Brain Research Bulletin, Vol. 53, No. 5, pp. 649 – 659, 2000 Copyright © 2001 Elsevier Science Inc. Printed in the USA. All rights reserved 0361-9230/00/$–see front matter

PII S0361-9230(00)00398-1

Contributions of intrinsic motor neuron properties to the production of rhythmic motor output in the mammalian spinal cord Ole Kiehn,* Ole Kjaerulff,† Matthew C. Tresch and Ronald M. Harris-Warrick‡ Section of Neurophysiology, Department of Medical Physiology, University of Copenhagen, Copenhagen N, Denmark [Received 16 May 2000; Revised 3 August 2000; Accepted 7 August 2000] ABSTRACT: Motor neurons are endowed with intrinsic and conditional membrane properties that may shape the final motor output. In the first half of this paper we present data on the contribution of Ih, a hyperpolarization-activated inward cation current, to phase-transition in motor neurons during rhythmic firing. Motor neurons were recorded intracellularly during locomotion induced with a mixture of N-methyl-D-aspartate (NMDA) and serotonin, after pharmacological blockade of Ih. Ih was then replaced by using dynamic clamp, a computer program that allows artificial conductances to be inserted into real neurons. Ih was simulated with biophysical parameters determined in voltage clamp experiments. The data showed that electronic replacement of the native Ih caused a depolarization of the average membrane potential, a phase-advance of the locomotor drive potential, and increased motor neuron spiking. Introducing an artificial leak conductance could mimic all of these effects. The observed effects on phase-advance and firing, therefore, seem to be secondary to the tonic depolarization; i.e., Ih acts as a tonic leak conductance during locomotion. In the second half of this paper we discuss recent data showing that the neonatal rat spinal cord can produce a stable motor rhythm in the absence of spike activity in premotor interneuronal networks. These coordinated motor neuron oscillations are dependent on NMDA-evoked pacemaker properties, which are synchronized across gap junctions. We discuss the functional relevance for such coordinated oscillations in immature and mature spinal motor systems. © 2001 Elsevier Science Inc.

final convergence pathway for the CPG networks. Motor neurons are, however, not merely passive interpreters of synaptic input from premotor neuronal circuits. Rather, vertebrate motor neurons are endowed with intrinsic and conditional membrane properties able to shape the final motor output [23–25]. These membrane properties are influenced by neuroactive substances, some of which are released during locomotor activity. In this paper, we will explore the role of such motor neuronal properties for rhythmic motor-pattern generation in the neonatal rat. In particular, we will present recent data on the contribution of a hyperpolarizationactivated inward current, Ih, to motor neuron activity during locomotion. Moreover, we will examine the capability of motor neurons to generate a motor pattern when synaptically isolated from premotor neuronal networks. THE ISOLATED NEONATAL RAT SPINAL CORD AS A MODEL SYSTEM FOR STUDYING LOCOMOTION The lower thoracic and lumbar spinal cord contains sufficient neuronal circuitry to control hindlimb locomotion in the newborn rat (reviewed in [27]). This part of the spinal cord can easily be isolated and can survive in vitro for extended periods of time. Rhythmic activity is readily induced in the cord by a variety of transmitter agonists, including excitatory amino acids, serotonin (5-HT), dopamine, noradrenaline and certain peptides [1,6,12,26, 29,33,57]. Although not all of these neuroactive substances produce locomotor-like activities, the rhythmic motor patterns induced by some of these substances resemble the pattern of walking or swimming in the intact animals and show the appropriate left-right alternation and intralimb coordination between flexor and extensor muscles [12,22,26]. It is therefore likely that the CPG in the relatively immature, isolated spinal cord has the essential features of the CPG controlling locomotion in intact adult animals. This, and the feasibility of combining electrophysiological recordings from interneurons and motor neurons with controlled pharmacological intervention makes the in vitro neonatal rat spinal

KEY WORDS: Motor neurons, Ih, 5-HT, NMDA, Bursting, Gap junctions.

INTRODUCTION The generation of rhythmic locomotor activity, such as swimming and walking in vertebrates, relies heavily upon activity in central pattern generating (CPG) networks. These networks are generally believed to play critical roles in generating the timing of the complex rhythmic motor patterns executed by motor neurons, the

* Address for correspondence: Ole Kiehn, Research Professor, MD, Section of Neurophysiology, Department of Medical Physiology, University of Copenhagen, Blegdamsvej 3, 2200 Copenhagen N, Denmark. Fax: ⫹45-35327499; E-mail: [email protected] † Present address: The Nobel Institute for Neurophysiology, Department of Neuroscience, The Karolinska Institute, Department of Neuroscience, S-171 77 Stockholm, Sweden. ‡ Present address: Department of Neurobiology and Behavior, Cornell University, Seeley Mudd Hall, 14853 Ithaca, NY, USA.

649

650

KIEHN ET AL.

FIG. 1. Ih in neonatal rat motor neurons. (A) Under voltage clamp activation of Ih is seen as a slowly developing inward rectification of the whole cell current (top trace). In the depicted case the motor neuron was stepped from ⫺50 mV to ⫺100 mV (bottom trace). The slow deactivation of the current is seen as an inward tail current (arrow) following the termination of the step command. All intracellullar recordings presented in this paper were performed using whole-cell tight-seal recordings (see [49]) (B) In current clamp activation of Ih by a long-lasting hyperpolarizing current pulse causes the membrane potential to slowly sag back towards the resting membrane potential. Note the depolarizing overshoot (arrow) following termination of the current pulse. Unpublished data from Kjaerulff and Kiehn. See also Fig. 2.

cord an excellent model preparation for the characterization of mammalian locomotor networks and investigating the role of cellular properties involved in rhythmogenesis and pattern generation. RHYTHMIC INHIBITION MAY ACTIVATE MOTOR NEURON CONDUCTANCES THAT FACILITATE PHASE TRANSITION FROM INHIBITION TO EXCITATION During locomotion the motor neuron membrane potential oscillates between periods of hyperpolarization and depolarization ([5,19]; see [28] for a broad review). It is generally accepted that the different phases of activity are largely defined by changes in the balance between synaptic excitation and inhibition. An essential element in the rhythmicity is the transition between the inhibitory and excitatory phase corresponding to the onset of muscle activity. This phase transition may be mediated predominantly by a change in presynaptic activity, e.g., a decrease in inhibition and/or an increase in excitation (see [49]). However, several neuronal ionic currents that are activated by hyperpolarization can change the timing of the escape from inhibition. These currents may facilitate the transition from inhibition to excitation during rhythmic motor activity (see [25]). One of these currents is Ih, a hyperpolarization-activated inward cation current (reviewed in [46]). Ih is present in spinal motor neurons from newborn rats, and, as in other central neurons, is strongly activated by hyperpolarization [32,59]. The current is carried by a mixture of K- and Na-ions [59], and at potentials more hyperpolarized than its reversal potential of about ⫺35 mV it will tend to depolarize the cell. Thus, the hyperpolarization originally activating Ih will be counteracted by that same activation. Ih in neonatal rat motor neurons activates and deactivates slowly and shows little inactivation. These properties result in a characteristic behavior of a neuron that possesses Ih. In voltage clamp, activation of Ih is seen as a slowly developing inward current in response to hyperpolarizing step commands (Fig. 1A), while in current clamp activation of Ih causes the membrane potential to slowly sag back towards the resting membrane potential (Fig. 1B). Because of its slow deactivation, Ih will contribute

to the formation of a slow inward tail current (voltage clamp; Fig. 1A, arrow) or a depolarizing overshoot (current clamp; Fig. 1B, arrow) that follows the offset of the hyperpolarizating step. Theoretically, the presence of Ih should therefore have two important consequences for the integrative properties of motor neurons during phasic inhibition: (1) Ih should limit the effect of sustained inhibitory inputs and help the cell escape from inhibition faster than when it is absent, and (2) the depolarizing overshoot caused by slow deactivation of Ih should trigger a rebound depolarization of the neuron at the end of the inhibition, causing cells to fire earlier and to produce more spikes than when Ih is absent. We are testing these hypotheses for neonatal rat motor neurons, which are rhythmically active during locomotion, using dynamic clamp or conductance clamp in combination with specific antagonists of Ih. Dynamic clamp allows artificial conductances to be inserted into real neurons, and relies on a computer program (Program name, Dyna-Quest Technologies) that monitors the membrane potential of the active neuron, simulates the corresponding time- and voltage-dependent activation of Ih and injects this simulated current into the cell. This method has been used successfully to evaluate the importance of the different synaptic and ionic conductances, including Ih, in the production of complex neuronal activity in other preparations [14,18,21,52,53,65]. One particular advantage of the dynamic clamp approach is the possibility to vary the intrinsic properties of the recorded neuron, without affecting the network activity and thereby the rhythmic synaptic input. This lack of effects on network properties contrasts the global effect of bath-applied drugs. Before describing these experiments, we will briefly discuss the biophysical parameters for Ih, its pharmacology and its modulation by 5-HT in neonatal rat motor neurons. BIOPHYSICAL PARAMETERS FOR Ih AND ITS PHARMACOLOGY IN RAT MOTOR NEURONS To perform the dynamic clamp experiments realistically, detailed knowledge of the biophysical parameters relevant for the time and voltage dependent activation of the native Ih is necessary (see [18,52,53] for a detailed description of the equations used by the program). These parameters were obtained from voltage clamp

CELLULAR PROPERTIES AND MOTOR PATTERNS

651

FIG. 2. Replacement of Ih by dynamic clamp after pharmacological block with ZD 7288. (A) Motorneuronal voltage responses (top) to hyperpolarizing current injections (bottom; ⫺150 pA steps; zero bias current) in control. Note the depolarizing sag evoked by activation of Ih and the fast instantaneous inward rectification due to activation of Ikir (see the text). (B) ZD 7288 (50 ␮M) blocked the expression of Ih, but not the fast instantaneous rectification. ZD 7288 caused the motor neuron to hyperpolarize and increased the cell’s input resistance, suggesting that Ih contributed to the resting conductance (⫺150 pA steps; zero bias current). (C) Replacement of Ih with dynamic clamp. The amplitude and shape of the injected dynamic clamp current (d-clamp current) is shown in the top traces, while the size of the injected current steps (p-clamp, ⫺90 pA steps; zero bias) are shown in the bottom traces. The following parameters where used for the dynamic clamp protocol: V1/2 ⫽ ⫺90 mV; Vrev ⫽ ⫺35 mV; Gmax ⫽ 15 nS; the time constant for activation was approximately 1.5 s a ⫺80 mV. Unpublished data from Kiehn and Kjaerulff.

studies performed in lumbar motor neurons in a thin slice preparation of the neonatal rat [59] and in lumbar motor neurons in the intact cord [32]. The main difference between the biophysical parameters determined in these two studies was that the maximal conductance, Gmax, was roughly 10 times higher in the intact cord (on average 12 nS) than in the slices. This is presumably because the motor neurons are mechanically less reduced in the intact cord than in the slice. The reversal potential (Vrev) and the voltage for half-maximal activation of Ih (V1/2) were in the same range in the two studies ([59]: ⫺42 mV and ⫺95 mV; [32]: ⫺33 mV and ⫺93 mV). A precise description of the time course for Ih activation was obtained in the intact cord. Activation of Ih was fit with single- or double exponential functions [32]. The goodness of the fit did not differ substantially at relative depolarized potentials (more positive than ⬃⫺80 mV). However, at more hyperpolarized levels than ⬃⫺80 mV, double-exponentials clearly fitted the current responses better than single-exponentials. We also found that Ih co-exists with a fast or instantaneous inward rectifier (IR) in 44% of the motor neurons [32]. The fast IR becomes apparent at ⫺80 mV, close to Ek (Fig. 2A), where it is seen as an apparent decrease in input resistance (smaller voltage responses to the same sized current at membrane potentials more hyperpolarized than ⫺80 mV). The fast IR and Ih can be separated pharmacologically. Thus, the fast IR is blocked by low concentrations (200 –300 ␮M) of barium ions, while the slow IR due to Ih is blocked selectively by ZD 7288 ([13,17,21,61]; see Figs. 2B,C). As in other neurons, both types of inward rectification are blocked by low concentrations of CsCl (1–2.5 mM). Together, these experiments demonstrate that spinal rat motor neurons possess two types of inward rectification: (1) a fast IR, which has a pharmacology and voltage-dependency similar to an inward rectifying current carried by potassium ions, called Ikir (i.e., [11,16,45]), and (2) a slow inward rectifier identical to Ih. The distinct pharmacological sensitivity of Ih allowed us to block it prior to induction of locomotor activity (below). Ih IS ENHANCED BY 5-HT Ih is modulated by biogenic amines in many nerve cells [46]. Such modulation is mediated through a change in one or several of

Ih’s biophysical parameters, typically changes in V1/2, Gmax and the time course of activation. Since we use 5-HT in combination with N-methyl-D-aspartate (NMDA) to induce locomotion [30,31, 48,49,62], we have tested whether 5-HT modulated Ih in neonatal rat motor neurons. We found that 5-HT enhanced Ih by causing a 7-mV depolarizing shift in the voltage range of activation, without increasing the maximal conductance [32]. 5-HT also increased the speed of activation of both the slow and fast time constant for activation of the current. Thus, at ⫺80 mV the slow time constant was decreased from ⬃4 s to ⬃2.5 s while the fast was reduced from ⬃1.7 s to ⬃0.8 s. The 5-HT enhancement of Ih is similar to that proposed by Takahashi and Berger [60], although they did not determine the exact mechanism responsible for the enhancement. The effect of 5-HT on Ih increases the resting motor neuron excitability and in theory could lead to an enhancement of rebound firing and phase-transition during rhythmic motor discharges. Interestingly, we find that Ikir is inhibited by 5-HT [32]. The implication of this effect for motor neuron firing awaits further study. REPLACEMENT OF Ih WITH DYNAMIC CLAMP WITHOUT LOCOMOTION Figure 2 shows that it was possible to obtain a relatively good, though not perfect, electronic replacement of Ih after it had been blocked pharmacologically. Ih was blocked with either ZD 7288 (100 –200 ␮M; Fig. 2B) or low concentrations of CsCl (1–2.5 mM). Either of these antagonists caused motor neurons to hyperpolarize (compare Figs. 2A,B) when initially held at their resting membrane potential (⫺75 to ⫺65 mV see also [19] and [38]). This hyperpolatization by ZD 7288 indicates that Ih is tonically activated at rest in neonatal rat motor neurons. Tonic Ih activation may also explain the observation that ZD 7288 increased the resting input resistance (Fig. 2B). Ih was replaced by running the dynamic clamp protocol (Fig. 2C; dynamic clamp current depicted in upper trace) with a biophysical parameter space originating from the voltage clamp experiments in 5-HT (see above). Replacement of the simulated Ih again depolarized the cell, alleviating the hyperpolarization instituted by the blockade of the native current. These

652

KIEHN ET AL.

FIG. 3. Replacement of Ih by dynamic clamp during locomotion. (A) Intracellular recording from lumbar motor neuron (top trace) during locomotion induced by a combination of N-methyl-D-aspartate (NMDA;6.0 ␮M) and serotonin (5-HT; 6.0 ␮M). Locomotion was recorded with suction electrodes attached to the left (L-L2) and the right R-L2 ventral roots (middle two traces). Ih was blocked with 1 mM CsCl prior to induction of locomotion. (B)–(C) Replacement of Ih with slow activation (a time constant for activation around 2.5 s at ⫺80 mV) and with fast activation (a time constant for activation around 0.7 s at ⫺80 mV). Introducing Ih depolarized the cell and increased motor neuron firing. Other D-clamp parameters: V1/2 ⫽ ⫺95 mV; Vrev ⫽ ⫺30 mV; Gmax ⫽ 10 nS. Unpublished data from Kiehn, Kjaerulff, Harris-Warrick and Johnson.

experiments shows that the dynamic clamp protocol can be used to reliably simulate the real time and voltage dependent activation of intrinsic conductances in identified neurons (see also 18,21,52, 53]). Ih ACTS AS A LEAK CONDUCTANCE DURING LOCOMOTION The effect of Ih on motor neuron firing during locomotion was tested by first blocking Ih pharmacologically. We then induced rhythmic activity with a combination of NMDA and 5-HT, and replaced Ih by running the dynamic clamp protocol (Fig. 3). Although Ih activation was accurately described by a slow time constant at depolarized membrane potentials ([32]; and above) as those observed in motor neurons during transmitter-induced locomotion, we wanted to emphasize the phasic aspects of Ih activation in our replacement experiments. Two parameter sets were therefore employed, one with a slow time constant for activation of Ih and one with a faster time constant. The slow and fast time constants were in the range of the average time constants for activation of Ih observe in the presence of 5-HT ([32]; and above). In both cases introduction of Ih in the motor neuron caused a depolarization accompanied by more intense firing (Figs. 3B,C; injected current depicted in the lower traces) compared to control (Fig. 3A). This suggests that the native Ih indeed is activated during locomotion and enhances motor neuron excitability. From these recordings it is, however, difficult to resolve whether the phase-transition in individual locomotor cycles was influenced by the introduction of Ih. This difficulty is because the shape of the locomotor drive potentials vary from cycle to cycle. We therefore low-pass filtered the intracellular recording (thick line in Fig. 4A) and then converted each time point of individual cycles to a locomotor phase by dividing the delay from cycle onset with the cycle duration and multiplying with 360. This normalization permitted the calculation of cycle averages, shown in Fig. 4B for control (solid red line) and after replacement of Ih with slow (black line) and fast (green line) parameters. Replacement of Ih caused a steady depolarization of the motor neuron and a phase advance of the rising phase of the depolarizing locomotor drive potential. This effect appeared more clearly when the control curve was aligned by shifting it in a depolarizing direction (red stippled line).

To quantify the phase transition between inhibition and excitation we determined the time (Thalf; Fig. 4A) or the phase (⌽half) at which the membrane potential has depolarized half way between the valley (Vval) and peak (Vpk) of the membrane potential fluctuation. There was a significant (p ⬍ 0.05; one-way analysis of variance; Fig. 4C) decrease in ⌽half after replacing Ih with both parameter sets, but no differences in ⌽half between the dynamic clamp experiments with different activation constants. These results suggest that the presence of Ih during locomotion limits the effect of rhythmic inhibition and causes the membrane potential to escape from inhibition faster than when Ih is absent. To observe if the presence of Ih caused motor neuron action potentials to appear earlier in locomotor cycle, we determined the time— expressed as locomotor phase—at which the first action potential occurred (⌽ap1, Fig. 4A). The results of this analysis are shown in Fig. 4D. There was a significant reduction (p ⬍ 0.05) of the phase of the first action potential for both parameter sets, but again no difference between the fast and slow activation of Ih. These experiments suggest that Ih accelerates the arrival of the first action potential. In conclusion, replacement of Ih (1) decreased the time for phase transition between the inhibitory and excitatory phases, and (2) decreased time to the first action potential in the locomotor cycle. These two observations are compatible with the idea that Ih—activated by hyperpolarization and contributing an inward, depolarizing current— helps the cell escape from inhibition faster than when Ih is absent; and that the depolarizing rebound caused by the slow deactivation of Ih phase advances cyclic spiking. These results are consistent with the common notion that the unique biophysical properties of Ih are critical for its shaping of motor neuron activity during rhythmic activity (see below). However, further observations strongly suggested that this canonical view may not capture the essence of the role of Ih in rhythmicity in our system. Introducing Ih caused a tonic depolarization of the average membrane potential (Figs. 3B,C and Fig. 4B), and the mean firing frequency throughout the cycle increased significantly compared to control (Fig. 4E). In general, action potentials centered around the phase corresponding to the average peak potential (Vpk), both in control (Fig. 4E, upper panel) and after replacement of Ih (Fig. 4E, middle panel). Only occasionally was the action potential

CELLULAR PROPERTIES AND MOTOR PATTERNS

653

FIG. 4. Ih decreases the time for transition between the inhibitory and excitatory phase during locomotion and increases motor neuron excitability. (A) Locomotor related parameters used to quantify the effect of Ih. After low-pass filtering the intracellular voltage trace the time (the valley, Tval) for the minimal voltage (Vval) in each locomotor cycle was determined. The peak voltage (Vpk) was also determined, as was the time or phase (Thalf or ⌽half) to the voltage half way between Vval and Vpk. Thalf was expressed as normalized cycle values (0 –360°) for each cycle. The time or phase from Tval to the first action potential (Tap1 or ⌽ap1) and consecutive action potentials (Tapn or ⌽apn) in each cycle was found from the unfiltered trace and expressed in normalized cycle values. The unfiltered trace was also divided into 10 equal-sized bins and the instantaneous firing frequency in each bin was calculated for each cycle by dividing the number of spikes in each bin by the duration the bin. The mean instantaneous frequency histograms were obtained by averaging the bins from all the locomotor cycles in a given experimental condition. Similar mean ⌽half, and mean ⌽ap1 were obtained by averaging the locomotor cycles in an episode. (B) Average locomotor drive potential for the cell shown in Fig. 3 in control (solid red line; N-methyl-D-aspartate (NMDA)/serotonin (5-HT) plus CsCl), and after replacement of Ih slow (black line) and fast (green line) dynamic clamp parameters. Note the tonic depolarization caused by replacing Ih. The stippled red line shows the control data after being shifted upward so that the through potential was aligned with those obtained after replacing Ih. (C)–(D) Replacing Ih caused a significant (p ⬍ 0.05; one-way analysis of variance) decrease in the time (phase) to ⌽half (C) and the first action potential (D; ⌽ap1). Same color code as in (B). (E) The mean instantaneous firing frequency during locomotion also increased after replacing Ih (compare to lower panels with the upper panel). All data shown in (B)–(C) are from the cell shown in Fig. 3 and using the dynamic clamp parameters in the legend to Fig. 3. Unpublished data from Kiehn, Kjaerulff, Harris-Warrick and Johnson.

distribution skewed with a peak in the beginning of the depolarizing phase (Fig. 4E, lower panel). This result is in contrast to a previous description of the firing of neonatal rat motor neurons during rhythmic activity, recorded in a split-bath setup [2]. Our data suggest that the importance of Ih relies simply on its effect on the average membrane potential. To test for this possibility we used the dynamic clamp to introduce a voltage-independent leak conductance. This conductance was given a reversal potential identical to that of Ih and a maximal conductance adjusted so that running the dynamic clamp caused a similar average membrane potential as when replacing Ih (Fig. 5A) Somewhat surprisingly, introducing the leak conductance caused a reduction in Thalf or ⌽half (Fig. 5B) and time (phase) to the first action potential (Fig. 5C). This reduction was significantly different from control but not significantly different from that caused by Ih. The changes in the action potential distribution throughout the locomotor phase were also similar with the leak and Ih (compare two lower panels in Fig. 5D). Similar effects were observed in other motor neurons. These data suggest that the main role of Ih during locomotion is to serve as a steady depolarizing ‘leak’ conductance. The effects on phase

transition and firing properties during locomotion are thus secondary to the steady depolarization and not a result of a continuous time- and voltage-dependent regulation of the conductance underlying Ih. This conclusion was emphasized by the fact than even when the activation of the artificial Ih was made three times faster than that of the native current it behaved like a depolarizing ‘leak’ conductance. The results of these studies are in contrast to what has been suggested from dynamic clamp studies in invertebrate motor neurons [18,34] or mammalian thalamic neurons [21] where a clear effect was observed, which could be ascribed directly to rebound firing. We believe that there are two main explanations for these different results. First, during locomotion the depolarizing component of the locomotor phase in mammalian motor neurons represents the summation of a barrage of NMDA/non-NMDA excitatory postsynaptic potentials (EPSPs) ([4,19] and Johnson, Enriquez-Denton and Kiehn, unpublished). In a critical voltage range, the slow depolarization can increase in duration and size solely due to the voltage dependence of the NMDA component of the EPSP [28,69]. This voltage dependence in itself will mimic a phase advance and increase firing. Second, firing in neonatal rat

654

KIEHN ET AL.

FIG. 5. A leak conductance has similar effect as Ih on phase transitions. (A) Average locomotor drive potential in control (same as in Fig. 4B; solid red line; N-methyl-D-aspartate/serotoin plus CsCl) and after replacement Ih with fast parameters (same as in Fig. 4B; green line) or adding a tonic leak conductance (Gmax ⫽ 10 nS and Vrev ⫽ ⫺30 mV; black line). The stippled red line shows the control data after being shifted upward so that the trough potential was aligned with those obtained after replacing Ih and the leak. (B)–(C) Replacing Ih or adding a leak conductance caused significant (p ⬍ 0.05) decreases in the time (phase) to ⌽half (B) and the first action potential [(C); ⌽ap1]. There were no differences between Ih and the leak. (E) The mean instantaneous firing frequency during locomotion increased in a similar way after replacing Ih and the leak (compare to lower panels). Same color code as in (A). Unpublished data from Kiehn, Kjaerulff, Harris-Warrick and Johnson.

neurons is not precisely determined from cycle to cycle and it is often necessary to average over many cycles to describe the firing profile in individual neurons [39,48,62,64]. This variable discharge pattern is mainly determined by a synaptic input that varies from cycle to cycle [49], although some of the variability in firing may also be due to unreliable transformation of slow potentials into neuronal firing [41]. We therefore propose that the main action of Ih in neonatal rat motor neurons is to increase the firing probability via a tonic depolarization, rather than actually determining the precise timing of action potentials in the locomotor cycle. In other words, Ih acts as a tonically depolarizing leak conductance. It is clear that the dynamic clamp cannot provide an exact replacement of Ih because our biophysical parameters are taken from the average response in a population of motor neurons. Moreover, current injection in the soma may not adequately compensate the contribution of the native dendritic Ih [40]. Finally, the dynamic clamp cannot mimic the dynamic modulation of Ih caused by intracellular messengers [36]. Despite these caveats we believe that the dynamic clamp can provide useful information about the contribution of cellular properties to neuronal firing during rhyth-

mic motor acts. This information is not easily obtained by other means. NMDA-INDUCED BURSTING IN VERTEBRATE MOTOR NEURONS Besides the ionic conductances involved in phase-transition, there are other neuronal membrane properties that are important for motor neuron firing during locomotion. One property in particular has attracted much attention during recent years: the NMDA receptor-mediated voltage oscillations which persist after action potential (AP)-mediated synaptic transmission has been blocked with tetrodotoxin (TTX). These NMDA-induced voltage oscillations are have been found in all vertebrates investigated: the lamprey [66], the chick [44], the turtle [15], the tadpole [47,54 – 56], the adult frog [50] and the neonatal rat [20,38,51]. Across these different species—and even within the same species—the expression of NMDA-induced voltage oscillations seems, however, to vary both with respect to the dependency on activation of other conductances, such as sustained calcium conductances [15,

CELLULAR PROPERTIES AND MOTOR PATTERNS

655

FIG. 6. Rhythmic motor coordination in the absence of action potentials mediated by N-methyl-D-aspartate (NMDA)-induced bursting and gap junctions. (A) A motor neuron recorded during NMDA/serotonin (5-HT) (12 ␮M NMDA/6 ␮M 5-HT) in the presence of TTX (1 ␮M). With no current injection the motor neuron was oscillating with a fast frequency around 3 Hz (top trace). With increasing hyperpolarization (from top to bottom), large amplitude oscillations were revealed. The frequency of the large amplitude oscillations was modulated by changing the membrane potential, while the frequency of the fast low amplitude oscillations remained constant. (B) Same motor neuron as in (A) with hyperpolarization (left) and with no current injection (right), along with the simultaneously recorded ventral root (VR), demonstrating the coupling between the fast low amplitude motor neuron voltage oscillations and the VR oscillations. (C)–(D) Same motor neuron as in (A) and (B), recorded after blocking gap junctions with carbenoxolone (100 ␮M, 20 min). Although the fast oscillations in the ventral root were abolished, the motor neuron still produced voltage oscillations, the frequency of which could be slowed by hyperpolarization of the neuron as shown in (D). (E) Large amplitude long-lasting membrane voltage shifts recorded in ventral root in the presence of NMDA/5-HT (10 ␮M/6 ␮M) and TTX (1 ␮M). Data in (A)–(D) from [63]. Data in (E), Tresch and Kiehn previously unpublished.

68] and the dependency on activation of 5-HT receptors [37,47, 55]. Another important difference between the NMDA-induced oscillations in different species has been whether they exhibit one of the main characteristics of intrinsic pacemaker oscillations, i.e., that the frequency of the oscillations can be varied by changing the neuron’s membrane potential. This characteristic has been demonstrated consistently for NMDA-induced oscillations in lamprey [43,66], chick [44], and turtle motor neurons [15]. However, there has not been a similarly consistent demonstration of this behavior in either the neonatal rat [38,51] or the tadpole [47,54,56]. Recent experiments from our laboratory [63] have provided an explanation for these inconsistencies. Moreover, they have suggested that NMDA-oscillations in individual motor neurons might play an important role in generating a coordinated output from a pool of motor neurons. NMDA-INDUCED BURSTING AND GAP JUNCTIONS CAN COORDINATE A MOTOR OUTPUT IN THE ABSENCE OF ACTION POTENTIAL-GENERATED SYNAPTIC TRANSMISSION As described above, a combination of NMDA and 5-HT evokes stable long-lasting locomotor activity in the neonatal rat (Fig. 3).

This activity is coordinated across the lumbar cord (left-right alternation; Fig. 3) and along the cord (rostrocaudal coordination [12,26]. We have examined whether such evoked rhythmic motor activity persists after blocking AP-mediated neurotransmitter release in rats 0 –2 days after birth [63]. This block of neurotransmitter release was performed either by removal of extracellular calcium or by adding TTX to the superfusate. After such blockade we found that application of NMDA in combination with 5-HT was still capable of producing a stable coordinated rhythmic motor output. In the case of calcium removal, this rhythm was observed in the usual AC ventral root recordings, while in the case of the TTX application it was observed in DC ventral root recordings (Fig. 6B; upper traces). This DC ventral root activity gives an attenuated measurement of the average membrane potential of motor neurons projecting in the recorded root. The DC ventral root rhythm observed in TTX therefore suggested that the NMDA/ 5-HT evoked membrane oscillations were synchronized across a population of motor neurons. These synchronous motor neuron oscillations were, however, not coupled across (e.g., L-L2/R-L2) or along the cord (e.g., L-L2/L-4). These results demonstrate that the activity of localized populations of motor neurons could still be coordinated into a stable rhythmic motor output in the absence of AP-mediated synaptic transmission [63].

656 These coordinated AP-independent motor rhythms were dependent on two conditions. First, the oscillations required coordination of neuronal activity across gap junctions. Application of gap junction blockers such as carbenoxolone abolished ongoing oscillations and prevented their expression (compare top ventral root traces in Figs. 6B–D). It is likely that the gap junction coupling mediating these oscillations was primarily between motor neurons since there does not appear to be any gap junction coupling between interneurons and motor neurons in this preparation [8]. Second, the oscillations required the expression of voltage dependent NMDA-induced oscillations (compare ref. [42]). This requirement was demonstrated by the block of the motor neuron oscillations after removing magnesium from the superfusate. Further, when NMDA receptors were blocked by application of AP-5 and the depolarization of motor neurons restored by increased potassium concentrations, no ventral root oscillations were observed. These results demonstrate that the generation of ventral root oscillations in the absence of action potential dependent neurotransmission resulted from the coordination of NMDA-induced oscillations across gap junctions between motor neurons. Taken together, these observations suggest that neonatal motor neurons might play a significant role in the generation of rhythmic motor output: not only do individual motor neurons exhibit NMDA-induced intrinsic pacemaker properties, but the rhythms in individual motor neurons can be assembled into a coordinated population output without spike activity in premotor networks [63]. We believe that our results can explain previous inconsistencies in demonstrating NMDA-induced pacemaker properties in neonatal rat motor neurons. Schmidt et al. [51] described three classes of NMDA-induced membrane oscillatory behaviors: (1) fast oscillations (0.5–3 Hz) where the frequency was unchangeable but the amplitude varied with membrane polarization, (2) plateau-like potential oscillations superimposed on fast oscillations, and (3) low frequency (0.1– 0.3 Hz) large amplitude long-lasting membrane voltage shifts (LLVSs). We suggest that these different classes of NMDA-induced membrane oscillatory are all different expressions of the interplay between intrinsic NMDA-induced pacemaker properties in individual motor neurons and extrinsic oscillations from other oscillating motor neurons across gap junctions. Such an interplay was suggested from intracellular motor neuron recordings during the NMDA/5-HT evoked oscillations [63] as illustrated in Fig. 6. When hyperpolarized, the motor neuron showed slow large voltage oscillations that co-existed with fast smaller oscillations (Fig. 6A; bottom trace). These fast oscillations at hyperpolarized potentials reflected the extrinsic input from the gap junction coupled network, as evidenced by their coupling to the DC ventral root oscillations (Fig. 6B) and by their abolishment following gap junction blockade (Figs. 6C,D). The slow large voltage oscillations reflected the NMDA-induced intrinsic pacemaker oscillations of the motor neuron, as evidenced by their increased frequency as the cell was depolarized (Fig. 6A, from bottom to top; Figs. 6C,D) and by their maintenance following gap junction blockade (Figs. 6C,D; bottom traces). With no current injection, the intrinsic oscillations were in the same frequency range as the extrinsic input and the two oscillations became coupled (Fig. 6A, top trace and Fig. 6B, right). Note also that the extrinsic input to this cell which was only a few millivolts in amplitude at hyperpolarized levels (consistent with other studies of the electrotonic coupling of neonatal rat motor neurons [8,67]) was substantially amplified due to the nonlinear voltage dependence of the NMDA channel at more depolarized levels. Based on this and other motor neuron recordings, we therefore suggest that the different classes of motor neuron oscillations described by

KIEHN ET AL. Schmidt et al. [51] reflect interactions between extrinsic network and intrinsic pacemaker oscillations. We also propose that gap junction coupling might explain the inability to demonstrate clear pacemaker activity in tadpole motor neurons. This explanation has already been proposed by Sillar and Simmers [56] and is also supported by the observation that synchronized DC ventral root oscillations can be observed in adult frog motor neurons in the presence of TTX and NMDA [50]. Like neonatal rat motor neurons, embryonic and adult frog motor neurons are known to be electrically coupled through gap junctions [3,10,56]. It thus appears that the demonstration of gap junction coupled motor neuron oscillations helps to reconcile the results from several different studies. However, there are several aspects of these motor neuron oscillations, which remain to be explicated. In particular, there was considerable hetereogeneity between different neurons in the balance of intrinsic and extrinsic properties. In some neurons, the extrinsic input completely entrained the intrinsic oscillations (as in Fig. 3B in [63] and in Fig. 3C [38]) while in other neurons, such as the motor neuron illustrated in Fig. 6 or the motor neuron illustrated in Fig. 2C in Schmidt et al. [51], the intrinsic oscillations were strong enough to escape from the extrinsic entrainment when the cells were hyperpolarized. Further, both the intrinsic and extrinsic oscillations often changed over the time course of the NMDA/5-HT application. Intrinsic oscillations generally became stronger with increased duration of application, ultimately resulting in large intrinsic oscillations such as those shown in Fig. 6. Extrinsic oscillations also became stronger but could also develop into irregular patterns, such as those illustrated in Fig. 6E. These irregular network oscillations are similar to the LLVS described by Schmidt et al. [51], suggesting that this class of motor neuron behavior might also be due to properties of the gap junction coupled network. Therefore, depending on the strength extrinsic and intrinsic oscillations at a particular time, a neuron could appear either entirely coupled to the extrinsic oscillation or entirely freely oscillating. This heterogeneity between neurons and the temporal changes in the balance between extrinsic and intrinsic oscillations suggests some sort of recruitment process, possibly in the development of the intrinsic oscillations, in the propagation of these oscillations across the motor neuron population, or in both. A complete understanding of the detailed dynamics underlying these motor neuron network oscillations will require further experiments. FUNCTIONAL RELEVANCE OF GAP JUNCTION COUPLED NETWORK Several previous studies have demonstrated electrical coupling between motor neurons in the neonatal rat [8,67]. The results reviewed here suggest that such coupling in the neonatal rat is capable of a surprising degree of neuronal coordination, even in the absence of AP-dependent neurotransmission. Further, such gap junction mediated coordination plays a role in some aspects of the basic function of spinal motor systems in the neonatal rat [63]. In the adult spinal cord, however, it has been much more difficult to demonstrate the physiological presence of gap junctions between motor neurons, even though the connexin proteins from which they are composed are still expressed [8,35]. While this difficulty suggests that gap junction coupling might simply play a small role in adults, a recent study has shown that gap junction coupling in the adult is reexpressed following axotomy of motor neurons [9]. Thus, it might be that coordination of neuronal activity by gap junction coupling in the adult is dynamically regulated, capable of being recruited in particular behavioral situations. The results of the study described here demonstrate that when such gap junction

CELLULAR PROPERTIES AND MOTOR PATTERNS

657 production of locomotion, synchronization between pairs of neurons was often directly apparent from visible inspection of the spike trains (Fig. 7A). This synchronization was also seen in cross correlations between motor neuron pairs (Fig. 7B). We are currently examining different aspects of motor neuron synchronization, including its pattern in different regions of the spinal cord and its underlying synaptic and cellular mechanisms. These observations clearly demonstrate that motor neuron firing is synchronized during the production of normal behavior in the neonatal rat. A separate issue is the role of the TTX-resistant and NMDA evoked motor neuron oscillations in the production of normal behaviors such as locomotion. It is clear that the NMDA-induced bursting by it self can provide and amplification mechanism for synaptic excitatory inputs to motor neurons during locomotion. However, the frequency of the oscillations evoked in the presence of TTX was usually faster (around 2.25 Hz; [63]) than the frequency of locomotor activity induced by NMDA/5-HT in normal Ringers (0.2–2 Hz; [30,51,58]). Therefore, for the fast oscillations to contribute to the slow rhythmic membrane potential oscillations observed in motor neurons during locomotion they would have to be entrained to a slower rhythm by the synaptic drive from the CPG. Since oscillators can easily be entrained by external inputs we consider this scenario a possibility. Another possibility, which has been suggested by Schmidt and co-workers [51], is that TTXresistant motor neuron oscillations evoked by NMDA correspond to the faster oscillations that can be superimposed on the depolarizing phase of the motor neuron drive during normal locomotion. Regardless of the precise behavioral correlate of the gap junction coupled motor neuronal oscillations, the presence of these intrinsic oscillations reinforces the view that the cellular properties of motor neurons contribute significantly to the normal production of behavior. CONCLUSION

FIG. 7. Motor neuron synchronization during locomotion. (A) Locomotor activity recorded in the L2 and L5 ventral roots (two top traces) with simultaneously spike activity in two L5 motor neurons (two bottom traces). The motor neuron activity was recorded extracellularly with tetrodes and the vertical lines indicate the spike arrival times. Locomotion was induced by a combination of N-methyl-D-aspartate and serotonin (6 ␮M/6 ␮M). (B) The cross correlation between spike arrival times of the two motor neurons illustrated in (A). Spike trains were first convolved with a gaussian kernel with standard deviation of 7 ms and the cross correlation performed using Matlab. Tresch and Kiehn, previously unpublished.

In this review we have presented recent data on the role of motor neuron membrane properties for rhythmic motor-pattern generation in the mammalian spinal cord. We demonstrate that such properties, them being inherent or conditional, contribute strongly to shape and coordinate rhythmic motor outputs. This contribution is expressed in a complex and non-intuitive interaction between the intrinsic motor neuronal properties and network activity. Our results stress the need for studying motor neuron properties in a behavioral context. ACKNOWLEDGEMENTS

coupling is recruited, it will make a strong contribution to spinal systems. One such contribution gap of junctions would be expected to make is the synchronization in action potential activity between motor neurons. Such synchronization in immature animals has been suggested to mediate the activity dependent development of muscle innervation [7]. The strong amplification of excitatory synaptic inputs due to the voltage dependence of the NMDA channel should emphasize this synchronization even further. In order to examine whether such synchronization between the activity in neonatal motor neurons does in fact take place during normal behavior, we have begun experiments recording the action potential activity in multiple motor neurons during the production of locomotion. Using tetrodes [62] we have recorded the spike activity of over 100 motor neuron pairs during the production of locomotion. Of these pairs, over half showed a significant synchronization in spike timing. An example is shown in Fig. 7. Because of the low firing rates of motor neurons during the

The work in Ole Kiehn’s laboratory is supported by the Danish Medical Research Council and the Novo Foundation. M.C.T. is a Postdoctoral Research Fellow supported by the Lundbeck Foundation. We thank Bruce Johnson for participating in the initial dynamic clamp experiments.

REFERENCES 1. Barthe, J. Y.; Clarac, F. Modulation of the spinal network for locomotion by substance P in the neonatal rat. Exp. Brain Res. 115:485– 492; 1997. 2. Bertrand, S.; Cazalets, J. R. Postinhibitory rebound during locomotorlike activity in neonatal rat motoneurons in vitro. J. Neurophysiol. 79:342–335; 1998. 3. Brenowitz, G. L.; Collins, W. F.; Erulkar, S. D. Dye and electrical coupling between frog motoneurons. Brain. Res. 274:371–375; 1983. 4. Brownstone, R. M.; Gossard, J.-P.; Hultborn, H. Voltage-dependent excitation of motoneurones from spinal locomotor centres in the cat. Exp. Brain Res. 102:34 – 44; 1994. 5. Cazalets, J. R.; Borde, M.; Clarac, F. The synaptic drive from the

658

6. 7. 8.

9. 10. 11. 12. 13. 14.

15. 16. 17.

18.

19. 20. 21. 22. 23. 24. 25.

26. 27.

KIEHN ET AL. spinal locomotor network to motoneurons in the newborn rat. J. Neurosci. 16:298 –306; 1996. Cazalets, J. R.; Sqalli-Houssaini, Y.; Clarac, F. Activation of the central pattern generators for locomotion by serotonin and excitatory amino acids in neonatal rat. J. Physiol. 455:187–204; 1992. Chang, Q.; Balice-Gordon, R. J. Gap junctional communication among developing and injured motor neurons. Brain Res. Rev. 32:242–249; 2000. Chang, Q.; Gonzalez, M.; Pinter, M. J.; Balice-Gordon, R. J. Gap junctional coupling and patterns of connexin expression among neonatal rat lumbar spinal motor neurons. J. Neurosci. 19:10813–10828; 1999. Chang, Q.; Pereda, A.; Pinter, M. J.; Balice-Gordon, R. J. Nerve injury induces gap junctional coupling among axotomized adult motor neurons. J. Neurosci. 20:674 – 684; 2000. Collins, W. F. Organization of electrical coupling between frog lumbar motoneurons. J. Neurophysiol. 49:730 –744; 1983. Constanti, A.; Galvan, M. Fast inward-rectifying current accounts for anomalous rectification in olfactory cortex neurones. J. Physiol. 335: 153–178; 1983. Cowley, K. C.; Schmidt, B. J. A comparison of motor patterns induced by N-methyl-D-aspartate, acetylcholine and serotonin in the in vitro neonatal rat spinal cord. Neurosci. Lett. 171:147–150; 1994. Gasparini, S.; DiFrancesco, D. Action of the hyperpolarization-activated current (Ih) blocker ZD 7288 in hippocampal CA1 neurons. Pflugers Arch. 435:99 –106; 1997. Gramoll, S.; Schmidt, J.; Calabrese, R. L. Switching in the activity state of an interneuron that controls coordination of the hearts in the medicinal leech (Hirudo medicinalis). J. Exp. Biol. 186:157–171; 1994. Guertin, P. A.; Hounsgaard, J. NMDA-Induced intrinsic voltage oscillations depend on L-type calcium channels in spinal motoneurons of adult turtles. J. Neurophysiol. 80:3380 –3382; 1998. Hagiwara, S.; Miyazaki, S.; Rosenthal, N. P. Potassium current and the effect of cesium on this current during anomalous rectification of the egg cell membrane of a starfish. J. Gen. Physiol. 67:621– 638; 1976. Harris, N. C.; Constanti, A. Mechanism of block by ZD 7288 of the hyperpolarization-activated inward rectifying current in guinea pig substantia nigra neurons in vitro. J. Neurophysiol. 74:2366 –2378; 1995. Harris-Warrick, R. M.; Coniglio, L. M.; Levini, R. M.; Gueron, S.; Guckenheimer, J. Dopamine modulation of two subthreshold currents produces phase shifts in activity of an identified motoneuron. J. Neurophysiol. 74:1404 –1420; 1995. Hochman, S.; Schmidt, B. J. Whole cell recordings of lumbar motoneurons during locomotor-like activity in the in vitro neonatal rat spinal cord. J. Neurophysiol. 79:743–752; 1998. Hochman, S.; Jordan, L. M.; Schmidt, B. J. TTX-resistant NMDA receptor-mediated voltage oscillations in mammalian lumbar motoneurons. J. Neurophysiol. 72:2559 –2562; 1994. Hughes, S. W.; Cope, D. W.; Crunelli, V. Dynamic clamp study of Ih modulation of burst firing and delta oscillations in thalamocortical neurons in vitro. Neuroscience 87:541–550; 1998. Iizuka, M.; Kiehn, O.; Kudo, N. Development in neonatal rats of the sensory resetting of the locomotor rhythm induced by NMDA and 5-HT. Exp. Brain Res. 114:193–204; 1997. Kiehn, O. Plateau potentials and active integration in the ‘final common pathway’ for motor behaviour. Trends Neurosci. 14:68 –73; 1991. Kiehn, O.; Eken, T. Functional role of plateau potentials in vertebrate motor neurons. Curr. Opin. Neurobiol. 8:746 –752; 1998. Kiehn, O.; Katz, P. S. Making circuits dance: Modulation of motor pattern generation. In: Katz, P. S., ed. Beyond neurotransmission. Neuromodulation and its importance for information processing. New York: Oxford University Press; 1999:275–317. Kiehn, O.; Kjaerulff, O. Spatiotemporal characteristics of 5-HT and dopamine-induced rhythmic hindlimb activity in the in vitro neonatal rat. J. Neurophysiol. 75:1472–1482; 1996. Kiehn, O.; Kjaerulff, O. Distribution of central pattern generators for rhythmic motor output in the spinal cord of limbed vertebrates. In: Kiehn, O.; Harris-Warrick, R. M.; Jordan, L. M.; Hultborn, H.; Kudo, N., eds. Neuronal mechanisms for generating locomotor activity, vol. 860. New York: New York Academy of Sciences; 1998:110 –129.

28. Kiehn, O.; Hounsgaard, J.; Sillar, K. T. Basic building blocks of vertebrate spinal central pattern generators. In: Stein, P. S. G.; Grillner, S.; Selverston, A. I.; Stuart, D. G., eds. Neurons, networks and motor behavior. Cambridge, MA: MIT Press; 1997:47–59. 29. Kiehn, O.; Sillar, K. T.; Kjaerulff, O.; McDearmid, J. R. Effects of noradrenaline on locomotor rhythm-generating networks in the isolated neonatal rat spinal cord. J. Neurophysiol. 82:741–746; 1999. 30. Kjaerulff, O.; Kiehn, O. Distribution of networks generating and coordinating locomotor activity in the neonatal rat spinal cord in vitro: A lesion study. J. Neurosci. 16:5777–5794; 1996. 31. Kjaerulff, O.; Kiehn, O. Crossed rhythmic synaptic input to motoneurons during selective activation of the contralateral spinal locomotor network. J. Neurosci. 17:9433–9447; 1997. 32. Kjaerulff, O.; Kiehn, O. 5-HT modulation of multiple inward rectifiers in intact preparations of the neonatal rat spinal cord. J. Neurophysiol. In press; 2001. 33. Kudo, N.; Yamada, T. N-methyl-D,L-aspartate-induced locomotor activity in a spinal cord-hindlimb muscles preparation of the newborn rat studied in vitro. Neurosci. Lett. 75:43– 48; 1987. 34. Le Masson, G.; Le Masson, S.; Moulins, M. From conductances to neural network properties: Analysis of simple circuits using the hybrid network method. Prog. Biophys. Mol. Biol. 64:201–220; 1995. 35. Lipski, J. Is there electrical coupling between phrenic motoneurons in cats? Neurosci. Lett. 46:229 –234; 1984. 36. Luthi, A.; McCormick, D. A. Ca2⫹-mediated up-regulation of Ih in the thalamus. How cell-intrinsic ionic currents may shape network activity. Ann. N.Y. Acad. Sci. 868:765–769; 1999. 37. MacLean, J. N.; Cowley, K. C.; Schmidt, B. J. NMDA receptormediated oscillatory activity in the neonatal rat spinal cord is serotonin dependent. J. Neurophysiol. 79:2804 –2808; 1998. 38. MacLean, J.; Schmidt, B. J.; Hochman, S. NMDA receptor activation triggers voltage oscillations, plateau potentials and bursting in neonatal rat lumbar motoneurons in vitro. Eur. J. Neurosci. 9:2702–2711; 1997. 39. MacLean, J. N.; Hochman, S.; Magnuson, D. S. Lamina VII neurons are rhythmically active during locomotor-like activity in the neonatal rat spinal cord. Neurosci. Lett. 197:9 –12; 1995. 40. Magee, J. C. Dendritic Ih normalizes temporal summation in hippocampal CA1 neurons. Nat. Neurosci. 2:848; 1999. 41. Mainen, Z. F.; Sejnowski, T. J. Reliability of spike timing in neocortical neurons. Science 268(5216):1503–1506; 1995. 42. Manor, Y.; Rinzel, J.; Segev, I.; Yarom, Y. Low-amplitude oscillations in the inferior olive: A model based on electrical coupling of neurons with heterogeneous channel densities. J. Neurophysiol. 77:2736 –2752; 1997. 43. Matsushima, T.; Grillner, S. Neural mechanisms of intersegmental coordination in lamprey: Local excitability changes modify the phase coupling along the spinal cord. J. Neurophysiol. 67:373–388; 1992. 44. Moore, L. E.; Chub, N.; Tabak, J.; O’Donovan, M. NMDA-induced dendritic oscillations during a soma voltage clamp of chick spinal neurons. J. Neurosci. 19:8271– 8280; 1999. 45. Nichols, C. G.; Lopatin, A. N. Inward rectifier potassium channels. Ann. Rev. Physiol. 59:171–191; 1997. 46. Pape, H.-C. Queer current and pacemaker: The hyperpolarizationactivated cation current in neurons. Annu. Rev. Physiol. 58:299 –327; 1996. 47. Prime, L.; Pichon, Y.; Moore, L. E. N-methyl-D-aspartate-induced oscillations in whole cell clamped neurons from the isolated spinal cord of Xenopus laevis embryos. J. Neurophysiol. 82:1069 –1073; 1999. 48. Raastad, M.; Kiehn, O. Spike coding during locomotor network activity in ventrally located neurons in the isolated spinal cord from neonatal rat. J. Neurophysiol. 83:2825–2834; 2000. 49. Raastad, M.; Johnson, B. R.; Kiehn, O. Analysis of EPSCs and IPSCs carrying rhythmic, locomotor-related information in the isolated spinal cord of the neonatal rat. J. Neurophysiol. 78:1851–1859; 1997. 50. Rioult-Pedotti, M.-S. Intrinsic NMDA-induced oscillations in motoneurons of an adult vertebrate spinal cord are masked by inhibition. J. Neurophysiol. 77:717–730; 1997. 51. Schmidt, B. J.; Hochman, S.; MacLean, J. N. NMDA receptor-mediated oscillatory properties: Potential role in rhythm generation in the mammalian spinal cord. In: Kiehn, O.; Harris-Warrick, R. M.; Jordan, L. M.; Hultborn, H.; Kudo, N., eds. Neuronal mechanisms for gener-

CELLULAR PROPERTIES AND MOTOR PATTERNS

52. 53. 54. 55. 56. 57. 58. 59. 60.

ating locomotor activity, vol. 860. New York: New York Academy of Sciences; 1998:189 –202. Sharp, A. A.; O’Neil, M. B.; Abbott, L. F.; Marder, E. The dynamic clamp: Artificial conductances in biological neurons. Trends Neurosci. 16:389 –394; 1993. Sharp, A. A.; Skinner, F. K.; Marder, E. Mechanisms of oscillation in dynamic clamp constructed two-cell half-center circuits. J. Neurophysiol. 76:867– 883; 1996. Sillar, K. T.; Simmers, A. J. Oscillatory membrane properties of spinal cord neurons that are active during fictive swimming in Rana temporaria embryos. Eur. J. Morphol. 32:185–192; 1994. Sillar, K. T.; Simmers, A. J. 5-HT induces NMDA receptor-mediated intrinsic voltage oscillations in embryonic amphibian spinal neurons. Proc. R. Soc. Lond. B 255:139 –145; 1994. Sillar, K. T.; Simmers, A. J. Electrical coupling and intrinsic neuronal oscillations in Rana temporaria spinal cord. Eur. J. Morph. 32:293– 298; 1994. Smith, J. C.; Feldman, J. L.; Schmidt, B. J. Neural mechanisms generating locomotion studied in mammalian brain stem-spinal cord in vitro. FASEB J. 2:2283–2288; 1988. Sqalli-Houssaini, Y.; Cazalets, J. R.; Clarac, F. Oscillatory properties of the central pattern generator for locomotion in neonatal rats. J. Neurophysiol. 70:803– 813; 1993. Takahashi, T. Inward rectification in neonatal rat spinal motoneurones. J. Physiol. 423:47– 62; 1990. Takahashi, T.; Berger, A. J. Direct excitation of rat spinal motoneurones by serotonin. J. Physiol. 423:63–76; 1990.

659 61. Thoby-Brisson, M.; Telgkamp, P.; Ramirez, J. M. The role of the hyperpolarization-activated current in modulating rhythmic activity in the isolated respiratory network of mice. J. Neurosci. 20:2994 –3005; 2000. 62. Tresch, M. C.; Kiehn, O. Coding of locomotor phase in populations of neurons in rostral and caudal segments of the neonatal rat lumbar spinal cord. J. Neurophysiol. 82:3563–3574; 1999. 63. Tresch, M. C.; Kiehn, O. Motor coordination without action potentials in the mammalian spinal cord. Nat. Neurosci. 3:593–599; 2000. 64. Tresch, M. C.; Kiehn, O. Population reconstruction of the locomotor cycle from interneuron activity in the mammalian spinal cord. J. Neurophysiol 83:1972–1978; 2000. 65. Ulrich, D.; Huguenard, J. R. GABA(A)-receptor-mediated rebound burst firing and burst shunting in thalamus. J. Neurophysiol. 78:1748 – 1751; 1997. 66. Walle´n, P.; Grillner, S. N-methyl-D-aspartate receptor-induced, inherent oscillatory activity in neurons active during fictive locomotion in the lamprey. J. Neurosci. 7:2745–2755; 1987. 67. Walton, K. D.; Navarrete, R. Postnatal changes in motoneurone electrotonic coupling studied in the in vitro rat lumbar spinal cord. J. Physiol. 433:283–305; 1991. 68. Wikstro¨m, M. A.; Bu¨schges, A. S.; Grillner, S.; El Manira, A. Roles of high-voltage activated calcium channel in a vertebrate locomotor network. Soc. Neuorosci. Abstr. 24:1668; 1998. 69. Ziskind-Conhaim, L. NMDA receptors mediate poly- and monosynaptic potentials in motoneurons of rat embryos. J. Neurosci. 10:125– 135; 1990.