All animal experiments were performed in accordance with the approved procedures of the Institutional Animal Care and Use Committee. All mice were maintained in the same conditions of 12-h light to 12-h dark, with unrestricted access to food and water. Lighting conditions were kept constant in all cages with illumination of 10 to 15 lux at the level of the cages.

The Rp2-DKO mice were generated by breeding the homozygous Nrl^-/- with the previously reported Rp2^null mice (Mears et al., 2001; Li et al., 2013). Male mice of appropriate genotypes were used in these studies. All mice were genotyped to confirm the absence of the rd8 allele.

Antibodies against β-tubulin, actin and acetylated α-tubulin were obtained from Sigma-Aldrich (St. Louis, MO, United States). Peanut agglutinin (PNA) was procured from Vector Labs (Burlingame, CA, United States); Rhodopsin and S-opsin antibodies were purchased from Millipore (Billerica, MA, United States) and SantaCruz, respectively.

The fundus and ERG recordings were performed essentially as described earlier (Chang et al., 2006; Rao et al., 2016). Only photopic (cone-mediated) ERGs were recorded as the Nrl^-/- mice develop a cone-only retina.

For immunofluorescence analyses, mouse eyes were enucleated and then fixed in 4% paraformaldehyde in PBS (pH 7.4) followed by cryosectioning and staining as recently described (Li et al., 2013). Primary antibodies were prepared in blocking solution and slides were further incubated overnight at 4°C. Sections were then washed three times with PBS and incubated for 1 h with goat anti-rabbit (or mouse) Alexa Fluor 488, 546, or 647 nm secondary antibody (1:500) at RT. Hoechst 33342 (Life Technologies, Corp.) was used to label the nuclei. The sections were then visualized using a scanning confocal microscope (Leica TCS SP5 II laser; Leica Microsystems).

For transmission electron microscopy (TEM), mouse eyes were enucleated and fixed in 2% glutaraldehyde, 2% paraformaldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) overnight at RT. The anterior portion was removed on the next morning and processed as described (Li et al., 2013). Ultrathin sections were observed in a Zeiss EM900 electron microscope and pictures were taken with a coupled digital camera using the ImageSP software.

Total RNA from the Nrl^-/- and Rp2-DKO mice was extracted using the RNeasy plus mini kit (Qiagen, United States) according to manufacturer’s instructions. The sample from neuroretina plus RPE was isolated from six retinas of each genotype in triplicates. The transcriptomic profile and data analysis of the differentially expressed transcripts were analyzed at Beijing Genomics Institute (Hong Kong), as described. Subsequent bioinformatic analysis was performed using the RPKM (reads per kilo base mapped) values, and the data were analyzed as log₂ ratio from the RPKM values. We selected genes that were significantly altered at least 1.5-fold. Pathway analysis was performed using the IPA system (Ingenuity Systems, Redwood City, CA, United States). The differentially expressed genes were validated by qRT-PCR, as described (Rao et al., 2016).

We previously showed that the ablation of Rp2 in mice (Rp2^null) results in cone OS extension and progressive cone dysfunction (Li et al., 2015). However, due to the overwhelming majority of rods in the mouse retina, it is difficult to investigate how cones respond to the loss of RP2. We therefore, generated Rp2^null:Nrl^-/- double knockout (Rp2-DKO) mice and confirmed the absence of RP2 by immunoblotting using an anti-RP2 antibody. We detected the RP2-immunoreactive band in the WT and the Nrl^-/- mice but not in the Rp2^null or the Rp2-DKO mice (Figure 1A). We also tested whether the loss of RP2 in cones resulted in alterations in the expression of other cone-specific phototransduction genes. We did not detect any differences in the expression of cone phosphodiesterase and S- opsin between the Rp2-DKO and Nrl^-/- mouse retinas. As predicted, rhodopsin expression was undetectable in these retinas (Figure 1B).

We next assessed the retinal morphology of the Rp2-DKO mice. Whereas the fundus examination of the Nrl^-/- mice at 1 and 3 months of age revealed the presence of white spots, which represent whorls and rosettes present in the outer nuclear layer (Roger et al., 2012), the Rp2-DKO mice did not exhibit such structures (Figure 1C). These findings were corroborated in the flat mounted retinas stained with peanut agglutinin (PNA; marker of cone extracellular matrix; red) (Figure 1D).

It was previously reported that the Nrl^-/- mice exhibit cone-mediated supranormal photopic b-wave responses up to ∼3 months of age (Roger et al., 2012). We therefore, analyzed the effect of loss of RP2 on cone function in these mice. We found that the loss of RP2 did not affect the supranormal responses of the Nrl^-/- mice. Moreover, the rate of age-dependent decline in photopic responses of the Nrl^-/- mice was indistinguishable from the Rp2-DKO mice (Figure 2A). These data indicated that the previously reported cone dysfunction observed in the Rp2^null mice or in the mice in which the Rp2 gene was conditionally ablated specifically in M-opsin expressing cones was largely due to their dysfunction (Li et al., 2013, 2015). The Rp2-DKO mice have an overwhelming majority of short wave-length S-opsin expressing cones, which could mask the effect on the M-opsin expressing cones.

As the Rp2-DKO mice did not exhibit major changes in photoreceptor dysfunction, we next assessed whether the long cone OS phenotype of the Rp2^null mice is phenocopied by the Rp2-DKO mice. Immunostaining of the Rp2-DKO mouse retinas with acetylated α-tubulin (microtubules) and S-opsin antibodies (outer segment marker) revealed the presence of elongated cone OS at 3 months of age as compared to the Nrl^-/- mice (Figure 2B). Furthermore, we did not detect significant differences in the localization of S-opsin between the Nrl^-/- and the Rp2-DKO retinas. The elongated OSs were detectable even at earlier stages (data not shown).

These results suggested that the longer cone OSs are functional in the S-cone rich Rp2-DKO mice. Thus, we predicted that the morphology of these cones would also be comparable to the Nrl^-/- mice. However, ultrastructural analysis demonstrated the presence of abnormal OS membrane vesicles in the OSs in 2 months old Rp2-DKO mice (Figure 2C). As previously described, the Nrl^-/- mice showed membranous vesicles in the OSs over wide areas of the retina and shorter OSs but with still discernible OS infoldings at 6 months of age (Roger et al., 2012). In contrast, the 6 months old Rp2-DKO mice exhibited abnormal infoldings loaded with membranous vesicles (Figure 2C). The connecting cilium appeared normal in the micrographs (data not shown).

The absence of rods in the Nrl^-/- mice provides a unique opportunity to investigate pure cone responses to the loss of a retinal disease gene. To delineate the cellular responses underlying the intriguing phenotypes of elongated cone OSs, we performed comparative transcriptomic analysis of the retinas from the Rp2-DKO and Nrl^-/- mice at 1 month of age. We chose this age to avoid degeneration-associated artifacts. We compared six retinas from each genotype and repeated the experiment three times. Our analysis revealed 154 differentially expressed genes (>1.5 fold) in the Rp2-DKO mice as compared to the Nrl^-/- mice (Supplementary Table S1). These genes represented major biological pathways, including developmental and ophthalmic disorders, cellular movement/trafficking, and stress responses (Figure 3A and Supplementary Table S2). We identified several crystallin subunits whose expression was downregulated in the Rp2-DKO mice (Figure 3B). The crystallin subunits belong to the chaperone family of proteins, respond to stress and inflammation and are known to be upregulated during retinal disease (Fort and Lampi, 2011). On the other hand, Lrat (lecithin retinol acyltransferase) Rpe65 (retinal pigment epithelium 65), Dcdc2a (Doublecortin domain containing protein), and Dnahc8 (dynein axonemal heavy chain 8) were the major representative genes of the ophthalmic disease and cilia length regulation and cell trafficking pathways (Redmond et al., 1998; Ruiz et al., 1999; Moiseyev et al., 2005; Samant et al., 2005; Grati et al., 2015). We therefore validated the expression of these genes by qRT-PCR. As shown in Figure 3C, whereas the expression of CryαA subunit was diminished in the Rp2-DKO mice, Dnahc8 exhibited ∼16-fold increase in expression. Lrat and Rpe65 expression was also elevated by ∼5 and 7 fold, respectively. As control, Gbp1 (guanylate binding protein 1) and Gbp10 did not show any change in the expression levels in the RNAseq and qRT-PCR analyses.