**6. Discussion**

### *6.1. Variability in Measuring PR Cell Loss*

We noted an extremely large variability in PR cell loss data presented in the various reports, which were based on several different types of measures, such as ONL thickness obtained from toluidine-stained plastic sections, hematoxylin and eosin-stained paraffin sections, or DAPI-stained cryosections; counts of rows of ONL nuclei in the same preparations; counts of the total number of ONL nuclei in a fixed retinal area; spider plots assessing ONL thickness over the full perimeter; ONL thickness from OCT B-scans. Cone numbers were assessed in sections or whole mounts stained with cone opsins, peanut agglutinin lectin, or cone arrestin. Methods to count cone cell nuclei efficiently might benefit from studies that examine the effect of mutations in genes that specifically affect cones. Perhaps more challenging is that many studies provided insufficient sample sizes or number of time points to assess the progression of PR cell loss. Some of this variability reflects the evolution of methods and quantitative approaches over several decades, and some may be attributed to the different choices each laboratory makes depending on what works best given resources and interests. However, to aid future efforts to compare PR cell loss between studies, we recommend that at least three ages be

assessed, one prior to the onset of PR cell loss, and at least two within the age range where PR cells are declining exponentially (that is, between 95% and 5% of wild-type values), and that the data be quantified relative to gender-matched controls at these same ages.

Although a decrease in PR cell numbers as estimated above is widely accepted as evidence for RD, an alternative explanation may apply in some instances. *Mcoln1tm1Sasl* homozygotes exhibit an apparent decrease in PR cells to 54% of wild-type at one month of age, a value that remained unchanged at two and six months [505]. TUNEL analysis revealed no increase in apoptosis at any age compared to controls, and both histology and OCT indicated a decrease in total retinal thickness. It is possible that the rapid initial decrease to a stable value may result from retinal thinning as the eye enlarges due to myopia during postnatal development, an occasionally observed feature of *MCOLN1*-associated disease in humans [506]. Methods are available to measure axial length in mice [507], which might be used to determine whether the observed change in nuclear layer thickness is due to ocular enlargement. Models in Table S1 with a similar phenotype include *Guca1atm1.1Hunt* and *Ctnna1Tvrm5*.

### *6.2. Correlation of PR Cell Loss with Gene Function*

As a fortuitous consequence of our inquiry, in some cases, the progress of PR cell loss could be correlated with gene function within categories. For example, in the ciliary function and tra fficking section we see trends in the onset and progression of RD depending on the role and location of gene products within the PR. Mice with mutations in genes involved in ciliogenesis or in transition zone protein complexes, typically result in a PR degeneration that begins at 2 weeks during OS biogenesis and progresses rapidly through OS maturation. Null mutations in genes that encode components of the IFT machinery tend to result in premature death or embryonic lethality, and conditional ablation of these genes in the retina typically leads to early and rapid PR loss. Models with disruptions in protein complexes with roles in BBSome assembly or regulation, protein/lipid tra fficking, axoneme extension, or disc morphogenesis tend to have a moderate to slow degeneration. Lastly, mice with mutations that disturb basal body and pericentriolar anchoring and integrity such as the Usher periciliary membrane complex undergo a very slow form of RD, which results in partial PR function throughout the life of the mouse. It remains possible that the slower progression of PR cell loss, especially when associated with members of protein complexes, may be the result of genetic redundancy where multiple genes encode proteins that have similar biochemical functions.

It was also interesting to note that disruption of chloride channels appeared to be particularly deleterious to PR survival. Many of the models we identified with early and rapid progression of PR cell loss, including those a ffecting a subset of the ciliary genes discussed above or the OS components *Prom1, Prph2,* and *Rho*, appear to be required for OS assembly. Chloride channel defects resulted in similar progression, raising the possibility that chloride homeostasis is important for OS development. This idea is supported by evidence that chloride transport by the chloride channel ANO1 is required for ciliogenesis [508] and that control of intracellular chloride ion levels by this channel regulates the membrane organization of phosphatidylinositol 4,5-bisphosphate [509], a prominent lipid that regulates ciliary development [131,510]. Characterization of early OS development in mouse models defective in chloride channels CLCN2, CLCN3, or CLCN7 may provide mechanistic clues on the role of intracellular chloride in ciliogenesis.

Finally, our analysis revealed links between PR cell loss and a network of 13 genes known to participate in the cellular response to DNA damage, four of which have been directly associated with transcription-coupled DNA repair. Based on canonical indicators of DNA repair, such as the colocalization of phosphorylated histone H2AX and TRP53BP1 (also known as 53BP1) at DSBs, it has been reported that rod PR cells in mice lack a robust canonical DNA damage response, [241,511]. An attenuated response may reflect an adaptation to improve rod cell survival [241,511]. Nevertheless, mechanisms are present in rod cells to repair DNA damage, as evidenced by the robust levels of DNA repair factors [241] and by our results indicating that a network of DNA damage response genes is required for maintaining PR cell viability. Together, these results support the idea that a non-canonical

DNA damage response pathway exists in rod PR cells [512]. Further study of the DNA damage response genes linked to PR cell loss in mice may be useful for elucidating this pathway.
