Interaction of baseline synaptic noise and Ia EPSPs: Evidence for appreciable negative correlation under physiological conditions

In the anesthetized cat, simultaneous intracellular recordings from pairs of spinal motoneurons were undertaken to see whether the amplitude of single-fiber excitatory postsynaptic potentials (EPSPs) in both cells fluctuated in a coordinated manner that would indicate correlative mechanisms at either pre- or postsynaptic level. Although these recordings revealed correlated fluctuations in the baseline, the single-fiber Ia/EPSPs recorded with the spike-triggered averaging technique exhibited no correlated fluctuations and, unexpectedly, virtually no increase in baseline variance associated with the EPSP. However, the fact that these experiments were carried out under conditions of high baseline synaptic noise (i.e., with muscle stretch) may have influenced the outcome because of interaction between EPSP and synaptic noise, and this possibility was evaluated explicitly. A given connection was studied under low noise by electrically stimulating a single Ia fiber in the absence of muscle stretch. The same connection was analyzed under conditions of high noise by activating the fiber and all other stretch receptor afferents with muscle stretch and by using spike-triggered averaging to extract the EPSP. The differences in mean EPSP amplitude at a given connection under conditions of low noise and high noise were minimal. Fluctuations in EPSP amplitude were then determined to see whether these were influenced by presence of baseline synaptic noise and whether the interaction was nonlinear. Two methods were used to measure EPSP fluctuations: measurement of the variance associated with the EPSP, and determination by the use of deconvolution methods of the discrete amplitude components associated with the EPSP. An increase in baseline variance was observed during the EPSP evoked under low noise conditions at all six connections studied in this way. This increase disappeared at two of these connections when examined under high noise. This may help to explain the results obtained in pairs of motoneurons (see 1). The deconvolution results were used to calculate the variance of the noise-free EPSP. This was found to differ from the variance of the EPSP amplitude distribution measured directly from the change in baseline variance associated with the EPSP. Analytic techniques suggested that this difference could be explained in most cases by negative correlation between the EPSP and baseline synaptic noise. These considerations led to an analytic method to assess the reliability of the deconvolution result. Simulation studies revealed that the baseline variance increase associated with the EPSP is also highly dependent on the correlation between signal and noise. If the correlation is negative, the baseline variance increase is reduced below levels obtained with no noise. The finding that the baseline variance increase measured experimentally is diminished as baseline synaptic noise is added can be interpreted as evidence for appreciable nonlinear interaction between synaptic noise and the EPSP under experimental conditions. The influence of preceding synaptic activity on the Ia fiber-evoked EPSP was studied by choosing trials in which a synaptic potential of a particular amplitude occurred at a particular interval in advance of the test Ia EPSP. In 4 of 12 motoneurons, such a conditioning EPSP could reduce the amplitude of the test EPSP significantly when it occurred 1-2 ms in advance of it. These interactions between EPSP and baseline synaptic noise must be considered in any attempt to use these methods to infer structure (e.g., number of release sites) from physiological measurements.