The precise role of the metabotropic glutamate receptor-mediated (mGluR) component of the corticothalamic EPSP in the regulation of thalamic output is still not fully understood. Using a multi-compartment model of a thalamocortical neurone (Antal et al. 2001), we have now investigated the ability of cortical EPSPs with different sizes of mGluR component to perturb the δ oscillation and to modulate the efficacy of the retinal input. Four spatial distributions of cortical synapses were used: symmetric distal, with synapses distributed among all distal dendrites; symmetric proximal, with synapses on all secondary dendrites; single-dendrite, with synapses on all distal segments of one dendrite; single-segment, with a single distal synapse. Retinal synapses had symmetric proximal location.

The presence of an mGluR component in the cortical EPSP increased its ability to perturb the δ oscillation. Without an mGluR component the cortical EPSP could only change the oscillation cycle when it arrived on the decay phase of the low threshold Ca²⁺ potential. With an mGluR component (▓ge│ 1 mV amplitude), however, it was able to completely reset the oscillation, independently from the time of arrival and synapse location.

The effect of subthreshold cortical EPSPs on retinal EPSPs arriving 10-500 ms after the cortical input was investigated. Cortical EPSPs facilitate retinal transmission when the delay was < 160 ms. The time-window for facilitation increased with the amplitude of the mGluR component, and depended on the location of cortical synapses, with the longer facilitation interval occurring with the single segment distribution. For delays ranging from 160 to 360 ms, the ability of retinal EPSPs to evoke low threshold Ca²⁺ potentials and action potentials decreased, as a retinal EPSP that alone induced a burst of action potentials did not evoke either potentials after a cortical EPSP. With increasing mGluR components, this suppressive effect of cortical ESPs became larger and less dependent on synaptic location.

The work was supported by the NIIF Supercomputer Project (grant 1043), and The Wellcome Trust (grant 37089/98).