**3. Results**

### *3.1. Generation and Validation of K42E DHDDS Knock-In Mutation*

K42E knock-in mice were generated commercially using CRISPR-Cas9 technology. The K42E knock-in mutations in both heterozygous and homozygous mice were confirmed by DNA sequence for one of the heterozygous F0 founder mice, which is shown in Figure 1. Both the A-to-G and G-to-A transitions that lead to the K42E mutation and the Q44Q silent polymorphism, respectively, are heterozygous (arrows). Intra-litter mating was done to establish at least fourth generation homozygous mice that were used for all subsequent analyses. Heterozygous mice were initially characterized by SD-OCT and histology and found not to differ from WT (not shown).

**Figure 1.** DNA sequence analysis of a tail DNA from a K42E/+ founder mouse. Tail DNA was amplified with primers that cover a 292 bp segmen<sup>t</sup> spanning the target region. The sequence analysis confirmed the presence (arrows) of the K42E (A-to-G) mutation and the Q44Q (G-to-A) polymorphism that was included to eliminate the CRISPR-related PAM site.

### *3.2. SD-OCT Analysis Reveals No Evidence for Retinal Degeneration in DhddsK42E*/*K42E Mice*

SD-OCT provides a non-invasive means of assessing retinal morphology *in vivo*. Qualitative SD-OCT images obtained from wild type (WT) and *Dhdds*K42E/K42E mice are presented in Figure 2. From these images, it is clear that the gross morphology of the retina in the homozygous knock-in animals, from PN 1 to 12 months of age, are comparable to that observed in fully mature, age-matched WT control mice. All retinal histological layers were intact and of normal appearance. Hence, there was no evidence of retinal degeneration, even up to one year of age.

We used SD-OCT to perform quantitative analysis of retinal morphology to compare ocular tissue layer thicknesses in WT and knock-in mice. Figure 3 compares data obtained at PN 1, 2, 3, 8, and 12 mos for *Dhdds*K42E/K42E mice, compared to age-matched WT control littermates. The data are shown both with respect to outer nuclear layer (ONL) thickness (yellow and gray lines) as well as total neural retina thickness (blue and orange lines) as a function of distance from the optic nerve head (ONH, point 4 in each graph) along the vertical meridian, for both the inferior and superior hemispheres. No differences in these quantitative metrics of retinal morphology were observed with respect to genotype, consistent with the representative OCT images shown in Figure 2.

**Figure 2.** Representative averaged SD-OCT images of retinas from (left panels: **A**,**C**,**E**,**G**,**I**) 1-, 2-, 3-, 8- and 12-months (mos) old wild type (WT), and (right panels: **B**,**D**,**F**,**H**,**J**) 1-, 2-, 3-, 8-, and 12-months old *Dhdds*K42E/K42E mice. Abbreviations: IPL, inner plexiform layer; ONL, outer nuclear layer; ROS, rod outer segmen<sup>t</sup> layer. No changes were observed at any age in retinas of *Dhdds*K42E/K42E mice compared to WT mice.

**Figure 3.** Analysis of the ONL thickness (yellow and gray lines) and total retinal thickness (blue and orange lines) in WT and *Dhdds*K42E/K42E mice ranging in age from PN 1 to 12 months. (**A**) 1 month, (**B**) 2 months, (**C**) 3 months, (**D**) 8 months, (**E**) 12 months. Outer nuclear layer (ONL) thickness and total retina thickness measurements (in microns), as a function of genotype and distance from the optic nerve head (ONH) along the vertical meridian in both the inferior and superior hemispheres. Genotypes: WT and *Dhdds*K42E/K42E mice. No significant differences were observed between the groups.

### *3.3. Gliotic Reactivity, Despite Lack of Overt Neural Retina Degeneration, in DhddsK42E*/*K42E Mice*

We performed immunohistochemical analysis on frozen sections of fixed, O.C.T.-embedded WT and *Dhdds*K42E/K42E mouse eyes at PN 2 months of age, probing with antibodies against GFAP

(polyclonal) and a C-terminal epitope of rod opsin (1D4 monoclonal). As shown in Figure 4, whereas WT control retinas only exhibited GFAP immunoreactivity (pseudocolored red) along the vitreoretinal interface, corresponding to astrocytes and Müller glia "end feet", retinas from *Dhdds*K42E/K42E mice exhibited extensive, robust anti-GFAP labeling in a radial pattern. This extended throughout the inner retinal layers to the outer plexiform layer (OPL), in addition to intense labeling along the vitreoretinal interface. The latter results are indicative of massive gliotic activation, which is remarkable considering the lack of overt retinal degeneration or loss of retinal neurons (per the SD-OCT data; see Figures 2 and 3). Gliosis in *Dhdds*K42E/K42E mouse retinas was also detected at PN one month and persisted even at PN six months of age (data not shown). Anti-opsin immunolabeling (pseudocolored green) was comparable in both WT and *Dhdds*K42E/K42E retinas. Notably, the label was confined to the OS layer; there was no mislocalization of opsin to the plasma membrane of the cell in the IS or ONL layer—unlike what is often observed in degenerating photoreceptor cells in various animal models—suggesting normal trafficking of opsin to the outer segment, and consistent with a lack of overt photoreceptor degeneration. It is worth noting that the green labeling in a few cells in the inner retina in Figure 4 is due to mouse-on-mouse binding of the monoclonal antibody to endogenous IgG in blood vessels. It does not represent true anti-opsin immunolabeling.

**Figure 4.** Laser confocal microscopy images of (**A**) WT control and (**B**) *Dhdds*K42E/K42E mouse retina frozen sections at PN 2 months of age, stained with antibodies to GFAP (pseudocolor: red) and rod opsin (pseudocolor: green), and counterstained with DAPI (blue). Scale bar (both panels) is 20 μm. Abbreviations: OS, outer segmen<sup>t</sup> layer; IS, inner segmen<sup>t</sup> layer; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. Scale bar (both panels) is 20 μm.

### *3.4. Lack of Defective Protein Glycosylation in DhddsK42E*/*K42E Mouse Retinas*

The *N*-linked oligo-saccharides of glycoproteins contain alpha-linked mannose residues as constituents, which are cognate ligands for the lectin concanavalin A (Con A) [17]. Hence, ConA lectin cytochemistry offers a reliable means of detecting the presence (or absence) of N-linked oligo-saccharides in tissue sections of *Dhdds*K42E/K42E mice, and a way to directly test the current hypothesis that RP59 is driven by lack of glycosylation. This is because the synthesis of oligosaccharide chains in cells and tissues obligatorily depends upon the presence of Dol-PP and Dol-P (which requires upstream DHDDS activity). Furthermore, N-linked oligosaccharide chains are selectively susceptible to hydrolysis by peptide:N-glycosidase F (PNGase-F) [18]; hence, tissue sections treated with PNGase-F should exhibit a marked loss of Con A binding (serving as a true negative control), thereby mimicking the scenario

where upstream DHDDS activity may be lacking. We performed ConA lectin cytochemical analysis on retinal sections from WT control and *Dhdds*K42E/K42E mice at PN six months of age, with and without pre-treatment with PNGase-F. The results are shown in Figure 5.

**Figure 5.** ConA lectin cytochemical analysis of retinas from (**A**,**B**) WT control and (**C**,**D**) *Dhdds*K42E/K42E mice at PN 6 months of age, with (**B**,**D**) or without (**A**,**C**) pretreatment with PNGase-F. ConA binding (green); PNA binding (magenta); DAPI counterstain (blue). Abbreviations are the same as in the Figure 4 legend. Scale bar (all panels): 20 μm.

Normally, *N*-linked glycoproteins are present throughout the retina, being notably enriched in photoreceptor cells and the synaptic endings of neurons (IPL, OPL). Hence, the inner and outer segmen<sup>t</sup> layers (IS and OS, respectively), including the glycoconjugate-rich interphotoreceptor matrix (IPM), as well as the inner and outer plexiform layers (IPL and OPL, respectively) were robustly labeled with fluor-tagged ConA in untreated WT retinal sections (Figure 5 A). As expected, treatment of WT retinal tissue sections with PNGase-F dramatically reduced the level of ConA binding throughout the retina (Figure 5B). Notably, retinal sections from *Dhdds*K42E/K42E mice also exhibited robust, pan-retinal ConA binding (Figure 5C), comparable to that of WT controls. Upon treatment with PNGase-F, most of the ConA staining was lost (Figure 5D). These results obviate any significant DHDDS loss-of-function in *Dhdds*K42E/K42E mice.

Peanut agglutinin (PNA) binds to the disaccharide Gal-β(1-3)-GalNAc in glycoproteins and glycolipids [17]. Oligosaccharides containing this disaccharide are highly enriched in the extracellular matrix surrounding cone photoreceptor outer segments (the "cone matrix sheath") [19,20]. Thus, PNA binding can be used to selectively label cone photoreceptors in retinal tissue sections, since rod

photoreceptors and their associated "rod matrix sheath" lack such glycan chains [21]. Furthermore, oligosaccharides containing Gal-β(1-3)-GalNAc are generally *O*-linked (e.g., through Ser or Thr residues), rather than *N*-linked, and their synthesis is not dolichol-dependent in mammalian cells [22]. Furthermore, the PNA-binding disaccharide epitope is not susceptible to PNGase-F hydrolysis [17]. Hence, we expected to observe no appreciable di fferences in the binding of PNA to retinal tissue sections from *Dhdds*K42E/K42E retinas vs. WT controls, nor e ffects of PNGase-F treatment on PNA binding, either with regard to labeling intensity or distribution. These expectations were realized, as illustrated in Figure 5 (magenta staining, all four panels). Both *Dhdds*K42E/K42E and control retinas exhibited comparable distribution of PNA-positive cone matrix sheaths, suggesting persistence of viable cone photoreceptors in the K42E mutants.
