LSPS/Optogenetics to Improve Synaptic Connectivity Mapping: Unmasking the Role of Basket Cell-Mediated Feedforward Inhibition

Neocortical pyramidal cells (PYRs) receive synaptic inputs from many types of GABAergic interneurons. Connections between parvalbumin (PV)-positive, fast-spiking interneurons (“PV cells”) and PYRs are characterized by perisomatic synapses and high-amplitude, short-latency IPSCs. Here, we present novel methods to study the functional influence of PV cells on layer 5 PYRs using optogenetics combined with laser-scanning photostimulation (LSPS). First, we examined the strength and spatial distribution of PV-to-PYR inputs. To that end, the fast channelrhodopsin variant AAV5-EF1α-DIO-hChR2(E123T)-eYFP (ChETA) was expressed in PV cells in somatosensory cortex of mice using an adeno-associated virus-based viral construct. Focal blue illumination (100–150 µm half-width) was directed through the microscope objective to excite PV cells along a spatial grid covering layers 2–6, while IPSCs were recorded in layer 5 PYRs. The resulting optogenetic input maps showed evoked PV cell inputs originating from an ∼500-μm-diameter area surrounding the recorded PYR. Evoked IPSCs had the short-latency/high-amplitude characteristic of PV cell inputs. Second, we investigated how PV cell activity modulates PYR output in response to synaptic excitation. We expressed halorhodopsin (eNpHR3.0) in PV cells using the same strategy as for ChETA. Yellow illumination hyperpolarized eNpHR3.0-expressing PV cells, effectively preventing action potential generation and thus decreasing the inhibition of downstream targets. Synaptic input maps onto layer 5 PYRs were acquired using standard glutamate-photolysis LSPS either with or without full-field yellow illumination to silence PV cells. The resulting IPSC input maps selectively lacked short-latency perisomatic inputs, while EPSC input maps showed increased connectivity, particularly from upper layers. This indicates that glutamate uncaging LSPS-based excitatory synaptic maps will consistently underestimate connectivity.