A computational model of intrathalamic signaling via open-loop thalamo-reticular-thalamic architectures

The thalamic reticular nucleus (TRN), a sheet of GABAergic neurons that partially envelops, receives excitatory input from, and projects inhibitory output to the dorsal thalamus, is known to form part of the thalamus9 intrinsic connectivity. In this capacity, the TRN has been shown to play a critical role in shaping physiological phenomena such as spindle-wave and absence-seizure activity, while it is also speculated to contribute to integrative neural functions such as arousal and attention. It was long supposed that pairs of thalamic relay (TC) and TRN neurons formed closed disynaptic loops, in which a TC neuron was inhibited by the TRN neuron it excited. Recent experimental observations and modeling studies, however, support both the existence and potential functional significance of thalamo-reticulo-thalamic (TC-TRN-TC) synaptic motifs, in which neurons from the TRN are not reciprocally excited by the TC neurons they inhibit. We hypothesized that these structural modules, when connected in series, might underlie certain modes of signal propagation from one part of the thalamus to another. In the present study, we sought to evaluate the relative capacities of closed- and open-loop TC-TRN-TC synaptic configurations to support both stimulus-evoked propagation and oscillation, both of which characterize a variety thalamic and thalamocortical waveforms, while simultaneously exploring the possibility that synaptic connections exclusive to the TRN, of which both chemical or electrical varieties have been identified, might cooperatively or separately underlie these wave properties. To this end, we generated and simulated permutations of a small thalamo-reticular-cortical network, allowing select synapses to vary both by class (homogeneously) and independently (heterogeneously), and examined how synaptic variations altered the propagative and oscillatory properties of the stimulus-driven responses arising in the networks. Our analysis revealed that 1) stimulus-evoked signal propagation was best supported in networks possessing strong open-loop TC-TRN-TC connectivity; 2) oscillation arose most commonly though one of two mechanisms, one of which involved periodically occurring post-inhibitory rebound induced by the TRN in the thalamus and required strongly closed-loop TC-TRN-TC motifs and the other of which was characterized by the propagation of oscillatory activity across a network and was dependent on uniformly strong reticulothalamic synapses; 3) intrareticular synapses were neither primary substrates of propagation nor oscillation, tending to interfere with the former and either attenuating or facilitating a weak, nondominant form of the latter; 4) neither the average propagative nor oscillatory efficiency of those network permutations best accommodating these properties significantly changed as a function of altering the duration of a fixed, external stimulus applied to them; and 5) heterogeneously synaptic networks tended to support more robust oscillation than their homogeneous counterparts, while the capacity to support propagation did not depend on the spatial uniformity synaptic weights. We relate these findings to those elucidated by related modeling studies constructed around exclusively closed-loop TC-TRN-TC connectivity and discuss the functional implications of both thalamic architectures relative to experimental data concerning normal and pathological processes in the thalamus.