1. The electroresponsive properties of neurons from layer II of the rat medial entorhinal cortex (MEC) were studied by intracellular recording under current clamp in an in vitro brain slice preparation. From a total of 184 cells that fulfilled our criteria for recording stability, two groups of projection neurons were distinguished on the basis of their intrinsic biophysical properties and morphological characteristics (demonstrated by intracellular biocytin injection; n = 34). 2. Stellate cells (SCs) were the most abundant (69%). They were highly electroresponsive, and minimal changes (1-3 mV) of membrane potential generated an active response. Subthreshold depolarizing or hyperpolarizing current pulse injection always caused the membrane potential to attain an early peak and then sag to a lower level. Depolarization-induced ''sags'' were larger and determined early firing in all cells. The voltage-current relationship of SCs was markedly non-linear, demonstrating robust inward rectification in the hyperpolarizing and depolarizing range. 3. SCs generated persistent rhythmic subthreshold voltage oscillations on DC depolarization positive to -60 mV. The mean frequency of the oscillations was 8.6 Hz (theta range) at a membrane potential of approximately -55 mV, at which level occasional single spiking also occurred. At slightly more positive potentials, a striking 1- to 3-Hz repetitive bursting pattern emerged. This consisted of nonadapting trains of spikes (''clusters'') interspersed with subthreshold oscillations that had a mean frequency of 21.7 Hz (beta range). 4. Nonstellate cells (39%; mostly pyramidal-like) displayed time-dependent inward rectification that was less pronounced than that of SCs, and minimal depolarization-induced sags. On threshold depolarization, firing was always preceded by a slowly rising ramp depolarization and thus occurred with a long delay. Inward rectification in the depolarizing range was very pronounced. However, non-SCs did not generate persistent rhythmic subthreshold oscillatory activity or spike clusters. 5. Of the electrophysiological parameters quantified, spike threshold, spike duration, depolarizing afterpotential amplitude and apparent membrane time constant demonstrated statistically significant differences between SCs and non-SCs. 6. The repetitive firing properties in response to square current pulses of short duration (<500 ms) were also different between SCs and non-SCs. First, most SCs displayed a bilinear frequency-current (f-I) relationship for only the first interspike interval, whereas most non-SCs displayed a bilinear relationship for all intervals. Second, SCs had a much steeper primary f-I slope for early intervals than non-SCs. Finally, SCs displayed more pronounced and faster spike frequency adaptation than non-SCs. Moreover, current pulse-triggered spike trains were followed by an afterhyperpolarization of larger amplitude and more complex waveform in the SCs than in the non-SCs. 7. Changes in spike trajectory during current-pulse triggered spike trains also followed a different pattern in SCs and non-SCs. In SCs, the action potential duration increased during the first three spikes and then progressively decreased to the initial value, although pronounced adaptation continued. Non-SCs behaved more similarly to other cortical neurons, with a progressive increase in spike duration during the adapting train. 8. Thus fundamental differences exist in the intrinsic electrophysiological architecture of SCs and non-SCs and, therefore, in the way these neurons transduce synaptic input into spike output toward the hippocampal formation via the perforant path. Accordingly, two parallel channels of information processing with different integrative properties may exist in layer II of the MEC. The SC channel would constitute a highly rhythmic processing system with striking pacemaker properties that may participate in the generation of theta and/or beta rhythmicities in the limbic system.
