INFLUENCE OF VPM AFFERENTS ON PUTATIVE INHIBITORY INTERNEURONS IN S1 OF THE AWAKE RABBIT - EVIDENCE FROM CROSS-CORRELATION, MICROSTIMULATON, AND LATENCIES TO PERIPHERAL SENSORY STIMULATION

1. Responses of thalamocortical projection neurons and suspected corti cal interneurons (SINs) to very brief peripheral stimuli were examined within the vibrissae, the sinus hair, the lip, and the chin represent ations of ventroposterior medial thalamus (VPM) and primary somatosens ory cortex (S1). VPM thalamocortical neurons (N = 40) were identified by their antidromic activation after electrical stimulation of S1. SIN s were identified by a high-frequency (>600 Hz) burst of three or more spikes elicited by suprathreshold stimulation of one or more afferent pathways. SINs also had spikes of very short duration. 2. Previous wo rk has shown that electrical stimulation of VPM elicits a very early a nd powerful synaptic response in many S1 SINs. Three experimental stra tegies were employed to test the hypothesis that such responses reflec t a monosynaptic VPM input onto SINs and to examine the effects of suc h input. 1) After a brief peripheral stimulus, the arrival times of VP M thalamocortical impulses in S1 were determined and compared with the initial response times of S1 SINs. 2) Shift-corrected cross-correlogr ams (CCGs) were constructed from the spike trains of pairs of VPM neur ons and SINs that were in precise topographic alignment. 3) Inferences of connectivity based on such CCGs were supported by applying very lo w-intensity (1-10 mu A) microstimulation pulses to the recording micro electrode in VPM and observing evoked responses in the cortical SIN. 3 . VPM thalamocortical neurons responded to a brief air puff stimulus a t a median latency of 5.05 ms, and the estimated arrival time of the V PM impulses at S1 had a median value of 5.97 ms. This estimate was obt ained by adding the antidromic latency of each VPM neuron to the laten cy of the peripheral stimulus and was supported by similar values obta ined from three VPM thalamocortical axons recorded near their terminat ion site within S1. SINs of S1 were among the first cortical neurons t o respond to the peripheral stimulus, responding to the air puff at a median latency of 6.6 ms (range 5.7-13.0 ms). The latency of SINs to t he peripheral stimulus was strongly related to the latency to moss ele ctrical stimulation of VPM (median value 1.52 ms, r(2) = +0.44, P < 0. 0001). 4. Many SINs(23 of 34) showed significant shift-corrected CCGs with VPM neurons that were in precise topographic alignment. Most sign ificant CCGs revealed a very brief increase in SIN spike probability ( half-amplitude response of similar to 1 ms) that reached a peak value at intervals of 1.4-2.0 ms after the VPM spike. Although the ''efficac y'' of VPM spikes in eliciting SW spikes was usually low (median 1.4%, range 0.6-6.7%), the reliability of this measure taken from repeated data sets was high (r(2) = 0.83, P < 0.0001). Similarly, the ''contrib ution'' of spikes elicited by a single VPM neuron to overall spikes of the SIN was low (median 1.9%, range 0.5 -4.8%), and was found to be r elatively stable over repeated measures (r(2) = 0.82, P < 0.0001). 5. The presence of a significant CCG for a given VPM-SIN pair was signifi cantly related to 1) a very short synaptic latency of the SIN to gross electrical stimulation of VPM, 2) a very short synaptic latency of th e SIN to air puff stimulation, and 3) an intermediate cortical depth. Thus significant CCGs were observed in nearly all (23 of 25) of those SINs that responded at very short latencies to both the peripheral sen sory stimulus (<7 ms) and electrical stimulation of VPM (<1.8 ms) and in 18 of 19 of those SINs estimated to lie either within or in close p roximity to layer 4. 6. A second type of SIN, with distinct properties , showed no sign of monosynaptic VPM input. Thus none of the 11 SINs t hat responded either to the VPM electrical stimulus at latencies of >2 ms or to the the air puff at latencies of >7 ms showed significant CC Gs with VPM neurons. Most of those SINs failing to show significant CC Gs with topographically aligned VPM neurons were located either more s uperficially or deeper in microelectrode penetrations than those that did shown cross-correlations. Previous studies have shown that SINs wi th such properties are very responsive to corticocortical input. 7. Al l SINs showing significant CCGs with VPM neurons responded to microsti mulation pulses delivered via the VPM recording electrode (N = 22, mea n threshold 4.9 mu A). In contrast, only four of the nine SINs that fa iled to show significant cross-correlations responded to microstimulat ion at intensities as high as 10 mu A. SINs that failed to show cross- correlations with VPM neurons, but that did respond to VPM stimulation had higher threshold intensities (mean 8.5 mu A) than those that were involved in cross-correlations (mean 4.9, P < 0.001). 8. Four conclus ions may be drawn from these results concerning the relationship betwe en VPM afferents and the SINs of S1 vibrissa cortex: 1) most of the mo nosynaptic input onto SINs occurs either within or in close proximity to layer 4, 2) virtually all of the SINs of this depth receive monosyn aptic input from VPM, 3) individual SINs of a layer 4 barrel receive a highly convergent input from a large proportion of VPM afferent fiber s of the corresponding barreloid, and 4) many VPM afferent fibers ente ring an S1 barrel diverge to synapse on nearly all of the SINs in the vicinity of layer 4. 9. Such a richly divergent and convergent network approaches the description of a ''complete transmission line.'' This network is characterized by a very ''high reliability,'' but at a cost of sacrificing ''complexity of task.'' The present and previous studi es have shown that S1 SINs indeed respond very reliably to peripheral stimulation, showing both very low thresholds and a very high temporal resolution. They appear to have sacrificed complexity of task, howeve r, in that they lack the directional selectivity seen in many VPM affe rents. The present results, therefore, are supportive of a very fast, highly reliable, sensitive, and temporally acute feed-forward inhibito ry mechanism within S1.