The neocortex of the brain develops from a simple germinal layer into a complex multilayer structure. To investigate cellular interactions during early neocortical development, whole-cell patch clamp recordings were made from neuroblasts in the ventricular zone of fetal rats. During early corticogenesis, neuroblasts are physiologically coupled by gap junctions into clusters of 15 to 90 cells. The coupled cells form columns within the ventricular zone and, by virtue of their membership in clusters, have low apparent membrane resistances and generate large responses to the inhibitory neurotransmitter gamma-aminobutyric acid. As neuronal migration out of the ventricular zone progresses, the number of cells within the clusters decreases. These clusters allow direct cell to cell interaction at the earliest stages of corticogenesis.

We have made patch-clamp recordings from slices of fetal and postnatal rat neocortex in order to study the initial expression and activation of NMDA channels. Recordings from both whole cells and outside-out patches indicated that functional NMDA channels are expressed on neurons within the cortical plate, but not on younger cells within the ventricular zone. The NMDA channels on cortical plate neurons had a unitary conductance of approximately 40 pS, had a mean open time of approximately 6 msec, required glycine to open, and were blocked in a voltage-dependent manner by magnesium. These precocious channels were present before the appearance of functional synaptic activity, yet like NMDA channels in the mature neocortex, they were spontaneously activated by an agonist within brain slices. These results demonstrate that NMDA channels are initially expressed on neocortical neurons some time between the last mitotic division within the ventricular zone and completion of migration into the cortical plate. These early NMDA channels have properties characteristic of NMDA channels on more mature neurons and are similarly activated by an endogenous agonist in situ. Their early appearance and activation indicate that NMDA channels may play a role during early stages of cortical development.

The electrophysiological properties of neurons in the three-layered dorsal cortex of the turtle, Pseudemys scripta elegans, have been studied in vitro. Intracellular recordings suggested two distinct classes of neuronal behavior. Cell labeling with either Lucifer Yellow or horseradish peroxidase revealed that these behaviors correlated with the two morphological classes of cortical neurons: pyramidal cells and stellate cells. Examination of Golgi-stained neurons of dorsal cortex did not uncover any other obvious classes. Pyramidal cells had their somata in the cell layer, and extended several densely spined apical dendrites through the molecular layer to the pia. They also had spiny basilar dendrites directed through the subcellular layer toward the ependymal border. Physiologically, pyramidal cells had relatively prolonged action potentials that showed marked frequency adaptation during a sustained suprathreshold current pulse. Their most striking characteristic was a tendency to fire two discrete sizes of action potential, one small (mean = 34 mV) and of relatively low threshold, the other large (mean = 76 mV) and of higher threshold. We hypothesize that at least some small spikes arise from distal dendritic sites, whereas large spikes are somatically generated. Both spikes were tetrodotoxin-sensitive, although calcium-dependent electrogenesis occurred when potassium channels were blocked. In contrast to pyramidal cells, the somata of stellate cells were found in the molecular and subcellular zones. Their dendrites tended to be horizontally oriented and spine-free. Stellate cells had relatively brief action potentials, each of which was followed by a large but short-lasting undershoot of membrane potential. Stellate cells showed little or no spike frequency adaptation. Spike amplitudes were always relatively uniform and large (mean = 73 mV). Thus, in the dorsal cortex of turtles, the pyramidal cells, which are projection neurons, and stellate cells, which are local GABAergic inhibitory neurons, have distinctly different membrane characteristics. The physiological properties of the two types of turtle cortical neurons are very similar to their counterparts in cortical structures of the mammalian telencephalon.

Neurons in "mature" (4- to 6-week-old) dissociated cell cultures of 15-day gestational age rat fetal cortex were injected with Lucifer Yellow in order to compare their detailed morphological features with those of cortical neurons in situ, and in order to determine which features of cellular morphology were dependent on local environmental conditions. Neurons were characterized by their cell form (pyramidal, multipolar, fusiform, etc.), dendritic branching pattern, spine density, and axonal projections. The neurons in culture appeared to display all the morphological features seen in cortical neurons in situ. These characteristics appeared to be independent of whether an individual neuron grew in a dense or sparse region of the culture. In addition, examination of neurons during early differentiation indicated that many of their morphological features developed as soon as the neurons could be recognized and before extensive synapse formation occurred.

We have examined the synaptic physiology of the isolated dorsal cortex of the turtle, Pseudemys scripta elegans. Electrical stimulation of afferent pathways elicited distinct, stereotyped responses in pyramidal and stellate neurons. Single shocks evoked a long-lasting barrage of excitatory postsynaptic potentials (EPSPs) in stellate cells, and led to a burst of several action potentials. Under the same circumstances, pyramidal cells displayed a small amount of short-latency excitation, but this was accompanied by a profound and prolonged set of inhibitory post-synaptic potentials (IPSPs). Synaptic excitation of the distal dendrites of pyramidal cells could evoke dendritic action potentials that were visible at the soma as small all-or-none spikes rising from the hyperpolarized level of the IPSP. There appeared to be two mechanistically different types of IPSPs in pyramidal cells. The first occurred at short latency, could produce a very large conductance increase, reversed polarity at -71 mV, and was chloride-dependent. The second was generally smaller and more protracted, had a relatively negative reversal potential of -85 to -95 mV, and was insensitive to chloride injection. Focal application of small doses of the putative inhibitory neurotransmitter gamma-aminobutyric acid (GABA) onto the somata of pyramidal cells caused a conductance increase and hyperpolarization. This response had features in common with the short-latency IPSP, including an identical reversal potential. Application of large doses of GABA to the somata of pyramidal cells or smaller doses to their dendrites elicited multiphasic or purely depolarizing responses that were at least partly due to time- or space-dependent shifts of the equilibrium potential of the response. Bicuculline methiodide, a potent GABA antagonist, depressed both the responses to GABA and the short-latency IPSP, but not the long-latency IPSP; synchronized epileptiform burst discharges also resulted. These findings, together with responses to locally applied electric shocks and the excitatory amino acid glutamate, suggested that inhibition of pyramidal cells was generated intrinsically by stellate cells, and that the cortical circuit provides pathways for both feedforward and feedback GABAergic inhibition. The data also suggest that pyramidal cells are mutually excitatory. These features are similar to the basic intrinsic circuitry in the telencephalic cortices of mammals.