Development and Serotonergic Modulation of NMDA Bursting in Rat Trigeminal Motoneurons

Chie-Fang Hsiao, Nanping Wu, Michael S. Levine, Scott H. Chandler


The development ofN-methyl-d-aspartate (NMDA)-induced burst discharge in rat trigeminal motoneurons (TMNs) between postnatal days P1 and P10 was examined using whole cell patch-clamp recording methods in brain slices. Bath application of NMDA (50 μM) induced a Mg2+-dependent rhythmical bursting activity starting around P8. Prior to the onset of bursting, the membrane potential depolarized and the input resistance increased. Hyperpolarization of the membrane potential with extrinsic current demonstrated a narrow window of membrane potential where maintained rhythmical burst discharge was evident. In P1–P4 neurons, NMDA application produced membrane depolarization and a minimal change in input resistance, but no burst activity at any membrane potential. Voltage-clamp analysis indicated that the bursting activity was related to the presence or absence of a voltage-dependent Mg2+ block and induction of a negative slope conductance (NSC) region in theI NMDA-V relationship. Regardless of age, reduction of extracellular Mg2+ from 1 mM to 30 μM enhancedI NMDA at voltages negative to −60 mV. However, in 1 mM Mg2+, P1–P4 neurons were devoid of a prominent NSC region compared with P8–P10 neurons, suggesting that the efficacy of depolarization in unblocking the NMDA receptors increased with age. NMDA bursting was not dependent on calcium influx through voltage-gated calcium channels (VGCC) but did require a minimal concentration of Ca2+ in the bath. Intracellular bis-(o-aminophenoxy)-N,N,N′,N′-tetraacetic acid application suppressed burst discharge completely, suggesting that intracellular Ca2+ directly, or via second-messenger systems, regulates NMDA receptor activity and bursting. Interestingly, NMDA bursting could be induced in P1–P4 neurons by simultaneous bath application of serotonin (5-HT, 10 μM), which by itself did not produce bursting, suggesting an “enabling” role for 5-HT. Voltage-clamp analysis demonstrated that the NMDA/5-HT bursting resulted from induction of an NSC in the I-Vrelationship of total membrane current. 5-HT by itself produced no such effect. The mechanisms for this effect were due to an enhancement of the NSC region of theI NMDA-V relationship and reduction of a presumed leak current by 5-HT. These data indicate that NMDA bursting in trigeminal motoneurons is developmentally regulated and subject to neuromessenger modulation. Control of the Mg2+ sensitivity of the NMDA receptor and voltage-dependent block by neuromessengers could be an effective means to control the efficacy of glutamatergic synaptic drive to motoneurons during rhythmical oral-motor activity at early postnatal ages.


Excitatory amino acid neurotransmission, and in particular,N-methyl-d-aspartate (NMDA) receptor activation, are critical for the production and maintenance of oral-motor activity and control of trigeminal motoneuronal output both in vivo (Chandler 1989; Katakura and Chandler 1990) and in vitro (Appenteng et al. 1995;Kim and Chandler 1995; Kogo et al. 1996;Trueblood et al. 1996). For instance, direct application of NMDA antagonists to guinea pig trigeminal motoneurons during cortically induced rhythmical jaw movements suppresses rhythmical motoneuronal discharge (Katakura and Chandler 1990). Furthermore, using the isolated neonatal brain stem preparation, which preserves the circuitry for rhythmical oral-motor activity, bath application of NMDA receptor antagonists blocks the ability to induce rhythmical oral-motor activity in response to bath application of the excitatory amino acid agonist AMPA (Kogo et al. 1996), indicating a critical role for NMDA receptors in rhythmical oral-motor activity.

NMDA receptors play important roles in diverse functions such as neuronal development, learning and memory, cell death, and motor control (Bliss and Collingridge 1993; Mori and Mishina 1995; Schmidt et al. 1998). The importance of NMDA receptors in these behaviors most likely results from the NMDA receptor channels high permeability to Ca2+ and voltage-dependent block by external Mg2+ ions (Mayer and Westbrook 1987). With respect to rhythmical motor behavior, the Mg2+ block in combination with activation of repolarizing conductances endows the neuron with bistable properties (the ability to switch between 2 stable membrane potentials) and, therefore the capability of generating maintained bursting behavior (Hochman et al. 1994; Kim and Chandler 1995; Schmidt et al. 1998; Sigvardt et al. 1985; Tell and Jean 1993). Additionally, Ca2+ entry through voltage-gated L-type Ca2+ channels is critical for NMDA oscillations in adult turtle spinal motoneurons (Guertin and Hounsgaard 1998) and may serve a similar role in other vertebrate species as well.

Rhythmical oral-motor activity such as sucking and chewing is thought to be under control of a network of brain stem neurons, typically referred to as a central pattern generator (CPG) (reviewed inGoldberg and Chandler 1990; Lund et al. 1998). However, the fact that trigeminal motoneurons exhibit nonlinear intrinsic membrane characteristics imparted by the presence of NMDA receptors (Kim and Chandler 1995) and that these receptors are functional on trigeminal motoneurons during oral-motor behaviors (Katakura and Chandler 1990), suggest that the final rhythmical motoneuronal output is determined by a complex interaction between rhythmically occurring trigeminal premotoneurons and NMDA receptor-mediated intrinsic motoneuronal membrane oscillations (Kim and Chandler 1995). Similar views for the contribution of NMDA receptors to rhythmical network function in rat, lamprey, and tadpole locomotion have been put forward (Guertin and Hounsgaard 1998; Hochman et al. 1994; Schmidt et al. 1998;Scrymgeour-Wedderburn et al. 1997; Wallen and Grillner 1987). Presently, there is a paucity of information on the development of the circuitry underlying oral-motor rhythmical activity and, in particular, the properties of the trigeminal motoneurons and NMDA receptors during the transition from rhythmical sucking behavior to adult-like mastication (Turman et al. 1999), a behavior that begins around P12 in the rat (Westneat and Hall 1992). This information is essential, at a minimum, for a complete understanding of the cellular factors controlling oral-motor behaviors.

NMDA receptors undergo changes in the expression of NR2A and NR2B subunits during development (Kirson et al. 1999;Monyer et al. 1994; Watanabe et al. 1993). These changes in the receptor subunits are thought to control the sensitivity of the receptor channel to Mg2+ block (Kuner and Schoepfer 1996), and hence the magnitude of the negative slope conductance region (NSC) of theI NMDA-V relationship that underlies the ability of NMDA to induce oscillatory phenomena (Kim and Chandler 1995; Tell and Jean 1993; among others). In some systems, development of the Mg2+ sensitivity of the NMDA receptor channel and the voltage dependence of the Mg2+ block are developmentally regulated, increasing with age (Ben-Ari et al. 1988; Burgard and Hablitz 1994; Kirson et al. 1999; Nabekura et al. 1994) and therefore engendering the neuron with enhanced capability for bursting behavior. Because sucking at birth requires rhythmically active trigeminal motoneurons, it is possibility that these NMDA receptor subunits are configured early on to support a Mg2+ block, NSC, and bursting (Tell and Jean 1993).

To address these possibilities, we examined the development of NMDA bursting in trigeminal motoneurons and their modulation by serotonin. This neuromessenger previously was shown to be critical for suckling behavior in neonatal rat pups (Ristine and Spear 1984) as well as for facilitating excitatory amino acid-mediated synaptic transmission to, and intrinsic membrane properties of, adolescent guinea pig trigeminal motoneurons (Hsiao et al. 1997;Katakura and Chandler 1990; Trueblood et al. 1996). In this study, we present evidence that in neonates NMDA-induced bursting is not present from day P1–P4 because of insufficient voltage dependence of the Mg2+ block of the NMDA receptor channel. However, NMDA bursting and induction of an NSC region in the steady-state total membrane current-voltage relationship (I tot-V) can be induced in that age group when in the presence of serotonin.


Preparation for whole cell recording

Whole cell patch-clamp experiments were performed on transverse slices of a neonatal rat brain stem (1–10 days). Rats were anesthetized by halothane inhalation (Halocarbon Laboratories, River Edge, NJ). The brain was removed and placed in oxygenated ice-cold cutting solution. Coronal sections (250–300 μM) through the trigeminal motor nucleus were obtained and then transferred into a holding chamber in incubation solution (see following text) at 37°C for 40 min.


Solutions were bubbled with 95% O2-5% CO2 and maintained at pH 7.25–7.30 (22–24°C). Cutting solution was composed of the following (in mM): 126 NaCl, 3 KCl, 1.25 NaH2PO4, 26 NaHCO3, 10 glucose, 1 CaCl2, 5 MgCl2, and 4 lactic acid. The recording solution consisted of the following (in mM): 124 NaCl, 3 KCl, 1.25 NaH2PO4, 26 NaHCO3, 10 glucose, 2 CaCl2, and 2 MgCl2.Incubation solution was identical to recording solution except for the addition of 4 mM lactic acid. To isolate NMDA currents during voltage-clamp experiments, the bathing solution contained tetrodotoxin (TTX) [0.5 μM; Sigma (St. Louis, MO)] to block Na+ currents; CdCl2 (100 μM; Sigma) to block voltage-gated Ca2+currents; 6-cyano-7-nitroquinoxaline-2,3-dione disodium (CNQX; 10 μM; RBI, Natick, MA) to block non-NMDA receptors; bicuculline methiodide (10 μM; Sigma) and strychnine hydrochloride (5 μM; Sigma) to block GABAA and glycine Cl-mediated currents, respectively; and glycine (5 μM; Sigma) to saturate the glycine-binding sites on NMDA receptors.

Normal pipette solution contained (in mM) 115 K-gluconate, 9 NaCl, 25 KCl, 1 MgCl2, 10 HEPES buffer, 0.2 EGTA, 3 K2-ATP, and 1 Na-GTP, with a pH of 7.25–7.30 and osmolarity of 280–290 mM. To isolate NMDA currents, we used a modified pipette solution containing (in mM) 110 CsF, 20 TEA, 10 HEPES buffer, 2 MgCl2, 1 CaCl2, and 10 BAPTA, pH of 7.25–7.30 (osmolarity, 280–290 mM). To investigate the modulation of NMDA-induced currents by 5-HT, we used a modified pipette solution containing (in mM) 108 CsMeSO3, 12 CsCl, 10 HEPES buffer, 0.2 EGTA, 20 TEA, 1 MgCl2, 2 K2-ATP, and 0.3 GTP-Tris salt. Lucifer yellow (0.1%, Sigma) was added to small volumes of electrode solution for fluorescent viewing in initial experiments.

Drug application

Drugs were bath applied at the following concentrations: NMDA (50 μM; RBI), d, l-2-amino-5-phosphonovaleric acid (APV; 10 μM; Sigma), and nimodipine (10 μM; RBI).

Identification of the trigeminal motor nucleus

The trigeminal motor nuclei were identified bilaterally in the coronal slice under low magnification (×5) as an opaque, pale oval region medial to the trigeminal principal sensory nucleus and ventral lateral to the periaqueductal gray and central canal (Chandler et al. 1994). In early experiments, retrograde labeling of TMNs from microinjections of Texas red (10%, Molecular Probes, Eugene, OR) into the masseter or mylohyoid jaw muscles was used to confirm that the area recorded was the trigeminal motor nucleus.

Whole cell recording

Slices were perfused with oxygenated recording solution (3 ml/min) while secured in a recording well mounted on a Zeiss fixed stage Axioskop microscope equipped with bright-field, fluorescence, and Nomarski optics in combination with infrared video microscopy for enhanced resolution of individual neurons. Patch recordings were obtained with the use of an Axopatch 1D (Axon Instruments, Foster City, CA) in concert with pCLAMP acquisition software (Version 8.0.2, Axon Instruments) operating on a Pentium-based personal computer. Signals were digitized on-line and filtered at 2 kHz (voltage clamp) or 5 kHz (current clamp). Patch pipettes were fabricated from conventional thin-wall glass (1.5 mm OD, 0.86 mm ID; Warner Instrument, Hamden, CT), pulled on a Brown/Flaming P-97 micropipette puller (Sutter Instruments, Novato, CA) and had bath resistances of 3–5 MΩ. Signals were grounded by a 3 M KCl agar bridge electrode (Ag/AgCl wire) mounted in the recording well. Liquid junction potentials were measured between the pipette and bath solutions and varied between 9 mV (normal pipette solution) and 7 mV (modified pipette solutions) and were corrected off-line. Cell capacitance (C M) for each trigeminal motoneuron recorded in voltage clamp was determined from the integral of capacity current in response to 15-ms hyperpolarizing voltage commands or directly obtained by the pClamp software during the experiment. Uncompensated series resistance (R s) was calculated from the decay time constant (tau) of the transient and was usually less than 20 MΩ. Sixty to 80% compensation was routinely employed.

Data analysis

Current- and voltage-clamp data were analyzed in Clampfit 8.0 (Axon Instruments). NMDA-induced bursting characteristics, such as time of burst onset and termination, intraburst spike frequency, cycle time, and burst duration were measured in Datapac III 1.61 (Run Technologies, Irvine, CA). Burst duration was defined as the mean time from burst onset to termination in three or more consecutive bursts. Burst cycle time was defined as the mean time from burst onset to the next onset for three or more consecutive cycles. Postburst after hyperpolarizing potential (AHP) was measured as the difference between the voltage level at the termination of the last spike in the burst and the peak of the trough immediately following a burst.

Isolation of NMDA currents

The NMDA current (I NMDA) was obtained in selected experiments by digital subtraction of the current traces obtained by applying voltage ramps in the absence (control) and presence of bath applied NMDA in 30 μM or 1 mM external Mg2+ using the dedicated solution described in the preceding text. Averages from at least three consecutive responses to ramps were obtained. TheI NMDA-V relation for each cell was then obtained for each of those Mg2+concentrations. The protocol is shown in Fig. 3. Briefly, the patched neuron was held at −80 mV then stepped up to +30 mV for 1.5 s to inactivate any transient voltage-gated currents. Subsequently, a ramp voltage command (0.13 mV/ms) was brought to −100 mV then returned to −80 mV. The protocol was initiated in control ACSF and then again after obtaining a steady-state response to bath application of NMDA (50 μM). The currents recorded in control were subtracted from those recorded in the presence of NMDA to obtain theI NMDA. The subtracted currents for each Mg2+ concentration were then normalized to the maximal outward current at +30 mV. TheI NMDA-V relationship was then obtained and reduced to 41 membrane potential points and plotted (Fig. 3).

Sigmaplot 4.0 (Jandel Scientific, San Rafael, CA), Excel (Microsoft, Redmond, WA), and StatView 5.0 (SAS Institute, Cary, NC) were used for additional analyses. Values are expressed as means ± SE. Significant differences were tested with Student's paired and unpairedt-tests. A significance level of P ≤ 0.05 was used in all tests unless otherwise stated.


The data presented in this paper are based on patch-clamp recordings from 203 rat trigeminal motoneurons (TMNs) ages P1–P10 obtained from brain stem slice preparations. The criteria for inclusion into the database were resting potential greater than −50 mV, action potential amplitude exceeding 80 mV and input resistance at least 100 MΩ. Prior to NMDA application, all neurons exhibited continuous spike discharge when artificially depolarized above spike threshold and were silent at potentials below threshold.

NMDA-induced burst discharge

Previously, we showed that bath application of NMDA produced rhythmical membrane potential fluctuations and bursting in adolescent guinea pig trigeminal motoneurons (Kim and Chandler 1995). To determine if such conditional bursting is present in neonatal rat TMNs, the effects of NMDA application were examined. A typical example of NMDA application is shown in Fig.1 taken from a P9 neuron. As shown, bath application depolarized the neuron and increased input resistance from 104 to 130 MΩ prior to spike onset (Fig. 1 A1). Once threshold was reached (−50 mV), continuous spiking occurred (Fig.1 A2) that could be transformed into bursting behavior by membrane hyperpolarization with extrinsic current application (Fig.1 A3). In this example, the onset of burst discharge started around −63 mV. This behavior was observed in 67/108 neurons examined (62%). Extrinsic polarization of the membrane potential in the absence of NMDA never produced bursting. In seven P8–P10 neurons, the precise range of membrane potentials where bursting discharge was initiated and terminated was measured and was between −64.1 ± 2.1 and −76.9 ± 2.8 mV. In the presence of TTX, the rhythmical bursting behavior was replaced by rhythmically occurring plateau potentials (Fig. 2 A, 1 and2) that were blocked by the specific NMDA antagonist APV (10 μM; Fig. 2 A3). These data indicate that the bursting behavior in the presence of NMDA resulted from specific activation of NMDA receptors and is not dependent on circuit activity as previously demonstrated in guinea pig trigeminal motoneurons (Kim and Chandler 1995), and other types of motoneurons (e.g.,Hochman et al. 1994; Scrymgeour-Wedderburn et al. 1997; Wallen and Grillner 1987).

Fig. 1.

Membrane potential responses toN-methyl-d-aspartate (NMDA) application from P2 and P9 trigeminal motoneurons. A: depolarization and bursting induced by NMDA (50 μM) in P9 neuron. Numbers indicate expanded segment of voltage trace (top). Current injection indicated by traces below voltage traces. Current pulses were induced to monitor changes in input resistance. B: same as above for P2 trigeminal motoneuron (TMN). Note the membrane depolarization but lack of bursting in response to hyperpolarizing current injection. Membrane potential prior to NMDA application is indicated at the top left side of the voltage trace. The threshold for spike onset was −53 mV and theR inp changed from 110 to 171 MΩ after NMDA application. Solutions used contained 6-cyano-7-nitroquinoxalene-2,3-dione (CNQX), strychnine and bicuculline unless otherwise stated (see methods) and applies for all figures.

Fig. 2.

Characteristics of NMDA bursting. A1: NMDA-induced bursting. A, 2 and 3: effects of tetrodotoxin (TTX; 2) and 2-amino-5-phosphonovaleric acid (APV; 3) on NMDA bursting. B1: NMDA bursting in control Mg2+ (B1) and in reduced Mg2+ (30 μM, B2). B3: recovery. Continuous spike discharge was always observed in nominal Mg2+ solutions.

Development of NMDA burst discharges

To determine if NMDA-induced bursting is developmentally regulated, the effects of bath application of NMDA on P1–P10 TMNs were examined. In 21 neurons examined prior to P8, only 2 neurons showed bursting activity in response to NMDA application. However, in nonbursting neurons, NMDA produced membrane depolarization and continuous spike discharge, as shown in Fig. 1 B1 for a P2 neuron. In contrast to that observed in neurons older than P7, alterations of the membrane potential within the same potential range did not produce bursting (Fig. 1 B, 2 and 3). In P1–P4 neurons, which rarely showed bursting, the mean membrane depolarization in response to 50 μM NMDA application prior to spike threshold was 14.5 ± 1.3 mV, n = 15; while that for the P8–P10 group, which usually showed bursting, was 15.0 ± 0.6 mV, n = 45. Table 1shows a summary of the effects of NMDA application on membrane potential and R inp for P1–P4 and P8–P10 groups. The data indicate that the lack of bursting in P1–P4 neurons must not reside in the lack of functional NMDA receptors but rather to a change in the underlying ionic mechanism and/or membrane properties responsible for NMDA bursting.

View this table:
Table 1.

Effects of NMDA and 5-HT on passive membrane properties of TMNs

Development of NMDA receptor Mg2+ sensitivity

NMDA-induced bursting is critically dependent on the presence of a region of negative slope conductance (NSC) in the steady-state total membrane current I tot-Vrelationship (Kim and Chandler 1995; Schmidt et al. 1998). In guinea pig TMNs in response to NMDA, bursting behavior results from the voltage-dependent Mg2+block and unblock of the receptor iontophore in combination with repolarizing outward currents (Kim and Chandler 1995). In nominal Mg2+ solutions, which remove the NSC in the steady-state I tot-Vrelationship, bursting in guinea TMNs is not possible (Kim and Chandler 1995). An example of the dependence on extracellular Mg2+ for NMDA-induced bursting in rat TMNs is shown in Fig. 2 B, 1–3. Reduction of the extracellular Mg2+ from 1 mM to 30 μM transformed the bursting pattern into continuous spike discharge that was easily reversed on washout.

To determine if the development of NMDA bursting is dependent on changes in the affinity of NMDA receptors for Mg2+ and development of a voltage-dependent Mg2+ block, in voltage clamp we isolated the NMDA current and examined the effects of two different Mg2+ concentrations on the underlying normalizedI NMDA-V (seemethods) relationship in P2–P4 and P8–P10 groups. Composite normalizedI NMDA-V relationships for these two groups are shown in Fig. 3,A and B. Although the reversal potentials for both groups were similar (approximately 0 mV), it is apparent that in the young group NMDA receptors have already developed a sensitivity to extracellular Mg2+; at voltages negative to around −25 mV, increasing Mg2+ reduced theI NMDA. However, the voltage dependence of the Mg2+ block was less apparent in the young compared with the older group, indicating that the efficacy of depolarization for relieving the Mg2+ block was less in the young group. Thus in the P2–P4 group, depolarization from −100 to −50 mV produced a small increase in normalizedI NMDA in 1 mM Mg2+ compared with the P8–10 group. These developmental changes in the voltage-dependent Mg2+ block were quantified as the ratio of peak I NMDA toI NMDA at −100 mV and plotted as histograms in Fig. 3 C for two different Mg2+ concentrations for each age group. As shown, in 1 mM Mg2+, the ratio is about three times larger in the P8–P10 group compared with that for the P2–P4 group, suggesting that the lack of a prominent NSC in the steady-stateI NMDA-V relationship in normal extracellular Mg2+ concentrations most likely accounts for the lack of NMDA bursting observed in the younger age group.

Fig. 3.

Normalized composite steady-stateI NMDA-voltage relationships in different Mg2+ solutions for P2–P4 and P8–P10 TMNs.A: normalized compositeI NMDA-V relationship for P2–P4 group for 2 different Mg2+ concentrations (seemethods). Inset: voltage command protocol.V indicates command potential. Arrows indicate current at V = −100 mV and peak current for 30 μM Mg2+. B: same as A except for P8–P10 group. Standard error bars were omitted when their size was equal to or smaller than symbols. C: histogram showing ratio of peak normalized current (I peak) to current measured at −100 mV (I −100). Statistical comparisons were performed between age groups for each Mg2+ concentration. A ratio of 1 would indicate the lack of an NSC region.

Role of calcium ions in the generation of NMDA bursting

In turtle spinal motoneurons, rhythmical NMDA bursting depends on L-type calcium channel activation (Guertin and Hounsgaard 1998). To determine the role of calcium entry for initiation and maintenance of NMDA-induced rhythmical bursting in rat TMNs, we performed a series of experiments that either blocked voltage-gated calcium channels (VGCC), reduced extracellular calcium, or maintained the intracellular [Ca i2+ ] to low levels by chelation with 10 mM BAPTA during NMDA application.

To investigate the possibility of L-type Ca2+channel participation in NMDA bursting, we first examined the effects of nimodipine (10 μM), an L-type calcium channel blocker, on NMDA bursting (Fig. 4 A, 1 and2). Although nimodipine altered burst duration and intraburst spike frequency (Table 2), maintained bursting was still present. In fact, when Cd2+, a general VGCC blocker was used, NMDA bursting was still evident in spite of the change in burst characteristics (Fig. 4 B, 1 and 2, and Table 2). These data suggest that VGCC contribute to shaping NMDA burst characteristics but are not necessary for initiation and maintenance of NMDA-induced rhythmical burst discharge.

Fig. 4.

High-threshold Ca2+ channels are not required for NMDA bursting. A: effects of an L-type Ca2+channel antagonist (nimodipine, 10 μM) on bursting. B: effects of the general Ca2+ channel antagonist Cd2+ on bursting. Note for both A andB that blockers alter burst characteristics but do not suppress rhythmogenesis or underlying plateau potential.C: lowering external Ca2+ (0.1 mM) eliminated bursting but did not suppress spike discharge. Membrane potential indicated at left side of voltage traces.

View this table:
Table 2.

Drug effects on burst cycle characteristics

Although calcium entry through VGCC was not essential for maintained NMDA bursting, calcium does permeate NMDA channels (Mayer and Westbrook 1987), and this could have been sufficient for NMDA bursting. To test this possibility, the following experiments were performed. First, in the presence of Cd2+ to block VGCC, the effects of reduced [Ca ext2+ ] on NMDA bursting were examined. Figure 4 C shows that during these conditions, only continuous spike discharge was observed (n = 9).

To determine whether a change in intracellular [Ca i2+ ], as opposed to Ca2+ as a charge carrier, is necessary for NMDA burst generation, BAPTA was applied directly through the pipette (10 mM) to chelate the intracellular free calcium to low levels. Figure5, A and B, shows a typical example observed in 5/6 neurons. NMDA was applied within the first 5 min of whole cell recording to induce maintained burst discharge (Fig. 5 A). Prior to applying NMDA, a short pulse (3 ms) elicited a single spike with a subsequent AHP (inset,BAPTA 4′). After NMDA burst discharge was established (Fig.5 A), NMDA was then washed out and then reapplied after 20 min (Fig. 5 B). To ensure that BAPTA was effective, a single spike was initiated again and a reduction in the slow phase of the AHP was observed (inset, BAPTA 30′), indicating a reduction in the Ca2+-activated K+conductance responsible for the slow AHP (Chandler et al. 1994). At that time, NMDA application produced spike discharge without plateau potentials or maintained bursting regardless of the level of membrane potential (Fig. 5 B). These data indicate that NMDA bursting is dependent on increases in the intracellular Ca2+ concentration, presumably through activation of a second-messenger system.

Fig. 5.

Maintenance of a requisite level of intracellular Ca2+ is essential for NMDA bursting. A: intracellular bis-(o-aminophenoxy)-N,N,N′,N′-tetraacetic acid (BAPTA) diffusion after 8 min did not eliminate NMDA bursting or the slow afterhyperpolarization (AHP) following an evoked spike (3-ms pulse;inset, —). B: BAPTA diffusion after 30 min eliminated the slow AHP following an evoked spike (inset, · · ·) and suppressed burst discharge indicating a role for intracellular Ca2+ in maintenance of NMDA bursting.

Modulation of NMDA-mediated bursting by serotonin

We examined whether 5-HT could enable NMDA bursting in the P1–P4 group because 5-HT receptor activation is necessary for NMDA bursting in tadpole spinal motoneurons (Scrymgeour-Wedderburn et al. 1997) and can directly induce rhythmical membrane potential oscillations in trigeminal motoneurons (Del Negro et al. 1999; Hsiao et al. 1998) as well as modulate rhythmical TMN discharge during cortically induced rhythmical jaw movements in vivo (Katakura and Chandler 1990).

To exclude the possibility that 5-HT directly produces bistability and bursting in rat TMNs (Del Negro et al. 1999;Hsiao et al. 1998), we examined the effects of 5-HT on TMN membrane properties in both current and voltage clamp. Figure6 shows the effects of bath application of 5-HT on membrane potential in current clamp (Fig. 6 A) and total membrane current in voltage clamp (Fig. 6 C). Typically, in current clamp, after 10 μM 5-HT the membrane potential depolarized and produced continuous spiking. In contrast to that observed in adult guinea pig trigeminal motoneurons (Hsiao et al. 1998), rhythmical burst discharge was never observed. In the presence of TTX, 5-HT depolarized the neuron (not shown) and, based on current pulse application, increased input resistance (Fig.6 B) with an apparent reversal potential of approximately −90 mV. Table 1 summarizes the effects of 5-HT on membrane properties for P1–P4 and P8–P10 groups. In voltage clamp, 5-HT produced inward current (Fig. 6 C). Figure 6 D shows the steady-state I tot-Vrelationship derived from ramp commands (see methods), before and during 5-HT application. The inset shows theI 5-HT obtained from digital subtraction. It is apparent that 5-HT produced an inward current throughout the voltage range examined and, more importantly, did not induce any prominent NSC in the steady-stateI tot-V relationship. Furthermore, the I 5-HT had an extrapolated reversal potential negative to −80 mV indicating that the increase in input resistance and depolarization resulted from a reduction of a leakage K+ current (Hsiao et al. 1997).

Fig. 6.

Effects of 5-HT on trigeminal motoneurons. A: current-clamp record showing membrane depolarization and spiking in response to serotonin (5-HT) from a P4 neuron (10 μM).B: membrane potential response to extrinsic current pulse application (V-I relationship) before (○) and during 5-HT application (●). Note the increase in steady-state R inp (slope) in the presence of 5-HT. The intersection of the curves indicate estimated reversal potential of approximately −90 mV. C: voltage-clamp record showing inward current in response to 5-HT application from a P3 neuron. Holding potential −50 mV.D: steady-state I-V relationship of total membrane current obtained in response to ramp voltage command before (○) and during (●) 5-HT application. Note the presence of outward rectification but lack of a NSC for both traces.Inset shows I 5-HT obtained by subtraction of control current from current obtained during 5-HT application.

The effects of 5-HT on NMDA bursting in a P2 and P8 neuron are shown in Fig. 7. In the absence of NMDA, 5-HT did not induce any rhythmical bursting activity in either age group (Fig.7, A1 and B1). In the P2 neuron, NMDA application produced continuous spike discharge (Fig. 7 A2). However, in the presence of 5-HT, this discharge was transformed into a robust, rhythmical bursting activity (9/12 neurons; Fig. 7 A3). Figure 7 C shows theI tot-V relationship prior to and during 5-HT application in the presence of NMDA for this neuron. Prior to 5-HT application, NMDA did not induce an NSC. However, after addition of 5-HT, a distinct NSC was observed in 4/5 neurons; a necessary condition for bursting. In the P8 neuron, which exhibited NMDA bursting (Fig. 7 B2), simultaneous application of 5-HT increased intraburst spike frequency and prolonged burst duration (Fig.7 B3, Table 2). Figure 7 D shows theI tot-V relationship obtained from voltage clamp in the presence of NMDA before and during simultaneous 5-HT application. As shown, 5-HT shifted the current downward, increased R inp and enhanced the NSC region. These changes most likely account for the increase in burst duration (Fig. 7 B3). This was observed in 4/5 neurons. These data demonstrate that neurons younger than P8 are, in fact, capable of maintained NMDA bursting and bistability in the presence of 5-HT.

Fig. 7.

5-HT enables NMDA bursting. A1: 5-HT (10 μM) induced spike discharge in a P2 neuron. A2: NMDA (50 μM) induced tonic spiking. A3: 5-HT plus NMDA induce burst discharge. B, 1–3: same as A1 but taken from a P8 neuron. Note the enhanced burst duration in the presence of 5-HT (B3). C: steady-stateI-V relationship of total membrane current taken in voltage clamp from neuron in A in the presence of NMDA (○) and during NMDA plus 5-HT (●). Note the lack of NSC prior to 5-HT. D: same asC except taken from the P8 neuron.

The synergistic effect of 5-HT and NMDA that resulted in the induction of bursting could have resulted from a direct voltage-dependent block and unblock of the NMDA channel by 5-HT, similar to that caused by Mg2+ (Chesnoy-Marchais and Barthe 1996). To test for this possibility, Mg2+was removed from the bath (Fig.8 A2) during NMDA/5-HT induced bursting in a P10 neuron (Fig. 8 A1). During these conditions, rhythmical bursting was eliminated and maintained membrane depolarization was produced (n = 3). This was tested more directly in voltage clamp from P8–P10 neurons. A representative example is shown in Fig. 8 B, 1 and 2. In 0 Mg2+ solutions, NMDA did not produce a region of NSC in the I tot-Vrelationship. After the addition of 5-HT, theI tot-V relationship was shifted downward without induction of an NSC (n = 6), indicating that 5-HT per se does not produce a voltage-dependent block of the NMDA channel (Chesnoy-Marchais and Barthe 1996) that could have created a condition for bistability.

Fig. 8.

Mg2+ is necessary for NMDA/5-HT bursting.A1: burst discharge induced by 5-HT (10 μM) plus NMDA (50 μM) in a P10 neuron. Removal of Mg2+ eliminated the rhythmical burst discharge (A2). Membrane polarization indicated by bottom current trace was not capable of restoring burst discharge. B1: steady-stateI-V relationship for I tot in 0 Mg2+ conditions for NMDA (○) and NMDA plus 5-HT conditions (●). Compare this figure to Fig.7 C. B2: composite data from 6 neurons ages P8–P10. Current density is plotted.

The preceding data suggest that in addition to the increase in total membrane current and input resistance produced directly by 5-HT, 5-HT also enables NMDA bursting in P1–P4 neurons through enhancement of the NSC region in the steady-stateI NMDA-V, presumably by alterations in the Mg2+ block of the NMDA channel. To test this directly, we isolated theI NMDA and examined the effects of 5-HT on the steady-state I NMDA-Vrelationship in 1 mM Mg2+ in the P1–P4 group. Figure 9 A shows an example from one neuron. Prior to 5-HT application the normalizedI NMDA-V relationship showed a very small region of NSC. In the presence of 5-HT, this region was enhanced most prominently between −60 and −40 mV, suggesting a reduction in the voltage-dependent block by Mg2+at depolarized potentials. These effects are summarized in Fig.9 B (n = 4) and are quantified in the histogram in Fig. 9 C where the ratio of peakI NMDA toI NMDA recorded at −100 mV prior to and during 5-HT application are depicted.

Fig. 9.

5-HT enhances NSC region ofI NMDA-V relationship.A: normalized I-V relationship for NMDA current prior to (○) and during 5-HT (10 μM) application (●) for a P2 neuron. Inset: command protocol. V, command potential.B: composite I-V relationship for 4 neurons (P1–P4). Note the increase in I NMDAwithin the voltage range of approximately −60 to −20 mV.C: the ratio of the normalized peakI NMDA (arrow in B) to NMDA current measured at −100 mV (I −100) is plotted as a histogram for both NMDA only conditions (□) and NMDA plus 5-HT (■). A value of 1 indicates the lack of an NSC. *, statistical difference,n = 4, P ≤ 0.03.


The present study demonstrates that NMDA-induced bursting is developmentally regulated within a narrow time window; TMNs less than P5 seldom exhibit NMDA bursting and this was coincident with the lack of a prominent voltage-dependent Mg2+ block of the NMDA iontophore channel. However, in the presence of 5-HT receptor activation, NMDA bursting was initiated. The “enabling” of the NMDA receptor/iontophore by 5-HT for ages less than P5 most likely resulted, in part, from enhancement of the negative slope conductance region of the total membrane current steady-stateI tot-V relationship and from an increase in net inward membrane current due to the reduction of a leakage conductance in the presence of 5-HT. Because trigeminal motoneurons, and oral-motor activity in general, are dependent on NMDA receptor activation for production of rhythmical discharge (Katakura and Chandler 1990; Kogo et al. 1996) and are subject to 5-HT modulation (Chandler et al. 1985; Hsiao et al. 1997), these finding become significant for understanding the factors controlling rhythmical sucking and the transition to adult-like masticatory behaviors in trigeminal motoneurons.

Development of NMDA-induced bursting activity

Bath application of NMDA produces membrane oscillations and rhythmical burst discharge in different types of neurons (Schmidt et al. 1998). The bistable behavior is generally attributed to the production of a negative slope conductance region in the steady-state I-V relationship in combination with voltage- and time- dependent repolarizing conductances (e.g.,Kim and Chandler 1995; Schmidt et al. 1998).

In the presence of synaptic blockade of rapid excitatory and inhibitory synaptic transmission, bath application of NMDA did not induce consistent bursting behavior until approximately P8 in the present study. Prior to that, NMDA application produced membrane depolarization and continuous spike discharge, providing physiological evidence for the anatomical observation (Turman et al. 1999) that functional synaptic and/or extra-synaptic receptors for NMDA are present on trigeminal motoneurons at birth. However, the inability to produce bursting most likely relates to the lack of a sufficiently developed NSC in the steady-stateI NMDA-V relationship. This is in contrast to that reported for neurons within the nucleus tractus solitari where bursting, albeit irregularly occurring, was observed at birth in conjunction with an NSC (Vincent et al. 1996). It is difficult to compare our data to theirs because they studied NMDA currents in the absence of synaptic blockade and the possibility of NMDA receptor modulation by background synaptic input, and a contribution from rhythm-generating networks, cannot be excluded.

The absence of rhythmical NMDA-induced bursting and a prominent NSC in P1–P4 neurons most likely resulted from insufficient development of a voltage-dependent Mg2+ block of the NMDA receptor channel. As shown in the present study and others (e.g., Kim and Chandler 1995; Tell and Jean 1993), NMDA bursting is critically dependent on the presence of Mg2+ in the bath. Although theI NMDA-V relationship for P2–P4 neurons showed sensitivity to changes in extracellular Mg2+ at hyperpolarized voltages, in the presence of 1 mM Mg2+ very little voltage dependence was observed as indicated by the small NSC in the steady-stateI NMDA-V relationship. In contrast, animals older than P7 exhibited a prominent NSC that provides the basis for bistability. These differences most likely contributed to the lack of bursting observed in the younger age group neurons. Although the exact postnatal time period may differ, similar developmental changes in the sensitivity to Mg2+and the NMDA voltage-dependent Mg2+ block have been reported in neocortex (Burgard and Hablitz 1994), tractus solitari (Nabekura et al. 1994), and hippocampal CA1 pyramidal neurons (Ben-Ari et al. 1988;Kirson et al. 1999), although, a lack of a postnatal developmental change in the voltage-dependent Mg2+ block has been reported (Kirson and Yaari 1996; Kleckner and Dingledine 1991).

The present study did not address the mechanism(s) for the change in voltage-dependent block by Mg2+ of the NMDA receptor. However, given that the expression profile of the different NMDA receptor subunits changes in trigeminal neurons during development (Turman et al. 1999) and that in heterologous systems, recombinant NMDA receptors containing NR2A or NR2B subunits (Kuner and Schoepfer 1996; Monyer et al. 1994) are more strongly blocked by Mg2+than those containing NR2C or NR2D, the developmental change in the voltage-dependent block by Mg2+ of the NMDA receptor channel in trigeminal motoneurons could be due to alterations in the expression of NMDA receptor subunits. Alternatively, developmental regulation of the Mg2+ block by intracellular messengers such as protein kinase C (PKC) (Ben-Ari et al. 1992; Chen and Huang 1992; Lan et al. 2001) must also be considered.

Intracellular calcium requirement for NMDA bursting

The present study provides evidence that NMDA bursting is dependent on intracellular messenger activity and in particular maintenance of a requisite level of intracellular calcium concentration. A previous study on spinal motoneurons of adult turtle, in vitro, showed that NMDA bursting depended on L-type Ca2+ channel activation (Guertin and Hounsgaard 1998). In the present study, L-type channel blockers or Cd2+, which blocks all trigeminal motoneuronal HVA Ca2+ currents (unpublished observation), did not block NMDA oscillations or the underlying plateau potentials during the bursts but did increase intraburst spike frequency and burst duration in the majority of neurons tested. This suggests that Ca2+, as a charge carrier per se through VGCCs, is not necessary for NMDA bursting but changes in [Cai] can alter NMDA burst cycle characteristics. This is not unexpected because TMNs posses a calcium-dependent K+ conductance that is responsible for the slow AHP following each action potential and contributes to regulation of spike discharge (Chandler et al. 1994; Inoue et al. 1999). However, when Ca2+ entry through both VGCC and NMDA channels was substantially reduced by simultaneous application of both Cd2+ and low external Ca2+media or when BAPTA was present in the pipette to buffer intracellular Ca2+ to low levels, bursting was not possible, suggesting that [Cai] acting as an intracellular messenger is necessary for NMDA bursting. This, of course, is reminiscent of the requirement for intracellular Ca2+ for some forms of LTP (Bliss and Collingridge 1993).

Serotonin enables NMDA bursting

In the presence of 5-HT robust NMDA-induced bursting was observed in most neurons, including those less than P8. Interestingly, in contrast to what was demonstrated in guinea pig TMNs where serotonin directly induces bistable bursting behavior (Hsiao et al. 1998), serotonin application in the absence of NMDA produced only tonic depolarization and an increase inR inp; thus eliminating the possibility that 5-HT directly induced nonlinear conductance changes, which could have supported bistability. A similar “enabling” function by 5-HT for NMDA oscillations was described in Xenopus spinal cord neurons where, in the absence of 5-HT receptor activation, NMDA bursting was not possible (Scrymgeour-Wedderburn et al. 1997). Similarly, in rat spinal motoneurons NMDA oscillations are dependent on endogenous serotonergic receptor activity (Maclean et al. 1998).

The present data indicate that the mechanism(s) for “enabling” of NMDA bursting by 5-HT in trigeminal motoneurons at birth is due, partly, to a reduction in the Mg2+ block at depolarized potentials, thus leading to enhancement of the NSC in the steady-state I tot-V. Based on current-clamp analysis, this was proposed for the “enabling” of NMDA oscillations by 5-HT in Xenopus motoneurons (Scrymgeour-Wedderburn et al. 1997). In the present study, in animals less than P5, theI tot--V relationship showed either a very small or no NSC in the presence of NMDA. After 5-HT application, the I tot-Vrelationship was shifted downward due to the reduction in leakage current and exhibited a prominent NSC (Fig. 7 C). It is unlikely that the NSC in theI tot-V relationship resulted from activation of L-type Ca2+ channels by 5-HT, as demonstrated in guinea pig trigeminal motoneurons (Hsiao et al. 1998), because in the absence of NMDA an NSC was not observed after 5-HT. Furthermore in the absence of external Mg2+ but in the presence of 5-HT, NMDA bursting was not possible. This excludes the possibility that 5-HT directly effected the NMDA iontophore to produce a voltage-dependent block of the NMDA receptor in a manner analogous to Mg2+as shown previously in embryonic spinal neurons (Chesnoy-Marchais and Barthe 1996).

A number of reports indicate that 5-HT modulates NMDA receptor function (Blank et al. 1996; Nedergaard et al. 1986; Trueblood et al. 1996), yet the precise mechanism for the effect is not clear. However, in Xenopusoocytes it was shown that 5-HT2 receptors potentiate NMDA responses via a PKC-dependent process (Blank et al. 1996). Because the Mg2+ block of the NMDA receptor channel in trigeminal sensory neurons is reduced by PKC (Chen and Huang 1992) and 5-HT couples to PKC in trigeminal motoneurons (Inoue et al. 1999), it is possible that a similar intracellular transduction mechanism is responsible for the increase in the NSC in theI NMDA-V and thus enabling of NMDA bursting by 5-HT in TMNs as well.

Physiological implications

NMDA receptors have been implicated in a number of behavioral activities such as synaptic plasticity and learning during development (Bliss and Collingridge 1993; Mori and Mishina 1995) and excitotoxic brain damage (Diemer et al. 1993), among others. Less attention has been placed on NMDA receptor participation in rhythmical motor acts (Kim and Chandler 1995; Schmidt et al. 1998;Scrymgeour-Wedderburn et al. 1997; Wallen and Grillner 1987). However, NMDA receptor activation is critical for rhythmical oral-motor activity in vivo (Katakura and Chandler 1990) and in vitro (Kogo et al. 1996). Although it is unlikely that NMDA activation per se is responsible for the motoneuronal rhythmogenesis during oral-motor behaviors, NMDA receptor activation does impart significant nonlinearity to the trigeminal motoneuron membrane I-V relationship, and as such, these membrane properties will undoubtedly have significant influence on synaptic integration during ongoing rhythmical activity (Kim and Chandler 1995). For instance inXenopus, it was proposed that the NMDA oscillations do not contribute to the genesis of the swimming oscillations but rather increase the responsiveness of the motoneurons to synaptic drive during the swimming cycle (ScrymgeourWedderburn et al. 1997).

Presently, there is little information on the factors responsible for the transition from sucking, a behavior present at birth, to adult-like mastication, which in the rat behaviorally emerges, rudimentarily, around P12. NMDA receptor activation is critical for early postnatal development of synaptic circuits, most likely due to the need for maintenance of a critical level of intracellular Ca2+ (Scheetz and Constantine-Paton 1994). The need for nutrient intake at birth is essential and, therefore rapid development of oral-motor circuitry occurs. The low Mg2+ sensitivity and lack of a prominent Mg2+ block of the NMDA iontophore in early postnatal development most likely contributes to the rapid development of the oral-motor circuitry by allowing the NMDA receptor to be functionally active at more hyperpolarized potentials in response to nonspecific synaptic activity, thus providing for increased levels of Ca2+ to enter the motoneuron through the NMDA iontophore. However, during conditions where stable rhythmical oral-motor activity is required, robust motoneuronal bursting could occur through integration of rhythmogenic network activity with enhanced NMDA burst activity produced by 5-HT modulation of the Mg2+ block and production of an NSC. A complete understanding of the mechanisms responsible for rhythmical oral-motor behavior, and in particular, the factors responsible for the transition from rhythmical sucking to masticatory behavior will necessitate inclusion of the developmental properties of the NMDA receptor.


We thank M. Castillo for technical assistance.

This work was funded by National Institute of Dental and Craniofacial Research Grant RO1 DE-06193.


  • Address for reprint requests: S. H. Chandler, Dept. of Physiological Science, UCLA, 2859 Slichter Hall, Los Angeles, CA 90095-1568 (E-mail: schandler{at}


View Abstract