Biophysical and histological determinants underlying natural firing behaviors of splanchnic sympathetic preganglionic neurons in neonatal rats

Neuroscience 150 (2007) 926 –937

BIOPHYSICAL AND HISTOLOGICAL DETERMINANTS UNDERLYING NATURAL FIRING BEHAVIORS OF SPLANCHNIC SYMPATHETIC PREGANGLIONIC NEURONS IN NEONATAL RATS C.-K. SU,* Y.-W. CHENG AND S. LIN

plays cardiac or respiratory rhythmicity (Gilbey et al., 1982; Backman and Henry, 1984; Boczek-Funcke et al., 1992; Pilowsky et al., 1994). Major determinants for these sympathetic rhythmicities are attributed to supraspinal sources (Dembowsky, 1995; Barman et al., 2005; Marina et al., 2006). Nonetheless, propriospinal neural ensembles also partly contribute to SND genesis by sustaining a low-level ongoing activity (Ardell et al., 1982; Dembowsky et al., 1985). In a decentralized and deafferented fragment of cat spinal cord, some SPNs reveal irregular, spontaneous firing (Mannard and Polosa, 1973). Such an SPN activity that is intrinsic to the spinal cord has been confirmed by observations obtained from in vitro approaches using spinal cord slices or hemisected spinal cords, showing a population of SPNs with spontaneous firing (McKenna and Schramm, 1983; Spanswick and Logan, 1990; Shen et al., 1994). One of the goals of this study was to determine whether the firing activities of SPNs in isolated spinal cords were rhythmic, which could be manifested as the quasiperiodicity of SND. From morphological viewpoints, anatomical features of SPNs tell how they can integrate synaptic influences to generate firing. SPNs are distributed in regions of the intermediolateral cell column (IML), the intermediomedial gray matter (IMM), and the central autonomic nuclei (Torigoe et al., 1985; Li et al., 1993). SPNs have somatal shape varying from fusiform to round and dendrites oriented longitudinally in the rostrocaudal direction, medially toward the central canal (cc), and laterally to the lateral funiculus (lfu); (Forehand, 1990; Pilowsky et al., 1992; Li et al., 1993; Tang et al., 1995). Distinct patterns of dendritic arborizations have been noted for SPNs with somata located in different regions, suggesting different neural connectivity (Forehand, 1990). In addition, few SPNs have intraspinal axon collaterals (Forehand, 1990; Bogan and Cabot, 1991; Llewellyn-Smith et al., 1995); these axon collaterals provide an anatomical substrate for local circuit interactions. Whether SPNs possess some anatomical features that correlate with their firing behaviors was not clear. Isolated thoracic spinal cords of neonatal rats spontaneously generate SND in splanchnic sympathetic nerves, which encodes a ⬃1-Hz quasiperiodic rhythm (Su et al., 2003). Using the in vitro nerve-cord preparation that made stable recordings of SPNs more feasible, we sought to determine the main biophysical factors and morphological features underlying their natural firing behaviors. Gaussian analyses of interspike intervals (ISIs) were used as a feature extraction of SPN firing. Using the dominant firing

Neuroscience Division, Institute of Biomedical Sciences, Academia Sinica, Taipei 11529, Taiwan, Republic of China

Abstract—Isolated thoracic spinal cords of neonatal rats spontaneously generate splanchnic sympathetic nerve discharge (SND) with a quasiperiodic rhythm ⬃1-Hz. Using in vitro nerve-cord preparations that retained T6 –T12 spinal segments, we investigated whether the natural firing behavior of sympathetic preganglionic neurons (SPNs) encoded the SND rhythm and what were the main biophysical and histological determinants of SPN firing. Under extracellular recording conditions, electrical stimulation of splanchnic nerves elicited antidromic responses in 212 SPNs. Among them, 92 SPNs were quiescent; 120 active SPNs had an average firing rate of 0.72ⴞ0.04 Hz, which was close to the quasiperiodic rhythm of SND. SPNs with rhythmic burst firing were rare. Probability plots of interspike intervals were constructed to extract mathematical features underlying SPN firing. Most active SPNs (88%) had a firing well described by unimodal Gaussian, suggesting a predominantly tonic pattern with normal variations. Biophysical properties of 112 SPNs were measured under whole-cell recording conditions. The charging time constant, ␶, is positively correlated with the average firing rate. Histological properties were examined in 45 SPNs with intracellular diffusion of Lucifer Yellow or biocytin. SPNs with pyramidal somata and multipolar dendrites tend to be spontaneously active. In contrast, those with bipolar somata and fewer dendritic branches were quiescent in firing. These observations suggest that activity levels of SPNs are correlated with their capacity for temporal and spatial summation of synaptic inputs. How the seemingly tonic firing of individual SPNs is integrated into whole-nerve SND with quasiperiodic rhythms is discussed. © 2007 IBRO. Published by Elsevier Ltd. All rights reserved. Key words: autonomic nervous system, spinal cord, rhythm, passive membrane property, somatal types.

Sympathetic nerve discharge (SND) is generally characterized with a quasiperiodic rhythm. How the activities of individual sympathetic preganglionic neurons (SPNs) were amalgamated to generate quasiperiodic SND was not fully understood. SPN firing in anesthetized cats or rats dis*Corresponding author. Tel: ⫹886-2-2789-9123; fax: ⫹886-2-2782-9224. E-mail address: [email protected] (C.-K. Su). Abbreviations: cc, central canal; Cm, whole-cell membrane capacitance; DAB, diaminobenzidine; IML, intermediolateral cell column; IMM, intermediomedial gray matter; ISI, interspike interval; lfu, lateral funiculus; Rm, whole-cell input resistance; RMP, resting membrane potential; Rs, access resistance of whole-cell recording; Rt, total resistance of whole-cell recording; SND, sympathetic nerve discharge; SPN, sympathetic preganglionic neuron; T, thoracic spinal segment; ␶, charging time constant.

0306-4522/07$32.00⫹0.00 © 2007 IBRO. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.neuroscience.2007.10.011

926

C.-K. Su et al. / Neuroscience 150 (2007) 926 –937

rates as the features of SPN firing, we constructed a probability plot of population firing. By revealing the firing properties at a population level, we attempted to resolve how amalgamation of individual SPN firing could generate a whole-nerve SND with quasiperiodic rhythms.

927

disappeared when a spontaneous spike appeared within the propagation time of antidromic action potentials. In practice, the fixed onset latency was taken as a conservative estimate of the critical delay for the occurrence of spontaneous spike to collide with the antidromic spike (Lipski, 1981; Cheng et al., 2005).

Measurements of biophysical properties

EXPERIMENTAL PROCEDURES General procedures In vitro sympathetic nerve–thoracic spinal cord preparations were obtained from 165 Sprague–Dawley rats (1–7 days old). All protocols were approved by the Institutional Animal Care and Utilization Committee of Academia Sinica (Protocol#: PRaIBMSC2003014), which conformed to the ILAR guidelines of the National Research Council (USA) on the ethical use of laboratory animals. Efforts were made to minimize the number of animals used and their suffering. Surgical procedures were as previously described (Cheng et al., 2005). During dissection, the thoracic spinal cord (T6 –T12) was immersed in 10 °C artificial cerebrospinal fluid (aCSF; in mM: 120 NaCl, 2.8 KCl, 1.4 CaCl2, 0.9 MgSO4, 22 NaHCO3, 0.5 NaH2PO4, 28 D-glucose, and 3 ascorbate; equilibrated with 95% O2–5% CO2). The distal ends of the splanchnic nerves were cut at a level proximal to the celiac ganglion. During experiments, the bath temperature was maintained at 24.5⫾1 °C.

Extracellular and whole-cell recording Splanchnic SPNs were recorded using extracellular or whole-cell recording techniques. Patch pipettes with a resistance of 5– 8 M⍀ were pulled from a borosilicate glass (AM-System, 6170, Carlsborg, WA, USA) using a horizontal puller (P-97, Sutter Instrument, Novato, CA, USA). The pipette solution contained (in mM) 132 potassium gluconate, 10 NaCl, 5 EGTA, 0.5 CaCl2, 4 MgATP, 0.3 Na3GTP and 10 Hepes (adjusted to pH 7.3 with 1 N KOH). In some experiments, pipette solution also contained 0.3% Lucifer Yellow or 0.6% biocytin for intracellular staining. Membrane currents or potentials were amplified using Axopatch 200B (Axon Instruments; Molecular Devices Corp., Sunnyvale, CA, USA) with CV-203BU head stage mounted on a 3-D micromanipulator. The patch pipette was step-advanced (2 ␮m) into the tissue and positioned by a motion controller (PMC100, Newport Corp., Irvine, CA, USA), which also read the depth of pipette tip from the dorsal surface of the spinal cord. Signals were low-pass filtered at 5 kHz and processed using a data acquisition system (pClamp 6.0, Axon Instruments) and analyzed using Axograph (version 4.9; Axon Instruments). Prior to achieving whole-cell recording configuration, extracellular signals were obtained when pipettes were gently attached to the recorded cells.

SND recording and antidromic activation of splanchnic SPNs Whole-bundle splanchnic nerves were introduced into a suction electrode for SND recording or antidromic activation of SPNs. Neural signals recorded from the nerves were amplified, filtered (bandpass: 0.1–1 kHz; WPI-DAM50, World Precision Instruments, Stevenage, Hertfordshire, UK), and stored in a pulse-code modulation tape recorder (Neuro-Corder DR-890; Cygnus Technology Inc., Delaware Water Gap, PA, USA) for off-line analysis. Procedures for power spectral analysis of the SND envelope were as previously described (Su et al., 2003). To verify the recorded neurons, splanchnic nerves were electrically stimulated (square pulse: ⱕ100 ␮A, 0.2 ms in duration, 0.5 Hz). As shown in Fig. 1A, a splanchnic SPN was confirmed using the following criteria: 1) the spinal neuron had an evoked spike with a fixed onset latency following the stimulation; and 2) the evoked spike was collided or

In voltage-clamp mode with a holding potential of ⫺60 mV (corrected for junction potentials), we measured whole-cell membrane capacitance (Cm) and input resistance (Rm). The total resistance (Rt) of whole-cell recording was determined by measuring the current responses elicited by voltage step pulses (⫺10 mV, 70 ms). Access resistance (Rs) and Cm were read from Axopatch 200B after compensation adjustment to reduce the transient current responses elicited by 5 mV step pulses. Rm was calculated by subtracting Rs from Rt. The charging time constant (␶) was calculated as a product of Cm⫻Rm.

Average firing rate and Gaussian analysis of SPN firing Spontaneous spike potentials that appeared under extracellular recording conditions for a period of 4 –10 min were taken for firing behavioral analyses. With the aid of software (Axograph 4.9), spike peaks or time locations of their occurrence were detected and used to obtain average firing rates or ISIs. Evaluation of moment-to-moment variations in firing was achieved by Gaussian analyses of ISIs. An ISI probability plot was constructed with the aid of Origin (version 7.5; OriginLab, Northampton, MA, USA). First, ISIs were classified into categories with step increments of 0.05– 0.25 s. Second, the probability of each ISI category was calculated accordingly. Third, the ISI probability plot was fitted using an amplitude version of Gaussian curve fitting algorithm, y⫽A EXP[⫺(x⫺xC)2/(2w2)]; y, the probability of ISI falling into a category; A, the maximal probability at the mode; xC, the mode, i.e. the ISI where the maximal probability occurs; w, half-maximal width, i.e. the ISI variation at the half-maximal probability. The goodness of fit was evaluated by the least chi-square protocol, setting the degrees of freedom (df) as: df⫽k⫺1⫺number of fitted parameters, where k is the number of ISI categories.

Histology and SPN image reconstruction At the end of electrophysiological experiments, thoracic spinal cords were immersed in 4% paraformaldehyde overnight and cryoprotected with 30% sucrose in 0.1 M phosphate buffer saline (pH 7.4). Frozen tissue blocks were cut into transverse or horizontal sections at 40 ␮m thicknesses. The diaminobenzidine (DAB) reaction was used to reveal biocytin. Signals were amplified by ABC solution (1:100; Vector Laboratories Inc., Burlingame, CA, USA) and revealed by a DAB reaction with peroxide generated by glucose oxidase (Minson et al., 1996). At the end of the DAB reaction, sections were washed and mounted onto gelatin-coated slides, dehydrated in ethanol, cleared in xylene, and coverslipped. Photos of sections were taken using a digital camera (Nikon Coolpix 4500, Tokyo, Japan) mounted on a light microscope with 6.3– 40⫻ objectives (Leitz Laborlux D, Wetzlar, Germany). Montage of high magnification photos used low magnification photos as guides. SPN images were reconstructed using the Magic Wand tool of Adobe Photoshop (version 8.0.1; Adobe Systems Inc., San Jose, CA, USA) for automatic selection of DAB signals on the photos. Terms and abbreviations of the histological structures of the cord were named according to the nomenclature described by Paxinos et al. (1991).

Statistics Student’s t-test was used for simple comparisons between properties of active and quiescent SPNs. Linear regression analysis

928

C.-K. Su et al. / Neuroscience 150 (2007) 926 –937

Fig. 1. Electrophysiological and histological identifications of a splanchnic SPN. (A) Extracellular spike potentials elicited by electrical stimulation of splanchnic sympathetic nerves. (Ai) Six traces were superimposed to show the appearance of evoked spike potentials at a constant onset latency (38 ms). (Aii–v) Traces show the appearance of spontaneous spike potentials in the critical period. Note the absence or collision of the evoked spike potentials. (B) Morphology of an SPN displayed in transverse sections of T9 spinal segment. The soma is located in the lateral region of the gray matter near the lfu, and at a horizontal level parallel to the cc. Two long dendrites coursing along the transverse plane project to the dorsal lamina X, i.e. the region dorsal to the cc. (C) Consecutive traces of spontaneously generated extracellular spike potentials, showing a tonic firing pattern with an average firing rate ⬃1 Hz. (D) Spontaneous action potentials under whole-cell recording conditions. Same time scale is used in (C, D). Average firing rate was increased to ⬃2.7 Hz after this SPN was brought into whole-cell recording conditions.

was used to test the correlation between average firing rates and biophysical properties. A non-random association of SPN firing behaviors with different somatal types was verified by chi-square tests. P values ⬍0.05 were considered significant. Unless otherwise mentioned, values are presented as means⫾S.E.M.

RESULTS Spontaneous firing behaviors observed under extracellular recording conditions Extracellular recordings were made on 212 splanchnic SPNs. Among them, 120 SPNs (56.6%) were spontaneously active under extracellular recording conditions. An example of active SPN is shown in Fig. 1. Average firing rate of these active SPNs was 0.72⫾0.04 Hz (range: 0.08 – 2.32 Hz). Quiescent SPNs did not reveal spontaneous firing under extracellular conditions; however, they always became active when they were brought into whole-cell recording conditions (Fig. 2). To avoid the interferences incurred by whole-cell recording techniques, observations obtained from extracellular recording were taken into ac-

count for firing behavioral analyses. The firing patterns of extracellular spike potentials for most active SPNs were apparently tonic (Fig. 1). Only 7 of 120 active SPNs (5.8%) revealed bursting firing. Fig. 3 shows an example of SPN with firing in bursts. Unimodal and bimodal firing properties of SPNs Mathematical features underlying the spontaneous firing behaviors were extracted by analyses of time intervals between consecutive firing. The ISIs varied considerably along the time course, which led to a probability plot in bell-shape (Fig. 4). We evaluated whether the plots could be described as Gaussian distributions. Fig. 4 shows examples that ISI probability plots are well fitted by unimodal or bimodal Gaussian curves. Most active SPNs (105/120⫽87.5%), including all with bursting firing, had plots best described by unimodal distributions. Only a few active SPNs (15/120⫽ 12.5%) had ISI probability plots that were bimodal. Table 1 summarizes the values of Gaussian fitting parameters for SPNs with unimodal or bimodal firing behaviors.

C.-K. Su et al. / Neuroscience 150 (2007) 926 –937

929

Fig. 2. A quiescent SPN with elongated soma. (A) Morphology of an SPN. The flat-triangular soma was located in the lateral region of the gray matter. Extensive dendritic arborizations around the soma mainly directed ventrally. Two long transverse dendrites formed extensive terminal branches in dorsal lamina X. Open arrowheads indicate an axon projected ventromedially. (B) Traces of firing activities. (Bi) Antidromic spike potentials appeared with constant latency. (Bii) Absence of spontaneous firing under extracellular recording conditions. (Biii) Waveform of action potentials showing prominent after hyperpolarization. The waveform was obtained from averaging 32 action potentials. (Biv) Spontaneous appearance of action potentials under whole-cell recording conditions. Fluctuations of membrane potential trajectory reveal abundant synaptic activities impinged upon this SPN. Arrows indicate excitatory synaptic potentials.

Bimodal population firing properties

Biophysical properties of SPNs

To describe the population firing property, the average firing rates of all active SPNs (n⫽120) or the ISI modes of those SPNs with unimodal firing (n⫽105) were used as features of individual SPNs to construct the plots of population firing. The inverse of ISI mode was calculated to obtain Gaussian firing rate. Fig. 5 shows that the probability plots of population firing obtained from average and Gaussian firing rates were well described by bimodal Gaussian distributions. The plot of average firing rates showed peak probabilities at 0.483⫾0.025 and 1.197⫾0.223 Hz. The plot of Gaussian firing rates showed peak probabilities at 0.634⫾0.016 and 1.226⫾0.195 Hz. These two plots also had comparable half-maximal widths with higher peak probability predominantly at lower frequency ranges. For comparison, Fig. 6 shows a typical example of power spectral analysis of wholenerve SND, revealing a dominant rhythm of population firing at 0.78 Hz.

Evaluation of biophysical properties of SPNs was based on a data pool of 212 SPNs, in which 112 SPNs were successfully brought into whole-cell recording conditions. We measured the biophysical properties, including antidromic stimulus threshold, onset latency of antidromic action potentials, resting membrane potential (RMP), Cm, Rm, and ␶. Table 2 summarizes the biophysical properties classified according to spontaneous firing behaviors. Stimulus threshold, Cm, Rm, ␶, and RMP did not differ between active and quiescent SPNs. However, active SPNs had significantly earlier onset latency (active vs. quiescent: 31⫾0.6 vs. 33.2⫾0.7 ms, P⬍0.01). Correlation analysis of SPN firing rates with their biophysical properties Whether activity levels of SPNs correlated with their biophysical properties was evaluated. For simplification, we sought to determine whether a linear relationship existed

930

C.-K. Su et al. / Neuroscience 150 (2007) 926 –937

Fig. 3. Spontaneously active SPN with rhythmic bursts. (A) Morphology of the SPN. The outline indicates the border edge of gray matter. This SPN has a pyramidal soma and two long transverse dendrites projecting toward dorsal lamina X. The axon coursing ventromedially gives rise to one collateral branch in the ventral horn (arrowheads) and exits gray matter via the branch projecting ventrolaterally (arrows). (B) Bursts of spontaneous spike potentials under extracellular recording conditions. The bursting rhythm was ⬃0.028 Hz with duration ⬃8.8 s. (C) Spontaneous bursts of action potentials under whole-cell recording conditions. Cessation of bursting firing was concomitant with a prominent after-burst hyperpolarization.

between average firing rates and Rm, Cm, or ␶. In the first data pool, as shown by the solid lines in Fig. 7, analyses

included both active and quiescent SPNs (n⫽112). Average firing rates showed a positive linear relationship with

Fig. 4. Gaussian analyses of SPN firing behaviors. (A) Unimodal firing pattern. (B) Bimodal firing pattern. In both panels, (i) consecutive traces of spontaneous spike potentials; (ii) superimposed traces showing the consistency in waveforms of evoked spike potentials (n⫽6, top) and spontaneous spike potentials (n⫽64, bottom); (iii) probability plots of ISIs. (Aiii) The probability plot was well described by unimodal Gaussian curve with mode at 1.411 s, half-maximal width 0.474 s, and maximal probability 0.405 (correlation coefficient: r⫽0.9956; degrees of freedom: df⫽5; P⬍0.00001). (Biii) The probability plot was well described by bimodal Gaussian curve with modes at 0.777 and 1.393 s, half-maximal widths 0.123 and 0.321 s, and maximal probabilities 0.188 and 0.174, respectively (r⫽0.9975; df⫽7; P⬍0.00001).

C.-K. Su et al. / Neuroscience 150 (2007) 926 –937

931

Table 1. Parametric values of Gaussian curve fitting for ISI probability plots of active SPNs Parameters/models

ISI modes (s)

Half-maximal width (s)

Maximal probability

r

n

Unimodal Bimodal

1.596⫾0.099 1.025⫾0.177 1.976⫾0.404

1.030⫾0.104 0.301⫾0.086 0.596⫾0.178

0.330⫾0.039 0.169⫾0.023 0.123⫾0.012

0.9462⫾0.0060 0.9890⫾0.0032

105 15

Values are means⫾S.E.M. r, Correlation coefficient.

Cm (P⬍0.05) or ␶ (P⬍0.01), but not Rm. In the second data pool, as shown by the dash lines in Fig. 7, analyses only included active SPNs (n⫽61). The linear relationship between average firing rate and Cm or Rm was not significant; however, the linear relationship between average firing rate and ␶ was significant (P⬍0.01). Therefore, regardless the size of data pool, average firing rate was linearly correlated with ␶. Histological features of SPNs Forty-five splanchnic SPNs were made visible by intracellular diffusion of Lucifer Yellow (n⫽16) or biocytin (n⫽29). Among them, 40 SPNs were observed in transverse sections and five in horizontal sections. Fig. 8 shows a Lucifer

Fig. 5. Bimodal Gaussian distribution of population firing. (A) Probability plot of average firing rates. This plot was based on average firing rates of 120 active SPNs. (B) Probability plot of Gaussian firing rates. This plot was based on 105 active SPNs with unimodal firing behavior. Gaussian firing rates were obtained from the inverse of ISI modes. Both probability plots in representing population firing were well described by bimodal Gaussian curves.

Yellow–labeled SPN in a horizontal section. This SPN had a conical soma and extensive long dendrites directed rostrally. In either transverse or horizontal sections, different morphological types of SPNs were recognized. Bipolar SPNs with spherical somata had two primary dendrites with one directed dorsomedially and another laterally (Fig. 9A; n⫽4). Fusiform SPNs had spindle-shaped somata and two primary dendrites; some showed a soma originated axon that projected ventromedially (Fig. 10A; n⫽16). Elongated SPNs had irregular or flat-triangular somata that tapered gradually becoming main dendritic trunks directed ventrolaterally (Figs. 9B and 10B; n⫽4). Pyramidal SPNs had conical somata and three or more multipolar-directed dendrites (Figs. 9C and 10C–D; n⫽20). One rectangular SPN had a rod-like soma that oriented in nearly parallel to dorsoventral axis (Fig. 9D). This rectangular SPN had multipolar dendrites.

Fig. 6. Power spectral analysis of whole-nerve SND. (A) Original traces of the SND envelope (obtained from leaky integration of SND). Under control conditions, quasiperiodic oscillations of SND were apparent. The background noise of SND recording (lower trace) was determined after depolarization-blockade of action potential generation by adding 100 mM KCl into the bath solution. (B) Power spectra of SND oscillations. Gray lines, spectra of original signals. Black line, the spectrum of true neural signals (obtained after subtraction). In this example, the power spectrum reveals a dominant rhythm embedded in SND oscillations at 0.78 Hz.

932

C.-K. Su et al. / Neuroscience 150 (2007) 926 –937

Table 2. Biophysical properties of SPNs classified according to their firing behaviors Biophysical properties/SPNs

Stimulus threshold (␮A)

Onset latency (ms)

Cm (pF)

Rm (M⍀)

␶ (ms)

RMP (mV)

All Active Quiescent P

21.9⫾1.0 (212) 20.6⫾1.2 (120) 23.6⫾1.9 (92) 0.0795

31.9⫾0.5 (212) 31.0⫾0.6 (120) 33.2⫾0.7 (92) 0.0096*

2.9⫾0.2 (112) 3.1⫾0.2 (61) 2.6⫾0.3 (51) 0.1055

446⫾18 (112) 448⫾26 (61) 442⫾25 (51) 0.4371

1.25⫾0.10 (112) 1.34⫾0.12 (61) 1.15⫾0.15 (51) 0.1581

⫺45.5⫾0.8 (85) ⫺45.5⫾0.9 (49) ⫺45.4⫾1.3 (36) 0.4910

Values represent means⫾S.E.M. P values, obtained from one-tailed t-test, indicate the significant levels of comparisons between values of active and quiescent SPNs. Values in parentheses are SPN numbers. * P⬍0.01.

Although SPNs had distinct somatal shapes and dendritic arborizations, they commonly had two long transverse dendrites that directed dorsomedially. The transverse dendrites formed complex dendritic terminals in dorsal lamina X (Fig. 2A). A ventromedial projected axon was seen in some SPNs (Figs. 1B, 2A and 3A). Interestingly, axonal collaterals were observed in two SPNs. One SPN was found in the IML showing an axonal branch at the ventral horn (Fig. 3A). This SPN had bursting firing behavior. The other SPN with fusiform soma and tonic firing behavior was located in the IMM. This SPN had an axonal branch at lamina VII. Somatal types and firing patterns The relationship between somatal types and modal firing patterns of 33 active SPNs were examined. Thirty unimodal SPNs had various somatal types: 18 pyramidal, nine fusiform, two elongated, and one rectangular. Somatal types in one of the three bimodal SPNs were fusiform; in the other two, pyramidal (an example shown in Fig. 8). Among the 33 active SPNs, three SPNs revealed bursting firing patterns. All bursting SPNs had pyramidal somata (an example shown in Fig. 3). In summary, somatal types did not clearly relate to firing patterns. Morphological comparisons between active and quiescent SPNs Correlation of somatal types with the spontaneous firing behavior was examined in 45 SPNs. Table 3 summarizes

the numbers of active and quiescent SPNs in different somatal types. All pyramidal SPNs were spontaneously active (average firing rate: 0.62⫾0.11 Hz; n⫽20). One rectangular SPN with multipolar dendrites was spontaneously active (average firing rate: 0.35 Hz). Somatal types of quiescent SPNs varied; 6 of 16 fusiform, two of four elongated, and four of four bipolar SPNs were without spontaneous firing. Chi-square analyses indicated that the observations of all pyramidal SPNs that were spontaneously active in firing, and all bipolar SPNs that were quiescent in firing could not have arisen by chance (P⬍0.0005).

DISCUSSION This study explored the mathematical features of SPN firing behaviors. Most SPNs (88%) have their natural firing behaviors best described by unimodal Gaussians, suggesting a random firing behavior with predictable ‘normal’ variations. ␶, a biophysical property reflecting the capacity for temporal summation, was found to be positively correlated with average firing rates. SPNs with pyramidal somata and multipolar dendrites are more likely to have spontaneous firing activity. These morphological features in combination with the biophysical property, ␶, render SPNs their capacity for synaptic integration, and consequently, are the main determinants for their active or quiescent firing properties.

Fig. 7. Linear regression analyses between average firing rates and Cm, Rm or time constant (␶). In (A, B, C), analyses were based on the data from active SPNs only (dashed lines, n⫽61) or the data from both active and quiescent SPNs (solid lines, n⫽112). In analyses including all SPNs (solid lines), the average firing rate was positively correlated with Cm (y⫽2.61⫹0.79 x, P⬍0.05) or ␶ (y⫽1.06⫹0.55 x, P⬍0.01). However, a linear relationship between average firing rate and Rm was not significant (y⫽424⫹62 x, P⫽0.13). In analyses including active SPNs only (dashed lines), the average firing rate was not significantly correlated with Cm (y⫽2.54⫹0.87 x, P⫽0.10) or Rm (y⫽376⫹113 x, P⫽0.06). However, a linear and positive correlation was evident between average firing rate and ␶ (y⫽0.83⫹0.79 x, P⬍0.005).

C.-K. Su et al. / Neuroscience 150 (2007) 926 –937

933

Fig. 8. A spontaneously active SPN with bimodal firing behavior has multipolar-oriented dendrites and a soma in conical shape. (A) Highermagnification photo shows a Lucifer Yellow–labeled SPN in a horizontal spinal cord section. (B) Photos show the soma located at the IMM region. The longitudinal dendrites directed predominantly rostrally. (C) Bimodal firing property. (Ci) Traces show the consistency of spike potential waveforms. Top: Six superimposed traces of antidromic spike potentials. Bottom: Ten superimposed traces of spontaneous spike potentials. (Cii) Spontaneous occurrence of extracellular spike potentials shows substantial ISI variations. (Ciii) ISI probability plot. The plot was well described by a bimodal Gaussian curve with modes at 1.027 and 1.623 s, half-maximal widths 0.191 and 0.371 s, and maximal probabilities 0.128 and 0.142, respectively (r⫽0.9938; df⫽10; P⬍0.00001).

Methodological considerations Sampling bias is an inherent difficulty in studies using microelectrode recording techniques. Even though blind whole-cell recording techniques have provided a mean as random sampling, a tendency in picking up the SPNs with relatively large somata is inevitable. This uneven sampling that is tentatively biased toward larger cells may distort the true proportion of SPNs with various biophysical parame-

ters. Taking the bimodal distribution of population firing as shown in Fig. 5 as an example, the higher proportion in observing SPNs with lower firing rates might simply result from sampling bias. In addition, we used an in vitro model. Although advantages in using a reduced system to unravel complex SPN firing behaviors are obvious, gaps still remain in the interpretations given below, when we attempt to correlate

934

C.-K. Su et al. / Neuroscience 150 (2007) 926 –937

Fig. 9. Photos show morphological types of SPNs with Lucifer Yellow labeling in transverse spinal cord sections. All photos were oriented in a way as indicated by arrows in (D). (A) Bipolar SPN with spherical soma and slender dendrites directed dorsomedially or laterally. (B) Elongated SPN with irregular soma and one primary dendritic trunk directed laterally. (C) Pyramidal SPN with conical soma. (D) Rectangular SPN with rod-like soma in a nearly dorsoventral orientation. (A, B) Quiescent SPNs. (C, D) Active SPNs with average firing rate 0.36 and 0.35 Hz, respectively.

our findings with most of the observations that have been obtained from in vivo studies.

SPNs with multipolar dendrites and pyramidal somata are likely to be spontaneously active

␶ Is the major biophysical property determining firing activity

Although the finding that SPNs are morphological heterogeneous is not new, in correlation with firing properties, a surprising finding in this study is that SPNs with pyramidal somata and multipolar dendrites tend to be spontaneously active under our experimental conditions. Many laboratories have reported that the somatal shapes of SPNs in neonatal rats are in fusiform, oval or round, triangular, and multipolar (McKenna and Schramm, 1983; Forehand, 1990; Shen and Dun, 1990). These morphological features are largely similar in adult rats and other species (Deuschl and Illert, 1981; Bacon and Smith, 1988; Pilowsky et al., 1992; Li et al., 1993). On the assumption that the orientation of dendrites tells sources of synaptic inputs (Griffin et al., 2001), the observation of multipolar dendrites in some active SPNs implies that multiple synaptic drives in the cord are responsible for activity initiation. SPNs receive more excitatory but less inhibitory inputs (Dembowsky, 1995). Moreover, the observation that SPNs with larger ␶ tend to be more active suggests a positive correlation between activity levels of SPNs and their capacity for synaptic integration. Therefore, it is likely that pyramidal SPNs with multipolar dendrites are active because of their capacity for synaptic integration of excitatory inputs from multiple sources in the spinal cord.

Conventional notions in addressing recruitment order of motor units are attributed to the size principle (Bawa et al., 1984; Prather et al., 2002). However, whether or not SPNs are recruited in a set order has been questioned (McAllen and Trevaks, 2003). Rm and Cm are often considered as size-related properties (Kado, 1993; Liu et al., 1996; McDonagh et al., 2002; Fukami and Bradley, 2005). In principle, large-size neurons tend to have low Rm and high Cm (McLachlan and Meckler, 1989; Grigaliunas et al., 2002; Greenwood and Fernald, 2004). Hence, in receiving equal amounts of synaptic current, larger neurons with smaller Rm are less activated. Paradoxically, larger neurons by providing more anatomical substrates for synaptic formation may receive more synaptic inputs (Dityatev et al., 2001; Gibbins et al., 2003). In this scenario, the nature of synaptic influences is likely to override intrinsic sizerelated properties. The uncertainty of Rm or Cm in predicting SPN firing behaviors was noticed in this study. Instead, in SPNs, ␶ as a product of Cm⫻Rm is linearly correlated with the average firing rate. Therefore, SPN firing behavior is not simply determined by size-related properties. ␶ is a factor facilitating temporal summation of synaptic events. We observed that SPNs with higher ␶ tend to be more active. This observation suggests that integration of synaptic influences plays a key role in determining SPN firing nature. Therefore, the SPNs that possess optimal sizes, which enable them to receive sufficient synaptic inputs without excessive loss of Rm or increase of Cm, could be spontaneously active.

Significance of modal firing behaviors Most SPNs have unimodal firing behaviors. Similarly, the sympathetic premotor neurons in the subretrofacial nuclei of anesthetized cats also reveal unimodal firing behavior (McAllen et al., 2001). This feature indicates that the firing of most SPNs is centered in a dominant frequency with predictable or normal variations. The underlying process

C.-K. Su et al. / Neuroscience 150 (2007) 926 –937

935

Fig. 10. Photos show morphological types of SPNs with biocytin labeling in transverse spinal cord sections. All SPNs have dendrites directed dorsomedially. (A) Fusiform SPN with spindle-shaped soma and one dendritic trunk directed ventrolaterally. (B) Elongated SPN with flat triangular soma and bipolar-oriented dendritic trunks. (C, D) Pyramidal SPNs with conical soma and multipolar-oriented dendrites. (A) Quiescent SPN. (B–D) Active SPNs with average firing rate 0.58, 1.32, and 0.32 Hz, respectively.

that caused the variations of firing was not a focus of studies here. Nonetheless, SPNs have been shown to receive abundant synaptic influences (Dembowsky et al., 1985; Dun and Mo, 1989; Spanswick et al., 1994) and exhibit prominent afterhyperpolarizations (Yoshimura et al., 1986; Shen et al., 1994; Sah and McLachlan, 1995). Different extents of afterhyperpolarizations could deter-

mine the dominant repetitive firing rates of individual SPNs, which would be varied by synaptic influences. On this assumption, the unimodal firing could result from a random variation in the strength of synaptic influences. In contrast to unimodal SPNs showing the firing pattern that could be mainly attributed to intrinsic properties, bimodal SPNs are likely to receive an extra command to

Table 3. Chi-square analysis of SPN firing behaviors in correlation to somatal types Firing properties/somatal types

Active

Quiescent

Total

E(active)

E(quiescent)

P

Bipolar Elongated Fusiform Pyramidal Rectangular Total

0 2 10 20 1 33

4 2 6 0 0 12

4 4 16 20 1 45

2.9 2.9 11.7 14.7 0.7

1.1 1.1 4.3 5.3 0.3

0.0009** 0.2913 0.3271 0.0070* 0.5465

Values are number of SPNs. E(active) and E(quiescent) are expected values. P values indicate the probability that the assortment of observations occurs simply by chance. * P⬍0.01. ** P⬍0.001.

936

C.-K. Su et al. / Neuroscience 150 (2007) 926 –937

initiate distinct firing behavior. Using average or Gaussian firing rates as features of individual neuronal firing, the population firing of SPNs was best described by bimodal Gaussian distribution (Fig. 5). Strikingly, dominant population firing rates as revealed by the two modes of population firing distribution (in Fig. 5; modes of average/Gaussian firing rates: 0.483/0.634 and 1.197/1.226 Hz) were comparable to the inverse of peak ISIs of those SPNs with bimodal firing (in Table 1; after conversion of bimodal peak ISIs: 0.51 and 0.98 Hz). This observation implies that the bimodal firing of individual SPNs is driven by dual sources, which also drive population firing. The sources for SPN firing initiation were not determined in this study. Some SPNs have pacemaker-like activities (Spanswick and Logan, 1990; Shen et al., 1994). Spinally generated synaptic inputs to SPNs are also abundant (an example as we have observed in Fig. 2Biv; arrows indicate). Whether the bimodal firing behavior is caused by the interplay of intrinsic pacemaking and extrinsic synaptic activities awaits further studies. Ensembles of SPN firing and SND rhythms Sympathetic activities are rhythmic (Gebber and Barman, 1980; Gootman and Cohen, 1981; Weaver and Stein, 1989; Allen et al., 1993; Ootsuka et al., 1995; Chang et al., 1999; Kunitake and Kannan, 2000; Barman et al., 2005). Power spectral analysis of splanchnic SND in conscious rats indicates sympathetic rhythms occurring at ⬃1 Hz (Persson et al., 1992). Using in vitro preparations, our previous study has demonstrated that the spinally generated splanchnic SND encodes a dominant rhythm of ⬃1 Hz (Su et al., 2003). Here, an example was shown in Fig. 6. Interestingly, this in vitro rhythm is comparable with the sympathetic rhythms obtained from other in vivo studies (Persson et al., 1992; Smith and Gilbey, 2000). The sympathetic rhythm encoded in a wholenerve activity seems to be a population phenomenon, which is not usually evident in the firing of individual neurons (McAllen and Malpas, 1997). Our observations here support this view. We found that the firing of 113 of 120 SPNs was apparently tonic. Only 7 of 120 SPNs exhibited rhythmic bursts. Ensemble of SPN firing as depicted by the distribution of population firing rates was prominent at ⬃0.5–1.2 Hz (Fig. 5). This is a frequency range falling into the rhythms of whole-nerve discharges (Su et al., 2003). We did not determine how the individual SPN activities are integrated into rhythmic discharges. In 8 to 14-day-old rats, electrical synapses between SPNs enable synchronized firing of action potentials (Logan et al., 1996; Nolan et al., 1999). Functional electrical synapses between fast-spiking GABAergic neurons are present in adult mouse neocortex (Galarreta and Hestrin, 2002). Although a functional link between electrical synapses and SPN firing in adult animals remains to be established, a coherent SPN firing resulted from signaling via electrical synapses may partly resolve the quasiperiodic rhythmogenesis in whole-nerve SNDs. This speculation remains to be confirmed.

CONCLUSION In isolated spinal cords in vitro, activity levels of SPNs are correlated with their capacity to integrate synaptic events. Their natural firing behaviors are best described by unimodal Gaussian. How the seemingly tonic firing of individual SPNs is integrated into quasiperiodic SND remains to be determined. Acknowledgments—We are grateful to Dr. C.-Y. Chai for general support. This work was supported by intramural funding of Academia Sinica, and partly supported by a grant from National Science Council of Republic of China (NSC 93-2320-B-001-024).

REFERENCES Allen AM, Adams JM, Guyenet PG (1993) Role of the spinal cord in generating the 2- to 6-Hz rhythm in rat sympathetic outflow. Am J Physiol Regul Integr Comp Physiol 33:R938 –R945. Ardell JL, Barman SM, Gebber GL (1982) Sympathetic nerve discharge in chronic spinal cat. Am J Physiol 243:H463–H470. Backman SB, Henry JL (1984) Physiological properties of sympathetic preganglionic neurones in the thoracic intermediolateral nucleus of the cat. Can J Physiol Pharmacol 62:1183–1193. Bacon SJ, Smith AD (1988) Preganglionic sympathetic neurones innervating the rat adrenal medulla: immunocytochemical evidence of synaptic input from nerve terminals containing substance P, GABA or 5-hydroxytryptamine. J Auton Nerv Syst 24:97–122. Barman SM, Orer HS, Gebber GL (2005) Role of medullary excitatory amino acid receptors in mediating the 10-Hz rhythm in sympathetic nerve discharge of cats. Brain Res 1049:249 –253. Bawa P, Binder MD, Ruenzel P, Henneman E (1984) Recruitment order of motoneurons in stretch reflexes is highly correlated with their axonal conduction velocity. J Neurophysiol 52:410 – 420. Boczek-Funcke A, Dembowsky K, Häbler H-J, Jänig W, McAllen RM, Michaelis M (1992) Classification of preganglionic neurones projecting into the cat cervical sympathetic trunk. J Physiol (Lond) 453:319 –339. Bogan N, Cabot JB (1991) Light and electron microscopic analysis of intraspinal axon collaterals of sympathetic preganglionic neurons. Brain Res 541:241–251. Chang H-S, Staras K, Smith JE, Gilbey MP (1999) Sympathetic neuronal oscillators are capable of dynamic synchronization. J Neurosci 19:3183–3197. Cheng Y-W, Ku M-C, Ho C-M, Chai C-Y, Su C-K (2005) GABABreceptor-mediated suppression of sympathetic outflow from the spinal cord of neonatal rats. J Appl Physiol 99:1658 –1667. Dembowsky K (1995) Integrative properties of sympathetic preganglionic neurones within the thoracic spinal cord. Clin Exp Hypertens 17:313–321. Dembowsky K, Czachurski J, Seller H (1985) An intracellular study of the synaptic input to sympathetic preganglionic neurones of the third thoracic segment of the cat. J Auton Nerv Syst 13:201–244. Deuschl G, Illert M (1981) Cytoarchitectonic organization of lumbar preganglionic sympathetic neurons in the cat. J Auton Nerv Syst 3:193–213. Dityatev AE, Chmykhova NM, Dityateva GV, Babalian AL, Kleinle J, Clamann HP (2001) Structural and physiological properties of connections between individual reticulospinal axons and lumbar motoneurons of the frog. J Comp Neurol 430:433– 447. Dun NJ, Mo N (1989) Inhibitory postsynaptic potentials in neonatal rat sympathetic preganglionic neurones in vitro. J Physiol (Lond) 410:267–281. Forehand CJ (1990) Morphology of sympathetic preganglionic neurons in the neonatal rat spinal cord: an intracellular horseradish peroxidase study. J Comp Neurol 298:334 –342.

C.-K. Su et al. / Neuroscience 150 (2007) 926 –937 Fukami H, Bradley RM (2005) Biophysical and morphological properties of parasympathetic neurons controlling the parotid and von Ebner salivary glands in rats. J Neurophysiol 93:678 – 686. Galarreta M, Hestrin S (2002) Electrical and chemical synapses among parvalbumin fast-spiking GABAergic interneurons in adult mouse neocortex. Proc Natl Acad Sci U S A 99:12438 –12443. Gebber GL, Barman SM (1980) Basis for 2– 6 cycle/s rhythm in sympathetic nerve discharge. Am J Physiol 239:R48 –R56. Gibbins IL, Teo EH, Jobling P, Morris JL (2003) Synaptic density, convergence, and dendritic complexity of prevertebral sympathetic neurons. J Comp Neurol 455:285–298. Gilbey MP, Peterson DF, Coote JH (1982) Some characteristics of sympathetic preganglionic neurones in the rat. Brain Res 241:43–48. Gootman PM, Cohen MI (1981) Sympathetic rhythms in spinal cats. J Auton Nerv Syst 3:379 –387. Greenwood AK, Fernald RD (2004) Social regulation of the electrical properties of gonadotropin-releasing hormone neurons in a cichlid fish (Astatotilapia burtoni). Biol Reprod 71:909 –918. Griffin JD, Saper CB, Boulant JA (2001) Synaptic and morphological characteristics of temperature-sensitive and -insensitive rat hypothalamic neurones. J Physiol (Lond) 537:521–535. Grigaliunas A, Bradley RM, MacCallum DK, Mistretta CM (2002) Distinctive neurophysiological properties of embryonic trigeminal and geniculate neurons in culture. J Neurophysiol 88:2058 –2074. Kado RT (1993) Membrane area and electrical capacitance. Methods Enzymol 221:273–299. Kunitake T, Kannan H (2000) Discharge pattern of renal sympathetic activity in the conscious rat: spectral analysis of integrated activity. J Neurophysiol 84:2859 –2867. Li YW, Ding ZQ, Wesselingh SL, Blessing WW (1993) Renal sympathetic preganglionic neurons demonstrated by herpes simplex virus transneuronal labelling in the rabbit: close apposition of neuropeptide Yimmunoreactive terminals. Neuroscience 53:1143–1152. Lipski J (1981) Antidromic activation of neurones as an analytic tool in the study of the central nervous system. J Neurosci Methods 4:1–32. Liu RH, Yamuy J, Engelhardt JK, Xi MC, Morales FR, Chase MH (1996) Cell size and geometry of spinal cord motoneurons in the adult cat following the intramuscular injection of adriamycin: comparison with data from aged cats. Brain Res 738:121–130. Llewellyn-Smith IJ, Pilowsky PM, Minson JB, Chalmers J (1995) Synapses on axons of sympathetic preganglionic neurons in rat and rabbit thoracic spinal cord. J Comp Neurol 354:193–208. Logan SD, Pickering AE, Gibson IC, Nolan MF, Spanswick D (1996) Electronic coupling between rat sympathetic preganglionic neurones in vitro. J Physiol (Lond) 495:491–502. Mannard A, Polosa C (1973) Analysis of background firing of single sympathetic preganglionic neurons of cat cervical nerve. J Neurophysiol 36:398 – 408. Marina N, Taheri M, Gilbey MP (2006) Generation of a physiological sympathetic motor rhythm in the rat following spinal application of 5-HT. J Physiol (Lond) 571:441– 450. McAllen RM, Malpas SC (1997) Sympathetic burst activity: characteristics and significance. Clin Exp Pharmacol Physiol 24:791–799. McAllen RM, Trevaks D (2003) Are pre-ganglionic neurones recruited in a set order? Acta Physiol Scand 177:219 –225. McAllen RM, Trevaks D, Allen AM (2001) Analysis of firing correlations between sympathetic premotor neuron pairs in anesthetized cats. J Neurophysiol 85:1697–1708. McDonagh JC, Hornby TG, Reinking RM, Stuart DG (2002) Associations between the morphology and physiology of ventral-horn neurons in the adult turtle. J Comp Neurol 454:177–191.

937

McKenna KE, Schramm LP (1983) Sympathetic preganglionic neurons in the isolated spinal cord of the neonatal rat. Brain Res 269:201–210. McLachlan EM, Meckler RL (1989) Characteristics of synaptic input to three classes of sympathetic neurone in the coeliac ganglion of the guinea-pig. J Physiol (Lond) 415:109 –129. Minson J, Arnolda L, Llewellyn-Smith I, Pilowsky P, Chalmers J (1996) Altered c-fos in rostral medulla and spinal cord of spontaneously hypertensive rats. Hypertension 27:433– 441. Nolan MF, Logan SD, Spanswick D (1999) Electrophysiological properties of electrical synapses between rat sympathetic preganglionic neurones in vitro. J Physiol (Lond) 519:753–764. Ootsuka Y, Xu T, Terui N (1995) The spinally mediated 10-Hz rhythm in the sympathetic nerve activity of cats. J Auton Nerv Syst 54:89 –103. Paxinos G, Törk I, Tecott LH, Valentino KL (1991) Atlas of the developing rat brain. San Diego: Academic Press, Inc. Persson PB, Stauss H, Chung O, Wittmann U, Unger T (1992) Spectrum analysis of sympathetic nerve activity and blood pressure in conscious rats. Am J Physiol 263:H1348 –H1355. Pilowsky P, Llewellyn-Smith IJ, Lipski J, Chalmers J (1992) Substance P immunoreactive boutons form synapses with feline sympathetic preganglionic neurons. J Comp Neurol 320:121–135. Pilowsky PM, Llewellyn-Smith IJ, Arnolda L, Minson J, Chalmers J (1994) Intracellular recording from sympathetic preganglionic neurons in cat lumbar spinal cord. Brain Res 656:319 –328. Prather JF, Clark BD, Cope TC (2002) Firing rate modulation of motoneurons activated by cutaneous and muscle receptor afferents in the decerebrate cat. J Neurophysiol 88:1867–1879. Sah P, McLachlan EM (1995) Membrane properties and synaptic potentials in rat sympathetic preganglionic neurons studied in horizontal spinal cord slices in vitro. J Auton Nerv Syst 53:1–15. Shen E, Dun NJ (1990) Neonate rat sympathetic preganglionic neurons intracellularly labelled with Lucifer yellow in thin spinal cord slices. J Auton Nerv Syst 29:247–254. Shen E, Wu SY, Dun NJ (1994) Spontaneous and transmitter-induced rhythmic activity in neonatal rat sympathetic preganglionic neurons in vitro. J Neurophysiol 71:1197–1205. Smith JE, Gilbey MP (2000) Coherent rhythmic discharges in sympathetic nerves supplying thermoregulatory circulations in the rat. J Physiol (Lond) 523:449 – 457. Spanswick D, Logan SD (1990) Spontaneous rhythmic activity in the intermediolateral cell nucleus of the neonate rat thoracolumbar spinal cord in vitro. Neuroscience 39:395– 403. Spanswick D, Pickering AE, Gibson IC, Logan SD (1994) Inhibition of sympathetic preganglionic neurons by spinal glycinergic interneurons. Neuroscience 62:205–216. Su C-K, Phoon S-L, Yen C-T (2003) Identification of active thoracic spinal segments responsible for tonic and bursting sympathetic discharge in neonatal rats. Brain Res 966:288 –299. Tang FR, Tan CK, Ling EA (1995) A light-microscopic study of the intermediolateral nucleus following injection of CB-HRP and fluorogold into the superior cervical ganglion of the rat. J Auton Nerv Syst 50:333–338. Torigoe Y, Cernucan RD, Nishimoto JAS, Blanks RHI (1985) Sympathetic preganglionic efferent and afferent neurons mediated by the greater splanchnic nerve in rabbit. Exp Neurol 87:334 –348. Weaver LC, Stein RD (1989) Effects of spinal cord transection on sympathetic discharge in decerebrate-unanesthetized cats. Am J Physiol Regul Integr Comp Physiol 257:R1506 –R1511. Yoshimura M, Polosa C, Nishi S (1986) Afterhyperpolarization mechanisms in cat sympathetic preganglionic neuron in vitro. J Neurophysiol 55:1234 –1246.

(Accepted 11 October 2007) (Available online 22 October 2007)