INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The lumbosacral parasympathetic preganglionic neurons (PGN) play an important role in regulating pelvic visceral organs, including urinary bladder, distal bowel, and sex organs (de Groat et al. 1981, 1982; de Groat and Steers 1990). PGN receives inputs from the brain via pathways in the dorsolateral funiculus and from segmental interneurons located in the intermediolateral and the dorsal commissure (DCM) regions of the spinal cord (Marson and Carson 1999; Nadelhaft and Vera 1995; Sugaya et al. 1997; Valentino et al. 2000; Vizzard et al. 1995, 2000).

Previous studies (Araki 1994; Araki and de Groat 1996; Miura et al. 2001b) have examined excitatory and inhibitory synaptic inputs to the PGN from interneurons located in the region of the sacral parasympathetic nucleus (SPN) as well as from axons in the lateral funiculus. The present study used whole-cell patch-clamp recording techniques in spinal cord slice preparations from the neonatal rat to examine excitatory inputs to PGN from interneurons and/or axons in the area of DCM adjacent to the midline and dorsal to central canal. Interneurons in the DCM receive afferent inputs from mechanoreceptors and nociceptors in the pelvic organs (Birder and de Groat 1992a,b, 1993; Cruz et al. 1994; Morgan et al. 1981; Nadelhaft and Booth 1984; Ohmori et al. 1987; Steers et al. 1991; Traub et al. 1996) and are presumed to play a role in urogenital and colorectal reflex mechanisms. Our experiments demonstrated the existence of monosynaptic and polysynaptic glutamatergic excitatory pathways from the DCM to the PGN that involve NMDA and non-NMDA synaptic mechanisms. Preliminary accounts of some of the observations have been presented in an abstract (Imaizumi et al. 1998).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Preparation

Sprague-Dawley rats, 5-16 days old, were killed by decapitation and the spinal cord was rapidly removed. The L₆-S₁ segments of spinal cord were embedded in 2% agar (Sigma) in an oxygenated external solution (see composition below) at 8°C. The spinal cord was sectioned into 150-µm transverse slices using a vibrating slicer (Vibratome, Technical Products International, St. Louis, MO). The slices were incubated at 37°C for 1 h in oxygenated external solution (see composition below) and then transferred to a recording chamber (0.5 ml) on an upright microscope equipped with fluorescent optics (Olympus BH-2 or BX50WI, Tokyo, Japan). Slices were perfused continuously with the external solution (1.5 ml/min). PGN in lumbosacral spinal cord slices were identified by retrograde axonal transport of a fluorescent dye (Fast Blue, EMS-Polyloy, GrossUmstadt, Germany) that was injected (5 µl of 4% solution) into the peritoneal space 3-14 days before the experiment. This procedure has been shown to efficiently label autonomic PGN in the spinal cord (Anderson and Edwards 1994).

Electrophysiological study

The basic procedures for recording whole-cell currents from individual neurons in slice preparations of the cord were identical to those described by Takahashi (1990). Each slice of lumbosacral cord was surveyed for Fast Blue-containing neurons along the intermediolateral border of the gray matter. Motoneurons in the ventral horn were often labeled, but it was easy to distinguish between PGN and motoneurons by their location. After identification, the surface of the cell was viewed with Nomarski optics and was cleaned by a stream of the external solution from a glass pipette, which was positioned near the cell. Whole-cell currents were recorded from the labeled neurons using an Axopatch 200A or B patch-clamp amplifier (Axon Instruments, Foster City, CA). The patch pipettes were made from borosilicate glass capillaries (1B150F-4, World Precision Instruments, Sarasota, FL) and had resistances of 2.5-3.5 M when filled with pipette solution (see Solutions) after the tip had been heat polished. Synaptic responses were evoked in PGN by electrical stimulation with a glass micropipette (1-2 M) filled with the external solution. The tip of the stimulation electrode was positioned on the surface of a large neuron in the DCM near the midline dorsal to the central canal while observing the field under 100× magnification and then the recording electrode was positioned on a PGN while observing the field with 400× magnification. A voltage pulse (70 µs, 0.2 Hz) of varying intensity (1-45 V) and negative in polarity relative to a reference electrode placed in the recording chamber was applied to the stimulating pipette. The latency of excitatory postsynaptic currents (EPSCs) was measured from the stimulus artifact to the onset of the synaptic currents. The time to peak was defined as the time between the start of the current inflection and the peak of the EPSC. The rising or decay time constant was determined from a fit of the rising or decay phase of the EPSC with a single or double exponential using a nonlinear simplex fit routine based on the least-squares method.

Bicuculline methiodide (10 µM) and strychnine (1 µM) were applied in the perfusion solution to block GABA_A and glycine receptor-induced synaptic inhibitory potentials (Araki and de Groat 1996). All experiments were performed at room temperature (20-25°C). Currents were filtered at 1-5 kHz, digitized using the Digidata 1200 interface (10 kHz, Axon Instruments), and stored on a ZIP drive disk connected to an IBM-compatible personal computer for off-line analysis using pClamp6 or 7 software (Axon Instruments). Numerical data are presented as mean ± SE. Statistical analysis was performed using a two-tailed t-test or Mann-Whitney test with a significance limit of P < 0.05.

Solutions

The standard external solution contained (in mM) 130 NaCl, 5 KCl, 2 CaCl_2, 1 MgCl_2, 10 HEPES, and 11 glucose. The pH was adjusted to 7.40 with NaOH. Glycine was not included in the external solution. However, in a few experiments the addition of glycine (0.01 mM) did not change the time course or amplitude of NMDA receptor-mediated EPSCs, suggesting that the spinal slices contain adequate glycine to saturate NMDA receptors. The pipette solution contained either (in mM) 140 CsCl, 10 NaCl, 15 CsOH, 5 EGTA, and 10 HEPES, pH adjusted to 7.3 with CsOH; or 140 CsF, 9 NaCl, 0.5 CaCl₂, 3 MgCl₂, 10 HEPES, and 5 EGTA, pH adjusted to 7.3 with CsOH; or 140 potassium gluconate, 4 NaOH, 3 MgCl₂, 10 HEPES, and 0.2 EGTA, pH adjusted to 7.3 with KOH. Bicuculline methiodide (Sigma) and strychnine sulfate (Sigma) were always present in the external solution to block spontaneous inhibitory postsynaptic currents (IPSCs). Glutamatergic receptor antagonists (5 µM, 6-cyano-7-nitroquinoxaline-2,3-dine, CNQX; Research Biochemical International or 50 µM, 2-amino-5-phosphonovalerate, APV; Sigma) were added to the external solution to examine the contribution of excitatory amino acids to the evoked EPSCs (Araki and de Groat 1996; Miura et al. 2000, 2001a,b).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Recordings were obtained from 102 PGN labeled by retrograde axonal transport with a fluorescent dye. The mean resting membrane potential and input resistance were 52.1 ± 0.3 mV (ranging from 60 to 48.5 mV) and 687.9 ± 33.4 M (ranging from 200 to 1600 M), respectively.

Evoked EPSCs in PGN

When the patch electrode contained potassium gluconate (Cl equilibrium potential, 80 mV) and the external solution did not contain glycine receptor or GABA_A receptor antagonists (strychnine and bicuculline, respectively) electrical stimulation of DCM evoked inward and/or outward synaptic currents in PGN at a holding potential of 45 mV. The outward currents were completely blocked by the combined administration of 1 µM strychnine and 10 µM bicuculline (data not shown), indicating that these outward synaptic currents were IPSCs. To study EPSCs in the absence of IPSCs, all subsequent experiments were conducted at a holding potential of 60 mV and in the presence of 1 µM strychnine and 10 µM bicuculline. Under these conditions electrical stimulation of the DCM elicited only inward synaptic currents in dye-labeled PGN. The evoked currents were completely blocked by applying tetrodotoxin (1 µM) to the external solution.

At room temperature the mean latency of the evoked synaptic currents was 3.8 ± 0.1 ms (n = 102, range 1.9 to 5.8 ms). This value is approximately two times the mean latency of the EPSCs evoked by electrical stimulation of interneurons in the vicinity of PGN (Araki and de Groat 1996) and is similar to that elicited from the lateral funiculus (Miura et al. 2001b). These findings suggest that the evoked EPSCs might occur via a monosynaptic pathway.

Two types of synaptic currents were elicited by stimulation of DCM (Fig. 1). The most common response (Type 1, n = 66 cells) consisted of a short, relatively constant latency (3.8 ± 0.1 ms, range 1.9 to 5.4 ms), large amplitude (47.9 ± 4.7 pA) inward current (fast EPSC) followed by a low-amplitude more prolonged current that was usually 10% or less of the amplitude of the initial current (Fig. 1A). Averages of multiple evoked responses yielded recordings (Fig. 1Ab) that closely resembled individual responses (Fig. 1Aa). A less common response (Type 2, n = 20 cells) consisted of relatively short latency (3.8 ± 0.3 ms, range 1.9 to 5.8 ms), large amplitude fast EPSCs (28.6 ± 4.3 pA, range 6.7 to 73.0 pA) followed by large amplitude (33.8 ± 7.4 pA, range 5.4 to 89.0 pA) inward current responses that occurred at variable latency (9.6-16.6 ms) (Fig. 1B). The average of a large number of EPSCs revealed a very prolonged current (Fig. 1Bb).

View larger version (18K): [in this window] [in a new window]   Fig. 1. Two types of excitatory postsynaptic currents (EPSCs) elicited in a preganglionic neuron (PGN) by stimulation of dorsal commissure (DCM). A: Type 1 EPSCs. Aa: 5 consecutive responses. Ab: average of 30 individual EPSCs to the stimulation of DCM. The arrow and arrowhead indicate the location of the stimulus artifacts and the response peak, respectively. At 60 mV Type 1 response was composed of a fast monosynaptic component with a short latency followed by a slow component. B: less common type of evoked response (Type 2 response). Ba: 5 consecutive responses. Bb: current trace is the average of 30 responses to the stimulation of DCM. At 60 mV Type 2 response was composed of an initial monosynaptic response having a short latency followed by a slow component with superimposed fast polysynaptic currents. Stimulus duration 70 µs; stimulus frequency 0.2 Hz.

The synaptic responses in different PGN occurred at threshold stimulus intensities ranging from 2 to 20 V. Commonly, all-or-none Type 1 synaptic responses, with frequent failures, appeared at stimulus intensities near threshold at the holding potential of 60 mV (Fig. 2). When the stimulus intensity was increased, the failures became infrequent. As shown in Fig. 2 the mean amplitude of the EPSCs exhibited an abrupt increase at a stimulus threshold of about 20 V and did not change with a further increase in stimulus strength. The amplitude of individual EPSCs fluctuated but their mean value in individual cells was virtually constant (ranging in different cells from 7.5 to 187 pA, 47.9 ± 4.7pA) in a limited range of stimulus strengths (1.25-2.0 T) above the threshold (Fig. 2). The latency and time from onset to peak of maximal Type 1 EPSCs were 3.8 ± 0.1 and 2.9 ± 0.2 ms in different cells (n = 66). Histograms of latency (3.6 ± 0.03 ms) and time to peak (2.4 ± 0.02 ms) of EPSCs (n = 76) for the cell illustrated in Fig. 2 exhibited unimodal distributions in a narrow range (3.0 to 4.2 and 2.0 to 3.0 ms, respectively (Fig. 3). Type 2 fast synaptic responses (n = 20), elicited at low stimulus intensities, appeared to consist of multiunit responses (Fig. 3) that gradually increased with increasing stimulus intensities and reached a maximum (32.8 ± 4.9 pA, range 15 to 73 pA) at two to four times (25-40 V) the threshold voltages (Fig. 3). The average latency and time from onset to peak of maximal Type 2 fast EPSCs were 3.8 ± 0.3 and 3.4 ± 0.7 ms in 20 cells.

View larger version (20K): [in this window] [in a new window]   Fig. 2. Type 1 EPSCs evoked by different DCM stimulus intensities. A: four superimposed traces showing unitary responses elicited at 2 stimulus intensities (25 and 40 V) and 1 stimulus intensity (15 V) below threshold at 60 mV. B: relationship between stimulus intensity and mean peak amplitude of the average of 30 EPSCs in the same cell as in A. Vertical bars represent mean ± SE. C and D: histograms of latency (C) and time to peak (D) of EPSCs elicited by DCM stimulation. Each graph represents 76 EPSCs in the same cell as in A.

View larger version (17K): [in this window] [in a new window]   Fig. 3. Type 2 EPSCs evoked by different DCM stimulus intensities. A: EPSCs evoked by stimulation (10, 15, 25, and 40 V) at 60 mV. Four superimposed traces at each stimulus intensity (V). B: relationship between stimulus intensity and mean peak amplitude of the average of 30 EPSCs in the same cell as in A. Vertical bars represent mean ± SE.

PGN location and Type 1 and Type 2 EPSCs

The relationship between the location of PGN (n = 28) and the types of EPSCs was examined. The SPN identified by the distribution of Fast Blue-labeled PGN was divided into four areas: medial, lateral, ventral, and dorsal. Type 1 and 2 EPSCs evoked by the stimulation of the DCM were recorded at the holding potential of 60 mV in PGN located in medial (n = 5 and 1), lateral (n = 5 and 2), ventral (n = 7 and 1), and dorsal areas (n = 6 and 1, respectively). Thus there was no obvious correlation between PGN location and the types of evoked EPSCs.

Glutamatergic EPSCs and their time course

The fast component of Type 1 synaptic currents recorded at a holding potential of 60 mV was blocked by CNQX (5 µM, n = 8), a specific antagonist of non-NMDA receptors (Fig. 4). The late component of EPSCs remaining after addition of CNQX was completely blocked by APV (50 µM), a specific antagonist of NMDA receptors (Fig. 4A). The effects of glutamatergic receptor antagonists were reversible 8-41 min after washout (Fig. 4A) (n = 4 cells). The fast component of Type 2 EPSCs recorded at a holding potential of 60 mV was blocked by CNQX (5 µM, n = 6) (Fig. 4B). The late component of EPSCs remaining after addition of CNQX was completely blocked by APV (50 µM) (Fig. 4B).

View larger version (13K): [in this window] [in a new window]   Fig. 4. Non-NMDA and NMDA receptor-mediated components of Type 1 (A) and Type 2 (B) evoked EPSCs. Averaged responses of 30 EPSCs from a PGN at 60 mV. a: control. b: EPSCs in the presence of 5 µM CNQX. c: EPSCs in the presence of 5 µM CNQX and 50 µM APV. d: recovery of EPSCs after washout of the drugs from external solutions.

In the presence of APV, the time to peak and decay time constant of the Type 1 EPSCs (n = 6), mediated by non-NMDA receptors, were 1.9 ± 0.2 and 4.1 ± 0.8 ms at the holding potential of 60 mV, respectively. The time to peak and decay time constant of the EPSCs mediated by NMDA receptors in the Mg²⁺-free external solution were 14.2 ± 2.0 and 91.0 ± 12.4 ms. In the presence of CNQX, the amplitude of the NMDA component of the Type 1 EPSCs was 34.7 ± 10.0 pA (n = 8) at the holding potential of 60 mV. The time course of these EPSCs was measured on averaged responses of 30 individual EPSCs.

Voltage dependence of EPSCs

The current-voltage relationships of the non-NMDA and NMDA components of evoked Type 1 EPSCs were examined by measuring the peak amplitude of the EPSCs at 4-10 and 100 ms after the stimulus. The early component was assumed to reflect mainly the non-NMDA component because the EPSCs mediated by NMDA receptors exhibited a slow rise time and small amplitude and should make a very small contribution to the EPSCs at this time point. On the other hand, the non-NMDA component would make little contribution to EPSCs at 100 ms after the stimulus (Araki and de Groat 1996). The amplitude of EPSCs at this point was assumed to reflect the NMDA component (Araki and de Groat 1996). The current-voltage relationship of the non-NMDA component had a linear conductance, whereas that of the NMDA component at 100 ms after the onset of response had a region of negative slope conductance at hyperpolarized holding potentials (more negative than 20 mV) (Fig. 5). The extrapolated reversal potentials of the non-NMDA and NMDA components were 7.6 ± 1.3 mV (filled circle) and 8.6 ± 1.6 mV (open circle, n = 5), respectively (Fig. 5).

View larger version (24K): [in this window] [in a new window]   Fig. 5. Current-voltage relationship for the evoked Type 1 EPSCs. A: averaged responses of 30 EPSCs. Responses were recorded at different holding potentials ranging from 80 to 40 mV in 20-mV steps. The amplitude at 100 ms after the onset of stimulus is assumed to reflect the NMDA component. B: current-voltage relationship of EPSCs from 5 cells. Peak amplitude of the currents () and the amplitude at 100 ms after the onset of stimulus artifacts () were measured. Responses were recorded at holding potentials ranging 80 to 40 mV in 10-mV steps. C: averaged responses of 30 NMDA EPSCs in the presence of 10 µM bicuculline, 1 µM strychnine, and 5 µM CNQX in the external solution (1 mM Mg2+) (Ca) or Mg2+-free solution (Cb). Responses were recorded at different holding potentials ranging from 80 to 40 mV in 20-mV steps. D: current-voltage relationship of NMDA EPSCs from 8 cells. Mean peak amplitudes of averaged responses at different holding potentials in the presence () and absence () of Mg2+, respectively. Vertical bars represent mean ± SE.

To further study the voltage dependence of the NMDA receptor-mediated EPSCs, the evoked responses were recorded in four cells in the presence of 5 µM CNQX and in the presence or absence of external Mg²⁺. As described above in normal external solution (1 mM Mg²⁺), the NMDA component of Type 1 EPSCs was prominent at depolarized membrane potentials (Fig. 5) and had a region of negative slope conductance at potentials more negative than 20 mV (Fig. 5C, open circle). In Mg²⁺-free solution, the NMDA-mediated EPSCs were prominent even at negative membrane potentials (Fig. 5, C and D). As shown in Fig. 5D, they had a nearly linear current-voltage relationship (filled circle) at potentials ranging from 60 to +30 mV. This indicates a voltage-dependent Mg²⁺ block of the NMDA receptor-mediated EPSCs. Thus the two components of the EPSCs in PGN, which are mediated by non-NMDA and NMDA receptors, were distinguished by the differences in their voltage dependence as well as their time course.

Synaptic facilitation in PGN

To examine the synaptic efficacy of interneuronal inputs to PGN, we determined whether Type 1 EPSPs evoked by low-intensity DCM stimulation were sufficient to evoke action potentials in PGN. EPSPs were recorded under current clamp conditions at 55 mV, which was near to the resting membrane potential. EPSPs evoked in PGN at 0.2 Hz induced action potentials infrequently (4.9 action potentials per 30 trials per cell in 13 cells). However, trains of three stimuli at a short interstimulus interval (50 ms) induced action potentials in 13 PGN (Fig. 6A). The first EPSPs evoked action potentials in 7.5 of 30 trials (average of 13 cells, Fig. 6A) and the second EPSPs induced firing in 10.7 of 30 trials. There was no significant difference between the number of action potentials evoked by first and second EPSPs, whereas the number was significantly increased in the third versus the first EPSPs (13.5 vs. 7.5 in 30 trials, P < 0.05). Temporal facilitation of the EPSCs at DCM-PGN synapses was also examined using two successive stimuli to the DCM at the holding potential of 60 mV. Paired-pulse facilitation was observed at interstimulus intervals ranging from 50 to 120 ms in PGN. When two stimuli were applied at interstimulus intervals of 50 (Fig. 6, B and C), 70, 100, and 120 ms, the facilitation of the peak amplitude of the second EPSC expressed as a percentage of the first EPSC was 141.0 ± 5.6%, n = 8 (Fig. 6, B and C); 153.3 ± 10.7%, n = 7; 121.0 ± 8.5%, n = 6; 128.0 ± 10.8%, n = 6, respectively), excluding 1 cell showing inhibition (46.6%). Paired-pulse stimulation was also examined in cells (n = 3) with Type 2 EPSCs. Facilitation was elicited at interstimulus intervals of 50 and 100 ms (115%, n = 2 cells and 148% of control, n = 2 cells, respectively).

View larger version (15K): [in this window] [in a new window]   Fig. 6. The synaptic efficacy of DCM inputs to PGN. A: under current-clamp conditions three successive stimuli (interstimulus interval of 50 ms) in the DCM-induced EPSPs and action potentials in a PGN that exhibited Type 1 EPSCs. Recordings were obtained in the presence of 1 µM strychnine and 10 µM bicuculline. Note: action potentials occurred more consistently following the 2nd and 3rd stimuli in the train. The resting membrane potential was 55 mV. B: paired-pulse facilitation of Type 1 EPSCs elicited by a pair of stimuli in the DCM at an interstimulus interval of 50 ms and at a frequency of 0.2 Hz. An averaged response of 30 EPSCs showing representative paired-pulse facilitation at 60 mV. C: relationship between the 1st EPSC and 2nd EPSC amplitudes (pA) of 30 individual paired responses in different PGN (n = 9).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present experiments revealed that parasympathetic PGN in the lumbosacral spinal cord of the neonatal rat receive glutamatergic excitatory synaptic inputs from neurons and/or axons in the DCM. These excitatory pathways activate NMDA and non-NMDA glutamatergic receptors. Latencies of evoked EPSCs revealed two distinct pathways: 1) a pathway mediating relatively short and fixed latency EPSCs (Type 1), probably reflecting monosynaptic connections, and 2) a pathway mediating longer and more variable latency EPSCs (Type 2 slow component), probably representing polysynaptic connections.

Type 1 EPSCs often had an all-or-none characteristic when the stimulus intensity was near threshold; whereas the fast component of Type 2 EPSCs usually had a graded stimulus intensity-response relationship indicating a multiunit input. The all-or-none responses, which varied in mean amplitude from 7.5 to 150 pA (average, 47.2 ± 5.1 pA), presumably represent "unitary EPSCs" evoked by a single DCM neuron and/or axon. This average amplitude is similar to the magnitude of unitary glutamatergic EPSCs evoked by stimulation of single interneurons in the region of the SPN (average amplitude 60 pA) (Araki and de Groat 1996) or unitary responses evoked by stimulation of axons in the lateral funiculus (average amplitude 59 pA) (Miura et al. 2001b). However, it has been noted that the magnitude of interneuronal inputs to PGN varies with the location of the interneurons. The EPSCs evoked by interneurons located within 100 µm medial to the PGN were considerably larger (average 84 pA) than the EPSCs elicited by interneurons located within 100 µm dorsal to the PGN (average 34 pA) (Araki and de Groat 1996). Thus the size of unitary EPSCs evoked by stimulation of the DCM is consistent with the size of other responses in PGN evoked by activation of a single neuron or axon.

The DCM-evoked EPSCs must be mediated exclusively by glutamatergic receptors, because combined administration of non-NMDA (CNQX) and NMDA (APV) antagonists completely blocked the EPSCs. This situation is very similar to the interneuronal and dorsolateral funiculus excitatory inputs to PGN that were also completely blocked by a combination of non-NMDA and NMDA antagonists (Araki and de Groat 1996; Miura et al. 2001b).

The non-NMDA and NMDA receptor-mediated synaptic currents evoked by DCM stimulation had properties similar to those elicited by local interneuronal projections to the PGN (Araki and de Groat 1996). For example, EPSCs activated by the two types of receptors could be distinguished on the basis of differences in time course, voltage dependence, and specific receptor antagonists. However, the mean time to peak of averaged DCM-evoked non-NMDA EPSCs was longer than that reported for unitary EPSCs evoked in PGN by stimulation of dorsal or medial interneurons (Araki and de Groat 1996) but was comparable to EPSCs in rat sympathetic preganglionic neurons evoked by axonal stimulation (Krupp and Feltz 1995). The current-voltage relationships for DCM-induced glutamatergic EPSCs correspond to data obtained at other synapses (Jonas et al. 1993), including interneuronal-PGN synapses in the neonatal rat spinal cord (Araki and de Groat 1996).

The organization of the DCM projections to the PGN could be complex, involving inputs from axonal and neuronal elements in the DCM directly to PGN and indirectly via other neurons. In future experiments it would be useful to try to distinguish between direct and indirect inputs by examining the effects of agents such as mephenesin, which have a more prominent depressant action on polysynaptic than on monosynaptic pathways (Floeter and Lev-Tov 1993). Even though we attempted to position the stimulating electrodes on the surface of large neurons in the DCM, it is possible that in some instances the stimuli activated adjacent axons located in close proximity to the neurons. This might explain the multiunit Type 2 responses, which consisted of both monosynaptic and polysynaptic components. It seems likely that it would be more difficult to activate a single axon than a cell body in the DCM; therefore the multiunit fast EPSCs could be evoked by stimulation of a bundle of axons passing through the DCM to the SPN. Furthermore these axons could make synaptic connections not only with PGN but also with interneurons that project to the PGN. Several types of axons might be involved, including 1) visceral primary afferent axons which are known to pass into the DCM after projecting laterally and medially around the dorsal horn (Morgan et al. 1981; Nadelhaft and Booth 1984; Steers et al. 1991); 2) axons from interneurons in the contralateral and ipsilateral DCM; and 3) bulbospinal or propriospinal axons involved in pelvic visceral functions.

If polysynaptic responses evoked in PGN by DCM stimulation involve activation of local interneurons in the SPN, it might be expected that the DCM- and interneuronal-evoked EPSCs would exhibit similar properties. This prompted us to examine the effects of paired-pulse stimulation, which previous experiments showed could distinguish between different excitatory pathways projecting to the PGN. Excitatory glutamatergic inputs from the lateral funiculus to the PGN exhibited paired-pulse inhibition, whereas interneuronal inputs to the PGN showed paired-pulse facilitation. However, the magnitude of the facilitation varied between different interneuronal pathways, being relatively modest (average 24% increase) in projections from medial interneurons, but much larger in dorsal interneuronal pathways (average 67% increase). The magnitude of the paired-pulse facilitation of EPSCs during electrical stimulation of the DCM was intermediate (40%) between these two values, suggesting that a different population of interneurons from those already studied might be involved in the polysynaptic pathway from the DCM to the PGN.

The excitatory glutamatergic monosynaptic and polysynaptic projections from the DCM to the PGN might have multiple functions. First, neurons in the DCM seem to have an important role in processing nociceptive as well as nonnociceptive afferent input from the pelvic viscera. Distension or chemical irritation of the bladder (Birder and de Groat 1992a, 1993; Vizzard 2000) or the distal bowel (Traub et al. 1996) induces c-fos expression in DCM neurons in the adult rat. Glutamatergic mechanisms involving NMDA and non-NMDA receptor contribute to this expression (Birder and de Groat 1992b; Kakizaki et al. 1996). Thus bladder and bowel reflexes induced under normal and pathological conditions by primary afferent input could be mediated by excitatory projections from the DCM to PGN.

Because somatic and visceral afferent inputs converge onto DCM neurons (Honda 1985) it is of interest to consider the possibility that the DCM to PGN pathway might play a role in viscerosomatic interactions. Of particular relevance to the present study, which was conducted in neonatal rat spinal cord, is the somatobladder reflex, which is essential for voiding in neonatal animals (de Groat et al. 1975). Micturition in neonatal rats is mediated by a spinal reflex pathway consisting of a somatic afferent limb in the pudendal nerve and a parasympathetic efferent limb in the pelvic nerve. Pudendal afferents project heavily into the DCM region as well as into the intermediolateral gray matter near the SPN and therefore could trigger polysynaptic voiding reflexes by activating interneurons in either region. The neonatal somatobladder reflex is downregulated during postnatal development but reemerges in adult animals and humans after spinal cord injury (de Groat et al. 1990; Yoshimura et al. 2000). Recent studies by Vizzard (2000) revealed that c-fos expression in the DCM induced by bladder distension in adult rats is markedly increased in chronic spinal cord-injured animals. Thus DCM to PGN excitatory pathways may play a role in bladder hyperreflexia after neural injury.

Pudendal nerve afferents also convey information from the sexual organs (penis, clitoris, vagina, and uterine cervix) and from urethral and anal sphincter muscles (Kawatani et al. 1990, 1994). Therefore pudendal nerve afferent projections to the DCM that in turn activate DCM to PGN excitatory pathways could play an important role in reproductive organ functions as well as in the coordination of bladder-urethra and colorectal-anal canal activity. In addition, coordination between efferent pathways on the right and left sides of the spinal cord is essential for proper function of organs such as the bladder, distal bowel, and penis, which are regulated by efferent neurons on both sides of the spinal cord. The midline location of the DCM, its receipt of afferent inputs from both sides of the cord, and its bilateral projection to the right and left SPN indicate that the excitatory glutamatergic pathways from the DCM to PGN demonstrated in the present experiments are likely to play an important role in the bilateral regulation of urogenital and distal bowel function.

As noted in a few of our preliminary experiments, stimulation in the DCM also elicited IPSCs in PGN. Thus the DCM may be involved in inhibitory as well as excitatory control of pelvic visceral function (de Groat et al. 1981) and in inhibitory reflex interactions between striated sphincter muscles and the viscera (de Groat and Steers 1990). These reflexes could be triggered by primary afferent projections to the spinal cord or by axon collaterals of PGN projecting to interneurons in the DCM (Morgan et al. 1991), which then could provide inputs back to the PGN and motoneurons innervating sphincter muscles.