Thus, Protein S does indeed function as a ligand for the Mer receptor expressed by RPE cells, and a fraction of this Protein S is produced by the RPE and CB. These effects notwithstanding, the PR loss seen in the Pros1fl/-/Trp1-Cre/Gas6−/− and Pros1fl/fl/Trp1-Cre/Gas6−/− mice is still less severe than that of the Mertk−/− mice ( Figure 2B). We therefore used a second, nervous-system-restricted Cre driver, Nestin-Cre ( Tronche et al., 1999), which should recombine

floxed Pros1 alleles in all cells check details of the retina, including the RPE and CB. We again crossed this driver with both Pros1fl/fl and Pros1fl/- mice, which were simultaneously either Gas6+/+, Gas6+/−, or Gas6−/−. Most dramatically, retinae from Pros1fl/-/Nes-Cre/Gas6−/−

mice, in which retinal expression of both ligands is eliminated, display a severe loss of ONL nuclei that is statistically identical to the PR death seen in the Mertk−/− selleck screening library mutants ( Figure 2C, solid dark green curve). Adding a single wild-type Gas6 allele back to this genotype—to generate Pros1fl/-/Nes-Cre/Gas6+/− mice—completely restores the ONL to a wild-type thickness ( Figure 2C, solid light green curve, outlined data points). Thus, a retina with no neural Protein S and no Gas6 displays the same severe PR loss and retinal degeneration seen in a retina with no Mer; but a retina with no neural

all Protein S and only half the normal level of Gas6 has a normal number of PRs ( Figure 2C). This is also the case for a retina of the reciprocal genotype, Pros1fl/-/Gas6−/−, which has no Gas6 and only half the normal level of Protein S; this retina also has an ONL of normal thickness that extends all the way to its ends ( Figures S1G and S1H). In summary, only half the normal retinal level of either ligand—in the complete absence of the other—is sufficient to maintain a normal number of PRs in the 12-week mouse retina. There is no difference in PR number across the retina between Pros1fl/-/Nes-Cre/Gas6+/− mice and Pros1fl/-/Nes-Cre/Gas6+/+ mice, both of which display a wild-type profile ( Figure 2C, light green curves). In contrast, Pros1fl/fl/Nes-Cre/Gas6−/− mice display PR degeneration that is comparable to the Mertk−/− and Pros1fl/-/Nes-Cre/Gas6−/− mice, but only in the center of the retina—from ∼35%–65% of the retinal DV axis ( Figure 2C, dark green dashed curve). At more peripheral positions—both ventral and dorsal from the center—PR degeneration becomes progressively less pronounced in these Pros1fl/fl/Nes-Cre/Gas6−/− mice. This effect is due to incomplete recombination of the floxed Pros1 allele under the influence of the Nestin-Cre driver at peripheral retinal locations.