BothN-methyl-d-aspartate (NMDA) and serotonin (5-HT) receptors contribute to the generation of rhythmic motor patterns in the rat spinal cord. Co-application of these chemicals is more effective at producing locomotor-like activity than either neurochemical alone. In addition, NMDA application to rat spinal motoneurons, synaptically isolated in tetrodotoxin, induces nonlinear membrane behavior that results in voltage oscillations which can be blocked by 5-HT antagonists. However, the mechanisms underlying NMDA and 5-HT receptor interactions pertinent to motor rhythm production remain to be determined. In the present study, an in vitro neonatal rat spinal cord preparation was used to examine whether NMDA receptor-mediated nonlinear membrane voltage is modulated by 5-HT. Whole-cell recordings of spinal motoneurons demonstrated that 5-HT shifts the region of NMDA receptor-dependent negative slope conductance (RNSC) of the current-voltage relationship to more hyperpolarized potentials and enhances whole-cell inward current. The influence of 5-HT on the RNSC was similar to the effect on the RNSC of decreasing the extracellular Mg2+concentration. The results suggest that 5-HT may modulate this form of membrane voltage nonlinearity by regulating Mg2+ blockade of the NMDA ionophore.
The contribution of intrinsic and conditional neuronal properties to motor pattern generation has been extensively studied in invertebrates (e.g., Hartline and Graubard 1992; Meech 1979; Miller and Selverston 1982) and lower vertebrates (e.g.,Grillner et al. 1991). In contrast, examination of these properties during rhythm generation in the mammalian spinal cord (e.g.,Brownstone et al. 1994; Gorassini et al. 1999; Kiehn 1991; Kiehn et al. 1996; Schmidt et al. 1998) is still in the early stages and much remains to be learned.
One active membrane property expressed by vertebrate spinal cord neurons is the voltage-sensitive conductance associated withN-methyl-d-aspartate (NMDA) receptor activation (Mayer and Westbrook 1987). This property is associated with a region of negative slope conductance (RNSC) in the current-voltage (I-V) relationship of the cell (Flatman et al. 1983; MacDonald et al. 1982) and is due to a voltage-dependent blockade of the NMDA receptor channel by Mg2+ (Mayer et al. 1984; Nowak et al. 1984). Mammalian spinal cord neurons generate rhythmic voltage oscillations in the presence of NMDA and synaptic blockade (with tetrodotoxin, TTX; Hochman et al. 1994a,b; Kiehn et al. 1996; MacLean et al. 1997), as was demonstrated in the lamprey spinal cord (Wallen and Grillner 1987). It is also established that activation of NMDA receptors in the synaptically intact cord produces rhythmic motor activity in the neonatal rat (e.g., Beato et al. 1997; Cazalets et al. 1992; Kudo and Yamada 1987; Smith and Feldman 1987), as well as in other vertebrate preparations (e.g., Dale and Roberts 1984; Douglas et al. 1993; Fenaux et al. 1991; Grillner et al. 1981; Roberts et al. 1995; Wheatley et al. 1992). In combination, these observations favor an important role for NMDA receptor-mediated events in the generation of rhythmic network activity in the mammalian spinal cord.
We recently observed that, in the presence of serotonin (5-HT) receptor blockade, NMDA application induces neither voltage oscillations (in synaptically isolated motoneurons) nor locomotor network activity (MacLean et al. 1998). This interplay between 5-HT and NMDA is similar to that reported in amphibian spinal neurons (Reith and Sillar 1998; Sillar and Simmers 1994). The exact mechanism of the interaction is unknown. However, in the amphibian preparation it appears that 5-HT enhances voltage-dependent Mg2+ blockade of the NMDA ionophore (Scrymgeour-Wedderburn et al. 1997). Thus, the present study examined whether 5-HT modulates the NMDA receptor-mediated RNSC. Some of the following data has been presented previously in abstract form (MacLean and Schmidt 1998).
Experiments were performed on 17 Sprague-Dawley rats (aged 2–8 days). Techniques for isolation of the spinal cord have been described previously (e.g., Cowley and Schmidt 1995). In brief, animals were anesthetized with ether, decapitated, eviscerated, and placed in artificial cerebral spinal fluid (ASCF) at 4°C containing as follows (in mM): 128 NaCl, 3.0 KCl, 0.5 Na2H2PO4, 1.5 CaCl2, 1.0 MgSO4, 21 NaHCO3, and 30 glucose, equilibrated to pH 7.4 with 95% O2-5% CO2. MgSO4 was not included in the ASCF during some experiments. In one experiment the ACSF Ca2+concentration was increased by 1 mM to maintain the same total divalent ion concentration. In six experiments, the spinal cord was bilaterally intact from C1 to the cauda equina. In 11 experiments, the cord remained bilaterally intact from C1 to T13; caudal to T13 the right half of the lumbosacral spinal cord was removed. Similar intracellular behavior was observed regardless of the type of preparation used. The spinal cord was stabilized, using insect pins, on the bottom of a recording chamber coated with Sylgard (Dow Corning). All recordings were obtained at room temperature.
Whole-cell patch recordings of motoneurons were obtained as previously described (Hochman et al. 1994b). Recording pipettes contained the following (in mM): 140 K-gluconate, 11 EGTA, 35 KOH, 10 HEPES, and 1 CaCl2. Electrodes were made from borosilicate glass (WPI) pulled on a vertical puller (Narishige PP-83). Internal tip diameters ranged from 2 to 4 μm, and resistances measured in ACSF ranged from 3 to 5 MΩ. Cells were approached either through a pial patch made over the ventrolateral surface of the spinal cord or through the medial surface of the lumbar spinal cord in those preparations where the right half of the lumbosacral cord had been removed. Cells were patched using a “blind” approach (Blanton et al. 1989) and were identified as motoneurons by their antidromic response to ventral root stimulation. The recordings were obtained with an Axopatch 1D amplifier (Axon Instruments) filtered at 2 kHz. Series resistance (22 ± 22 MΩ) was monitored continuously and compensated. In voltage-clamp mode, series resistance was compensated up to 80%. In current-clamp mode, series resistance was compensated by adjusting the series resistance potentiometer such that the make-and-break points of voltage transients in response to current steps were balanced (i.e., bridge balance). The electrode-bath solution liquid junction potential (10 mV) was corrected in all recordings. Data were collected at 4 KHz and analyzed with the pCLAMP acquisition software (v6.0; Axon Instruments).
Input resistances and time constants were estimated using Clampfit software (pCLAMP v6.0, Axon Instruments). Input resistance and time constants were calculated from an average voltage response to a series of hyperpolarizing current steps. The tau was fit using a first-order exponential (Chebyshev) function. Current-voltage plots were generated by applying a series of voltage steps (500 ms duration), in 2.5 mV increments, from −130 to +20 mV. A holding potential of either −80 or −90 mV was used. The current plotted represents the average current measured during the final 100 ms of the voltage step. In some experiments, long duration (4 s) voltage ramps (from −130 to +40 mV) were used to determine steady-state I-V curves. In other experiments, long duration depolarizing current ramps (5 s), averaging 500 pA, were injected producing a peak depolarization of at least 20 mV. The current ramps were always preceded and followed by a 100-pA hyperpolarizing current pulse.
Neurochemicals were applied from concentrated stock solutions (10 mM) in 1–5 μM increments to a static bath (volume = 30 ml) that was continuously oxygenated and agitated. Concentrations in the following text refer to final bath concentrations of NMDA (3–20 μM) and 5-HT (5–60 μM). The final concentration of TTX was 1.5 μM in all applications (stock solution 100 μM). Recordings were obtained after a stable response to the applied neurochemical was obtained (usually this required 5–15 min).
Seventeen antidromically identified motoneurons were examined from L3 (n = 4), L4 (n = 4), and L5 (n = 9). The mean biophysical data obtained in the absence of applied neurochemicals were as follows: input resistance 179 ± 76 MΩ (range 81–303 MΩ), time constant 11 ± 5 ms, resting membrane potential −72 ± 6 mV.
Nonlinear membrane properties in the presence of NMDA
With all currents preserved, except the fast Na+ current blocked by TTX, NMDA alone (5–10 μM) induced a RNSC in the whole-cell I-V relationship of 11/17 motoneurons (Fig. 1); six motoneurons required 5-HT (40–50 μM) in addition to NMDA (10–20 μM), as described below. The mean potential at which the RNSC initiated was −63 ± 13 mV. The whole-cell mean maximal inward current in response to NMDA application was 200 ± 102 pA. The development of a RNSC depended on the presence of Mg2+ (1 mM), as expected (Mayer et al. 1984; Nowak et al. 1984). Thus the RNSC was abolished by removal of Mg2+ (n = 10), which enabled a persistently enhanced inward current, via NMDA receptor channels, even at relatively hyperpolarized membrane potentials (Fig. 5 C). The RNSC was also abolished by bath application of the dl-2-amino-5-phosphonovaleric acid (AP5, 10 μM, n = 3; Fig. 1), which specifically blocks NMDA receptors (Davies et al. 1981).
Current-clamp recordings of the response to depolarizing ramp current injection in the presence of NMDA (5–10 μM) demonstrated a nonlinear jump in membrane voltage (n = 10, Fig.2 A1). Eight of these motoneurons were capable of developing TTX-resistant voltage oscillations or rhythmic plateau potentials (Fig. 2 A2). Only one of these cells required constant current injection to elicit the oscillations. Removal of Mg2+ from the ACSF abolished the nonlinear voltage jump during ramp current injection in all 10 cells (Fig. 2 B1), as well as oscillations and plateau potentials (n = 4, Fig. 2 B2). All motoneurons examined in this study for the presence of voltage oscillations in TTX were subjected to a range of holding potentials between −70 and −50 mV.
5-HT modulates the NMDA receptor-mediated RNSC
With all currents preserved, except the fast Na+ channels blocked by TTX, 5-HT (50–60 μM) alone enhanced net inward current at potentials more depolarized than −80 mV (Fig. 3 A1) in all five motoneurons examined (1 cell was recorded in the absence of TTX, using instead a QX-314 filled electrode to block cell firing), but failed to elicit a RNSC. Addition of NMDA (5–10 μM) was required to produce the RNSC (Fig. 3 A1, trace c). However 6/17 motoneurons developed neither a RNSC in their I-Vrelationship (Fig. 3 A2) nor a nonlinear voltage response to ramp current injection (Fig. 3 B1) after bath application of NMDA (10–20 μM) alone. Subsequent application of 5-HT (40–50 μM) was necessary to induce a RNSC in the whole-cell current (Fig.3 A2), or nonlinear voltage response to ramp current injection (Fig. 3 B2). Induction of the RNSC after application of 5-HT was observed in conjunction with the development of membrane voltage oscillations, recorded in current-clamp mode (Fig.3 B2).
5-HT application (30–50 μM) to motoneurons that initially displayed an RNSC in the presence of NMDA alone (n = 8 cells) shifted the onset of the RNSC significantly leftward by 18.3 ± 15.0 mV (P < 0.001, Fig.4 A1). This leftward shift was also evident in the current response to depolarizing voltage ramps (Fig. 4 A2). 5-HT significantly increased the maximal inward current associated with the RNSC by an average of 107 ± 85 pA (P < 0.005). The mean threshold for activation of the RNSC during co-application of 5-HT and NMDA was −79.5 ± 16.2 mV. Thus the mean RNSC threshold level was more negative than the mean resting membrane potential of TTX-treated motoneurons in the absence of applied neurochemicals (−72.5 ± 6.1 mV, n = 17). The negative shift of the RNSC and the increase of the maximal negative slope current by 5-HT depended on the concentration of 5-HT in the bath (Fig. 4 B1), as well as the time elapsed after 5-HT application (Fig. 4 B2). The effect of 5-HT was partly reversed by application of the 5-HT receptor antagonist mianserin (80 μM, n = 4), as shown in Fig. 4 B3.
5-HT may regulate the voltage-dependent blockade of the NMDA channel
Decreasing the Mg2+ concentration in the bath solution was associated with a shift of the RNSC to more hyperpolarized potentials, similar to the effect of 5-HT on the RNSC (n = 3, Fig.5 A). The RNSC was ultimately completely abolished in Mg2+-free ACSF (Fig.5 C, bottom trace), requiring approximately 15 min to develop. During the Mg2+ washout period, serialI-V plots were obtained (Fig. 5 C). The RNSC shifted increasingly leftward toward more hyperpolarized potentials. This shift also occurred despite adding an extra 1 mM Ca2+ to the bath to maintain divalent cation charge balance (n = 1). The reversal of the 5-HT-induced leftward shift of the RNSC, observed after addition of mianserin, was itself reversed after subsequent washout of Mg2+ (Fig. 5, B and C). Thus the data suggest that 5-HT receptors may modulate the RNSC by regulating the voltage-sensitive Mg2+-dependent blockade of NMDA receptors.
The main finding of this study is that NMDA receptor channel nonlinear voltage sensitivity is modulated by 5-HT.
NMDA receptor activation has a prominent role in the production of vertebrate locomotor rhythms (Beato et al. 1997;Cazalets et al. 1992; Dale and Roberts 1984; Douglas et al. 1993; Fenaux et al. 1991; Grillner et al. 1981; Guertin and Hounsgaard 1998; Hernandez et al. 1991;Kudo and Yamada 1987; Smith and Feldman 1987; Smith et al. 1988; Wheatley et al. 1992), including transmission of network excitatory drive during locomotion (Brownstone et al. 1994;Cazalets et al. 1996; Hochman and Schmidt 1998; Moore et al. 1987). NMDA receptors also mediate TTX-resistant voltage oscillations in spinal neurons (Hochman et al. 1994a,b; MacLean et al. 1997; Prime et al. 1999; Sillar and Simmers 1994; Wallen and Grillner 1987). Thus, given the appropriate neurochemical milieu, some neurons are likely endowed with the capacity to develop voltage oscillations, or at least certain active membrane properties that are well-suited to the needs of a rhythmogenic network. However, the phasic discharge induced by exogenous application of NMDA alone is often nonlocomotor-like in pattern (Cowley and Schmidt 1994). Thus activation of additional receptor systems (such as 5-HT receptors) are important for the promotion of a stable locomotor-like pattern of network activation. If NMDA receptor-mediated voltage nonlinearity is important for the production of rhythmic activity, its enhancement by 5-HT may explain, at least in part, why co-application of 5-HT and NMDA is more effective in producing rhythmic locomotor activity than application of either neurochemical alone (Cowley and Schmidt 1994;Kjaerulff et al. 1994; Sqalli-Houssaini et al. 1993).
In the present study, 35% of the motoneurons exposed to NMDA alone failed to develop a RNSC. Although the exact reason for this observation is unknown, it seems clear that 5-HT application to such cells promotes the expression of an NMDA-dependent RNSC. It should be noted that in the present series no currents, other than the fast Na+ current, were blocked. Among other motoneurons, which did display a RNSC in response to NMDA alone, 5-HT shifted the RNSC leftward in the hyperpolarizing direction, similar to the effect on the RNSC of decreasing the concentration of Mg2+. Therefore the data suggest that the influence of 5-HT on the RNSC may be due to a decrease in the efficacy of Mg2+ blockade of NMDA-gated channels.
Previous studies support the possibility that 5-HT facilitation of NMDA currents may be due to reduced Mg2+ blockade. Protein kinase C (PKC) modulates NMDA currents (Ben-Ari et al. 1992; Blank et al. 1996). More specifically,Chen and Huang (1992) showed that PKC potentiates NMDA-activated currents and produces a negative shift of the RNSC by decreasing Mg2+ blockade of the NMDA channel. This potentiation of NMDA currents was greatest (60–80%) in the range of −60 to −80 mV; smaller amounts of facilitation occurred at more depolarized (e.g., 23% increase at −20 mV) and hyperpolarized (e.g., 28% increase at −100 mV) membrane potentials (Chen and Huang 1992). Some serotonergic actions (5-HT2in particular) are mediated through the PKC pathway (e.g.,Martin and Humphrey 1994). Indeed, serotonin has been shown to directly potentiate NMDA currents by a PKC-dependent mechanism (Blank et al. 1996). We observed that the 5-HT2 receptor antagonist mianserin reversed the effect of 5-HT on the RNSC. However, the mianserin reversal was only partial, allowing for the possibility that other 5-HT receptor subtypes may be involved. Moreover, it is possible that 5-HT may modulate NMDA responses via the PKC pathway independent of any effect on Mg2+ blockade, as is the case for substanceP modulation of NMDA responses in lamprey spinal cord neurons (Parker et al. 1998).
In addition to classical 5-HT receptor actions, and analogous to the effect of Mg2+ ions, 5-HT has been shown to directly block NMDA-gated cationic channels in cultured embryonic rat spinal neurons (Chesnoy-Marchais and Barthe 1996). This voltage-dependent effect is most prominent at relatively hyperpolarized holding potentials (−60 to −100 mV) and in the absence of Mg2+ions. It is hypothesized that 5-HT competes with Mg2+ ions in the open channel (Chesnoy-Marchais and Barthe 1996). Therefore, the possible contribution of a direct 5-HT-mediated blockade of NMDA-gated channels cannot be excluded in the present series. However, if this type of voltage-dependent 5-HT influence is present, it is not evident in the highly linear whole-cell I-V relationship plotted during NMDA receptor activation in the absence of Mg2+ ions (Fig. 5 C, bottom trace).
If 5-HT does influence Mg2+ blockade of the NMDA ionophore, this mechanism is unable to fully suppress Mg2+ blockade. That is, complete abolishment of the RNSC, similar to that recorded after 15 min of Mg2+ washout, was never observed during application of 5-HT, even after using relatively high concentrations of 5-HT and observing for prolonged periods (up to 90 min). Thus the 5-HT effect under these conditions appears truly modulatory in nature.
Because the mean voltage threshold for activating the RNSC is shifted to more hyperpolarized values relative to mean resting membrane potential, 5-HT appears to promote inward currents through the NMDA ionophore in neurons that would otherwise develop only weak currents near resting potential. 5-HT enhancement of the RNSC also facilitated the expression of voltage oscillations in the presence of TTX. Conversely, 5-HT receptor blockade, which was shown to block locomotor network activity and NMDA-induced oscillations (MacLean et al. 1998), shifted the RNSC to more depolarized levels. This shift would decrease conductance through the NMDA ionophore and may contribute to the observation that 5-HT receptor antagonists mimic the effect of AP5 in abolishing rhythmic activity and TTX-resistant oscillations (MacLean et al. 1998).
NMDA receptor-mediated voltage oscillations elicited in embryonic and larval Xenopus spinal cord neurons have been shown to display 5-HT dependency (Sillar and Simmers 1994), although NMDA alone can produce voltage oscillations in this preparation, the expression of which is facilitated by 5-HT (Prime et al. 1999). Several differences are noted comparing the Xenopus and neonatal rat. In theXenopus, 5-HT1A-like rather than 5HT2 receptors are implicated and 5-HT enhances rather than diminishes the voltage-dependent blockade of NMDA channels by Mg2+ (Scrymgeour-Wedderburn et al. 1997). In Xenopus spinal neurons, NMDA produces tonic depolarization. Subsequent application of 5-HT results in superimposed rhythmic hyperpolarizing potentials, consistent with enhanced Mg2+ blockade of the NMDA channel (Sillar and Simmers 1994). In contrast, application of 5-HT to the neonatal rat spinal cord facilitates the development of depolarizing oscillations (see also MacLean et al. 1998), compatible with a reduction in the Mg2+ blockade of NMDA ionophores. It appears that the same neuromodulator produces opposite actions yet achieves a similar behavior (oscillations) in the two species. Both excessive Mg2+ blockade as well as insufficient blockade (e.g., total removal of Mg2+ ions) of NMDA ionophores may inhibit the optimal expression of the RNSC. Therefore, different modulatory actions of 5-HT may be required by different systems.
Although not specifically examined in the present series, 5-HT is known to increase or decrease several other membrane currents in vertebrate preparations. For instance, recordings of various brain stem and spinal motoneurons have shown that 5-HT enhancesI h, low- and high-voltage-activated Ca2+ currents, and persistent Na+ currents and reduces leak K+ and calcium-dependent K+currents (e.g., Berger and Takahashi 1990; Hsiao et al. 1997, 1998; Larkman and Kelly 1992;Takahashi and Berger 1990; Wallen et al. 1989). Some of these currents, such asI h, high-voltage-activated Ca2+, and persistent Na+currents, may also contribute to the production of rhythmic activity in mammalian networks, as suggested by Hsiao et al. (1998)and Bertrand and Cazalets (1998). Thus, although 5-HT alone failed to produce a RNSC in this series (in contrast to guinea pig trigeminal motoneurons, Hsiao et al. 1998) and the RNSC was Mg2+-dependent, other non-NMDA receptor-mediated currents were presumably activated by 5-HT since no attempt was made to block them. Some of these currents, if sensitive to 5-HT at membrane potentials overlapping with the RNSC, could have a synergistic effect on total inward current in this part of theI-V plot and contribute to RNSC enhancement independent of any action on Mg2+ blockade. Indeed, we observed that 5-HT applied alone did enhance net inward current at all potentials depolarized relative to −80 mV (Fig. 3 A1). However, 5-HT application decreased the slope of the I-Vplot in some (e.g., Fig. 3 A), but not all (e.g., Fig.3 B), motoneurons, indicating an overall increase in cell input resistance. In these neurons, a 5-HT-mediated decrease in outward current must have been greater than any increase of inward currents. It is possible that the resulting increased input resistance in these cells favorably influenced space clamp conditions and thereby facilitated the detection of NMDA currents by the patch electrode.
Although the present data suggest one mechanism through which 5-HT may influence NMDA currents in the rat spinal cord, these experiments do not define which specific locomotor network elements might possess this property. The precise identity of mammalian locomotor network-related interneurons is largely unknown. Therefore, we limited our whole-cell recordings to a functionally identifiable group of cells (i.e., motoneurons). It is quite probable that some voltage-sensitive events characterized in motoneurons also exist in locomotor network-related interneurons. Although motoneurons are last-order elements of the network, the ability to modulate their nonlinear membrane properties offers an important mechanism for shaping rhythmic output. For example, 5-HT receptor-mediated facilitation of active membrane currents may limit the effects of temporal dispersion among phasic synaptic input received by motoneurons, thereby decreasing the performance requirements of premotor network elements. In future studies, it will be important to determine the role of 5-HT-NMDA receptor interactions and other nonlinear conductances in locomotor network-related interneurons.
The authors thank Drs. R. Brownstone, S. Hochman, L. Jordan, S. Shefchyk, and K. Sillar for helpful comments.
This study was supported by the Manitoba Medical Services Foundation and the Canadian Institutes of Health Research. J. N. MacLean was supported by the Rick Hansen Man-in-Motion Legacy Fund.
Address for reprint requests: B. J. Schmidt, Dept. of Physiology, University of Manitoba, 730 William Ave., Winnipeg, Manitoba R3E 3J7, Canada.
- Copyright © 2001 The American Physiological Society