Locomotor rhythm generation is a fundamental characteristic of neural networks in the spinal cord. Identifying the synaptic interactions between neurons in the locomotor circuitry is key to our understanding of the mechanisms that underlie the production of rhythmic motor outputs. Using transgenic mice in which the homeobox gene HB9 drives the reporter green fluorescent protein (GFP), we have demonstrated that a genetically distinct cluster of Hb9/GFP-expressing interneurons (Hb9 INs) can generate locomotor-like rhythms in the newborn mouse spinal cord (Hinckley et al., 2005b). Processes of Hb9 INs are in close apposition to adjacent Hb9 INs, raising the possibility that the interneurons are synaptically interconnected. To test this hypothesis, whole-cell paired recordings were performed from visually identified Hb9 INs. High-incidence bidirectional electrical coupling was evident between Hb9 INs in spinal cords of newborn and juvenile mice. The coupling strength varied from 2 to 32% with an average of 12%. Our data suggested that the variability was not correlated with the distribution of electrical synapses at different electronic distances. Electrical synapses behaved as low-pass filters, reducing currents passing at frequencies >3 Hz. Episodes of spontaneous bursts of EPSCs were synchronous in coupled Hb9 INs, indicating that common synaptic inputs coordinated their activity. However, non-NMDA receptor-mediated synaptic transmission was not required to synchronize neurochemically induced membrane oscillations between electrically coupled interneurons. The finding that electrical transmission persists in mice that can walk is indicative of its importance in coordinating the activity of this neuronal population in functionally mature spinal networks.
- electrical coupling
- gap junction-mediated transmission
- rhythm coordination
- locomotor-related interneurons
- excitatory spinal interneurons
- mouse spinal cord
Electrical transmission is rare in the mature mammalian nervous system but is evident among neurons in which synchronous activity is important for function, such as between hippocampal pyramidal neurons (MacVicar and Dudek, 1981) inferior olivary neurons (Llinás and Yarom, 1986), cerebellar inhibitory neurons (Mann-Metzer and Yarom, 1999), and GABAergic neurons in the neocortex (Galarreta and Hestrin, 1999; Gibson et al., 1999). Motor behavior requires temporally coordinated activity between motoneurons innervating homonymous muscles. It has long been recognized that before the establishment of strong synaptic inputs in newborn rodents, motoneuron activity is synchronized by gap junction-mediated electrical coupling in the spinal cord (Fulton et al., 1980; Chang et al., 1999) and brainstem (Rekling and Feldman, 1997). In spinal motoneurons, as in numerous other preparations, gap junction connectivity gradually decreases as neuronal activity increases (Walton and Navarrete, 1991), and inappropriate synapses onto motoneurons are eliminated (Seebach and Ziskind-Conhaim, 1994).
Electrical synapses between interneurons are crucial for generating rhythmic motor outputs in invertebrates (Norekian, 1999; for review, see Marder, 1998), but little is known about electrical transmission between rhythmic interneurons in networks mediating motor activity in tetrapods (for review, see Kiehn and Tresch, 2002). The limited information can be at least partially attributed to difficulties identifying homogeneous populations of locomotor-related spinal interneurons. Using paired recordings, it has been shown that functionally identified rhythmogenic respiratory interneurons are electrically coupled in the pre-Bötzinger complex of newborn mice (Rekling et al., 2000).
Recently, we have characterized a genetically distinct cluster of lamina VIII interneurons expressing the Hb9 protein [Hb9 interneurons (Hb9 INs)]. Neurochemically induced locomotor-like rhythms in green fluorescent protein-positive (GFP+) Hb9 INs are in-phase with rhythmic motor outputs (Hinckley et al., 2005b), raising the intriguing possibility that they are integral components of the locomotor central pattern generator. Based on morphological and immunohistochemical observations, we have proposed that Hb9 INs are premotor, excitatory interneurons.
Neurobiotin-filled processes of Hb9 INs are in close apposition to somata and proximal dendrites of adjacent Hb9 INs (Hinckley et al., 2005b), suggesting that the clustered interneurons are synaptically interconnected. To test this hypothesis, whole-cell paired recordings were performed in Hb9 INs in spinal cords of newborn stationary mice [1–4 d after birth (P1–P4)] and juvenile mice that can walk (P10–P11). Bidirectional electrical coupling was evident between the majority of Hb9 INs, and it persisted in juvenile mice with functionally mature motor activity. Moreover, our findings suggested that electrical transmission contributed to rhythm coordination between coupled Hb9 INs, independently of fast glutamatergic transmission.
A preliminary report of this study was published in an abstract form (Ziskind-Conhaim and Hinckley, 2005).
Materials and Methods
Whole-cell and population recordings.
Experiments were performed using the HB9/eGFP transgenic mouse line (Wichterle et al., 2002). Newborn mice (P1–P4) were anesthetized by hypothermia, and juvenile mice (P10–P11) were anesthetized by exposure to halothane or ether. Mice were then decapitated, and spinal cords were extracted in ice-cold oxygenated extracellular solution. The procedures for whole-cell recordings from visually identified Hb9 INs in the hemisected spinal cord were identical to those described previously (Ziskind-Conhaim et al., 2003; Hinckley et al., 2005b). Simultaneous whole-cell recordings from Hb9 INs (Multiclamp 700B amplifier; Molecular Devices, Union City, CA) were performed using electrodes pulled to tip resistances of 5–7 MΩ using a multi-stage puller (P-97; Sutter Instruments, Novato, CA). Intracellular potentials/currents were filtered at 3 kHz and digitized at 10–20 kHz. Only interneurons with resting membrane potentials more negative than −50 mV and overshooting action potentials were included in this study. Membrane potentials were corrected for a 10 mV liquid junction potential (Gao et al., 2001). Population recordings from lumbar motoneurons (L1–L3 ventral roots) were identical to those described previously (Hinckley et al., 2005a).
Retrograde labeling of neurons.
To retrogradely label motoneuron somata, the ventral root was placed in a tight-fitting glass pipette filled with crystals of rhodamine dextran-amine dissolved in extracellular solution (Glover et al., 1986). Incubation was for 5–8 h at room temperature. Spinal cords were then fixed in 4% paraformaldehyde in PBS overnight, and neuron labeling was examined in 100-μm-thick transverse sections.
Neurons expressing choline acetyltransferase (ChAT) were identified by staining with anti-ChAT antibody. Fixed sections (100 μm; see above) were incubated in 3% hydrogen peroxide and 10% methanol for 30 min and in 0.2% Triton for 10 min, followed by a 1 h incubation in 10% horse serum. All dilutions were in 0.1 m PBS. This was followed by incubation in goat anti-ChAT antibody (1:200; Chemicon, Temecula, CA), 0.1% Triton, and 0.5% horse serum for 1 h at room temperature and overnight at 4°C. Sections were then washed, blocked in 10% horse serum for 10 min, and incubated in horse biotinylated anti-goat antibody (1:200; Vector Laboratories, Burlingame, CA), 0.05% Triton, and 0.5% horse serum for 1 h. To visualize the staining, sections were incubated in ABC reagent (Vector Laboratories) and developed with DAB (Ziskind-Conhaim, 1990).
The coupling coefficient was measured from the ratio between the amplitude of steady-state hyperpolarizations in the postjunctional and prejunctional neurons (V2:V) in response to a 30 pA/1 s negative current pulse in the prejunctional neuron (I1). The junctional conductance was estimated from the current generated in the postjunctional neuron in response to a voltage command of −50 mV in the prejunctional neuron. The calculations were based on the method described by Neyton and Trautmann (1985). Junctional delays were measured between the peak action potential in the prejunctional neuron and the peak transient inward current in the postjunctional neuron. The time constant of the junctional current was measured from the onset of the current, which was best fit with a single-exponential function (Origin 6.0).
Cross-correlograms were used to evaluate the correlation between spontaneous action potentials in paired Hb9 INs. Data were resampled at 1–2.5 kHz (Hinckley et al., 2005a), and cross-correlograms were computed using Clampfit 9.2 (Molecular Devices). Histograms of instantaneous spike frequency tested the correlation between firing in Hb9 INs and the phase of the cycle period.
Solutions and chemicals.
The extracellular solution contained the following (in mm): 128 NaCl, 4 KCl, 1.5 CaCl2, 1 MgSO4, 0.5 NaH2PO4, 21 NaHCO3, and 30 glucose. The solution was adjusted to pH 7.3 using NaOH, and the osmolarity was 315–325 mOsm. In extracellular solution containing tetraethylammonium (TEA; 10 mm), the NaCl concentration was reduced to 118 mm and CaCl2 was increased to 3 mm. The whole-cell pipette solution contained the following (in mm): 140 K-gluconate, 9 KCl, 10 HEPES, 0.2 EGTA, 1 Mg-ATP, and 0.1 GTP. The solution was adjusted to pH 7.2 using KOH, and the osmolarity was 290–305 mOsm. Neurobiotin (0.5%) was included in the pipette solution. All chemicals were obtained from Sigma (St. Louis, MO).
Identification of Hb9 INs
The homeodomain transcription factor HB9 is primarily expressed in motoneurons (Arber et al., 1999; Thaler et al. 1999). Our findings that the Hb9 protein is distinctly expressed in a group of lamina VIII interneurons in segments L1–L3 (Hinckley et al., 2005b) raised the interesting possibility that these are, in fact, a subpopulation of motoneurons. To determine whether axons of Hb9 INs projected to the periphery, motor axons were retrogradely filled with rhodamine-conjugated dextran-amine. Somatic and sympathetic motoneurons, but not Hb9 Ins, were labeled with rhodamine (Fig. 1A), indicating that Hb9 INs did not innervate peripheral targets. One of the criteria for motoneuron identification is the generation of antidromic action potentials in response to ventral root stimulation. The failure of ventral root stimulation to evoke action potentials in Hb9 INs (n = 8; data not shown) provided additional evidence that Hb9 INs do not constitute a subpopulation of motoneurons.
ChAT, the enzyme for acetylcholine biosynthesis, is frequently used as a marker for motoneuron identification. Somatic and sympathetic motoneurons were prominently stained with anti-ChAT antibody, but the staining was absent in Hb9 INs (Fig. 1A). These findings ruled out the possibility that Hb9 INs are cholinergic neurons. Similar data have been reported recently in juvenile Hb9/GFP mice (Wilson et al., 2005).
The morphological properties of the spindle-shaped Hb9 INs and their clustering in medial lamina VIII facilitated their visual targeting for whole-cell recordings. Common electrophysiological properties served as additional measures for their identification. These properties included the following: (1) relatively high input resistance, ranging from 700 MΩ to 1.3 GΩ (Fig. 1B); (2) linear current–voltage relationships at potentials more negative than −50 mV. Hyperpolarization-dependent depolarization sags were frequently recorded in adjacent interneurons but not in Hb9 INs (Fig. 1B); (3) small afterhyperpolarizing potentials (<10 mV); and (4) a maximal firing rate of ∼25 Hz in response to prolonged positive currents and characteristic spike-frequency adaptation. For example, at 20 Hz, the interspike interval increased by an average of ∼85% during the 800 ms current pulse. Based on their genetic, morphological, and electrophysiological properties, we concluded that Hb9 INs constitute a homogenous population of locomotor-related interneurons.
Bidirectional electrical coupling between Hb9 INs
To determine whether Hb9 INs were synaptically interconnected, interneurons with visible GFP+ processes between them were targeted for paired recordings. Prolonged membrane depolarization and a train of action potentials in the injected neuron produced a relatively large DC component and spikelets correlating with each action potential in the noninjected neuron (Fig. 2). Similarly, hyperpolarization of one neuron hyperpolarized the second neuron. The transfer of positive and negative currents between the paired neurons suggested that the neurons were electrically coupled. Electrical transmission was evident between 84% of Hb9 INs in newborn mice (n = 31 of 37 pairs). In all coupled interneurons (n = 31 pairs), current passing was bidirectional so that current injection in the second neuron also produced potential changes in both neurons (Fig. 2). Electrical coupling was not detected between Hb9 INs and adjacent GFP− interneurons (n = 7).
The coexistence of electrical and chemical synapses has been demonstrated in various preparations. To examine whether Hb9 INs are coupled by excitatory glutamatergic synapses, action potentials were evoked in prejunctional neurons to generate inward currents in postjunctional neurons (Fig. 3). Currents with latencies <0.3 ms and an average amplitude of 6.5 ± 1.4 pA (SE; n = 16) were recorded in electrically coupled interneurons. The amplitude of the evoked currents did not significantly change at various membrane potentials (e.g., −40 to −80 mV) (Fig. 3A). Moreover, 6-cyano-7-nitroquinoxalene-2,3-dione (CNQX, 10 μm), a non-NMDA receptor (NMDAR) antagonist that suppressed spontaneous and dorsal root-evoked EPSCs, did not block the inward currents (n = 8) (Fig. 3B). To confirm that the short-latency currents were mediated via gap junctions, recordings were performed in the presence of carbenoxolone, a gap junction blocker (Draguhn et al., 1998). Carbenoxolone did not alter the resting membrane potential or action potential properties. Exposure to carbenoxolone (100 μm) reversibly suppressed the postjunctional currents (Fig. 3C), with an average current reduction of 79.2 ± 0.07% (SE; n = 7).
The fraction of electrically coupled spinal motoneurons and the strength of transmission between them significantly decrease during the second week after birth, when juvenile rodents exhibit functionally mature motor behavior (Walton and Navarrete, 1991). To determine whether a similar developmental trend affected gap junction coupling between Hb9 INs, paired recordings were performed at P10–P11. Whole-cell recordings were difficult to obtain in older mice. In contrast to the transient coupling between spinal motoneurons, electrical transmission persisted in 83% of Hb9 INs (n = 5 of 6 pairs) in juvenile mice (Fig. 4A).
Functional properties of electrical coupling
The average distance between somata of coupled Hb9 INs was 33 μm (±3.8 SE; n = 25). The estimate was based on the distance between the tips of the recording electrodes. The mean time constant of the junctional currents was 2.9 ± 0.6 ms (n = 12). The strength of electrical synapses as estimated by calculating the coupling coefficient varied considerably (Fig. 4B), ranging from 2 to 32% with an average of 12.4 ± 1.6% (SE; n = 27) in newborn mice. Similar ratios were estimated in both directions. It is unlikely that the variability in coupling coefficients was related to the distance of electrical synapses from the somata, because neither the time constant nor the distance between paired interneurons was correlated with the coupling coefficient (r = 0.36 and 0.32, respectively). An average coupling coefficient of 13.3 ± 2.4% (n = 5) was estimated in juvenile mice, indicating that the strength of electrical transmission did not decrease during the first 2 weeks after birth.
The junctional conductance was estimated from currents generated in postjunctional neurons in response to voltage commands of −50 mV in prejunctional neurons (Fig. 4A). An average conductance of 139 ± 20.7 pS (SE; n = 12) was calculated for coupled Hb9 INs in both newborn and juvenile mice.
To examine whether the coupling coefficient was frequency dependent as has been demonstrated in other systems (Galarreta and Hestrin, 1999; Rekling et al., 2000, Gibson et al., 2005), coupling ratios were measured at variable sine-wave frequencies (Fig. 5). The coupling coefficient decreased as a function of increased frequency to ∼50% at 2–3 Hz. The functional implication of the low-pass filtering was that waveforms of fast signals such as action potentials were heavily distorted (Fig. 5B).
Episodes of spontaneous bursts of EPSCs and action potentials were infrequent in Hb9 INs, but in all cases they were strongly coordinated between electrically coupled Hb9 INs (Fig. 6A,B). It is reasonable to assume that common synaptic inputs triggered the onset of spontaneous bursts. Firing at 0.5–2 Hz, which was not associated with bursts of EPSCs, was also synchronized between coupled Hb9 INs (Fig. 6C,D). In the example shown, spikelet amplitudes varied from 0.8 to 2.6 mV in one interneuron and from 0.7 to 2.4 mV in the second interneuron, with averages of 1.8 mV (n = 38) and 1.4 mV (n = 44), respectively. Similar potentials that were not associated with prejunctional action potentials had mean amplitudes of 1.5 mV (n = 28) and 1.0 mV (n = 8). It is conceivable that these were spikelets generated by firing in another electrically coupled Hb9 INs. However, we could not definitively identify these as postjunctional potentials because their waveforms were similar to those of EPSPs (Fig. 5B).
Exposure to NMDA (4–7 μm), 5-HT (7–10 μm), and dopamine (25–50 μm) (Hinckley et al., 2005a) produced membrane oscillations and rhythmic firing in Hb9 INs that were in-phase with ventral root bursts (Fig. 7A). A rhythmic, twofold increase in EPSC frequency was evident during ventral root bursts, but the low frequency of IPSCs (<0.5 Hz) did not change during the cycle period Hinckley et al., 2005b, their Fig. 7. The increased EPSC frequency preceded the burst onset, raising the possibility that excitatory synaptic inputs triggered locomotor-like membrane oscillations. To determine whether rhythm onset was dependent on non-NMDAR-mediated EPSCs, experiments were conducted in the presence of CNQX. CNQX-resistant membrane oscillations were recorded in 70% of Hb9 INs (n = 7 of 10), but ventral root bursts were suppressed. This finding suggested that neurochemically induced rhythm initiation was independent of non-NMDAR-mediated synaptic transmission.
Large, neurochemically induced membrane oscillations were generated more consistently in a low concentration of TEA (10 mm) and increased extracellular Ca2+ concentration (3 mm) (Fig. 7B). Synchronous rhythms were evident in the presence of CNQX (n = 6), indicating that non-NMDAR-mediated synaptic transmission was not required for rhythm coordination between electrically coupled Hb9 INs. The similar amplitude and waveform that characterized the induced rhythms in electrically coupled Hb9 INs were attributed to their homogeneous electrophysiological properties. Under these conditions, the average cycle period was 16.3 s (with a coefficient of variance of 0.2; n = 5), longer than the average period of 9 s reported previously (Hinckley et al., 2005b). It is conceivable that blocking TEA-sensitive K currents contributed to the longer cycle period.
Bidirectional electrical coupling plays an important role in synchronizing the rhythms generated in the population of Hb9-expressing interneurons in the mouse spinal cord. Electrical transmission has been reported previously between somatic (Walton and Navarrete, 1991; Chang et al., 1999) and sympathetic (Logan et al., 1996) motoneurons in the rat spinal cord. However, to the best of our knowledge, this is the first report that demonstrates a high-incidence electrical transmission between excitatory, locomotor-related interneurons in the mammalian spinal cord. Bidirectional electrical coupling provides an effective mechanism of coordinating the activity of clustered Hb9 INs independently of non-NMDAR-mediated synaptic transmission.
Properties of electrical coupling between Hb9 INs
The properties of electrical transmission between rhythmic Hb9 INs will be discussed with reference to known characteristics of electrical synapses between other spinal neurons and neurons that comprised integral components of networks mediating rhythmic motor behavior. Electrical coupling was evident between the majority of Hb9 INs in the spinal cords of newborn (84%) and juvenile (83%) mice. The incidence of coupled Hb9 INs was significantly higher than the percentage of coupled neurons reported in other neuronal populations. Ten percent of sympathetic, preganglionic neurons were electrically coupled in the juvenile rat spinal cord (Nolan et al., 1999). A low percentage of coupling (25%) was also documented between pairs of inspiratory XII motoneurons (Rekling and Feldman, 1997) and among rhythmic type-1 inspiratory interneurons (13%) in the newborn rat (Rekling et al., 2000). The ability to visually target the GFP+ Hb9 INs for paired recordings and confirming their identity based on electrophysiological properties decreased the probability of recording from adjacent, non-Hb9 INs.
The location of gap junctions has functional implications in signal transmission. Somatic or proximal dendritic contacts are more effective than distal contacts in current transmission because they tend to reduce signal filtering and the time for its propagation between cells. Clustered, paired Hb9 INs were closer to each other than distances reported for electrically coupled inspiratory-modulated motoneurons (50 μm) and rhythmogenic type-1 interneurons (60 μm). The relatively short time constant of currents transmitted between Hb9 INs might be indicative of electrical synapses close to the somata, but other factors, not examined in this study, cannot be ruled out.
The strength of electrical coupling between Hb9 INs varied considerably, but we concluded that the variability was not related to differences in electrotonic distances. The average coupling coefficient of 12–13% was similar to that reported in various preparations including preganglionic neurons (Nolan et al., 1999), spinal interneurons in embryonic zebrafish (Saint-Amant and Drapeau, 2001), inspiratory motoneurons (Rekling and Feldman, 1997), and rhythmogenic inspiratory interneurons in newborn rats (Rekling et al., 2000). The average junctional conductance was ∼140 pS. Assuming that the channels are formed by the neuronal gap junction protein connexin 36, and that the unitary channel conductance is 14 pS (Teubner et al., 2000), then only ∼10 channels are generally open. It is conceivable that because of the high input resistance of Hb9 Ins, only a small number of channels need to be open to allow for an effective current transfer between the coupled neurons.
Similar to previous reports, we found that electrical synapses behaved as low-pass filters, greatly attenuating signals traveling between coupled interneurons at frequencies higher than 2–3 Hz. Because low-pass filtering allows the passing of slow potential changes, it is likely that gap junction coupling between Hb9 INs contributes to the synchronization of slow membrane oscillations such as those underlying locomotor-like rhythms (Fig. 7). A similar mechanism has been proposed for the coordination of membrane oscillations in inspiratory neurons at frequencies corresponding to the inspiratory synaptic drive (∼5 Hz) (Rekling et al., 2000).
Functional significance of electrical transmission between Hb9 INs
It is generally believed that electrical synapses enhance the synchronization of subthreshold potentials and spiking among clusters of neurons (for review, see Bennett, 1997; Galarreta and Hestrin, 2001; Connors and Long, 2004). Providing direct experimental evidence that electrical transmission coordinates network activity is complicated because its role in rhythm coordination has to be isolated from the contribution of chemical synaptic inputs. Our findings that membrane oscillations persisted in the presence of CNQX implied that electrical coupling played an important role in synchronizing the activity of this homogenous population of excitatory interneurons. Gap junction-mediated electrical transmission coordinates bilateral spontaneous bursts in embryonic chick spinal cord (Milner and Landmesser, 1999), and motoneuron coordination that is independent of firing in premotor interneurons has been reported in the newborn rat spinal cord (Tresch and Kiehn, 2000).
Determining the number of electrically interconnected interneurons provides a qualitative estimate of the extent of output synchronization within the population of Hb9 INs. Numerous reports have calculated the size of coupled networks based on dye coupling (e.g., neurobiotin injection) (Devor and Yarom, 2002; Lin et al., 2003; Hornstein et al., 2004). In our studies, indirect labeling of neurons by neurobiotin diffusion during whole-cell recordings from Hb9 INs was apparent in <7% of the preparations. Therefore, dye diffusion through gap junctions cannot be used as a reliable technique to label electrically coupled Hb9 INs. It is possible to assess the number of coupled interneurons based on distinct waveforms and multiple peak amplitude histograms of postjunctional potentials (Devor and Yarom, 2002). This analytical approach could not be implemented in our study because the waveforms of postjunctional potentials in Hb9 INs were similar to those of EPSPs (Fig. 5B).
Electrical transmission is widespread throughout the developing mammalian nervous system (for review, see Dermietzel and Spray, 1993; Kandler and Katz, 1995), with possible roles not only in synchronizing the activity of a functional neural population but also in providing channels for intercellular diffusion of molecules such as second messengers. In the developing mammalian spinal cord and brainstem, electrical coupling between motoneurons has been proposed as the primary mechanism for coordination of motoneuron firing during movement (for review, see Kiehn and Tresch, 2002) and breathing (Rekling and Feldman, 1997; Rekling et al., 2000). However, the coupling is transient and is reduced from 77% occurrence in spinal motoneurons of newborn rats (P0–P3) to only 31% at P8–P13 (Walton and Navarrete, 1991). Moreover, its strength is significantly weaker in juvenile rats. Although electrical transmission disappears in mature spinal motoneurons, connexin expression persists into adulthood (Chang et al., 1999). The developmental trend to eliminate functional gap junctions between neonatal spinal motoneurons can be delayed by transient, botulinum-induced paralysis (Pastor et al., 2003). Furthermore, nerve injury restores the coupling among axotomized adult rat motoneurons (Chang and Balice-Gordon, 2000), suggesting that dysfunctional synaptic connections with their target muscles re-establish electrical transmission between mature motoneurons.
Based on studies of synaptic communication between motoneurons, the general consensus is that electrical coupling is widespread in the immature mammalian spinal cord, but it has a smaller role in rhythm coordination in the adult spinal cord. Our findings that bidirectional electrical transmission persists between locomotor-related excitatory interneurons in mice with functionally mature motor behavior, raises the intriguing possibility that gap junction coupling provides a mechanism of synchronizing their activity in the adult spinal cord. It will be of great interest to examine whether electrical coupling is prevalent among other spinal interneurons and serves as a fundamental mechanism of rhythm coordination in neuronal populations in the mammalian spinal cord.
This work was supported by National Institutes of Health Grant NS23808 to L.Z.-C. We dedicate our paper to the memory of Prof. Robert Werman (1929–2005), a renowned scholar who made fundamental contributions to our understanding of synaptic transmission in the spinal cord. We thank Linying Wu for expert technical assistance with rhodamine staining.
- Correspondence should be addressed to Dr. Lea Ziskind-Conhaim, Department of Physiology, 129 SMI, University of Wisconsin Medical School, Madison, WI 53706. Email: