Gao, Bao-Xi, Gong Cheng, and Lea Ziskind-Conhaim. Development of spontaneous synaptic transmission in the rat spinal cord. J. Neurophysiol. 79: 2277–2287, 1998. Dorsal root afferents form synaptic connections on motoneurons a few days after motoneuron clustering in the rat lumbar spinal cord, but frequent spontaneous synaptic potentials are detected only after birth. To increase our understanding of the mechanisms underlying the differentiation of synaptic transmission, we examined the developmental changes in properties of spontaneous synaptic transmission at early stages of synapse formation. Spontaneous postsynaptic currents (PSCs) and tetrodotoxin (TTX)-resistant miniature PSCs (mPSCs) were measured in spinal motoneurons of embryonic and postnatal rats using whole cell patch-clamp recordings. Spontaneous PSC frequencies were higher than mPSC frequencies in both embryonic and postnatal motoneurons, suggesting that even at embryonic stages, when action-potential firing rate was low, presynaptic action potentials played an important role in triggering spontaneous PSCs. After birth, the twofold increase in spontaneous PSC frequency was attributed to an increase in action-potential–independent quantal release rather than to a higher rate of action-potential firing. In embryonic motoneurons, the fluctuations in peak amplitude of spontaneous PSCs were normally distributed around single peaks with modal values similar to those of mPSCs. These data indicated that early in synapse differentiation spontaneous PSCs were primarily composed of currents generated by quantal release. After birth, mean mPSC amplitude increased by 50% but mean quantal current amplitude did not change. Synchronous, multiquantal release was apparent in postnatal motoneurons only in high-K+ extracellular solution. Comparison of the properties of miniature excitatory and inhibitory postsynaptic currents (mEPSCs and mIPSCs) demonstrated that mean mEPSC frequency was higher than mIPSC frequency, suggesting that either excitatory synapses outnumbered inhibitory synapses or that the probability of excitatory transmitter release was higher than the release of inhibitory neurotransmitters. The finding that mIPSC duration was several-fold longer than mEPSC duration implied that despite their lower frequency, inhibitory currents could modulate motoneuron synaptic integration by shunting incoming excitatory inputs for prolonged time intervals.
Numerous studies have examined the properties of action potential-independent spontaneous synaptic transmission in the adult mammalian spinal cord (Edwards et al. 1976; Lüscher 1990; Redman 1990; Walmsley and Bolton 1994), but little is known about the nature of spontaneous synaptic transmission at early stages of neuronal differentiation in vivo. Studying developmental changes in the properties of action-potential–independent miniature postsynaptic currents (mPSCs) is important for our understanding of the relative roles of pre- and postsynaptic factors in the differentiation of synaptic transmission.
Characteristically, the frequency of spontaneous synaptic potentials is low in embryonic motoneurons(<0.2 Hz), but an appreciable increase in their frequency is apparent immediately after birth (Nishimaru et al. 1996; Xie and Ziskind-Conhaim 1995; Ziskind-Conhaim 1988). The low-frequency spontaneous synaptic potentials recorded in embryonic motoneurons are in sharp contrast to the relatively large, long-lasting dorsal root-evoked synaptic potentials generated in embryonic spinal cords as early as day 15 of gestation (Saito 1979; Ziskind-Conhaim 1990). At that age, the evoked synaptic potentials are produced by excitation of polysynaptic pathways with inputs from excitatory glutamatergic interneurons, and inhibitory glycinergic and GABAergic interneurons (Kudo and Yamada 1985, 1987; Saito 1979; Seebach and Ziskind-Conhaim 1994; Wu et al. 1992). By embryonic day 17–18, primary afferent projections directly synapse onto motoneurons, giving rise to glutamate-mediated monosynaptic potentials. At the same age, descending serotonin-containing axons project into the ventral horn (Ziskind-Conhaim et al. 1993), but their role in modulating synaptic transmission in the immature spinal cord remains largely unknown.
Infrequent spontaneous synaptic activity during embryonic development could be attributed to numerous presynaptic factors: low-frequency action-potential firing along newly established neuronal pathways (Fitzgerald 1987; Gao and Ziskind-Conhaim 1994; Xie and Ziskind-Conhaim 1995; Ziskind-Conhaim 1988), strong inhibitory synapses that transiently dominate synaptic transmission along developing sensorimotor pathways (Nishimaru et al. 1996; Wu et al. 1992), and the undifferentiated state of presynaptic components, which modulate action potential-independent transmitter release.
In this study, we analyzed the properties of spontaneous postsynaptic currents (PSCs) and action potential-independent mPSCs at a critical developmental period during which the level of spontaneous synaptic transmission is enhanced significantly. The roles of presynaptic action potentials and pre- and postsynaptic components at the release site in regulating the frequency and amplitude of spontaneous PSCs were examined at embryonic days 17–18, when dorsal root afferents first formed excitatory monosynaptic inputs on motoneurons, and a few days after birth. The contribution of excitatory and inhibitory synapses to synaptic integration in developing motoneurons was studied by simultaneous recording of excitatory and inhibitory currents. Basic kinetic properties of excitatory and inhibitory mPSCs also were examined to determine whether the developmental increase in mPSC frequency was correlated with changes in their properties.
Preliminary data were published in an abstract form (Gao and Ziskind-Conhaim 1995b).
Spinal cord preparation
Lumbar spinal cords were isolated from Sprague-Dawley rat embryos at 17–18 days of gestation (E17–18, birth is at E21–22), and from 1- to 3-day–old postnatal rats (P1–3). Pregnant rats were anesthetized lightly and decapitated. Embryos were decapitated quickly, and their spinal cords removed and placed into ice-cold dissecting solution (Ziskind-Conhaim et al. 1993). Postnatal rats were anesthetized by hypothermia. Spinal cord slices were prepared using a procedure described previously (Gao and Ziskind-Conhaim 1995a). Transverse slices (350 μm) were cut using a Vibrotome. Slices were transferred into a recording chamber and held on the glass-bottom chamber with a grid of nylon threads. The submerged slices were perfused with extracellular solution gassed with 95% O2-5% CO2 and maintained at room temperature (21–23°C). The extracellular solution contained (in mM) 113 NaCl, 3 KCl, 2 CaCl2, 1 MgCl2, 25 NaHCO3, 1 NaH2PO4, and 11 glucose (pH 7.2).
The following substances were added to the extracellular solution at known concentrations: tetrodotoxin (TTX, Sigma), d−2-amino-5-phosphonovaleric acid (d-APV), 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), strychnine, and bicuculline methchloride from Research Biochemicals.
Whole cell recordings
Whole cell recordings were made from visually identified motoneurons using infrared DIC-videomicroscopy (Dodt and Zieglgansberger 1990; MacVicar 1984). The largest multipolar or round cells (15–25 μm diam) in the lateral and medial ventral horn were identified as motoneurons (Gao and Ziskind-Conhaim 1995a; Takahashi 1978). Patch pipettes with a tip diameter of 2–3 μm and DC resistance of 4–8 MΩ were fabricated using a Flaming-Brown P-97 puller (Sutter Instruments). The pipette solution contained (in mM) 140 Cs-gluconate, 9 CsCl, 1 Mg-ATP, 0.1 GTP, 10 N-2-hydroxyethylpiperazine-N′−2-ethanesulfonic acid, and 0.2ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid(buffered to pH 7.2 with CsOH). After the formation of a giga seal and breaking through the membrane, the cell capacitance was compensated by adjusting the compensation dial. The series resistance was 8–15 MΩ, and it was compensated >60%. To record both excitatory and inhibitory synaptic currents, experiments were performed at a holding membrane potential of −40 mV. All recordings were corrected for the liquid junction potential (13 mV) (Gao and Ziskind-Conhaim 1995a).
Synaptic currents were recorded continuously using either Axopatch-1D or Axopatch 200A amplifiers (Axon Instruments). Currents were filtered at 1 kHz, digitized at 5 kHz, and stored on an optical disk. To analyze amplitude distributions of synaptic currents, recordings in a given motoneuron were carried out for 10 min or until >150 events were recorded at E17–18 and >250 events were recorded at P1–3.
The threshold for synaptic current detection was set at 2 pA above the background noise. Synaptic currents were isolated from the contaminating noise by using both DATAPAC III software (Run Technologies) and visual detection. The same peak detection windows were used for spontaneous PSCs and mPSCs. Examples of histograms of spontaneous PSC and mPSC peak amplitudes and corresponding histograms of peak values of baseline noise are illustrated in Fig. 1. The overlap between these histograms was insignificant therefore, it is unlikely that an appreciable number of small mPSCs were mistakenly regarded as noise events and that mPSC amplitudes were overestimated.
Synaptic currents were analyzed off-line using pClamp software (Axon Instruments) and DATAPAC III. Only currents with monotonic rising phase were included in the analyses. To minimize the probability of analyzing attenuated currents generated at distal synapses, only excitatory currents with rising time ≤3 ms were included in amplitude analyses. The cutoff for inhibitory currents was 5 ms. The cutoff criteria for inclusion of currents were based on preliminary data demonstrating that the rising times for large miniature excitatory PSC (mEPSCs; ≅10 pA) varied between0.3 and 3 ms, whereas those for large miniature inhibitory PSCs (mIPSCs) varied between 0.3 and 5 ms. There was no correlation between the amplitude and rise time of those currents, indicating that they were not affected by cable attenuation.
Amplitude histograms were plotted based on a binwidth of 1 pA. This binwidth was chosen so that when two consecutive peaks were apparent in an amplitude histogram, they were separated by ≥5 bins (Edwards et al. 1990). However, amplitude histograms constructed with binwidths of 2–3 pA were not significantly different from those based on binwidths of 1 pA (see also Edwards et al. 1990).
To determine whether amplitude fluctuations were normally distributed, Gaussian function was tested using Microcal Origin software. The peaks in multiple-peak amplitude histograms were determined by eye, and Gaussian fitting was performed using Microcal Origin software. There was no attempt to fit normal and skewed amplitude distributions with any other curves.
Analyses of mPSC properties included: peak amplitude, rise time from 10 to 90% peak amplitude, decay time constant (decay τ), and duration. mPSC decays were best fitted with single exponential curves that were superimposed on mPSCs. mPSC duration was measured as the time from 10% of peak current to 90% return to baseline. Analyses of mPSC properties included all currents with monotonic rise and decay phases, regardless of the rise time. Data are presented as means ± SE. Student's t-test was used to determine the statistical significance of the results. The level of statistical significance was 5%.
Postnatal increase in the frequency of spontaneous postsynaptic currents
Developmental changes in the frequency of spontaneous PSCs were examined during early stages of synapse formation when a significant increase in the frequency of spontaneous synaptic potentials has been reported (Xie and Ziskind-Conhaim 1995; Ziskind-Conhaim 1988). Currents wererecorded at E17–18 and P1–3. To distinguish between excitatory and inhibitory currents, motoneurons were held at a membrane potential of −40 mV (Fig. 2 A). The reversal potential for both glycine- and γ-aminobutyric acid (GABA)-mediated inhibitory currents is −66 mV (Gao and Ziskind-Conhaim 1995a). The inward currents were blocked by d-APV (20 μM) and CNQX (10 μM; Fig. 2 B), antagonists of N-methyl-d-aspartate (NMDA) and non-NMDA glutamate receptor subtypes, indicating that they were glutamate-mediated excitatory synaptic currents. The outward currents were eliminated by strychnine (5 μM) and bicuculline (20 μM, Fig. 2 B), suggesting that they were generated by activation of glycine- and GABAA-gated receptors.
In both embryonic and postnatal motoneurons, spontaneous PSCs were heterogeneous in size, varying between 3 and >40 pA. At E17–18, the mean frequency of spontaneous PSCs was >0.5 Hz, significantly higher than the frequency of spontaneous synaptic potentials recorded using high-resistant intracellular microelectrodes (<0.2 Hz) (Xie and Ziskind-Conhaim 1995; Ziskind-Conhaim 1988). It is likely that the higher signal-to-noise ratio of the whole cell recording technique provided a better resolution for detecting smaller synaptic events than those recorded using intracellular microelectrodes.
From the onset of formation of excitatory monosynaptic connections, and at least until 1–3 days after birth, the mean frequency of spontaneous excitatory postsynaptic currents (EPSCs) was higher than that of spontaneous inhibitory postsynaptic currents (IPSCs, Table 1). At E17–18, spontaneous EPSC and IPSC frequencies were ∼0.5 and 0.2 Hz, respectively (Table 1). Despite the significant difference in their frequencies, spontaneous EPSC and IPSC amplitudes were similar. At that age, the fluctuations in their peak amplitudes were normally distributed around single peaks with mean modal values of 6.3 and 5.7 pA, respectively (n = 7). Our finding that the mean frequency of spontaneous IPSCs in embryonic motoneurons was twofold lower than that of spontaneous EPSCs does not support the hypothesis that inhibitory inputs onto motoneurons dominate synaptic transmission during embryonic development.
After birth, the mean frequency of spontaneous EPSCs and IPSCs increased to ∼1.2 and 0.6 Hz, respectively (Table 1). The more than twofold postnatal increase in spontaneous PSC frequencies was associated with larger currents as evident by broader normal amplitude distributions and amplitude distributions skewed toward large currents (Fig. 4, P1: sIPSCs). Several factors might have contributed to the developmental increase in frequency and amplitude of spontaneous PSCs: higher frequency of action potentials invading presynaptic terminals, presynaptic axonal growth and formation of new synaptic contacts onto motoneurons, and differentiation of pre- and postsynaptic components at the release site.
Developmental changes in the frequency and amplitude fluctuations of mPSCs
To test the hypothesis that the postnatal increases in frequency and amplitude of spontaneous PSCs resulted from an increase in action potential firing in presynaptic pathways, action potentials were blocked by TTX. In the presence of extracellular Mg2+, most TTX-resistant mEPSCs were blocked by CNQX (10 μM), indicating that they were generated by activation of non-NMDA receptors (Cheng et al. 1997). The outward currents were blocked partially by either strychnine (5 μM) or bicuculline (20 μM), suggesting that mIPSCs were composed of both glycine and GABAA receptor–mediated currents (Gao et al. 1997).
mEPSC and mIPSC frequencies were about twofold lower than spontaneous EPSC and IPSC frequencies, indicating that generation of spontaneous PSCs was regulated partially by presynaptic action potentials. In embryonic motoneurons, fluctuations in peak amplitudes of spontaneous PSCs and mPSCs were normally distributed and were well fitted by Gaussian curves with similar mean amplitudes (Figs. 3 and 4). The theory of quantal transmitter release assumes that each mPSC is generated by the release of a single quantum of transmitter, therefore during embryonic development spontaneous PSCs consisted mostly of quantal events.
Blocking presynaptic action potentials in spinal cords of postnatal rats not only reduced the frequency of spontaneous PSCs but also significantly decreased their amplitudes. mPSC unimodal amplitude distributions were narrower (Fig. 3, P2) and their mean modal value (6.4 pA, n = 5) was 3 pA smaller than the modal value of normally distributed spontaneous PSCs (9.3 pA, n = 5). Moreover, in some postnatal motoneurons in which spontaneous PSC amplitude distributions were skewed toward large currents (Fig. 4, P1: sIPSC), blocking action potentials eliminated the large currents. These findings implied that the large spontaneous PSCs resulted primarily from action potential-dependent release.
After birth, there was a twofold increase in mPSC frequencies (Table 1), which probably reflected the formation of new synaptic inputs onto motoneurons and/or the increased probability of transmitter release. In 70% of motoneurons, mPSC amplitude fluctuations were normally distributed with a mean mPSC amplitude of 6.4 pA, which was not significantly different from the mean mPSC amplitude of 5.3 pA in embryonic motoneurons (Figs. 3 and 4). This finding suggested that the size of the quantal current did not change in most motoneurons. However, in ∼30% of postnatal motoneurons, the higher frequencies were associated with larger mean mPSC (9.8 pA) with skewed amplitude distributions. Several factors might have contributed to the skewed amplitude distribution including the relatively small fraction of large currents generated by synchronous, multiquantal release and the intrinsic fluctuations in quantal size and in the number of postsynaptic receptors available at each release site (Isaacson and Walmsley 1996).
Similar to the ratio of spontaneous EPSC to IPSC frequencies, mean mEPSC frequency was about twofold higher than mIPSC frequency in both embryonic and postnatal motoneurons (Table 1). This finding indicated that either excitatory synapses contacting motoneurons outnumbered inhibitory synapses and/or the inherent probability of excitatory transmitter release was higher than the probability of inhibitory transmitter(s) release.
The relatively low-frequency spontaneous PSCs at early stages of synapse formation might be indicative of an undifferentiated state of presynaptic membrane components that couple presynaptic depolarization to vesicular transmitter release. To test this possibility, we examined the effects of high-K+–induced depolarization on mPSC frequency and amplitude (Finch et al. 1990; Gottmann et al. 1994; Liu and Tsien 1995). Various extracellular K+ concentrations were examined (10–32 mM), and 18 mM appeared to be the optimal concentration for increasing mPSC frequency without causing a significant overlap of synaptic currents (Fig. 5, →). Overlapping currents were excluded from analyses of amplitude distributions.
Continuous recordings of mPSCs before and after elevating extracellular K+ were carried out in embryonic and postnatal motoneurons (n = 12). At both ages, the mean frequency of high-K+–evoked mPSCs was fourfold higher than mPSC frequency, suggesting that cellular components coupling presynaptic depolarizations to vesicular release were not the limiting factors in generating low-frequency spontaneous currents in embryonic motoneurons.
High-K+–induced increase in mPSC frequencies was associated with larger mPSC peak amplitudes (Figs. 5 and 6). Mean amplitudes of high-K+–evoked mPSCs were 55 and 78% higher than mPSC amplitudes in embryonic and postnatal motoneurons, respectively. The increased amplitude was correlated with significant changes in amplitude distribution. Before high-K+ application, unimodal amplitude distributions were apparent in embryonic motoneurons (n = 6), but in 50% of those neurons, amplitude distributions of high-K+–evoked mPSCs were skewed toward large currents (Fig. 5,mEPSCs). In the postnatal motoneurons tested, most amplitude histograms of high-K+–evoked mPSC amplitudes were skewed, and 30% of those had multiple, clearly separated peaks. In the latter group, the approximately equidistant peaks corresponded to integer multiples of the first peak, and the mean amplitude of the first peak was similar to the mean mPSC measured in a single Gaussian curve (Fig. 6). Based on the quantal theory, the multiple-peak histograms represented synchronous release of several quantal. Therefore it is reasonable to assume that high-K+–induced presynaptic depolarization increased the probability of synchronous multiquantal release in postnatal motoneurons.
Developmental changes in the properties of mEPSCs and mIPSCs
To determine whether the postnatal increase in mEPSC and mIPSC frequencies was associated with developmental changes in their properties, we compared mPSC peak amplitudes, rise times and decay τs at E17–18 and P1–3 (Table 2). In the presence of extracellular Mg2+, the majority of mEPSCs were non-NMDA receptor-mediated currents (Cheng et al. 1997). Infrequently (<0.01 Hz), slow-rise, long-duration NMDA receptor-mediated mEPSCs were recorded, but those were excluded from data analyses. mIPSCs were comprised of both glycine- and GABA-mediated currents, but because of their similar properties (Gao et al. 1997), both mIPSC populations were pooled for quantitative analyses.
In embryonic motoneurons, mean amplitudes of mEPSCs and mIPSCs were ∼5 pA (Table 2), and they increased postnatally by 42 and 57%, respectively. The larger currents were not correlated with changes in mEPSC and mIPSC mean rise times of ∼1 and 3 ms, respectively. After birth, mEPSC mean decay τ significantly increased from 2.2 ms at E17 to ∼4.0 ms at P1–3 (Table 2). However, the developmental changes in mIPSC decay τs were not significant, and values of 24.0 and 15.8 ms were measured in embryonic and postnatal motoneurons, respectively.
Our data illustrated that mEPSC and mIPSC basic kinetic properties were significantly different at both ages. mEPSC rise time was about twofold faster, and its mean decay τ was at least fourfold shorter than mIPSC rise time and decay τ. The slower rise time and longer decay τ of mIPSCs resulted in significantly longer duration mIPSCs than mEPSCs. At E17, mean mIPSC and mEPSC durations were 45.9 and 7.8 ms, respectively. Smaller but significant difference between mIPSC and mEPSC durations persisted after birth, when their mean durations were 46.3 and 23.3 ms, respectively.
A possible explanation for the slow-rise, long-duration mIPSCs is that they arise at distal dendrites, and dendritic filtering introduced a significant error in mIPSC shape and size. Motoneurons receive synaptic inputs from different electrotonic sites, which are not equally spaced or voltage clamped. Therefore dendritic field is likely to filter synaptic events that originate distal to the soma, and the magnitude of the error caused by voltage clamping of distal synapses depends on numerous factors such as the distance of the synaptic site from the soma, and the ratio between the time course of synaptic conductance and the membrane time constant (Redman and Walmsley 1983; Ulrich and Lüscher 1993; reviewed by Redman 1990).
To determine whether dendritic filtering introduced a significant error in mIPSC properties, the correlations between their amplitudes and rise times and between decay τs and rise times were examined (Fig. 7). Correlation between these parameters within a given neuron would indicate that cable attenuation introduced a significant distortion in mIPSC properties (Rall 1969). The lack of correlation between these parameters in all motoneurons studied (n = 9) suggested that mIPSC slow kinetics resulted from intrinsic properties of glycine and GABA receptors rather than distortion due to dendritic filtering. Similarly, there was no correlation between mEPSC amplitudes and rise times and between their decay τs and rise times (not shown). We cannot rule out the possibility that the lack of correlation between these properties might be indicative of larger intrinsic variability than the variability expected from electrotonic locations.
Development of synaptic transmission is a fundamental event in the establishment of functional integrated synaptic activity. In this study, we examined the relative roles of presynaptic action potentials and action-potential–independent quantal release in regulating spontaneous synaptic transmission in motoneurons developing in vivo. This study is the first to analyze the properties of excitatory and inhibitory postsynaptic currents simultaneously recorded in motoneurons differentiating in vivo and to determine the relative contribution of excitatory and inhibitory inputs to synaptic integration during early stages of synapse formation.
Mechanisms underlying developmental changes in spontaneous synaptic transmission
Spontaneous PSC frequency is determined by numerous factors, including the firing rate of action potentials invading presynaptic terminals, the number of synaptic inputs on a given neuron, factors that modulate action-potential–independent transmitter release, and receptor activation at the postsynaptic site. Our findings that blocking action potentials reduced spontaneous PSC frequency by ∼50% in both embryonic and postnatal motoneurons implied that the functional contribution of presynaptic action potentials to the frequency of spontaneous PSCs did not increase after birth. Therefore the low-frequency spontaneous PSCs in embryonic motoneurons could not be attributed primarily to the low rate of action-potential firing. The finding that action potential contribution to the generation of spontaneous PSCs did not increase after birth was somewhat surprising because of the developmental increase in the frequency of action potential firing along newly established sensorimotor pathways (Fitzgerald 1987). Our data suggested that the twofold postnatal increase in mPSC frequency resulted from either an increase in the number of synaptic inputs converging onto motoneurons or an enhanced probability of transmitter release.
It is unlikely that low-frequency spontaneous PSCs in embryonic motoneurons resulted from relatively low-density expression of voltage-gated Ca2+ channels on presynaptic terminals because high-K+–induced presynaptic depolarization produced a similar increase in mPSC frequency in embryonic and postnatal motoneurons. We cannot rule out the possibility that other components that couple intracellular Ca2+ to vesicular release are undeveloped in embryonic neurons, resulting in a limited transmitter release.
This study illustrated that blocking presynaptic action potentials in embryonic spinal cords changed neither spontaneous PSC unimodal amplitude distributions nor their mean amplitudes, suggesting that at that age, PSCs were mostly composed of unitary currents generated by quantal release. However, after birth, spontaneous PSCs consisted of currents larger than the unitary currents, indicating that at later stages of synapse differentiation, presynaptic action potentials contributed preferentially to the large amplitude spontaneous events probably by increasing the probability of multiquantal release. The contribution of action potentials to the frequency and amplitude of spontaneous synaptic events varies considerably in different populations of developing motoneurons. For example, TTX mostly eliminates large spontaneous currents in phrenic motoneurons of neonatal rats (Liu and Feldman 1992), but it significantly reduces the frequency of both small and large spontaneous potentials in motoneurons developing in organotypic spinal cord slices (Streit and Lüscher 1992).
Developmental changes in properties of mEPSCs and mIPSCs
The finding that mEPSC frequency was significantly higher than mIPSC frequency implied that either excitatory synapses outnumbered inhibitory synapses and/or the probability of excitatory transmitter release was higher than the release of inhibitory transmitter. Similar findings have been reported in other studies that examined the nature of newly formed synapses between mammalian neurons developing in vitro (Benson and Cohen 1996; Gottmann et al. 1994; Jackson et al. 1982). Several studies have suggested that early in spinal cord differentiation, inhibitory synapses transiently dominant synaptic transmission along sensorimotor pathways (Nishimaru et al. 1996; Wu et al. 1992). However, based on our data, it is unlikely that such predominant inhibitory synapses were formed directly on motoneurons.
The finding that in a given motoneuron there was no correlation among mPSC rise time, peak amplitude, and decay τ suggested that dendritic filtration did not introduce a major error in mPSC shape and size. Similarly, it has been suggested that the variability in mPSC amplitude and shape in developing neurons in the hippocampus and neocortex is not attributed to cable attenuation (Burgard and Hablitz 1993; McBain and Dingledine 1992). Moreover, large variability in mPSC amplitudes was apparent in neurons that lack dendrites (Callister and Walmsley 1996). In bushy cells of the anteroventral cochlear nucleus, the substantial variability in mPSC sizes was attributed to intrinsic fluctuations in quantal release and in the number of receptors at each release site (Isaacson and Walmsley 1996). Current fluctuations could arise from a number of other factors including the time course of diffusion and uptake of the transmitter and receptor affinity (Kitzing et al. 1994).
In our study, the postnatal increase in mPSC frequency was associated with a 50% increase in the mean amplitude of mPSCs. Large amplitude mPSCs did not characterize all postnatal motoneurons, and they were prominent in only 30% of motoneurons. We cannot rule out the possibility that those motoneurons represented a more differentiated population of neurons, in which either receptor density at the postsynaptic membrane was higher and/or the quantal content was significantly larger than in the rest of motoneurons. Unlike our findings, in dissociated central neurons developing in vitro, developmental increases in synaptic transmission were not temporally correlated with changes in miniature current amplitudes (Gottmann et al. 1994; Kraszewski and Grantyn 1992).
If the increase in receptor density was the only factor responsible for the increased mPSC amplitude, it is reasonable to assume that in embryonic motoneurons quantal release did not saturate the postsynaptic receptors (Faber et al. 1992). In other neurons differentiating in culture, it has been shown that quantal release activates only a small fraction of receptors. For example, only a fraction of glutamate receptors clustered opposite release sites contributes to mEPSCs in differentiating motoneurons (Vogt et al. 1995). Similarly, it has been suggested that in dissociated spinal neurons differentiating in vitro, activation of only 10 NMDA channels is required for generating a 20 pA mEPSC (Ascher et al. 1988), and openings of only a few GABAA receptors are necessary for the production of mIPSCs (Edwards et al. 1990).
Developmental changes in kinetic properties of mPSCs might reflect changes in diffusion and uptake of the transmitter in the synaptic cleft and changes in receptor affinity and gating kinetics. During embryonic development, there was no change in mEPSC rise time, but mEPSC mean decay τ increased almost twofold. Therefore, it is conceivable that channel kinetics and/or factors involved in transmitter uptake changed immediately after birth. It is unknown whether additional changes in mEPSC properties occur at later stages of spinal cord maturation, but the time course of mEPSCs in postnatal motoneurons is within the range of values reported for primary afferent-evoked excitatory postsynaptic potentials in spinal cords of adult cats (Redman and Walmsley 1983).
mIPSC mean rise time and decay τ did not change during the period studied. mIPSC mean decay τ was 15.9 ms in postnatal motoneurons, similar to the decay τ of evoked IPSCs recorded in sensory spinal neurons of newborn rats (17.8 ms) (Takahashi et al. 1992). In those neurons, the decrease in decay τs occurs only 2–3 wk after birth.
Our study illustrated that in both embryonic and postnatal motoneurons, the mean duration of mIPSCs was significantly longer than mEPSC duration. Similar findings have been illustrated in developing thalamic neurons, in which mIPSCs and mEPSCs were recorded simultaneously (Gottmann et al. 1994). The relative slow rise and decay times of mIPSCs compared with those of mEPSCs are probably indicative of differences in receptor gating kinetics rather than their generation at distal synaptic sites. Moreover, substantial electrophysiological and morphological evidence indicates that in mammalian spinal cords, inhibitory synapses concentrate on motoneuron somata and proximal dendrites (Burke et al. 1971; Örnung et al. 1996).
This study investigated the relative roles of presynaptic action potentials and quantal transmitter release in regulating the frequency and amplitude of spontaneous PSCs during early stages of synapse formation in vivo. Our data suggested that from the onset, action potentials invading presynaptic terminals contributed significantly to the frequency of spontaneous PSCs. However, the developmental increase in spontaneous PSC frequency was independent of presynaptic action potentials and was at least partially the result of a higher frequency of quantal release. Most mPSCs in embryonic and postnatal motoneurons consisted of single unitary currents, but synchronous, multiquantal release could be triggered by high-K+–induced presynaptic depolarization in postnatal spinal cords. Our findings indicated that developmental increases in mPSC frequencies and amplitudes were not temporally correlated with appreciable changes in mPSC basic kinetic properties.
We thank Drs. Peter Lipton, Meyer Jackson, and Robert Conhaim for critical comments on the manuscript.
This work was supported by the National Institute of Neurological Disorders and Stroke Grant NS-23808 to L. Ziskind-Conhaim.
Address for reprint requests: L. Ziskind-Conhaim, Dept. of Physiology, 129 SMI, University of Wisconsin Medical School, 1300 University Ave., Madison, WI 53706.