Visual Processing by Calretinin Expressing Inhibitory Neurons in Mouse Primary Visual Cortex

Abstract

Inhibition in the cerebral cortex is delivered by a variety of GABAergic interneurons. These cells have been categorized by their morphology, physiology, gene expression and connectivity. Many of these classes appear to be conserved across species, suggesting that the classes play specific functional roles in cortical processing. What these functions are, is still largely unknown. The largest group of interneurons in the upper layers of mouse primary visual cortex (V1) is formed by cells expressing the calcium-binding protein calretinin (CR). This heterogeneous class contains subsets of vasoactive intestinal polypeptide (VIP) interneurons and somatostatin (SOM) interneurons. Here we show, using in vivo two-photon calcium imaging in mice, that CR neurons can be sensitive to stimulus orientation, but that they are less selective on average than the overall neuronal population. Responses of CR neurons are suppressed by a surrounding stimulus, but less so than the overall population. In rats and primates, CR interneurons have been suggested to provide disinhibition, but we found that in mice their in vivo activation by optogenetics causes a net inhibition of cortical activity. Our results show that the average functional properties of CR interneurons are distinct from the averages of the parvalbumin, SOM and VIP interneuron populations.

Figure 1

Colocalization of fluorescent label with calretinin. (A) Left panels: calretinin (CR) immunoreactive cells (green) and tdTomato (TOM) expressing cells (red) in a coronal slice of visual cortex of a CR-Cre mouse injected with cre-dependent tdTomato virus. Right panels: magnification of left panels. In the lower panel, a TOM−/CR+ cell (arrow head), TOM+/CR+ cell (star), TOM+/CR− cell (arrow). (B) Normalized intensities for CR staining and TOM expression. Cells expressing only CR are green, cells expressing only TOM are red, cells expressing both are yellow, and background samples are black. (C) Pie chart of the colocalization between CR immunoreactivity and TOM expression. The percentages are of the total number of cells that had TOM and/or CR label. (D) Colocalization of TdTomato with VIP and SOM. Higher panel: VIP immunostaining on a coronal slice of a CR-cre mouse injected with a cre-dependent tdTomato virus. From left to right, VIP immunoreactive cells in green, TOM+ cells in red, merge of the two showing a TOM+/VIP− cell (arrow), a TOM+/VIP+ cell (star), a VIP+/TOM− cell (arrow head). Lower panel: SOM immunostaining and colocalization with tdTomato. From left to right, SOM immunoreactive cells (green), TOM expressing cells (red), merge of the previous panels showing a TOM+/SOM- cell (arrow), a TOM+/SOM+ cell (star) and a SOM+/TOM− cell (arrow head).

Figure 2

Responses and selectivity measured by calcium fluorescence and cell-attached spiking. (A) Example CR+ neuron. In red tdTomato, in green GCaMP6s and dye filled cell-attached pipette. (B) Neuronal activity measured as GCaMP6s fluorescence changes (red) and electrophysiological recording (blue) from the cell in A. Grey indicates visual stimulus period. (C) Average peristimulus time calcium responses to the different directions for the cell in A. Colors correspond to directions shown in D. (D) Polar plot of the responses to the different directions in calcium fluorescence (red) and in spiking (blue) for the cell in A. (E) Example normalized responses to different grating contrasts (for cell in A). Normalization was done by dividing by the maximum response over all stimuli. (F) Correlation between normalized calcium response and normalized spiking response. Each color represents a different cell. Green represents a CR− cell, cyan, blue and magenta CR+ cells. Black line is the identity line. Inset: correlation between the non-normalized calcium response and spike rate. (G) Correlation between orientation selectivity indexes computed from calcium and spiking responses. Different colors represent different cells. Green dot represents OSI data for CR− cell, cyan, blue and magenta the OSI for CR+ cells. The identity line is drawn in black.

Figure 3

Orientation selectivity is lower in cells expressing CR. (A) Two-photon image showing the most orientation-selective CR− cell (green), an averagely tuned CR− neuron (blue) and an untuned CR+ neuron (red). On the right, average responses and polar plots for the same cells. Colors represent stimulus directions. Grey indicates stimulus period. (B) Distribution of OSI for CR+ and CR− neurons. Red and green outlined dots are the examples shown in A. Vertical bars show mean. **Indicates p < 0.01. (C) Percentage of CR+ and CR− neurons with an OSI higher than 0.33. *Indicates p < 0.05. (D) Distribution of DSI is similar for CR+ and CR− neurons. Vertical bars show mean.

Figure 4

Lower surround suppression in CR+ cells than in CR− population. (A) Example average time courses and tuning of responses to stimuli of different sizes for a CR+ (top) and a CR− (bottom) neuron. Grey indicates stimulus period. Colors represent different stimulus sizes. Error bars indicate mean and SEM. (B) Normalized response to stimuli of increasing sizes for the CR+ population and the CR− population. Responses were normalized for each cell by dividing the responses by the maximum response over all sizes. *Indicates t-test p < 0.05. **Indicates p < 0.001. In the inset, normalized response to stimuli of size relative to preferred size. Error bars indicate mean and SEM. (C) Suppression index for the CR+ population and the CR− population. Bars show mean. *Indicates p < 0.05. (D) Cumulative probability profile for CR+ neurons and CR− neurons. (E) Preferred diameter for the CR+ group and CR− group. Bars show mean.

Figure 5

Spatial and temporal frequency and contrast tuning are equal for CR+ and CR− populations. (A) Example time courses and mean responses to sinusoidal gratings of different spatial frequency. Colors represent different spatial frequencies. Error bars indicate mean and SEM. Grey indicates stimulus period. (B) Normalized responses to various spatial frequency for the CR+ and CR− populations. Responses were normalized for each cell by dividing the responses by the maximum response over all spatial frequencies. (C) Means of high spatial frequency at half max. response for CR+ and CR− neurons are equal. Bars show mean. (D) Percentages of low-pass cells for the CR+ and CR− groups are equal. (E) Example time courses and mean response to stimuli of different temporal frequencies. Colors represent temporal frequencies. (F) Temporal frequency tuning for CR+ and CR− populations are similar. Responses were normalized for each cell by dividing the responses by the maximum response over all temporal frequencies. (G) Means of high temporal frequency at half max. response for CR+ and CR− neurons are equal. (H) Example time course and mean response to gratings of different contrasts. Colors represent contrasts. (I) Normalized responses to different contrasts for CR+ and CR− populations are the same. Responses were normalized for each cell by dividing the responses by the maximum response over all contrasts. (J) Mean C₅₀ values for the CR+ and CR− populations are equal.

Figure 6

Activation of CR+ cells reduces gain in anesthetized and awake mouse. (A) Activation of ChR2 transfected cells by fibre-coupled blue laser. (B) Example transfection of ChR2-YFP (green) in DAPI (blue) stained slice of V1. (C) Example contrast tuning and spike histogram of cell showing decreased responses when laser is turned on. (D) Change in mean responses for contrast tuning curve test for cells when activating CR+ cells for cells in awake (red/pink) and anesthetized (blank/grey) mice plotted versus approximate depth, measured from the top channel in the brain. Dark markers show cells that were significantly modulated (Friedman test), lines show averages per depth. (E) Most effect of CR+ activation on mean maximum responses occurs in top layers (awake and anesthetized combined). (F) Population contrast tuning curve in top layers with and without CR+ activation in awake mice. Responses were normalized for each cell by dividing the responses by the maximum response over all contrasts for the condition when the laser was off. (G) C₅₀ in top layers is changed only little by CR+ activation in awake and anesthetized mice. Arrows indicate means. (H) Population size tuning curve in top layers with and without CR+ activation in awake mice. (I) Surround suppression in top layers is unchanged by CR+ activation. Arrows indicate means.