NMDA Receptor-Mediated Currents are Prominent in the Thalamocortical Synaptic Response Before Maturation of Inhibition

: 1. The N-methyl-D-aspartate subtype of glutamate receptor (NMDAR) is thought to underlie synaptic plasticity in both adult and developing CNS; however, its involvement in the thalamocortical synapse has not yet been directly demonstrated. 2. Whole-cell, thalamus-evoked synaptic currents were recorded from layer IV cells in slices of immature mouse somatosensory

1. The N-methyl-D-aspartate subtype of glutamate receptor (NMDAR) is thought to underlie synaptic plasticity in both adult and developing CNS; however, its involvement in the thalamocortical synapse has not yet been directly demonstrated.
2. Whole-cell, thalamus-evoked synaptic currents were recorded from layer IV cells in slices of immature mouse somatosensory cortex.
3. Earlier than postnatal day 9 the majority of responses were monosynaptic and purely excitatory, with both non-NMDAR and NMDAR-mediated glutamatergic components.
4. In older animals, disynaptic inhibitory currents summated with the excitatory ones and lowered the reversal potential of the response to voltages at which the NMDAR conductance is mostly blocked.
5. These findings suggest a cellular basis for the transient plasticity observed in layer IV during early postnatal development. INTRODUCTION Electrical activity plays a major role in sculpting synaptic circuits in the nervous system. Among the more striking examples are segregation of thalamocortical terminals into eye-specific stripes in cat and monkey visual cortex ( LeVay et al. 1978( LeVay et al. , 1980 and into whisker-specific clusters, associated with structures called barrels, in rodent somatosensory cortex (Killackey et al. 1990). Each of these processes is dependent on normal sensory input during a time-window of plasticity restricted to early postnatal development (Jeanmonod et al. 198 1;Mower et al. 1985). Analogy with other systems (Constantine-Paton et al. 1990) suggests that these phenomena are dependent on activation of the N-methyl-D-aspartate subtype of glutamate receptor (NMDAR).
One would therefore expect NMDAR-mediated activity in the thalamocortical system to be timecorrelated with the period of plasticity. Indeed, evoked single-unit responses in layer IV of kitten visual cortex are NMDAR-dependent initially but lose this dependency after the end of thalamocortical terminal segregation (Fox et al. 1989); other studies in cat (Hagihara et al. 1988) and turtle  suggest that thalamocortical responses in the adult visual cortex are mediated mainly by non-NMDA receptors. The cellular basis for the apparent loss of NMDAR-mediated activity in the adult animal remains unknown because the only reported loss of NMDAR binding sites in layer IV of cat visual cortex occurs several weeks after terminal segregation ( Bode-Greuel and Singer 1989), and no loss has been reported in layer IV of the rodent barrel cortex (Jaarsma et al. 199 1). We utilized a novel slice preparation of the mouse thalamocortical system (Agmon and Connors 199 1) to follow, at the intracellular level, the developing roles of both NMDA and non-NMDA receptors in thalamocortical neurotransmission during early postnatal life.

RESULTS
Whole-cell synaptic currents evoked by electrical stimulation of the thalamus were recorded from 39 layer IV neurons of somatosensory (barrel) cortex in slices taken from 22 mouse pups, 3-17 days old. Responses were recorded as early as postnatal day 3 (P3, PO being the first 24 h after birth), the day on which layer IV differentiates and barrels are formed (Rice and Van der Loos 1977). Note that I-?%.trve is linear at 3 ms but has a zone of pronounced negative slope at 35 ms after response onset. C: pharmacological analysis of an EPSC in a P7 cell. The fast component, which dominated the two most negative potentials in control artificial cerebrospinal fluid ( ACSF), was blocked by 2.5 PM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX). The remaining slow component, which was activated at depolarized potentials, was blocked by 50 PM DL-2amino-5-phosphonovalerate (APV). Both components partially recovered 1 h after return to control ACSF. Vertical scale is 200 pA for control, 100 pA for the other panels.
positive potential, identifying it as an excitatory postsynaptic current (EPSC). Its short and invariant latency indicated that it was a direct (monosynaptic) thalamocortical response ( Agmon and Connors 1992). The current-voltage (Z-V) relationship of the response (Fig. 1 B) was linear at early times after response onset (0) but exhibited a region of negative slope conductance typical of NMDAR-mediated currents (Hestrin et al. 1990) at late time points (0). The involvement of both NMDA and non-NMDA receptors in the neonatal response was tested pharmacologically in six cells (Fig. 1 C) . In all tested cells 6-cyano-7-nitroquinoxaline-2,3-dione (RBI) (CNQX), a specific non-NMDAR antagonist, blocked the fast component of the response, and DL-2-amino-5-phosphonovaleric acid (RBI) (APV), a specific NMDAR antagonist, blocked the remaining slow component. We conclude that, like various other monosynaptic pathways in the CNS (e.g., Hestrin et al. 1990), the neonatal thalamocortical response is a dualcomponent EPSC, mediated by both non-NMDA and NMDA receptors.
The majority (62%) of neonatal cells exhibited a pure excitatory response. In contrast, none of the juvenile cells exhibited responses that were exclusively excitatory. Rather, 65% of them, as well as the remaining 38% of neonatal cells, exhibited a composite, biphasic response. This response consisted of two distinct components, the onset of the second delayed by 2-3 ms relative to the first and thus identified as disynaptic ( Fig. 2A, arrows). The reversal potential of the monosynaptic component of the composite response [ 18 t 20 (SD) mV, n = 221 was not significantly different from that of the neonatal EPSC ( 13 t 11 mV, n = 8) and did not show any age dependency. The reversal potential at the disynaptic peak, however, was significantly more negative, indicating activation of an inhibitory post-  Fig. 2. Increasing the frequency of stimulation suppressed the IPSC and revealed a nearly pure EPSC with a pronounced voltage-dependent component (Fig. 2B) that was blocked by APV (not shown). Digital subtraction of the high-frequency from the low-frequency response revealed the inhibitory component in isolation (Fig. 2C). The reversal potential at the disynaptic peak (Fig. 20, 0) was intermediate between the reversal potentials of the pure excitatory (A) and pure inhibitory (0) components. We conclude that the juvenile thalamocortical response is a composite synaptic response, consisting of a monosynaptic dual-component EPSC followed by a disynaptic IPSC and that the time course of the IPSC overlaps considerably with that of the NMDAR-mediated component of the EPSC. The reversal potential at the disynaptic peak exhibited pronounced age-dependent changes. Figure 3 shows the reversal potential at the disynaptic peak for all 22 cells with a composite response ( n ) and, for comparison, the reversal potential at equivalent time points for all 8 cells with a pure EPSC (0). In P9 and younger animals the disynaptic reversal potential values were widely scattered around a low nega- values to -58 t 14 mV (n = 11) in PI 1 and older animals.
Because of the voltage-dependent blockade of the NMDAR, at voltages below -60 mV less than 10% of the total NMDAR-mediated conductance is available for activation (Hestrin et al. 1990). Since under physiological conditions inhibitory inputs in the neocortex tend to bring the membrane potential of the cell very close to the inhibitory equilibrium potential (Connors et al. 1988 ) , we conclude that in PI 1 and older animals activation of the thalamocortical synapse will result in very little recruitment of NMDAR-mediated currents in layer IV.  3. Age-dependent change in the reversal potential at the disynaptic peak. Reversal potentials were measured at the disynaptic peak for all 22 cells that exhibited a composite response ( n ) and at equivalent time points after response onset for all 8 cells with a pure monosynaptic response ( q ) . Nine additional juvenile cells exhibited a pure or nearly pure inhibitory postsynaptic current and are not included. DISCUSSION Our results indicate that thalamocortical synaptic responses can be elicited in mouse layer IV neurons as early as p3, and in deeper layers as early as PO (Agmon and O'Dowd 1990), a week earlier than single-unit responses were previously recorded in the rat (Armstrong-James 1975 ) . These responses have both NMDAR and non-NM-DAR-mediated components. NMDAR-mediated spontaneous and evoked activity was previously reported in neonatal rat neocortex ( LoTurco et al. 199 1;Yuste and Katz 199 1 ), and sensory stimulation evokes NMDAR-dependent responses in kitten visual cortex (Fox et al. 1989;Tsumoto et al. 1987 ) . Our data localize the NMDAR-mediated activity to the thalamocortical synapse and provide intracellular evidence that thalamocortical neurotransmission is mediated by both major types of excitatory amino acid receptors. The NMDAR-mediated component of this synapse is most prominent during the first postnatal week and is thus present at the appropriate place and time to participate in segregation of thalamocortical terminals and in morphogenesis of barrels.
Consistent with previous reports from cat (Komatsu 1983 ) and rat ( Luhmann and Prince 199 1) neocortex, we found that maturation of inhibitory synaptic responses was delayed relative to excitatory ones; however, in layer IV we encountered immature disynaptic IPSCs as early as P4, again -1 wk earlier than previously reported in the rat. The IPSCs apparently mature in the first one-half of the second postnatal week, as can be judged by the steep negative shift in the reversal potential of the disynaptic peak. Part of this shift could have been due to developmental changes in the inhibitory equilibrium potential (Luhmann and Prince 199 1); however, the major factor was most likely an increase in inhibitory conductance, due to establishment and / or maturation of corticocortical inhibitory synapses ( Lund and Harper 199 1;Miller 1986). By mid second postnatal week, IPSCs were strong enough to bring the reversal potential of the thalamocortical response below threshold for NMDAR activation and thus presumably reduce the capacity of the thalamocortical synapse to undergo anatomic and physiological reorganization. Indeed at least one study (Seo and Ito 1987) suggests that the barrel cortex loses its capacity for reorganization around PI 0, coincident with the maturation of inhibition described here. A link between maturation of inhibition and loss of polysynaptic NMDAR-mediated activity in upper layers of rat neocortex has previously been proposed (Luhmann and Prince 1990a,b); our data suggest that a similar process occurs, -1 wk earlier, in the thalamocortical synapse in layer IV. One corollary of this hypothesis is that even beyond the age range studied here, any process that reduces the efficacy of inhibition may cause unblocking of the NMDAR-mediated component and thus potentially restore synaptic plasticity in the adult animal.