| Summary: | While sensory environments can vary dramatically in their statistics, neurons have a limited dynamic range with which they can encode sensory information. In sensory cortex, this problem is in part resolved by the systematic adjustment of neural gain in accordance with the contrast of sensory input. This computation can produce contrast invariant representations in cortex that are also more resilient to static background noise. In visual cortex shunting inhibition by parvalbumin (PV) expressing interneurons and contrast-dependent membrane potential variance have been shown to contribute to contrast gain control (CGC), but whether these mechanisms underlie CGC in auditory cortex (AC) is currently unknown. I aimed to investigate the contributions of these mechanisms to CGC in mouse AC, as optogenetic methods for the circuit level interrogation of the nervous system are well established in that species. First, I characterised CGC in the anaesthetised mouse by performing large scale extracellular recordings across all layers of A1. I found that CGC is present in units recorded in all layers of mouse AC and is marginally stronger in deep layers, implicating intracortical mechanisms in the generation of this CGC. In order to investigate whether PV interneuron activity is capable of modulating the gain of sensory responses, I performed extracellular recordings of sensory evoked multi-unit responses in AC while I manipulated the activity of the PV interneurons optogenetically using Channelrhodospsin (ChR2) or Archaerhodopsin (Arch). PV interneuron activation with ChR2 did not alter spectrotemporal tuning of neuronal responses in AC but reduced both their gain and baseline activity. PV interneuron suppression with Arch left receptive field structure similarly unchanged. The strongest effect of PV suppression was an increase in the gain of sensory evoked responses. Thus, PV interneuron activity does appear capable of modulating the gain of AC auditory responses. However, I found that the activity of PV interneurons did not increase systematically with increasing stimulus contrast, indicating that these neurons may not implement CGC. Finally, I performed whole cell recordings from AC neurons in anaesthetised mice in order to assess the contribution of shunting inhibition and membrane potential variance to CGC. I found that CGC exists at the level of the membrane potential responses in AC, but neither input conductance nor membrane potential variance appeared to contribute to this. This canonical computation therefore appears to be implemented by non-canonical mechanisms in different cortical areas.
|
|---|
