Figure 3.

When horizontal cell coupling is prevented, the shift to long integration times is impaired at both the behavioral level and the ganglion cell level. (A) Horizontal cell coupling in a retina from a Cx57 knockout versus horizontal cell coupling in a retina from a wild-type sibling control. In each retina, a single horizontal cell was injected with dye, and the extent of dye spread was measured for >1 h. Consistent with the results in (Hombach et al., 2004; Shelley et al., 2006; Dedek et al., 2008), coupling is abolished. Scale bar = 50 μm. (B) Behavioral performance curves measured from Cx57 knockouts and wild-type sibling controls under the night condition. The shift to long integration times (low temporal frequency responses) is significantly impaired (p < 10⁻⁴). (C) Ganglion cell performance curves measured from Cx57 knockout animals and wild-type sibling controls under the night condition. As in (B), the shift to long integration times is significantly impaired (p < 10⁻³). All measurements were taken as in Figure 1; for the behavioral experiments, n = 5 wild-type mice, 5 knockout mice, and for the ganglion cell measurements, n = 20 cells from wild-type retinas, 24 cells from knockouts. (D) Left, performance for all animals shown individually. In daylight conditions, the performances of the knockouts are essentially identical to those of the wild-type animals. In night conditions, they diverge: the wild-type animals make the shift toward longer integration times, while the knockouts do not. Right, performance for all ganglion cells. Similar to plots on the left, the performances of the ganglion cells from the knockout and wild-type animals are the same in daytime conditions but diverge at night: the ganglion cells from the wild-type animals undergo the shift toward longer integration times, while those from the knockout are left behind.