Detection of input sites in scanning photostimulation data based on spatial correlations

Scanning photostimulation is a well-established method for studying the functional microcircuitry in brain slices. Light-evoked responses are thereby taken as an indicator for a connected presynaptic partner. Such an approach thus requires a clear distinction between the photo-evoked and the spontaneous responses. Here we show that, for a data set from entorhinal cortex layer II with high spontaneous synaptic rates of up to 10 Hz, it is possible to identify presynaptic sites. The underlying detection algorithm is based on the finding that a presynaptic cell has several neighboring activation sites, resulting in the clustered appearance of specific photo-evoked inputs. The main idea behind this approach is to identify “hit” locations at which the number of intracellularly recorded synaptic events is significantly larger as expected from the hypothesis of statistical independence. The algorithm works without making use of EPSC amplitude information and for single trials, i.e., each site is stimulated only once. The hit maps are tested upon reliability by repeated stimulations and by blocking synaptically mediated responses via TTX. Furthermore, based on the hit density of surrogate data, we devise a Bayesian formalism to estimate the number of presynaptic partners. In these simulations we find good agreement between estimated and real number of input cells, which shows that the hit density can be used as a reliable measure for afferent connectivity.

Research highlights▶ Spatial oversampling is a valid strategy in scanning photostimulation experiments. ▶ The method works for up to 10 Hz of background activity. ▶ A “hit density” measure for quantifying afferent connectivity is derived from counting statistics.