The excitatory input to PV1 cells did not show a discontinuous decrease in strength (Figure 4D), suggesting that horizontal cells are not responsible for the switch. Since amacrine cells mediate inhibitory input to ganglion cells, we conclude that the switch involves the activation of GABAergic spiking amacrine cells that can act from a distance and are directly connected to PV1 cells. To confirm that far reaching amacrine cells directly connect to PV1 cells, we carried out monosynaptically restricted viral tracing using G-deleted rabies virus in which the G protein is supplied to the PV ganglion cells by a conditional adeno-associated

PD0332991 concentration (Marshel et al., 2010; Stepien et al., 2010; Wickersham et al., 2010) or Herpes virus (Yonehara et al., 2011) (Figure S6). We reconstructed Autophagy inhibitor the transsynaptically labeled amacrine cells around three PV1 cells, each in a different mouse (Experimental Procedures), and found amacrine cells with long processes, some reaching over 1 mm across the retina, connected to PV1 cells (Figures 5, S6, and S7). These “wide-field” amacrine cells, revealed by monosynaptic tracing, are probably the inhibitory cells that are activated by the switch. Note that PV cells other than PV1 also receive input from wide-field

cells and, therefore, the PV1 connecting amacrine cells must have special properties that allow the implementation of the switch (Lin and Masland, 2006). How could inhibition be differentially activated in two different regimes of vision? The retina incorporates two kinds of photoreceptors, rods and cones, which provide the sensory interface for image-forming vision. The more sensitive rods and the less sensitive cones have overlapping light intensity ranges of signaling (Figure S2) and, therefore, three ranges can be defined: vision mediated by rods only, rods and cones, and cones only. In order to determine whether the transition between switch-OFF and switch-ON states corresponds to the transition

from vision mediated by rods only to rods and cones, or rods and cones to cones only, we recorded from rod and positive contrast-activated cone bipolar cells in a retinal slice preparation (Figures 6A–6C). We presented the slice with full-field steps of illumination with fixed contrast across different light intensities, Casein kinase 1 incorporating rod only and cone only intensity ranges. The critical light intensity at which the switch was turned on corresponded to those light intensity values in which cone bipolar cells became strongly activated. At this light intensity, rod bipolar cells have already been fully activated. The critical light intensity was within the range reported to activate cones in mice (Nathan et al., 2006; Umino et al., 2008). These experiments are consistent with a view that the activation of cones toggles the switch (see Discussion for an alternative explanation).