistance, and resting membrane potential required for spontaneous action potential generation, especially in vivo, needs to be investigated.

Functional significance. Sympathetic vasomotor tone depends on a relatively small number of spinally projecting, monosynaptic, mostly excitatory inputs to sympathetic preganglionic neurons (presympathetic neurons) (9). Contrary to the motoneuronal outputs to skeletal muscles, the sympathetic vasomotor outflow is very resistant to anesthesia. One possible explanation for this resistance is that under anesthesia the discharges of brain stem presympathetic neurons depend minimally on synaptic inputs because the cells have autoactive properties. Previous work (16, 17) has suggested that the C1 adrenergic cells and other nonadrenergic cells of the rostral ventrolateral medulla may have such characteristics. The present work adds the A5 neurons to the list of spinally projecting presympathetic neurons that may have autoactive properties. More work is needed to verify that the autoactivity of A5, C1, and many other medullary cells recorded in the neonate persists in the adult.