**5. Analysis**

### *5.1. Progression of PR Cell Loss*

While a host of effects can occur as a result of disruptions in genes expressed in PRs and ancillary cell types that functionally impair vision, such as night blindness, or color vision defects, the focus of the models described here are those that bear single gene mutations that lead to actual PR cell loss. From a review of the models in Table S1, significant PR cell loss is reported as early as postnatal day 7 (P7) and can extend throughout the lifetime of animals examined. The progression of PR cells loss is also highly variable and includes models that progress rapidly with complete ablation within several weeks to models with extremely slow progression where only <10% PR cell loss is noted over the span of time in which animals were examined. Generally, while rapid to moderate progression led to almost complete PR ablation, slow and very slow progression, or degeneration in models that primarily affect cone PRs left a substantial number of PR cells intact.

**Figure 5.** Characterization of mouse models from the Translational Vision Research Models (TVRM) program at JAX. A fundus image (circular panels) and corresponding OCT B-scan are shown for homozygous (**a**) *Rpgrip1nmf247*, at one month of age; (**b**) *Nmnat1tvrm113*, at two months; (**c**) *Alms1Gt(XH152)Byg*, at one year; and (**d**) *Ctnna1Tvrm5*, at two years. Age-matched OCT and fundus images for C57BL6/J control mice are shown to the right of the mutant images. PR cell loss is indicated by a decreased ONL thickness. Fundus images were acquired using a Micron III or IV retinal camera. The vertical dimension of OCT images was doubled to emphasize changes in retinal layer thicknesses.

To compare cell loss among functionally similar genetic models, models in each category of Table S1 were sorted based on the estimated age at which the PR cell population had degenerated by 50% compared to control values, defined as D50 (Figure 6). This quantity represents neither the rate of PR cell loss nor the delay before loss commences, although both parameters are used to calculate it. Rather, D50 provides a common measure of progression that allows both complete and sparse datasets to be evaluated. With sufficient data, delay, and exponential decay constants were calculated and are reflected by the shaded bars in Figure 6. Many datasets with fewer measurements, some containing a single point, allowed only an estimate of D50 (Figure 6; filled circles, range lines indicate estimated limits). Figure 6 may be used to identify models in which overall PR cell loss progresses at an earlier age (lower D50), proceeds at a higher rate once initiated (shorter bar), or is accompanied by a substantial delay (bar starts farther to the right), which may aid in experimental design. Values for D50, the exponential decay constant k, and the delay, when available, are also included in Table S1. We relate qualitative descriptions of progression to D50 as follows: rapid, <2 months; moderate, 2 to <6 months; slow, 6 to <12 months; and very slow, ≥12 months.

**Figure 6.** *Cont.*

**Figure 6.** Progression of PR cell loss in mouse models by functional category. Models in each category were sorted by the age at which the estimated number of PR cells reached 50% of wild-type (D50) as determined by fitting reported data to an exponential decay function combined with a delay. Models are identified by gene symbol, MGI allele symbol, and PMID. Models are homozygous except as indicated by \* (heterozygous) or by the presence of two allele symbols (compound heterozygous). Left- and right-hand margins of shaded bars represent the delay and D50 values, respectively. Bars with a delay value of ≤0.1 month derive from datasets in which no values at 100% of wild type were reported. The midpoint and range of estimated D50 values for datasets with only one point in the 5–95% range is indicated by closed circles and range lines, respectively. In cases where the dataset consisted of a single point at 50%, no estimate was needed.

Figure 6 does not include models in which the age of the animal at the time of measurements was not provided, or was reported as "adult", although these models were included in Table S1. We also omitted models where only cone PR cell degeneration was reported, as limited data and a lack of cone nuclei counts made mathematical modeling of cell loss unreliable. Finally, we omitted all conditional alleles from Figure 6, as the efficiency and specificity of inactivation of genes, as well as the temporal expression of different transgenic Cre drivers utilized varies, making comparison of these alleles difficult.

### *5.2. Biological Processes A*ff*ected by Mutations*

Categorization of PR cell loss genes (Figure 3) relied on current functional knowledge, similarity to other genes, and localization, if known. Genes are often expressed in multiple cell types and serve different functions within the retina. We placed genes in categories that were most pertinent to their roles in the PR or effect on PR survival. The data in Table S1 can be explored after reassigning models to different categories if desired. In the descriptions below, we provide an overview for the first three categories and then a detailed description of the effect of gene disruptions on biological functions included in these categories. Subsequent categories are described without overview.

### 5.2.1. Category 01: Ciliary Function and Trafficking

Overview. Disruptions in a vast number of genes associated with PR sensory cilia result in RD [24,44]. The onset and rate of PR cell loss may vary depending on the role of a given gene in maintaining ciliary structural integrity and function during early PR development and maintenance in adulthood. Retinal defects within any one of the key ciliary structures and/or processes such as ciliogenesis, OS morphogenesis, or PR homeostasis may result in the loss of rod and cone PR cells. Understanding the pathophysiological mechanisms involved in PR cell loss requires a detailed examination of the components and functionalities of the PR sensory cilium.

The mouse PR sensory cilium is a specialized structure comprised of a tubulin-rich axoneme with a symmetrical "9+0" arrangemen<sup>t</sup> of microtubular doublets [45]. The connecting cilium (CC) is akin to the transition zone (TZ) of primary cilia [46,47] and serves as a passageway for the movement of proteins from the organelle-rich IS to the photosensory OS. The OS is composed of an elongated distal axoneme with adjacent stacked membranous discs decorated with proteins necessary for phototransduction. It has been hypothesized that extension of the PR plasma membrane towards the apical RPE provides a convenient sink in the OS for the storage of a large number of membrane proteins [48]. In the CC-TZ, axonemal microtubule doublets interconnect the distal axoneme and basal body, and also connect to the periciliary membrane via Y-linkers. The CC-TZ harbors membrane-associated and soluble proteins that coordinate as gatekeepers to regulate the entry, retention, and exit of proteins in and out of the OS [47,49,50]. At the axonemal base, the ciliary membrane is anchored by the nine microtubular triplets of the basal body and its associated appendages.

In developing murine PRs, the connecting cilium and OSs assemble through a series of coordinated events. Shortly after cell cycle exit of retinal progenitor cells, the mother centriole docks to a primary ciliary vesicle initiating its expansion and fusion with the plasma membrane. Concurrently, the microtubular axoneme elongates towards the apical end of the neuroblastic layer [51]. In rod PRs, ciliogenesis typically begins shortly after birth with the appearance of a mature centriolar-bound ciliary vesicle at around P4 [52]. Subsequently, OS biogenesis occurs asynchronously between P8 and P14 [52–54] while axoneme and PM extension continue until OS maturation around P19–25 [52,53]. During this process, intraflagellar transport (IFT) provides an efficient mechanism for the movement and delivery of crucial proteins to the developing OS [55,56], as discussed in greater detail below.

Over the past several decades, two disparate mechanisms of rod disc morphogenesis have been debated, a vesicular fusion model [57,58], which postulates that membrane discs originate from rhodopsin bearing vesicles that undergo intracellular membrane fusion, and the classic evagination model [59], which proposes that new discs result from the evagination of the plasma membrane at the base of the OS. Recent ultrastructural studies in mouse rod PRs have provided compelling evidence that support the classic evagination model. By high resolution microscopy, several groups demonstrated the plasma membrane origin of the evaginated rod disc membranes [60], which subsequently flatten, elongate, and become enclosed [52,60,61].

OS membranous discs undergo a rapid and continuous turnover with approximately 75 rod discs being shed daily, corresponding to 10% of the OS [53,62]. At the ciliary tip, aged discs are removed by the adjacent RPE through phagocytosis. Consequently, OS renewal requires a continuous flow of new proteins from IS to the OS through the connecting cilium, a process that requires careful regulation. For the CC-TZ gate to function properly, e fficient mechanisms are needed to prevent the entry of undesired proteins and to remove the non-OS proteins improperly targeted to the OS. At the base of the CC-TZ lie transition fibers, which act as a barrier in conjunction with other CC-TZ components, such as the membrane-associated Meckel syndrome (MKS) complex. The BBSome, an octameric coat complex, is thought to coordinate the delivery and removal of proteins from the OS during ciliary formation and maturation [63,64].

Finally, protein tra fficking through the CC-TZ is important for PR development and maintenance [51,65,66]. The movement of protein cargo from the IS to the OS requires a highly regulated passage of vesicles along microtubules that dock and fuse with the periciliary membrane to deliver their cargo at the CC base. Movement of targeted proteins through the cilia to the OS may be facilitated by IFT transport machinery [65] or by a lipidated protein tra fficking system [67,68]. During IFT, protein cargo associate with IFT particles that attach to kinesin and dynein motors and move along the axonemal microtubules in both anterograde and retrograde directions, respectively. The BBSome is known to associate with IFT particles and may provide a mechanism for the removal of non-targeted protein accumulation in the OS [69].

Ciliogenesis. As the ciliary axoneme serves as an important conduit for the IFT movement of signaling molecules, it comes as no surprise that the disruption of ciliogenesis genes may result in significant developmental abnormalities causing early lethality and/or rapid PR degeneration. Genes essential for the elongation of the proximal axoneme (A and B tubules) include *Kif3a* [70], which encodes a subunit of kinesin 2, *Iqcb1* [71], *Arl3* [42], and *Arl13b* [72]. While patients with missense mutations in *ARL13B* present with Joubert-associated features [73], a null mutation, *Arl13bhnn*, results in embryonic lethality in mice [74]. Axonemal disturbances and a failure to form OS discs are observed in developing retinas with conditional *Arl13b* disruption [72].

Axonemal and ciliary membrane extension. Disruptions in genes that a ffect ciliary extension include *Rp1*/*Rp1l1*/*Spata7* (distal axoneme)*, Mak,* and *Pcare (C2orf71;* ciliary membrane). Mice with knockout alleles of *Rp1* (*Rp1tm1Eap* and *Rp1tm1Jnz*) and *Spata7* (*Spata7tm1Mrd*) show a progressive, moderate loss in PR cells through the first year of life. *Rp1m1Jdun* mice homozygous for a Leu66Pro missense mutation experience a much slower degeneration with 30% of PRs left at 26 months of age. Conditional ablation studies of *Spata7* in PRs and in the RPE have shown that the disruption of SPATA7 in rod and cone PRs, but not in the RPE, is the molecular basis of the retinal degenerative phenotype [75].

Ciliary Gate and the CC-TZ. Sensory/primary cilia and their gatekeepers (CC-TZ) are found abundantly in most cell types [76]. Thus, the disease spectrum of ciliary proteins is extensive given their roles in ciliary tra fficking, signaling, and development. Disruptions in CC-TZ genes may result in isolated cases of inherited retinal dystrophies such as Leber congenital amaurosis or in multisystemic, ciliopathies such as Joubert, Meckel, or Senior-Løken Syndrome. Such syndromic ciliopathies may include a multitude of disease phenotypes such as brain malformations, renal cysts, nephronophthisis, and retinal dystrophy.

Within the CC-TZ reside MKS and NPHP modules that closely interact and form multiple distinct protein complexes [47,50,77–79]. The MKS complex includes membrane-associated proteins, such as MKS1 and TMEM67, while NPHP complex proteins, such as NPHP1 and NPHP4, associate in closer proximity to the ciliary axoneme.Mice harboring mutations in genes coding for these complex-associated proteins form normal cilia, however, display early abnormalities in OS morphogenesis. After ciliary biogenesis, retinas in these mutant mice quickly degenerate, eliminating most PRs by 3–4 weeks of age. Genes whose disruptions a ffect the ciliary gate functions of the CC-TZ and cause rapid degeneration include *Nphp1*, *Nphp4*, *Ahi1*, *Iqcb1*, *Tmem67*, and *Cep290*. In humans, mutations in *CEP290* can lead to primarily single-organ diseases such, as retinitis pigmentosa and nephronophthisis, or pleiotropic diseases, such as the Joubert, Meckel, and Bardet–Biedl syndromes. The most studied allele is *rd16,* which harbors a 297 basepair in-frame deletion in *Cep290*. Compared to *Cep290tm1.1Jgg* knockout mice, which show a rapid 78% loss at P14, *Cep290rd16* homozygotes have a longer disease progression with a 60% ONL loss at three weeks of age.

Basal bodies and associated pericentriolar material (PCM). The basal body is a structure derived from the mother centriole and resides at the base of the cilium along with the daughter centriole, neighboring centriolar satellites and other related PCM. Proteins positioned at the ciliary base can also be seen in centrosomes of dividing cells (ALMS1, CEP250, and C8ORF37). Specifically, ALMS1 and CEP250 (CNAP1) localize in close proximity to each other at the proximal ends of centrioles [80]. *ALMS1* encodes a 460kDa protein that when disrupted results in the Alström syndrome (ALMS) [81,82]. Mice with a gene trap, frameshift, and nonsense mutations recapitulate human ALMS disease features such as obesity, diabetes, and neurosensory deficits [21,83,84]. The proper formation of the connecting cilium and the slow progression of PR cell loss in *Alms1Gt(XH152)Byg* [83], *Alms1foz* [84], and *Alms1tvrm102* [21] models suggests that ALMS1 is not essential for ciliary biogenesis but necessary for overall PR homeostasis.

The ciliary base contains supportive structures necessary for the proper docking of cargo to the ciliary membrane. Targeted *Macf1* null mutants fail to develop the ciliary vesicle needed for basal body docking while conditional ablation of *Macf1* in the developing retina disrupts retinal lamination and maturation [85]. Mutations in *Cdcc66*, which encodes a component of centriolar satellites and *Sdccag8*, which encodes a recruiter of PCM, result in an early onset but slow-moderately progressive disease [86,87]. The slower RD makes these alleles attractive models for therapeutic investigations.

Genetic mutations in *CC2D2A*, which encodes a component of the subdistal appendages of mother centrioles and basal bodies [88], have been observed in patients with Meckel Syndrome [89], Joubert Syndrome [90], and non-syndromic rod-cone dystrophy [91]. Mice with null mutations in *Cc2d2a* experience embryonic lethality due to the absence of subdistal appendages and nodal cilia [88]. The retinas of adult mice with tamoxifen-induced deletion of *Cc2d2a* in PRs have a significantly diminished ONL (2–3 layers) 12 weeks post-injection [92], suggesting that CC2D2A is necessary for ciliary homeostasis.

Periciliary membrane complex. At the periciliary membrane complex of PRs lies an Usher protein interactome complex that provides a scaffold for the anchoring of fibers to the periciliary membrane [93]. Mutations in genes encoding members of this complex, *Ush1c*, *Whrn*, and *Ush2a*, result in Usher syndrome, a disease that results in progressive hearing and vision loss. Multiple forms of Usher syndrome exist resulting in different degrees of the onset and severity of disease symptoms. In the mouse, targeted mutations in Usher genes results in a late-onset and very slow progression of PR degeneration. Homozygous *Ush2atm1Tili* mice have normal retinas at 10 months of age and lose 70% of their PRs by 20 months of age [94]. PR degeneration in *Whrntm1Tili* retinas is protracted with only 30% loss observed at 28 months of age [95].

Disc morphogenesis. Although the molecular mechanisms involved in OS disc morphogenesis are not completely understood, there has been considerable progress within the past decade with the emergence of refined ultrastructural methods. The OS protein, peripherin-2 (PRPH2), localizes to the rims of rod and cone discs and functions to establish and maintain the membrane rim curvature during disc formation and maintenance [96,97]. Recent investigations using the *Prph2Rd2* (*rds*) mouse model [10] have suggested another role for PRPH2 during disc morphogenesis [52]. Using transmission electron microscopy, Salinas et al. [52] demonstrated that like other forms of cilia [98,99], PR sensory cilia have an innate ability to spew off ectosomes at the OS base. During normal development of the OS, ectosome release is inhibited and the retained membrane at the CC-TZ is transformed into discs upon membrane evagination. In the homozygous *Prph2Rd2* mice, discs fail to form resulting in the accumulation of ectosomes at the OS base. This finding led the authors to propose that PRPH2 may play a role in inhibiting ectosome release during normal rim formation [52].

Knock-in mice carrying heterozygous alleles of *Prph2* (Tyr141Cys [100] and Lys153Δ [101], mimic the dominant RP disease observed in human patients. It is interesting that PR degeneration rates vary among *Prph2* mutant alleles. While homozygous *Prph2Rd2* mice gradually lose their PRs within the first year [102,103], mice harboring a homozygous null mutation, *Prph2tm1Nmc* undergo a faster degeneration with most PRs lost by 4 months of age [104]. Heterozygous *Prph2tm1Nmc* mice also

experience PR loss, however the rate of decline is much slower [104]. Comparative studies of *Prph2Rd2* with rhodopsin double-knockout mice have suggested that abnormal accumulation of mislocalized rhodopsin may contribute to PR degeneration in *Prph2Rd2* [105]. Hence, the zygosity differences in degeneration rates may be a result of varying rhodopsin: PRPH2 ratios. In addition, the onset and severity of the disease may be influenced by the location of the mutation, as different PRPH2 domains have been implicated in dual roles during disc morphogenesis, the tetraspanin core in rim membrane curvature, and the *C*-terminal domain in ectosome release suppression [52].

ROM1 is thought to be involved in the regulation of OS disc formation. PRPH2 and ROM1 are closely associated at the disc rims in the OS. In humans, a double heterozygous disruption in both *ROM1* (a PRPH2-interacting protein) and *PRPH2* results in digenic RP [106]. While it is not clear whether defects in ROM1 alone causes RP in humans, mice with a monogenic *Rom1* disruption show signs of dominant RP. At one month of age, *Rom1* knockout OS discs are visible but appear enlarged and slightly disorganized [107]. PRs slowly degenerated, reducing the ONL by 34% at 1 year of age. In contrast, *Rom1Rgsc1156* mice with a heterozygous missense mutation, p.W182R, show a 55% loss of PRs at 35 weeks of age [108]. Furthermore, RD was more pronounced in homozygous mice. The degeneration in *Rom1Rgsc1156* mice may be a consequence of an early reduction in endogenous PRPH2 and ROM1 levels, which may interfere with PRPH2-mediated stabilization of disc outer rims.

PRCD, progressive rod-cone degeneration, is a rhodopsin-binding protein [109] that localizes to the OS disc rims [110]. Patients and canines with *PRCDC2Y* mutations have a slowly progressive form of rod-cone degeneration [111]. The Cys2Tyr mutation results in mislocalization of PRCD from the OS to the ONL where it is actively degraded [109]. In the PRs of mice with homozygous *Prcdtm1Vya* mutations, loss of PRCD results in the formation of bulging discs that do not properly flatten and in the accumulation of extracellular vesicles that originate at the OS base [112]. Interestingly, mutant PRs are able to form membrane discs and the distribution of OS proteins and light response do not appear to be perturbed. While activated microglia infiltrate the interphotoreceptor space to remove extracellular vesicles and debris, removal is insufficient and PRs undergo a very slow degeneration. Homozygous *Prcdtm1Vya* mice show only 36% ONL loss at 17 months [112] while in *Prcdtm1(KOMP)Mbp* homozygotes a similar loss is observed at 30 weeks of age [110]. Both models are knockout alleles that target the 5 end of *Prcd* but are on different genetic backgrounds. Further investigations are necessary to determine whether gene modifiers affect progressive PR cell loss in these two models.

IFT trafficking. IFT is essential for ciliogenesis in mammals [113] and disruption of this process often leads to abnormalities in embryonic development. In the mouse, null mutations in genes encoding subunits of the IFT-A (*Ift122* [114], *Ift88* [115], and *Ttc21b* [116]) and IFT-B (*Ift172* [117], *Ift80* [118], and *Traf3ip1* [119]) complexes result in embryonic lethalities, many of which are attributable to ciliary-related disturbances in hedgehog signaling [120,121]. These findings further highlight the integral role of cilia and IFT machinery during embryogenesis.

Hypomorphic and conditional alleles have been useful for elucidating the roles of IFT components in retinal disease. The hypomorphic allele, *Ift88Tg737Rpw*, contains a transgenic insertion resulting in a 2.7 kb intronic deletion. Homozygous *Ift88Tg737Rpw* mice exhibit disorganized OSs as early as P10 and a progressive degeneration of PRs that reduces the ONL to one layer at P77 [66]. Rod-specific ablation of *Ift172* [122] leads to mislocalization of rhodopsin, RP1, and TTC21B (IFT139) and rapid degeneration of PRs. Conditional depletion of *Ift20* in M cones and mature rods both results in opsin mislocalization suggesting that proper opsin trafficking hinges on functional IFT components [123]. To gain a clearer understanding of the roles that IFT molecules play in both rod and cone PRs, additional studies using conditional models are warranted to elucidate the contributions of impaired IFT components to PR cell loss.

Lipidated protein trafficking. Lipid modification of proteins, such as prenylation or acylation, helps direct intracellular protein targeting and regulates protein activity [124]. These hydrophobic modifications help tether their protein partners to the surface of specific membranes throughout the cell, such as the ER, Golgi, transport vesicles, or plasma membranes. Improper trafficking of lipidated

proteins can result in RD. RP2 is a GTPase activating protein that interacts with ARL3 to regulate assembly and movement of membrane-associated protein complexes [125]. Homozygous mice with mutations in the gene encoding RP2 exhibit a slowly progressive rod-cone degeneration [126,127]. ARL3, a small GTPase, tra ffics lipidated membrane-associated proteins to the rod OS [128]. Although *Arl3* knockout mice exhibit early postnatal lethality and Joubert-like features [42], mice with hypomorphic mutations survive to post-wean and display OS abnormalities as early as P9 [129]. Conditional ablation of *Arl3* in the developing retina results in the absence of cilia, and therefore PR cells are rapidly lost [42]. In contrast, depletion of *Arl3* in mature rods leads to mislocalization of lipidated OS proteins, shortened OS, and a moderate progressive PR loss. These results are consistent with roles for ARL3 in ciliogenesis during development and cargo displacement during lipidated protein tra fficking.

The ciliary TZ-associated protein, RPGR, binds and directs the ciliary targeting of INPP5E [130], a phosphoinositide phosphatase that is important for ciliogenesis [131]. Ciliary localization of RPGR itself requires modification with a prenyl group, which interacts with PDE6D [130], a prenyl-binding protein first discovered as a copurifying component of cGMP phosphodiesterase 6 (PDE6) [132]. Like *Rpgrrd9* [133] and *Rpgrtm1Tili* knockout mice [134], mice with *Pde6dtm1.1Wbae* null mutations [135] undergo a very slow degeneration with at least 50% of PR cells remaining at 20 months of age. Rao et al. have demonstrated reduction of INPP5E in RPGR-deficient axonemal OSs [130]. Altogether, these observations validate RPGRs role in ciliary tra fficking and homeostasis and sugges<sup>t</sup> that other players may be involved in ciliary targeting of INPP5E.

Mutations in *Aipl1* result in early and rapid loss of PR cells (Table S1). D50 < 0.55 months for the four germline alleles shown in Figure 6, *Aipl1tm1Mad*, *Aipl1tvrm127*, *Aipl1tm1Visu*, and *Aipl1tvrm119* [21,39,136]. AIPL1 is a protein chaperone that mediates the folding of phosphodiesterase 6 (PDE6), a key component of the visual transduction pathway that regulates cGMP levels (see Section 5.2.2. below) [137]. AIPL1 binding is promoted by prenylation of PDE6 subunits [137]. In *Aipl1* mutants, PDE6 subunits are greatly diminished [136], providing further evidence for the importance of lipid modification in PR viability and vision.

BBSome assembly and regulation. Disruptions in the octameric BBSome complex or associated chaperonins may cause syndromic ciliopathies such as the Bardet–Biedl syndrome and McKusick– Kaufman syndrome. In mice, most gene disruptions that a ffect the BBSome [138] (BBS1, BBS2, BBS4, BBS7, BBIP1, TTC8, and ARL6), and its regulators (LZTFL1, MKKS, BBS10, and BBS12) result in a moderate degeneration of PRs. For instance, PRs in homozygous mice harboring gene trap or null mutations of *Bbs4*, *Bbs4Gt1Nk* [139], and *Bbs4tm1Vcs* [140] appear to progressively decline after maturation with >90% loss at 7 months of age. The delay and lack of ciliogenesis defects suggests that there may be some functional redundancy amongs<sup>t</sup> components of the BBSome.
