1. In Mg2+-free external solution, rat cortical neurons in cultured networks entered a stable firing mode, consisting of regular bursts of action potentials superimposed on long-lasting depolarizations. The average separation between bursts varied from culture to culture, but was usually between 5 and 20 s. The distribution of burst intervals followed a Gaussian or normal distribution, with a standard deviation of typically 10% of the average burst period. 2. A gradually depolarizing pacemaker potential was never observed between bursts, but the threshold for action potentials during the quiescent phase was greater-than-or-equal-to 10 mV above the resting potential. No progressive change in conductance or excitability was observed during the quiescent period. Intracellular stimulation of action potentials did not reproduce the long-lasting depolarization. 3. Switching from current clamp to voltage clamp at the resting potential revealed large postsynaptic currents, mainly excitatory but with a small inhibitory component, at the same phase and frequency as the spike bursts, showing that periodic synaptic input is responsible for the burst-depolarizations. The current could be eliminated by local application of 2-amino-5-phosphonovaleric acid (APV) or 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) to the postsynaptic cell. In the presence of tetrodotoxin, irregular miniature excitatory postsynaptic currents were observed. 4. A fluorescent calcium indicator (fluo-3, 100 muM) was included in the whole-cell pipette solution, to allow simultaneous electrical and calcium measurements in the same cell. In current clamp, transient intracellular calcium increases were found, which were synchronized to the spike bursts. The Ca2+ rise lasted as long as the action potential burst, and was followed by an exponential decay considerably slower than that of the membrane potential. Calcium transients disappeared during voltage clamp at the resting potential, suggesting that calcium influx through voltage-dependent calcium channels greatly exceeds that through synaptic channels. 5. Multisite Ca2+ recording, after loading with fluo-3 acetoxymethyl (AM) ester, revealed that the onsets of burst-related calcium transients were synchronized in all active cells of each view-field, to within approximately 20 ms. Occasionally, secondary rhythms were observed in which only a subset of cells participated. The times to peak and the decay times of calcium transients varied among synchronized cells. 6. The pharmacology of the burst-related calcium transients was investigated by bath application of a variety of compounds. Tetrodotoxin (1 muM) produced reversible inhibition of transients, as did APV (100 muM) and Mg2+ (1 mM), implying that voltage-dependent sodium channels and N-methyl-D-aspartate (NMDA) receptor-mediated channels are essential. Bicuculline (20 muM) and strychnine (20 muM) both produced a reversible increase in the burst period, sometimes with a slight increase in the amplitude of transients, whereas muscimol (10 muM) reversibly arrested the transients. 7. To investigate the timing of action potential firing at many sites, we cultured cells on substrates with embedded arrays of transparent electrodes. Simultaneous recordings at up to eight sites showed extracellular action potential bursts coincident with intracellular action potentials. Differences in burst initiation times of approximately 20 ms were observed at physical separations of approximately 1 mm. The order of burst initiation at different channels and the detailed firing pattern changed from burst to burst, implying that the wave of excitation was initiated randomly. 8. The phase of periodic bursting could be locked to stimulating current pulses passed through individual sites in the electrode array, and periodic bursting could be initiated in silent cultures. 9. These results indicate that the periodic spike bursts and intracellular calcium transients are generated by periodic excitatory synaptic conductance transients, with a slow NMDA receptor-mediated component. The depolarization opens voltage-dependent calcium channels and the resulting calcium elevation persists during the interburst quiescent period. The random direction and timing of excitation and the lack of any observed pacemaker potential in cells could be explained if the prevalent spontaneous miniature excitatory synaptic events and tonically active NMDA channels act as random sources of excitation.
