Cross-correlation and microstimulation: complementary tools in the extracellular analysis of synaptic interactions

Abstract

The utility of extracellular microstimulation as an ancillary tool to compliment cross-correlation analyses is explored. This technique aims at demonstrating the ‘sufficiency’ of the spike train generated by the ‘reference’ neuron in eliciting spikes in the ‘target’ neuron. It involves activating the reference neuron with very low-intensity current pulses (1–10 μA) delivered through the recording microelectrode. Microstimulation can provide independent evidence that can either support or refute notions of synaptic connectivity generated by correlation analyses. Data are presented from antidromically identified descending corticofugal neurons and putative inhibitory interneurons of S1 ‘barrel’ cortex of the rabbit and from simultaneously recorded neurons in the topographically aligned ‘barreloid’ of ventroposterior medial thalamus (VPM). A very strong relationship was seen between brief, statistically significant peaks in the cross-correlograms (CCGs) suggestive of monosynaptic excitatory input and the responses of cortical neurons to VPM microstimulation. Thus, 18 19 cortical neurons that responded to microstimulation pulses of < 6 μA showed a significant CCG with the VPM neuron. Conversely, each of 16 cortical neurons that failed to respond at 10 μA also failed to show a significant CCG. CCGs compiled from independent data sets collected before and after hundreds of microstimulation pulses were very similar, showing that such stimulation can be relatively benign. The method has many limitations, which are discussed. The most serious problem is that the effects of extracellular current pulses cannot be limited to the reference neuron. Nevertheless, microstimulation can provide independent experimental support or refutation of hypotheses of synaptic connectivity that are generated by cross-correlation analyses. In addition, since the technique is rapidly implemented and has considerable power to predict significant cross-correlations, it can be useful in deciding which neuron pairs to study when limited time is available for data collection.