INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

THERE IS NOW SUBSTANTIAL evidence that vasomotor sympathetic tone and arterial blood pressure largely depend on the activity of a specific subpopulation of neurons located in the rostral and ventrolateral part of the medulla oblongata (the rostral ventrolateral medulla, RVL; for review see Refs. 10, 16, 39). Studies conducted in vivo have demonstrated that these RVL vasomotor neurons exhibit a tonic activity, project to thoracic segments of the spinal cord where they excite preganglionic sympathetic neurons, and are powerfully inhibited after activation of arterial baroreceptors (e.g., Refs. 6, 25, 31, 32, 36, 47). Many of these spinally projecting RVL neurons have a catecholaminergic phenotype and belong to the C1 adrenergic cell group (e.g., Refs. 1, 19, 32, 38).

The RVL neurons have also been investigated in tissue slices. These in vitro studies revealed that many neuronal properties are affected by experimental conditions such as the thickness of the slice, temperature, age of the animal, type of microelectrode used, and the criteria for neuron selection. For example, recordings with intracellular microelectrodes in thick slices (500 µm, 30-31°C) obtained from adult rats demonstrated spontaneous activity only in nonadrenergic cells (42-44). On the other hand, recordings with patch electrodes in thin slices (120-250 µm, 21-23°C) obtained from neonatal rats showed spontaneous firing in both catecholaminergic and noncatecholaminergic neurons (23, 24, 29, 30). In neurons that displayed spontaneous activity, action potentials were triggered by pacemaker-like depolarizations (e.g., Refs. 24, 29, 41, 43). This finding differs from the results obtained in the intracellular in vivo studies, which demonstrated that the ongoing activity of both catecholaminergic and noncatecholaminergic RVL neurons results from synaptic inputs (31, 32).

To further characterize the properties of spinally projecting RVL neurons, we used an acute dissociation technique that leads to elimination of most, if not all, cell-to-cell interactions (synaptic and nonsynaptic). Apart from providing a precisely controlled extracellular environment, neuron isolation offers an excellent means of membrane visualization for patch-clamp recording. In addition, dissociated neurons are more suitable for single-cell reverse transcription (RT)-polymerase chain reaction (PCR), a technique that allows linking of electrophysiological data to a molecular analysis of the mRNAs expressed in individual cells (5, 48, 49). This new approach allowed us to examine whether or not the dissociated, spinally projecting RVL neurons display pacemaker-like activity similar to that previously described in tissue slices and to study other properties of these cells.

Preliminary results were presented elsewhere (33).

	METHODS

Top Abstract Introduction Methods Results Discussion References

Retrograde labeling of RVL neurons. The upper thoracic spinal cord was surgically exposed in Wistar rat pups (10 to 13 days old) under halothane anesthesia. A suspension of rhodamine-labeled latex microspheres (Lumafluor) was injected unilaterally or bilaterally into the T₂-T₄ segments using a 10-µl Hamilton syringe (30-gauge needle), with one or two deposits made per side (0.5 µl each). After the skin incision was closed with acrylic glue, the pups were returned to lactating females.

Tissue dissociation. Three to six days after the spinal injection, rats were decapitated, and their brains were rapidly removed and immersed in an ice-cold artificial cerebrospinal fluid (aCSF) equilibrated with 95% O₂-5% CO₂ (carbogen). The aCSF contained (in mM) 124 NaCl, 3 KCl, 2.6 CaCl₂, 1.3 MgSO₄, 2.5 NaH₂PO₄, 26 NaHCO₃, and 20 glucose, pH 7.3-7.4, 300 ± 10 mosmol. Transverse sections (350 µm) were cut with a Vibratome from a block of medulla oblongata submerged in the aCSF. A single slice containing the RVL area was selected under a dissecting microscope with illumination from below, using the following criteria: 1) location immediately caudal to the facial nucleus; 2) clearly visible compact formation of the nucleus ambiguus; and 3) presence of the rostral tip of the inferior olive. Neurons were acutely dissociated using a procedure similar to that described by Kay and Wong (26). Mild enzymatic digestion was performed for 35-45 min (35°C) with 140 units papain (Worthington), 1.6 mM L-cysteine, 0.2 mM EDTA, and 13.4 µM -mercaptoethanol (Sigma) in a total volume of 7.0 ml of aCSF. The digestion and pre- and postdigestion incubations were conducted in a custom-built incubation chamber, with continuous gentle stirring and bubbling with carbogen. After digestion, one RVL region was cut out of the section on a Sylgard-coated petri dish under a dissecting microscope. The rest of the section was returned to the incubation chamber (aCSF, room temperature), allowing cells from the contralateral RVL to be harvested 2-4 h later. The dissected RVL tissue was dissociated by trituration with fire-polished Pasteur pipettes in 0.5-ml Eppendorf tubes. The dissection, trituration, and subsequent recording were conducted in a "standard" external solution that contained (in mM) 155 NaCl, 3 KCl, 1 CaCl₂, 1 MgCl₂, 10 N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES), and 10 glucose, pH 7.3-7.4, 300 ± 10 mosmol. Cells were plated at low density on poly-L-lysine-coated coverslips that were placed in an RC-13 recording chamber (Warner Instruments). The chamber was mounted on an inverted microscope (TMD; Nikon) equipped with phase-contrast, fluorescence attachment (G-2A filter block; Nikon), two independent micromanipulator systems (Burleigh PCS-750/Newport M-423 and Leitz), and a video camera (DXC-930; Sony). The use of the second (Leitz) micromanipulator was necessary for collecting cells for RT-PCR and for application of ligands. The chamber was perfused under gravity with external solutions at 1.5-2.0 ml/min (22-24°C). Images of selected neurons were saved to computer (Quadra 950: Macintosh) at 656 × 504 pixel resolution, using a frame grabber (RasterOps, 24 XLTV).

Whole cell patch clamping. Recordings were made only from cells that had a neuronal appearance (see RESULTS). Cells that were swollen, dark, or had grainy membranes were considered to be injured and were not patched. Pipettes were pulled from thin-walled borosilicate glass capillaries (outside diameter 1.5 mm; Clark Electromedical) using a horizontal puller (P-97; Sutter Instruments). They were filled with a solution containing either (in mM) 150 KCl (or KF), 10 HEPES, 10 ethylene glycol-bis(-aminoethyl ether)-N,N, N',N'-tetraacetic acid (EGTA), and 10 glucose, pH 7.3, 280 mosmol (K⁺-based pipette solution; pH adjusted with KOH) or 140 CsF, 5 NaCl, 10 tetraethylammonium (TEA) chloride, 10 HEPES, 5 EGTA, and 10 glucose, pH 7.3, 268-280 mosmol (Cs⁺-based solution; pH adjusted with CsOH). In addition, some pipettes contained lucifer yellow (LY; dipotassium salt, 0.05%; Sigma). Electrode resistance was 3-5 M. Neurons retrogradely labeled with rhodamine microspheres, as well as some nonlabeled neurons, were patched using the EPC-7 amplifier (List Electronic) and a tight-seal whole cell voltage- or current-clamp configuration (17). After formation of a gigaseal (5 G), a short period of recording was made in most cells in a cell-attached mode to examine for the presence of extracellular unit activity (2, 29). After rupture of the cell membrane by suction, the recorded currents were low-pass filtered at 4 kHz or 400 Hz, digitized at 10 kHz (12-bit resolution), and stored on computer disk for analysis (pCLAMP 5.5.1 and AxoGraph 3.5; Axon Instruments). Access resistance (R_s), membrane resistance (R_N), and membrane capacitance (C_m) were measured by applying hyperpolarizing voltage pulses (10 mV) from a holding potential of 65 mV. R_s was <15 M. The capacitance artifact was compensated using the circuitry of the amplifier. In some cases, linear leak subtraction was performed on-line (4). Junction potential was nulled before seal formation. L-Glutamic acid (monosodium salt) and kainic acid (Sigma) were pressure injected through a pipette (tip diameter 10-20 µm) placed 60-100 µm away from the examined cell body for 6-8 s. Na⁺ current was blocked by adding 25-500 nM tetrodotoxin (TTX; RBI) to the standard external solution. Ca²⁺ current was blocked by perfusing the cells with a modified external solution containing 1 mM CoCl₂ (nonselective Ca²⁺ channel blocker). Values are expressed as means ± SE. An independent Student's t-test was used for statistical analysis.

Single-cell RT-PCR. After whole cell recording, cells were collected by aspiration into a separate pipette (tip diameter ~3.0 µm) that contained (in mM) 140 KCl, 3 MgCl₂, 0.1 CaCl₂, 2 EGTA, and 10 HEPES, pH 7.6, 305 mosmol (total volume 8.0 µl). Suction was performed in two stages. An initial gentle suction applied to the cell surface allowed lifting of the cell several hundred micrometers above the bottom of the recording chamber, and a second stage of stronger suction was used for complete aspiration of the cell into pipette. Care was taken to avoid aspiration of adjacent cells or cell debris. Subsequent procedures were similar to those described by Lambolez et al. (27; also see Ref. 8). In brief, RT was performed for 50 min at 42°C with Superscript II (200 units; GIBCO) in 11.0 µl of incubation mixture containing 1 mM dNTPs, 1.25 mM MgCl₂, 20 units ribonuclease (RNase) inhibitor (RNAguard; Pharmacia), 5 mM dithiothreitol, 0.5 µl random hexamers, plus 8.0 µl pipette solution. A seminested PCR protocol (50) was used to identify mRNA species present in individual cells. PCR reactions were performed in a 50-µl volume containing (in mM) 20 tris(hydroxymethyl)aminomethane (Tris) · HCl (pH 8.4), 50 KCl, 1-3 MgCl₂, 0.25 dNTPs, and 1.25 units Taq DNA polymerase (GIBCO). The concentration of primers was 20 nM in the first round of amplification and 200 nM in the second round. The primers used targeted the following three genes: neuron-specific enolase (NSE; product length of the second round of amplification 361 bp), tyrosine hydroxylase (TH; 220 bp), and phenylethanolamine N-methyltransferase (PNMT; 543 bp). NSE, which is expressed in all mature neurons (34), was used as a positive control. The primer sequences are given in a report by Comer et al. (8). Two microliters of RT product were used for PCR with NSE primers and 4 µl for PCRs with TH or PNMT primers. In each PCR, 5 µl of the first reaction product were used in the second round of amplification. Control tubes (RT-PCR with the external solution taken from near the bottom of the recording chamber, without aspiration of neurons or cell debris) were included for each experiment. The thermal cycler (PTC-100; MJ Research) was programmed for 30-35 cycles of 45 s denaturation (94°C), 2 min annealing (54-58°C), and 2 min and 30 s elongation (72°C). PCR products from the second amplification round were separated by electrophoresis on an ethidium bromide-stained 2% gel [1% agarose (GICBO) and 1% NuSieve (FMC)]. Bands were visualized by ultraviolet transillumination and photographed with a digital camera (Kodak DCS200), and the images were saved to a Macintosh (Quadra 950) computer file.

Immunocytochemisty. After electrophysiological recording, the cells that remained attached to the coverslip were fixed in freshly prepared 4% paraformaldehyde in 0.1 M phosphate buffer for 30-60 min and then washed in 10 mM phosphate-buffered saline (PBS, pH 7.4). The cells labeled with LY and rhodamine microspheres were identified using a fluorescence microscope (filter blocks D and N2; Leitz Diaplan) and photographed (Leitz Vario-Orthomat). After overnight incubation at 4°C with a monoclonal TH antibody (1:400; Boehringer Mannheim), 10% normal goat serum and 0.3% Triton X-100 in PBS, they were rinsed with PBS and incubated at 37°C for 2 h in biotinylated sheep anti-mouse antibody (1:500; Sigma), 10% normal goat serum, and 0.3% Triton X-100 in PBS. This was followed by another wash and 1 h incubation at 37°C in ExtrAvidin peroxidase (1:1,000; Sigma). A dark reaction product was obtained by exposing the cells to diaminobenzidine (0.05%), ammonium nickel sulfate (0.6%), and hydrogen peroxidase (0.005%) in 0.05 M Tris · HCl buffer (pH 7.6).

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Identification of dissociated RVL neurons. Neurons were dissociated from the RVL region dissected from transverse medullary sections. The dissected area was bounded laterally by a line extending between the compact formation of the nucleus ambiguus and the ventral pole of the spinal trigeminal tract and was bound medially by a line extending between the compact formation and the ventral medullary surface, immediately lateral to the pyramid (Fig. 1A). In spite of the small size of the area (0.3 mm³), dissociations resulted in several dozen healthy looking, neuron-like cells from each side of the medulla. Viable cells had a bright appearance and displayed a characteristic "halo" around the cell body. Fluorescent microspheres were identified in ~5-10% of such cells. Examples of phase-contrast micrographs (video camera images) of retrogradely labeled neurons are shown in Fig. 1, B-E. The concentration of fluorescent beads was high in the soma and lower in proximal processes (Figs. 2B and 3C). Spinally projecting neurons had two to six processes (mean, 3.5). Their cell bodies were multipolar, triangular, oval, or fusiform in shape, with the average major axis 24.5 ± 1.2 µm and the minor axis 13.4 ± 0.4 µm (n = 23).

View larger version (87K): [in this window] [in a new window]   View larger version (105K): [in this window] [in a new window]   Fig. 1.   Photomicrograph of a transverse medullary section showing the area used for neuron dissociation (A) and examples of dissociated neurons under phase-contrast optics (B-E). Arrow in A indicates the compact formation of nucleus ambiguus. Neurons shown in B-E were retrogradely labeled with fluorescent microspheres.

View larger version (95K): [in this window] [in a new window]   Fig. 2.   Example of identification of a catecholaminergic rostral ventrolateral (RVL) neuron with reverse transcription (RT)-polymerase chain reaction (PCR). A: video image of a patched neuron under phase contrast. B: rhodamine microbeads identified with epifluorescence (Nikon filter block G-2A). C: PCR products obtained from the same neuron after agarose gel electrophoresis. Three lanes on right show bands obtained after cDNA amplification with primers specific for neuron-specific enolase (NSE), tyrosine hydroxylase (TH), and phenylethanolamine N-methyltransferase (PNMT). Lane in middle shows 100-bp molecular weight ladder (GIBCO). Three lanes on left show controls obtained with external solution aspirated in the vicinity of the patched neuron after RT-PCR reactions with appropriate primers. Note the presence of low-molecular-weight primer dimers.

View larger version (87K): [in this window] [in a new window]   View larger version (53K): [in this window] [in a new window]   View larger version (40K): [in this window] [in a new window]   View larger version (55K): [in this window] [in a new window]   Fig. 3.   Identification of a catecholaminergic neuron with immunocytochemistry. A: video image of dissociated RVL neurons, with one cell patched with pipette containing lucifer yellow. B: lucifer yellow fluorescence observed with epifluorescence (Leitz filter block D). C: rhodamine microbeads visualized with Leitz filter N2. D: diaminobenzidine reaction product observed in the same neuron after immunocytochemistry with a monoclonal anti-TH antibody.

Catecholaminergic neurons belonging to the spinally projecting C1 group were identified in the following two ways: by using the single-cell RT-PCR technique or with immunocytochemistry. For RT-PCR analysis, neurons were aspirated under visual control into glass pipettes with a negative pressure, immediately after the end of the whole cell patch-clamp recording (Fig. 2A). Nineteen of 23 aspirated cells (all showing neuron-like morphology, fluorescent microspheres, and a fast Na⁺ current during whole cell recording; see below) were shown to express NSE (Fig. 2C). Each NSE-positive cell was then tested in two separate PCR tubes for TH and PNMT mRNA. The 220-bp product obtained after PCR with primers specific for TH was identified in 11 cells, whereas the 543-bp product obtained with primers specific for PNMT was found in 10 cells. Nine cells showed expression of both genes (Fig. 2C).

Some whole cell recordings were made with patch pipettes containing LY. After formaldehyde fixation, the position of cells that were fluorescent for both rhodamine and LY was identified on the coverslip. Their shapes were compared with video camera images taken immediately after whole cell recordings. A subsequent immunoperoxidase reaction with an antibody to TH revealed immunoreactivity in 6 of 14 neurons that were labeled with both LY and fluorescent beads (Fig. 3). The remaining eight cells were TH negative.

Electrophysiological properties of spinally projecting RVL neurons. No clear differences in electrophysiological properties were found between neurons identified as catecholaminergic and noncatecholaminergic; therefore, both groups are mostly presented together. Whole cell recordings were made from 54 retrogradely labeled RVL neurons with patch pipettes containing either CsF (n = 42) or KCl/KF (n = 9 KCl; n = 3 KF). In general, use of F resulted in better recording stability. To get appropriate voltage control, only neurons with processes shorter than 60-80 µm were used. In 35 neurons, extracellular recordings were made in cell-attached mode before rupturing of the patch. Such recordings did not reveal any spontaneous firing in 31 cells (Fig. 4, A1). Fast extracellular currents, indicative of action potentials, were observed in four remaining neurons (Fig. 4, A2). However, in all of these cases, extracellular activity occurred only transiently during seal formation and stopped after the final resistance (5 G) was obtained (not illustrated).

View larger version (27K): [in this window] [in a new window]   Fig. 4.   Examples of extracellular recording in the cell-attached mode from two retrogradely labeled neurons (A1 and A2), and intracellular recording in the current-clamp mode after patch rupturing (B and C). Cell in B is the same as that in A1 (identified as catecholaminergic with RT-PCR), whereas the cell in C is another spinally projecting neuron. A2 was obtained before full seal was established. B and C: responses to hyperpolarizing and depolarizing steps, respectively. Note the lack of firing when no current pulse was applied (B, arrow in middle trace) and also during a 10-pA depolarizing pulse (C; arrow).

After rupturing of the membrane and a short stabilization period (1-2 min), the resting membrane potential measured under current-clamp with K⁺-containing pipettes was 61.5 ± 2.3 mV (n = 12). No spontaneous action potentials were observed (Fig. 4B). Eight cells exhibited repetitive spiking in response to 300- to 400-ms depolarizing current pulses of up to 100 pA (Fig. 4B). The remaining four cells responded with a single or double spike (Fig. 4C) or a short burst of three to four spikes (not illustrated). Spike height was 76.9 ± 3.5 mV and spike width, measured at one-half amplitude, was 1.2 ± 0.15 ms. No attempt was made to measure the resting membrane potential with Cs⁺-based pipettes, and a variable amount of holding current (up to 50 pA) had to be used to maintain the membrane potential at 60 to 65 mV. Depolarizing current pulses applied through CsF-containing pipettes resulted in prolonged action potentials (not illustrated). Similar recordings with KF-containing pipettes resulted in short-lasting action potentials and, in two out of three cases, repetitive spiking similar to that recorded with KCl-containing pipettes (not illustrated).

In voltage-clamp mode, the holding current measured during the holding potential of 65 mV was 54 ± 9.7 pA (n = 12) when recorded with K⁺-containing pipettes and 58.2 ± 13.8 pA (n = 42) for Cs⁺-based pipettes. Several characteristics were measured from current responses evoked by a hyperpolarizing step from 65 to 75 mV (Fig. 5, A1 and B1). In nearly all tested cells (20/21), such responses could be fitted to the first exponential with the least square residual higher than 0.999. Table 1 gives the values of the R_N, C_m, and membrane time constant (₂ = R_N × C_m) measured with Cs⁺- and K⁺-based pipettes in 12 catecholaminergic and 8 noncatecholaminergic neurons. The differences between the groups were nonsignificant, with the exception of R_N (P < 0.001) when measured with Cs⁺- or K⁺-containing pipettes. For comparison, data are also given for a group of nonspinal neurons (n = 17).

View larger version (30K): [in this window] [in a new window]   Fig. 5.   Currents in two spinally projecting neurons recorded with a K+ (A1-A3)- or Cs+ (B1-B3)-containing pipette. A1 and B1: current relaxation after a hyperpolarizing step shown in top of each panel. Current data points () are fit to a 1st-order exponential (1, charging time constant; see text for other abbreviations). A2 and B2: families of currents recorded using the step protocols shown. Records were obtained after on-line linear leak subtraction. A3 and B3: responses to 180-ms voltage commands (steps from 100 to +50 mV in 15-mV increments; holding potential 70 mV). Current-voltage (I-V) relationships are shown (, peak outward currents; , outward currents recorded at the end of each pulse). Note that the outward currents were reduced but not abolished when recorded with Cs+-based pipette.

Depolarizing voltage steps (10 or 20 ms long, stepped to a command potential ranging from 60 to 10 mV in increments of 10 mV) elicited a fast-activating, rapidly inactivating (spike-like) inward current that could be reversibly abolished by 0.5 µM TTX. The peak amplitude of this current, its duration at one-half amplitude, and the threshold were, respectively, 3.7 ± 1.2 nA, 2.4 ± 0.5 ms, and 42.3 ± 4.8 mV (n = 11) when recorded with K⁺-based pipettes (Fig. 5, A2) and 3.5 ± 1.7 nA, 3.9 ± 2.7 ms, and 44.5 ± 6.7 mV (n = 22) when recorded with Cs⁺-containing pipettes (Figs. 5B2, 6A, and 7A3). During recording with K⁺-containing pipettes and voltage steps from 100 to +50 mV (180-ms pulses), this fast inward current was followed by a sustained outward current (Fig. 5A3; I_K; see Ref. 22). In two out of five cells, this I_K was preceded by a larger and faster-decaying component [transient A-type current (I_A), Fig. 5A3]. In the remaining cells, only the sustained component could be identified. The amplitude of the sustained current measured at the end of the depolarizing steps to a command potential +50 mV was 2.1 ± 0.7 nA (n = 5). The current was largely reduced when the Cs⁺-based pipettes were used (Fig. 5B3). Its amplitude, measured at the end of the depolarizing pulse, was 295 ± 66 pA (n = 3).

View larger version (19K): [in this window] [in a new window]   Fig. 6.   Example of currents recorded with Cs+-containing pipette during ramp voltage commands. A: inward transients recorded with step commands from 40 to 10 mV (on-line linear leak subtraction). B1 and B2: currents during ramp protocols indicated in inset. Two superimposed sweeps are shown in each record. B1, transient (* activation threshold about 26.5 mV) and slow (approximate activation threshold -44 mV) inward currents during a 0.125-mV/ms ramp. B2, current recorded in the same cell during a 0.07-mV/ms ramp. Note absence of fast inward transients during the slower ramps.

View larger version (19K): [in this window] [in a new window]   Fig. 7.   Identification of sustained inward currents in RVL neurons during recording with Cs+-based pipettes. Cell in A was catecholaminergic as demonstrated with RT-PCR. A1: control, slow inward currents during the ramp protocol indicated in inset (0.125 mV/ms); Co, current induced by the same ramp during cell exposure to 1 mM cobalt; Co + TTX, current observed during exposure to 1 mM cobalt and 25 nM tetrodotoxin (TTX). Two superimposed sweeps are shown in each record. A2: cobalt and TTX-sensitive components obtained after subtraction of currents recorded with cobalt from control currents (Co-sensitive component) and after subtraction of currents recorded with cobalt and TTX from currents recorded during exposure to cobalt only (TTX-sensitive component). Currents were averaged before subtraction. A3: inward transients recorded in the same neuron with step commands from 40 to 10 mV (on-line linear leak subtraction). B: persistent current recorded in another RVL neuron during a step command from 100 to 30 mV. Control, average of 3 control episodes; TTX, average of 3 episodes during exposure to 25 nM TTX. * Inward transient (truncated) that was reduced in amplitude, but not abolished, by this concentration of TTX.

Depolarizing voltage ramps were used to identify sustained inward currents mediated by Na⁺ or Ca²⁺. The pipette solution containing Cs⁺ and TEA was used to block K⁺ channels. Ramps were applied from a starting level of 85 mV to a peak ranging from 10 to +15 mV, with a slope ranging from 0.05 to 0.125 mV/ms. When relatively fast ramp commands were used (0.125 mV/ms), depolarization often induced a large (>2.0 nA), TTX-sensitive, transient inward current (or, in a few cases, a short burst of such transients; not illustrated), which was often preceded by a smaller and slowly rising component (Fig. 6B1). With 600-ms ramps to a command potential of +15 mV, the threshold range for triggering the transient current was 35 to 25 mV. If necessary, ramps with a smaller slope were used to eliminate this transient current by accommodation (Fig. 6, B1 and B2). In 16 neurons tested with TTX and/or cobalt, three components of the sustained inward current were identified by using depolarizing ramps that triggered no transient current (Fig. 7, A1 and A2). The first, which was activated in the range of 65 to 50 mV and peaked at 45 to 35 mV (30-180 pA), was TTX sensitive. Interestingly, this current could be blocked by a low concentration of TTX (25 nM), which did reduce, but not abolish, the amplitude of the action potential-like, transient inward current evoked by suprathreshold depolarizing voltage steps (n = 4, not illustrated). The two other components were cobalt sensitive. The lower-threshold component, which largely overlapped with the slow, TTX-sensitive component, activated at 65 to 42 mV and peaked around 50 to 30 mV (10-160 pA; Fig. 7A2). The higher-threshold component activated at 40 to 30 mV and peaked around 10 mV. If this latter component was prominent, it could be seen in control records as a second (higher-threshold) phase of the slow inward current (Figs. 7A1 and 8C). Its amplitude could not be measured reliably, as the inward current was opposed, at this level of depolarization, by residual K⁺ and Cs⁺ currents. This higher-threshold slow inward current could not be clearly identified with K⁺-containing pipettes due to a large I_K. It was reduced considerably in amplitude, but often not totally abolished, by 1 mM cobalt (Fig. 7A1). The relative amplitudes of the three sustained components varied between neurons. These components could be identified in catecholaminergic (4 out of 6 tested) and noncatecholaminergic (3 of 5) spinal neurons, unidentified spinally projecting neurons (6 of 8), and in nonspinal cells (3 of 4). The TTX-sensitive sustained inward current could also be demonstrated in both spinal (n = 3, 1 catecholaminergic) and nonspinal (n = 1) neurons with step commands from 100 to 30 (or 40) mV (Fig. 7B).

During voltage clamping in both spinal and nonspinal RVL neurons, extracellular application of L-glutamic acid or kainic acid induced a dose-dependent sustained inward current and a large reduction in R_N (Fig. 8). This effect was observed in four out of four cells tested with L-glutamate (1 spinally projecting and 3 nonspinal) and in three out of three cells tested with kainic acid (1 spinal and 2 nonspinal).

View larger version (15K): [in this window] [in a new window]   Fig. 8.   Examples of currents induced in RVL neurons by activation of ligand-gated channels (recordings with Cs+-containing pipettes). A and B: currents evoked by kainic acid (50 µM) at different holding potentials (Vh). C: I-V relationship tested with ramp indicated in inset. C, control; KA, kainic acid application; R, recovery. D: inward current generated in another neuron by monosodium L-glutamate (L-Glu; 50 µM).

Figure 8, A and B, shows that the current, induced by 50 µM kainic acid and recorded with a Cs⁺-containing pipette, could be reversed in polarity at the command potential of ~6.3 mV.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

Although the technique of acute dissociation has been extensively used to examine neurons isolated from a large number of brain regions, to our knowledge this is the first study of putative vasomotor neurons dissociated from the RVL medulla. Isolated neurons were viable and largely retained their morphological, chemical/molecular, and functional identity. They were isolated from a small but well-defined medullary region, identified as spinally projecting or nonspinal with a fluorescent tracer, and tested for the presence of specific mRNAs or an enzyme involved in catecholamine synthesis. Methodological aspects of the techniques will be considered before discussing the properties of these neurons.

Methodological considerations. Our dissociation technique was similar to that of Kay and Wong (26), with the following two modifications: 1) bicarbonate-buffered aCSF was used in all initial stages, including the enzymatic treatment, and 2) we used a low concentration of papain instead of trypsin. The choice of papain was based on the data demonstrating an increased neuronal survival after this enzyme (e.g., Refs. 12 and 15, but see Ref. 26). It has been reported that exposure of neuronal membrane to a high concentration of papain may remove Na⁺ inactivation (15). However, the short duration of the transient Na⁺ current induced by depolarizing steps (2.4 ± 0.5 ms in cells patched with K⁺-based pipettes) indicates that this was not the case in our experiments.

One of the advantages of neural dissociation, in comparison with recording from tissue slices, is a superior space-clamp due to truncated processes. Good space-clamp was confirmed by showing that capacitive transients could be well fitted to the first exponential. Further electronic shortening of the processes was achieved by blocking K⁺ channels with Cs⁺ and TEA. The input (membrane) resistance of dissociated RVL neurons was significantly higher than that measured in tissue slices with intracellular microelectrodes (88-200 M; see Refs. 41 and 43) but comparable to the values recorded in slices with patch electrodes (0.5-1.6 G; see Refs. 29 and 30). High input resistance facilitated voltage-clamp analysis by reducing the series resistance error. In agreement with others, we observed that the use of F in the patch pipette resulted in better recording stability (e.g., Refs. 11 and 14). Although the mechanism of this effect is not completely understood, it may be due to blockage of the activity of lysosomal enzymes (3).

We successfully applied the single-cell RT-PCR technique to identify expression of mRNA coding for NSE, as well as for TH and PNMT, which were used as two markers specific for catecholaminergic neurons. By using dissociated neurons, we could overcome the major problem associated with the use of this technique in tissue slices: the possibility of false-positive results due to pipette contamination with mRNA originating from adjacent cells. This limitation is practically eliminated when cells plated at relatively low density are collected and when the two stages for cell aspiration (as described in METHODS) are applied. The possibility of cross-contamination with cDNA was also excluded, since the control samples containing the external solution only were negative. Another problem associated with single-cell RT-PCR is the possibility of false-negative results due to an insufficient number of mRNA copies taken into the suction pipette, with mRNA degradation resulting from the residual RNase H activity of the reverse transcriptase, or inadequate PCR amplification. In all cells tested, we first examined expression of mRNA for NSE. This was performed not only to obtain additional evidence for the neuronal character of collected cells but mainly as a positive control for mRNA "survival" over many hours of the experiments, its transfer to the collection pipette, and the efficiency of RT. RT was performed with SuperScript II, which has no detectable RNase H activity. Our findings demonstrate that false-negative results were infrequent. Most of the examined cells (83%) were shown to express NSE. TH mRNA was found to be expressed together with PNMT in 9 out of 19 (47%) of these cells. Only two TH-positive neurons (of the total 11) that were PNMT-negative were found, and one cell was positive for PNMT but negative for TH. Collectively, the proportion of dissociated, retrogradely labeled cells that expressed mRNAs specific for catecholaminergic neurons (TH and/or PNMT; 63%) was similar to that reported in previous studies conducted in vivo or in tissue slices for catecholaminergic spinally projecting RVL neurons (see below).

The catecholaminergic phenotype of some cells was identified using immunocytochemistry, which also provided evidence that the TH mRNA was being translated into protein. However, this approach was less practical than single-cell RT-PCR. The main problem was the lifting of some cells from the coverslip on completion of patching (in spite of the improved cell attachment to a poly-L-lysine-coated surface) and consequently losing them for immunocytochemical analysis. When this occurred, it was necessary to aspirate a lifted cell to a second pipette mounted in an independent micromanipulator and conduct RT-PCR. Another disadvantage with immunocytochemistry was the indirect way in which the cells were identified, which occasionally resulted in uncertainty when matching the cells. The procedure involved comparison of the shape and coverslip location of patched cells with LY- and rhodamine bead-labeled cells found on the coverslip after fixation. A similar comparison was then required after conducting immunocytochemistry.

Properties of spinally projecting RVL neurons. After acute dissociation, retrogradely labeled RVL neurons showed many morphological features similar to those described in the studies conducted in vivo or in tissue slices, including the size and shape of the cell body and the number of primary processes (24, 32, 42). Fifty-two percent of these cells were catecholaminergic, as demonstrated with RT-PCR or immunocytochemistry. The proportion of catecholaminergic bulbospinal RVL neurons found in previous anatomic studies ranged from 39 to 80% (24, 29, 30, 38, 45). One reason for the relatively large differences between these studies is the method used to identify spinal projection (electrophysiologically or with retrograde tracers) and the inconsistent definition of the borderlines of the RVL region. We dissected a triangular area ventromedial and ventrolateral to the compact formation of the nucleus ambiguus (cf. Ref. 19). In the study by Li et al. (29), which reported the highest proportion of catecholaminergic neurons (78-80%), a smaller rectangular area was selected below the compact formation, which did not fully overlap with the region investigated in our study.

The isolated spinally projecting RVL neurons are referred to as putative vasomotor (or presympathetic) neurons, since we could not establish 1) whether they receive an inhibitory baroreceptor input and 2) whether they directly project to preganglionic sympathetic neurons. However, previous studies conducted in the in vivo rat demonstrated that a large proportion of nonrespiratory RVL neurons projecting to the upper thoracic segments of the spinal cord are inhibited by baroreceptors (for review see Ref. 16). In addition, the anterograde tract-tracing studies showed that, in the thoracic spinal cord, axons of RVL neurons terminate almost exclusively in the intermediolateral and intermediomedial sympathetic cell columns, and not within the ventral horn where respiratory motoneurons are located (e.g., see Ref. 37). The possibility that a significant proportion of neurons included in our study belonged to the Bötzinger complex (a group of respiratory neurons located in the proximity of RVL presympathetic neurons; see Ref. 25) is unlikely, as these neurons only rarely project below cervical segments of the spinal cord (7, 25). In agreement with others (23, 24, 29, 30), we observed no clear differences between the spinally projecting catecholaminergic and noncatecholaminergic neurons with respect to their basic electrophysiological properties. Therefore, the following discussion applies to both groups pooled together.

A major difference between our results and previous studies conducted in RVL slices was the absence of spontaneous firing in dissociated neurons, in contrast to its presence in cells recorded in tissue slices (see introduction for references). The tonic activity observed in slices was not due to residual synaptic interactions (it was not abolished by low Ca²⁺/high Mg²⁺; see Ref. 28) and therefore is thought to be of a pacemaker type. The comparison between our results and previous results is mainly based on extracellular recordings in the cell-attached mode. This approach is less likely to induce tonic firing in otherwise silent cells due to changes of the membrane potential associated with intracellular impalement or rupturing of the membrane patch with subsequent cell dialysis during whole cell recording. Li et al. (29) observed extracellular activity in a vast majority of bulbospinal RVL neurons recorded in slices, whereas, in our study, 31 out of 35 examined cells were silent. A transient firing observed in the remaining cells was probably due to the contact of the patch pipette with the cell and membrane deformation (or an increase in extracellular K⁺ concentration) rather than to inherent properties of these cells. The resting membrane potential recorded in our study (mean 61.5 mV when measured with K⁺-containing pipettes) cannot be directly compared with previous data obtained from tonically active neurons. When the midpoint of interspike trajectory was used, the mean values of the membrane potential recorded with sharp intracellular microelectrodes or patch pipettes in both thick and thin slices were, at similar K⁺ gradients, around 51 to 59 mV (23, 24, 29, 30, 41, 43). After TTX, the membrane potential was about 50 to 51 mV (29, 40). It is therefore possible that the spontaneous activity recorded in slices is due to the somewhat lower level of membrane potential in neurons studied in this type of preparation. The depolarization could be induced by some unidentified (synaptic or nonsynaptic) cell-to-cell interactions, a factor that is not present in isolated neurons. At this lower level of membrane potential, the firing threshold can occasionally be reached, resulting in a slow and irregular firing pattern. The cells that are more depolarized would be expected to fire at higher frequency and more regularly, with a pacemaker-like activity (cf. Refs. 23, 29, 42). Alternatively or additionally, a sustained Ca²⁺ (29) or Na⁺ current (24) may be activated that could also depolarize the cells to threshold.

Li et al. (29) observed TTX-resistant, subthreshold oscillations of the membrane potential in a proportion of slowly firing RVL neurons, which were eliminated by hyperpolarizations to 70 mV. They suggested that this may be a basis for autorhythmicity seen at more depolarized levels (see also Ref. 23). No such oscillations were observed in our experiments, even though the dissociated neurons retained a capacity to generate sustained Na⁺ and Ca²⁺ currents when appropriately depolarized. Small and random fluctuations of the membrane potential observed in the current-clamp mode (and of the membrane potential in the voltage-clamp mode) were not decreased in size when the potential was shifted from the threshold level to 85 mV and were not affected by TTX and cobalt. Such fluctuations are typical of dissociated neurons, in which the plasma membrane has been largely stripped of synaptic and glial contacts. The absence of spontaneous firing in dissociated neurons corresponds to the results of our studies conducted in vivo (31, 32), which showed that the activity of these neurons is evoked by synaptic inputs and not by autodepolarizations.

The fact that no pacemaker-like activity was observed in the present study cannot be interpreted as evidence that these neurons are not capable of generating such activity under different experimental conditions. However, the factors necessary for expressing autodepolarizations still remain to be identified. Such factors may include a changed balance between excitatory and inhibitory postsynaptic potentials, leading to depolarization (31), the presence of a particular chemical messanger(s) in the extracellular space, and the developmental stage of the animal (cf. Ref. 29). It is possible that RVL presympathetic neurons undergo a developmental transformation displaying mainly pacemaker behavior at the late embryonic and early postnatal stage, whereas in the mature brain their activity is largely determined by synaptic inputs. Interestingly, a similar transformation has been proposed for medullary neurons generating respiratory movements (21).

In most dissociated RVL neurons examined with K⁺-containing pipettes, repetitive firing could be induced by rectangular depolarizing current pulses. The remaining neurons showed a rapid adaptation, probably due to the insufficient amplitude and/or duration of the afterhyperpolarizations needed for removal of Na⁺ channel inactivation. As expected, the threshold for activation of the transient Na⁺ current (I_NaT), recorded in the voltage-clamp mode, was dependent on the rate of depolarization and was less negative during depolarizing ramps than when measured with step commands. In agreement with Kangrga and Loewy (24), we observed the complete inactivation of I_NaT when the rate of depolarization was reduced further. This, together with the use of Cs⁺-based patch pipettes to block the outward K⁺ current, allowed us to examine sustained inward currents. The TTX-sensitive component of these currents resembled the persistent Na⁺ current (I_NaP) previously identified in RVL neurons patched in tissue slices (24). I_NaP has also been identified in other classes of neurons and is believed to be an important factor affecting the cell's excitability and, in some cases, subthreshold oscillations of the membrane potential (9, 46). Interestingly, in RVL neurons, a low concentration of TTX (25 nM) abolished the I_NaP but not I_NaT. This observation is consistent with the possibility that I_NaP flows through a distinct subtype of Na⁺ channel that is more sensitive to TTX than the I_NaT channel (see also Ref. 18).

The cobalt-sensitive inward current had two components, resembling the low-voltage activated (LVA) and high-voltage activated (HVA) Ca²⁺ currents (22). The LVA component was present at the level of depolarization close to firing threshold. This indicates that this current, together with the I_NaP, may affect the excitability of RVL neurons. In view of the fact that LVA Ca²⁺ currents often inactivate with a relatively fast time constant after depolarizing steps (13, 21), it remains to be examined to what degree the low-threshold cobalt-sensitive component observed in our experiments was affected by the quasi-steady-state conditions associated with the generation of current-voltage curves by depolarizing ramps. The LVA Ca²⁺ current (as defined by sensitivity to nickel) has been implicated in depolarization of pacemaker-like RVL neurons by hypoxia (41). This suggests that this current may function in RVL neurons for a prolonged period of time. The HVA component is likely to be associated with ion channels involved in neurotransmitter release or contribute to currents during action potentals. It should be noted that, in some cells, the HVA current was reduced but not totally abolished by cobalt.

When K⁺-containing pipettes were used, step or ramp depolarizations of sufficient amplitude activated a Cs⁺ and TEA-sensitive I_K. Along with Kangrga and Lowey (24), we observed the following two components of this current during steps: one resembling transient A-type current (I_A) and the other delayed rectifier (I_K). A contribution of Ca²⁺-dependent K⁺ conductance would have to be assessed using a faster Ca²⁺ chelator in the patch pipette, such as 1,2-bis(2-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid. As expected (22), the I_A current was small or absent in neurons stepped from the holding potential of 70 mV, which is insufficient for full activation of this channel. Finally, our results show that, after dissociation, RVL neurons retain at least some ligand-gated channels. The increase of membrane conductance and the inward current observed after application of glutamate or kainic acid corresponds to the results of the previous studies that demonstrated high sensitivity of vasomotor RVL neurons to extracellularly applied excitatory amino acids (e.g., Ref. 35).

In conclusion, our results demonstrate that acute dissociation of RVL neurons provides a new and useful model for studying properties of these cells. We found that such neurons can be successfully labeled with fluorescent microspheres and therefore positively identified as bulbospinal and that single-cell RT-PCR provides an extremely useful tool to study their mRNA expression. The RT-PCR approach was found to be more practical than immunocytochemistry when dealing with isolated neurons. Dissociated RVL neurons retain the major classes of voltage- and ligand-sensitive ionic channels, previously identified in studies conducted in tissue slices, but future electrophysiological and pharmacological studies are needed to analyze these currents in more detail. Finally, we found that most isolated RVL neurons can fire tonically when depolarized with current pulses but normally show no pacemaker-like activity. It remains to be examined why the autorhythmicity reported in RVL neurons studied in tissue slices is not present in vivo or in isolated neurons.