The intrinsic membrane properties of thalamic neurones, the ionic conductances underlying them, and the mechanisms responsible for oscillatory behaviour of thalamic neurones in the thalamo-cortico-thalamic circuit have been extensively characterised both in vivo and in vitro, and recent reviews have covered these aspects of thalamic physiology at length and they will not therefore be repeated here (McCormick 1992; Steriade & Llinas, 1988; Steriade 1993; Steriade et al., 1993). However, it is evident that the interaction of synaptic ligand-gated conductances with the membrane potential-dependent intrinsic membrane conductances will be one of the principal determinants of the firing pattern of thalamic neurones evoked by sensory stimuli.
Burst firing patterns arise following membrane hyperpolarisation sufficient to de-inactivate a low threshold Ca2+ current, IT. This results in a depolarising Ca2+ spike (low-threshold spike: LTS) which can depolarise the membrane potential to the threshold to elicit one or a burst of conventional Na+-action potentials: this is often called a burst of rebound spikes (Jahnsen & Llinas, 1984; Steriade & Llinas, 1988). Spontaneous or sensory-evoked inhibitory input to thalamic relay neurones in vivo can evoke IPSPs which can provide the hyperpolarising stimulus which generates low threshold burst spiking activity. During EEG-synchronised states (e.g. slow wave sleep, barbiturate or urethane anaesthesia) relay neurones can receive a spontaneous barrage of IPSPs at 7-14Hz every 0.1 - 0.3Hz (Steriade & Deschênes, 1984). There is substantial evidence to suggest that these IPSP sequences are derived from the nrt, which acts as a pacemaker, sending a rhythmic sequence of bursts of action potentials into the thalamus. IPSP-elicited low threshold Ca2+ spikes can also be evoked by sensory stimuli. Thus, in the VB short duration stimulation of the peripheral receptive field elicits a short latency EPSP plus one or two spikes and is followed by a sequence of one or more nrt-derived IPSPs, which can give rise to one or more LTS and associated action potentials (Salt & Eaton, 1990), and similar rebound spikes are evident following visual stimulation in the LGN (Lo et al., 1991). It is important to recognise that two fundamentally different mechanisms are responsible for generation of the short latency and rebound spikes. Glutamate receptor-mediated conductances underlie the EPSPs which evoke the short latency spike(s) (see section 5.1), whilst the intrinsic membrane properties, de-inactivated by an IPSP, generate the LTS and associated rebound spike(s). Nevertheless, the LTS can contribute to the short latency response if the membrane potential is sufficiently hyperpolarised (Salt & Eaton, 1990; Turner et al., 1994).
In relay mode, the interplay of intrinsic membrane conductances prevalent in this state results in a characteristic tonic, non-inactivating firing pattern of action potentials. Such firing patterns can be experimentally produced by simple injection of sufficient depolarising current (Steriade & Llinas, 1988). One of the membrane conductances involved in the generation and maintenance of tonic firing patterns is the persistent, non-inactivating Na+ conductance, NaP, which is activated in a regenerative fashion by depolarisation (Jahnsen & Llinas, 1984). Excitatory glutamate receptor-mediated synaptic input to the thalamic relays can elicit such tonic firing patterns, typically evoked by long duration stimulation (Salt, 1986, 1987; Sillito et al., 1990a, Turner et al., 1994), and thus it is feasible that synaptic glutamate receptor-mediated depolarisation activates NaP during sensory responses and NaP contributes to sensory-evoked responses in the relay thalamic nuclei. Both NMDA receptors and AMPA/kainate receptors contribute to the mediation of this type of sensory response (section 5.1). It is particularly notable that NMDA receptors make a very substantial contribution to maintained responses in the thalamic relays, including responses to visual stimuli in the LGN and the response of neurones in the VB to noxious and non-noxious somatosensory stimulation (section 5.1). In several other brain regions (Eaton & Salt, 1991; Flatman et al., 1983; Herrling et al., 1983), NMDA receptor activation by exogenous agonists evokes a burst firing pattern, which is likely to result from the unusual voltage-dependency associated with the well known interaction of Mg2+ with NMDA receptors. In the thalamus, whilst Mg2+ appears to normally gate NMDA receptors, agonists acting at NMDA receptors or AMPA/kainate receptors (Eaton & Salt, 1991) and at metabotropic glutamate receptors (McCormick & Von Krosigk, 1992; Salt & Eaton, 1991b and unpublished observations), elicit a tonic firing of spikes. We have previously suggested that the intrinsic membrane properties of thalamic neurones may take precedence over the ligand gated conductance properties in the determination of such tonic firing patterns in the thalamus (Eaton & Salt, 1991). The functional significance of this dominance of the intrinsic membrane conductances over the properties of ligand-gated conductances may be to maintain the fidelity of NMDA receptor-mediated response components through the thalamus when thalamic neurones are in relay mode, since the introduction of burst firing patterns by NMDA- receptor-mediated sensory input could disrupt the integrity of thalamocortical sensory information flow through the thalamic relays. Considering the marked utilisation of NMDA receptors in sensory responses in the thalamus, this mechanism may represent an important feature of sensory processing at the level of the thalamus. Similarly, at hyperpolarised membrane potentials, the dominance of intrinsic membrane properties will govern responses of the relay neurones, but again, such responses can be triggered by glutamate-gated currents (Leresche et al., 1991; Soltesz et al., 1991; Turner et al., 1994).
Several amine transmitter systems (section 6.2) and L-glutamate acting via Group I mGluRs can regulate the same K+ leak current via intracellular second messenger systems, and this represents a final common path with a maximal response (McCormick 1992). This K+ current is important in biasing the thalamic neurone membrane towards either relay (tonic firing) mode or burst mode (McCormick 1992; Steriade et al., 1993). Thus, activity in any of these transmitter systems can bias the neurone towards tonic firing, but in different ways: for example, the cholinergic system will cause widespread activation, whereas the glutamatergic system is likely to provide a more discrete regulation of this system. This would represent a type of 'OR' gate, so that cells could be switched into relay mode either globally or discretely, depending upon circumstances.
The importance of the voltage-dependent Mg2+ block of the NMDA channel is also difficult to assess. The resting potential of thalamic neurones in vivo is likely to be in the range -55 to -65 mV, at which level the block of the NMDA channel is thought to be substantial (Mayer et al., 1984; Nowak et al., 1984). It is however evident from many studies that NMDA can evoke depolarising responses in the thalamus at such membrane potentials and that exogenous Mg2+ can block NMDA responses, thus indicating that the Mg2+ channel block is not complete under these conditions (Eaton & Salt, 1991; Leresche et al., 1991; Turner et al., 1994). It is possible that, during synaptic activation of NMDA receptors, the depolarisation leads to some relief of the Mg2+ blockade, although this may not be very large when the level of maintained depolarisation upon repetitive stimulation is considered (Turner et al., 1994). Whether the membrane potential changes recorded at the soma are an accurate reflection of potential changes in the dendritic tree/spines of thalamic neurones is not known, and it may be that the potential changes in these regions of the neurone are such that relief of Mg2+ blockade is of functional significance.
Less is known about the mechanisms of action of the cortico-thalamic input, although it is reasonable to assume that it is glutamatergic, possibly acting at least via NMDA receptors and metabotropic receptors. Both of these mechanisms are ideally suited to provide a modulatory input onto thalamic cells which could enhance other depolarising inputs (e.g. those mediated by AMPA receptors originating from sensory afferents) without having large effects in their own right. The function of NMDA receptors would again be critical on the type of subunit composition and the local membrane potential (see above).
The cortico-thalamic input could be the basis of selective enhancement or gating of sensory transmission through the thalamus, a concept which has been suggested by several groups of workers (Koch, 1987; McCormick & Bal, 1994; Mumford, 1991; Sillito et al., 1994; Singer, 1977). An interesting question is why the cortico-thalamic projection would impinge upon both NMDA receptors and mGluRs. It could be that these receptors are activated under different conditions of input, for example different frequencies and/or durations of activation. It is certainly evident that mGluR-mediated EPSPs are of longer duration than EPSPs mediated by ionotropic receptors (Batchelor et al., 1994; Gerber et al., 1993), and thus the different receptors could contribute to different time courses of activation. A further distinction is however also possible, as the postsynaptic mGluR response results in a decreased membrane conductance: this will not only tend to enhance responsiveness to ionotropic glutamate receptor inputs, but will also enhance the effects of GABAergic inhibitory inputs which are mediated via increased Cl- and K+ conductances (Crunelli & Leresche, 1991). Thus activation of mGluRs on thalamic neurones will allow the GABAergic influence from nrt and Golgi type II neurones (where present) to exert a greater influence, and this would also be of potential importance in the selective gating of thalamic sensory transmission. The precise physiological conditions under which such inputs are activated remain to be elucidated. However it is evident that certain types of sensory stimulation protocols may be appropriate for activation of the cortico-thalamic circuits, as indicated by the cortical inactivation protocols and cross-correlation analysis work (Sillito et al., 1994).
It has recently been shown on cerebellar neurones that AMPA/kainate receptors and mGluRs are located adjacent to each other in such a way that suggests that AMPA/kainate receptors are activated in the synaptic cleft, and that mGluRs are activated by glutamate if sufficient glutamate is released from the pre-synaptic terminal (Nusser et al., 1994). If this is the case in other brain areas, such as the thalamus (Kharazia et al., 1995), then it would offer a mechanism for the differential activation of ionotropic glutamate receptors and mGluRs in the cortico-thalamic system, and open up the possibility of pattern-dependent modulation of thalamic transmission by the sensory cortex. Clearly, it is important to determine the relative distributions of glutamate receptor types on different parts of the thalamic neurones in order to progress in this area. It would be particularly interesting if, for example, different NR2 subtypes were differentially located subsynaptically opposed to cortico-thalamic terminals and sensory afferents. A speculative scenario might be that NR1-NR2A receptors serve sensory afferent input whereas NR1-NR2B serve the cortical input: this could provide the basis for EPSPs with faster decay for the sensory input and EPSPs with slower decay which might be more appropriate for a modulatory input from the cerebral cortex.
The role of glutamate receptors in the mediation of excitatory input to GABAergic inhibitory neurones in the thalamus has been discussed in section 5.2. It is becoming evident, however, that metabotropic glutamate receptors may have a function in directly modulating inhibitory transmission. This may be via presynaptic metabotropic receptors on GABAergic terminals (see section 5.3). The source endogenous agonist which may activate presynaptic mGluRs on GABAergic terminals is also unclear. In the absence of direct innervation by glutamatergic terminals, other possibilities such as glia and volume transmission might be considered (see section 7.3). Another intriguing possibility is that presynaptic mGluRs may be activated by NAAG, which recently has been shown to activate mGluR3 (Wroblewska et al., 1993, 1995) and which is known to be localised in nrt neurones (Henderson & Salt, 1988). It is thus possible that NAAG is released from GABAergic terminals together with GABA, and thus exerts a presynaptic action on what are in effect autoreceptors. This hypothesis remains to be tested.
The GABAergic inhibitory processes in the thalamus are of importance in maintaining the temporal and spatial representations of sensory stimuli within the thalamus, and are important in contributing to the network properties of the thalamus during bursting activity in slow-wave sleep and absence seizures (section 7.1). Thus, specific modulation of inhibitory transmission by mGluRs is likely to reduce stimulus-specific inhibition, and thus possibly stimulus-discrimination. Furthermore, reduction of inhibition will also reduce the tendency to burst firing in the thalamus, and thus regulation of the presynaptic mGluRs could influence the overall state of the thalamus in transition from sleep to arousal.
It is evident from the above that there are great similarities in transmitter mechanisms across the various thalamic nuclei and species. It is thus likely that many common principles of operation will apply to the various thalamic relay nuclei, even though they receive their afferent inputs from different modalities and sub-modalities. However, it is conceivable that the thalamic circuitry and transmitter mechanisms are utilised in different ways by different modalities. For example, the nociceptive somatosensory response of certain VB neurones is largely dependent upon NMDA receptors and Group I mGluRs rather than on AMPA/kainate receptors (Eaton & Salt, 1990; Eaton et al., 1993). These responses are relatively slow in onset and of long duration as might be expected from responses showing such a synaptic pharmacology. Corresponding thalamic responses of other sensory modalities have not been described. However, it has been suggested that the nociceptive responses of some thalamic neurones may be dependent upon the recruitment of a cortico-thalamic input (Condes-Lara & Zapata, 1988; Nothias et al., 1988; Salt & Eaton, 1995b), and thus it may be that the dominance of mGluR and NMDA receptors in this response reflects a cortical input (Salt & Eaton, 1995b). Furthermore it is possible to speculate that such use of circuitry and receptors reflects a specialisation of the thalamic nociceptive system which may be the equivalent of the 'selective attention' mechanisms which have been proposed for other modalities.
It has been suggested recently that activation of AMPA/kainate receptors on astrocytes can release D-serine from these glia so that it can enhance neuronal NMDA-receptor-mediated responses (Schell et al., 1995). The presence of D-serine and D-serine binding sites in the thalamus (Schell et al., 1995) and the enhancement of NMDA receptor mediated responses in vivo in the thalamus by D-serine (Salt, 1989), suggests that this mechanism may be important in thalamic transmission. It is possible that this allows enhancement of NMDA-mediated transmission during repetitive input onto AMPA/kainate receptors, and this may partly explain why responses to repetitive activation are so dependent on both NMDA and non-NMDA ionotropic receptors.
The release of arginine in the thalamus in vivo upon sensory afferent stimulation may also reflect the participation of glia in synaptic processes (Do et al., 1994). It is known that arginine in the thalamus is located in astrocytes, and it appears that the post-synaptic activation of VB relay neurones via ionotropic glutamate receptors is necessary for the release of arginine (Salt et al., 1995). Exogenously-applied L-arginine can enhance thalamic relay cell responses, apparently via the synthesis of NO (Do et al., 1994). The increase in extracellular arginine levels may represent a transfer of arginine between two cellular compartments (e.g. from glia to brain stem afferents, which are known to contain NO-synthase (section 6.4)), in order to supply the NO synthase with its substrate. It is thus interesting to speculate that the release of NO may be a local modulatory system which, at least in the thalamus, enables rapid enhancement of synaptic transmission upon stimulation of sensory afferents: in effect a positive feed-back system. If this is the case, then the local NO system could act to facilitate transmission through a restricted part of the thalamus in order to enhance transmission of an afferent signal which may be deemed important because of its persistence. It is intriguing that the cholinergic fibres ascending from the brainstem may be the source of NO (section 6.4): this would represent a dual function for these cholinergic neurones: firstly to provide widespread cholinergic modulation of thalamic transmission, and secondly to provide the substrate for a more local modulation of transmission which may be independent of, or related to, the first function.
Clearly, we are only beginning to appreciate the widespread role of glia in transmission, and the presence of functional receptors on these elements is indicative that they do not have a merely supportive role (Blankenfeld & Kettenmann, 1992; Chiu & Kriegler, 1994). Furthermore, the widespread nature of glial processes in the ensheathing of terminals and dendrites indicates an intimate role in synaptic transmission, not only in the thalamus, but in all brain areas (Blankenfeld & Kettenmann, 1992). This may be by the release of neuro-active substances onto neurones (as in the case of D-serine) or via the control of access to receptors by glial processes.
Please note: this document is part of the HTML version of a paper
originally published in print in Progress in Neurobiology.
Contents 
Previous Section 
Reference List
Return to Tom Salt's Home Page