The roles that the vast array of inhibitory neuron subtypes play in cortical circuits remain to be elucidated, largely because of the historic difficulty in targeting these cells for physiological recordings or manipulations in vivo. Recently, however, genetic labeling methods have allowed the targeted recording of identified subtypes of inhibitory interneurons. Several different strategies have been used to label individual cell classes, including knock-in mouse lines expressing fluorophores under control of inhibitory-specific promoters, as well as mice expressing Cre under subtype-specific promoters, such as Parvalbumin (PV), which are then injected with a floxed viral fluorescent reporter construct, or crossed with a floxed reporter mouse line. The cells can then be targeted with patch pipettes under two-photon guidance, or loaded with calcium dye, allowing the comparison of these cells’ response properties to those of other cell classes. Conflicting results using subsets of these methods have led to controversy regarding the receptive field properties of inhibitory neurons, and in particular, whether inhibitory neurons in the mouse visual cortex can be selective for orientation. In order to help resolve this issue, we have recorded from a large number of PV+ neurons expressing RFP (n = 93), using two-photon guided loose-patch recording. Briefly, adult PV-Cre mice were injected with an LS2L-RFP AAV 2/9 construct. Two to four weeks later, mice were re-anesthetized, a craniotomy was performed, and a patch pipette containing Alexa 488 dye was guided to RFP+ neurons. Gratings drifting in 18 directions were displayed at the preferred spatial and temporal frequencies for each cell, in order to characterize the orientation tuning. Next, each cell was filled with the Alexa 488, and a z-stack of the cell’s dendritic tree was collected. Finally, the dendritic tree was reconstructed off-line. Although PV+ neurons are less tuned than unlabeled neurons on average, we found a clear bimodal distribution of tuning in the PV+ population, that included a large untuned subgroup, and a second highly tuned subgroup. We then compared the dendritic morphology of the untuned and tuned PV+ neurons, and found that the untuned PV+ neurons have significantly larger dendritic trees than do highly tuned PV+ neurons. Thus, by recording from a large number of PV+ neurons, we have been able to define two subclasses that differ both functionally and morphologically: untuned PV+ neurons with wide dendritic trees, and highly tuned PV+ neurons with smaller dendritic trees, possibly corresponding to large and small basket cells.

Society for Neuroscience Abstract, 2011.