**1. Introduction**

Tetraspanins represent a family of highly conserved membrane proteins involved in a variety of functions. These functions include membrane organization and compartmentalization, cell signaling, adhesion and migration [1–4]. Structurally, tetraspanins consist of a short cytoplasmic N-terminus, four transmembrane domains (TM), one small and one large extracellular loop (EC1 and EC2, respectively), as well as a short cytoplasmic C-terminus [3,5,6]. The EC2 loop can be further divided in a conserved and a variable region [7,8]. The highly conserved part of the EC2 loop was shown to mediate the dimerization of tetraspanin proteins, while the variable region is required for specific interactions of the tetraspanins with their variable interaction partners [7–10]. Tetraspanin interactions can be divided into three levels of interactions [1]. The primary level consists of direct associations between two tetraspanin proteins as well as between tetraspanin and a non-tetraspanin interacting

partners, interactions that resist the treatment with strong detergents [1,4]. Binding of these primary complexes to each other represents the secondary level which is indirect and in some cases supported by palmitoylation [1,4,11]. In the third level, tetraspanins can indirectly associate with their interacting partners to form large insoluble complexes, which only resist mild detergents like CHAPS [1,4,12,13]. The ability of tetraspanins to interact with each other and with other partners allows them to form tetraspanin-enriched microdomains known as the tetraspanin web [4,14].

Peripherin 2 (Prph2, formally known as retinal degeneration slow, RDS) represents a photoreceptor-specific tetraspanin. It is a structural protein that is critical for the proper development of rod and cone outer segments (ROS and COS, respectively) and thus for vision [15,16]. The main function of Prph2, in the morphogenesis of ROS and COS, is to promote membrane curvature, flattening and fusion; processes that are required for the rim formation of the OS discs and lamella [15,17–20]. Like the other members of the tetraspanin family, Prph2 contains two loops, referred to as intradiscal loops (D1 and D2, respectively). The D2 loop is equivalent to the EC2 loop of other tetraspanins and mediates the bulk of interactions of Prph2 [21–23]. Prph2 interacts with another photoreceptor-specific tetraspanin, called rod outer segmen<sup>t</sup> membrane protein 1 (Rom1) [24]. Prph2 forms non-covalently associated homo- and hetero-tetramers with Rom1. These tetramers associate and form covalently linked intermediate complexes. In addition to that, Prph2 is able to form higher order complexes consisting of several Prph2 homo-tetramers. Interestingly, hetero-tetramers including Rom-1 are excluded from these higher order complexes [21,25,26]. Formation of these complexes is required for Prph2 to promote the development of rim domains essential for OS formation [18,27,28]. In line with its key role in OS formation, mutations in *PRPH2* are connected with a multitude of retinal diseases ranging from autosomal dominant retinitis pigmentosa (ADRP) to macular degeneration (MD) [29]. The associated phenotypes show a high degree of variability of onset and phenotype even between patients carrying the same mutation. In addition to the degeneration of the retina, some Prph2 related diseases were found to cause secondary e ffects on the retinal pigment epithelium (RPE) or the choroid [29–31].

This review describes the importance of Prph2 in the development of both ROS and COS. Recent data provided new insight into the steps necessary for OS morphogenesis and the role of Prph2 in orchestrating these events. Furthermore, we will summarize how the precise formation and regulation of Prph2/Rom1 complex a ffects the function of Prph2 in rods versus cones. Finally, we will discuss the animal models expressing pathogenic Prph2 mutations and provide a short summary about therapeutic approaches so far aimed to treat Prph2 related diseases.

### **2. The Role of Prph2 in Photoreceptor Outer Segment Morphogenesis**

Since mice lacking Prph2 (*Prph2*−/−) fail to form ROS, it became clear that the major function of Prph2 is in the initialization and elaboration of ROS [32]. Recent data confirmed this function and identified the C-terminus of Prph2 as the motif responsible for the initialization of ROS formation [33,34]. The photoreceptor outer segmen<sup>t</sup> (OS) represents a highly specialized form of a modified primary cilium [35]. The initial steps in the development of the OS are similar to the development of other primary cilia, starting with the attachment of the ciliary vesicle to the basal body [36–38]. After this initial step, the elongation of the ciliary axoneme and the expansion of the plasma membrane happen in a synchronous fashion. A recent study showed that these initial steps in ciliogenesis take place in both wild type (WT) and *Prph2*−/− photoreceptors, an observation which was in line with previous data showing the presence of an intact connecting cilium (CC) in the *Prph2*−/− photoreceptors [32,33,39]. Additionally, the same study revealed that the Prph2 C-terminus mediates suppression of the formation of ectosomes at the distal part of the CC in rod photoreceptors [33]. Ectosomes are thought to play an important role in the processes of ciliary resorption, regulation of the ciliary protein content and regulation of signal transduction [40–44]. The ability of Prph2 C-terminus to suppress the release of ectosomes o ffers an explanation to why photoreceptor cilia are able to form the highly complexed intra-ciliary membranous structures like that of the ROS [33]. Further evidence for a key role of the Prph2 C-terminus in the initialization of ROS formation was provided by a recent study that generated a chimeric protein composed of the C-terminus of Prph2 fused to the body of Rom1, the non-glycosylated homologue of Prph2 [34]. Expression of the chimeric protein was sufficient to initialize the formation of ROS structures but failed to elaborate them proving that the Prph2 C-terminus is responsible for the formation of the initial membrane outgrowth while the other parts of Prph2 are required for the development of the highly organized structure of the OS.

*Prph2*+/− mice show defects in both function and structure of photoreceptor OS, which proves that the role of Prph2 is not restricted to the initialization of OS formation but also important for its structural development and maintenance [16]. In mature ROS discs, Prph2 is restricted to the rim [15]. In addition, previous studies demonstrated that Prph2 displayed membrane fusion activity, which led to the theory that Prph2 is involved in the shaping of the closed rim of fully developed discs in ROS [45,46]. Indeed, a function of Prph2 in inducing the curvature of the plasma membrane was observed in several studies [25,33,47,48]. While all these studies agree that Prph2 is involved in the generation of membrane curvature, they are contradictory in defining the exact region of Prph2 responsible for this activity. Two studies pinpointed the membrane curvature inducing activity to an α-helix motif located in the C-terminus of Prph2 [47,48], while a third study argues that this motif is rather preventing membrane curvature than inducing it [25]. A recent study found evidence for a function of the tetraspanin core of Prph2 in generating membrane curvature while the C-terminus was unable to perform this function [33]. The localization of Prph2 in newly formed ROS discs provided further evidence for the importance of Prph2 in membrane curvature and disc closure [17]. Newly formed discs in the ROS have a close rim at the site adjacent to the axoneme of the photoreceptor cilium while the opposing side of the disc remains open [49]. In these newly formed discs, Prph2 is restricted to the axonemal closed side of the disc [17]. A recent study confirmed the localization of Prph2 on the axonemal side of newly formed discs in the ROS [50]. An additional finding in this study was that Prph2 is more abundant at the closed rim of newly formed discs than at the rim of mature discs, indicating that Prph2 is not only involved in the closure of this rim but also required for the initialization of the formation of new discs.

The membrane curvature of the open end of newly formed discs in ROS is opposite to the curvature on the closed rim, and thus unlikely to allow the binding of Prph2 [26]. This led to the question of whether a different transmembrane protein may promote the curvature of the open ends. Prominin-1 was identified as a potential candidate involved in the shaping of newly synthetized discs in ROS. Prominin-1 is a membrane protein with five transmembrane domains that localizes at the open end of newly synthesized discs in ROS, thus opposite to the localization of Prph2 [51–55]. The vital role of Prominin-1 in the structural integrity of the photoreceptor OS is supported by the finding that mutations in Prominin-1 are connected to retinal diseases causing degeneration of the photoreceptors [54,56–59]. Earlier studies on Prominin-1 demonstrated that it shows a binding preference for curved membranes, but does not induce the curvature itself [60,61]. Thus, Prominin-1 is a promising candidate for the maintenance of the curvature on the open end, while the protein responsible for inducing the curvature at these ends remains elusive. A second question, which remains open thus far, is what mechanism restricts Prominin-1 to the newly synthesized discs? When ROS discs mature and reach their final diameter, the open ends close; presumably through the activity of Prph2 and become separated from the plasma membrane of ROS. While Prph2 is located on the entire rim of these closed discs, Prominin-1 is no longer found after the closure of the now matured disc [55].

While rod OS discs are closed and separated from the plasma membrane, cone OSs on the other hand contain open discs also referred to as lamellae and are contiguous with the plasma membrane. Although amphibian COS consists exclusively of open discs, the number of open discs in mammalian COS varies depending on the species [38,62–65]. The localization of Prph2 in the open discs of COS is comparable with its localization in the newly synthesized ROS discs, with Prph2 being restricted to the side adjacent to the axoneme of the photoreceptor cilium [55]. The discs in COS display a closed rim formation on the axonemal sides. Prominin-1 is localized on the open side, in line with its localization

in the newly formed ROS. Studying the role of Prph2 in the development of cones in mice has proven to be di fficult due to the low percentage of cone photoreceptors in the murine retina, whereby only ~3–5% of photoreceptors are cones. We have analyzed the e ffect of loss of Prph2 in the background of an Nrl knockout (*Nrl*−/−), in which all rods are converted to S-cone-like cells [66–68]. Here the cones in the double knockout mice (*Prph2*−/−/*Nrl*−/−) displayed disorganized COSs lacking the flattened lamellae characteristic of WT COS [69]. Furthermore, this study provided evidence for the tra fficking of many OS proteins to this altered OS structure, indicating that the transport of those proteins towards the OS does not rely on Prph2 [68]. The ability of *Prph2*−/− retinas to form disorganized COSs argues for a di fferent role of Prph2 in cones than in rods. As mentioned above, ROSs fail to develop in the *Prph2*−/− mice, thus demonstrating a vital role for Prph2 in the initialization of the formation of this complex intra-ciliary membrane structure in ROS. The discovery of a role for the C-terminal region of Prph2 in inhibiting the release of ectosomes from the rod photoreceptor cilia [33] added new interesting aspect to the process of the rod OS formation. Future studies are needed to determine if this function of Prph2 is preserved in cones.

The identification of Rom1 as a Prph2 interactor initiated a multitude of studies investigating the complex formation between Prph2 and Rom1 [24,26,70]. Like Prph2, Rom1 is a photoreceptor-specific membrane protein and a member of the tetraspanin superfamily [71]. While only sharing a sequence identity of 35% with Prph2, both proteins form a highly conserved and similar secondary and tertiary structures with four transmembrane domains and two intradiscal loops [24]. Prph2 and Rom1 form heteromeric and homomeric tetrameric complexes, held together by non-covalent bonding mediated by the second intradiscal (D2) loop [21–23]. These tetramers can assemble into intermediate complexes consisting of both hetero- and homotetramers to establish a mix of covalent and non-covalent linked complexes [72]. These intermediate complexes were found to include at least two tetramers. In addition to the intermediate complexes, higher order complexes containing only Prph2 were also identified [72]. The core of both intermediate and higher order complexes consists of tetramers [21,25]. Interestingly Rom1 is excluded from these higher order complexes [21,25]. How this exclusion is mediated is not fully understood. A possible explanation is that Prph2 can utilize both the conventional secretory pathway through the Golgi apparatus as well as the unconventional pathway which bypasses the Golgi apparatus [26,73]. The presence of Rom1 results in more Prph2 being transported through the conventional pathway [34]. It is likely that this level of sorting during the initial steps of protein tra fficking provides a possible mechanism mediating the exclusion of Rom1 from higher order complexes. Studies verifying this mechanism are a possible direction for future work. The higher order complexes are localized at the closed rim of discs indicating an important role for them in the membrane folding necessary to close the disc rim [25,74]. Biochemical evidence supported the idea that during the formation of Prph2/Rom1 complexes, intermolecular disulfide bonding between Prph2 and Rom1 are formed [72,75,76].

The ability of Prph2 and Rom1 to form complexes was proven to be vital for their function in OS formation [18,27,74]. The cysteine at position 150 (C150) in the D2 loop of Prph2 was found to be critical for the formation of higher order Prph2/Rom1 complexes [21,28,77]. C150 of Prph2 forms an intermolecular disulfide bond with C150 of another Prph2 to make homodimers or with C153 of Rom1, thus allowing the formation of Prph2/Rom1 heterodimers and heteromeric tetramers. A knock-in mouse expressing a point mutation at C150 of Prph2 (C150S) provided further evidence for the importance of this cysteine in the formation of higher order complexes and thus for the role of Prph2 and Rom1 in the development of the OS [78]. The formation of intermediate as well as higher order complexes was fully abolished by the C150S mutation and both ROS and COS were severely disorganized. Despite being disorganized, the OS structure of the C150S mice were somewhat better when compared to *Prph2*+/− while no functional improvement was observed, indicating that the ability of Prph2 to form intermediate complexes with Rom-1 as well as its ability to form higher order complexes consisting of Prph2 homo-tetramers is not only relevant for maintaining the OS structure but also for its function [78].

The development of higher order complexes seems to be expendable for the initiation of OS formation, since generation of both ROS and COS is initiated in the C150S mutant retinas that are devoid of these complexes. Comparable results were obtained for a mutation deleting lysine at position 153 in the D2 loop of Prph2 (K153 Δ) [79]. K153 Δ represents a mutation found in patients resulting in variable phenotypes ranging from ADRP to more cone specific MD [80]. This mutation also obliterates the ability of Prph2 to form higher order complexes [79]. The initiation of ROS and COS structures was una ffected while both displayed structural and functional decline. While the C150S and K153 Δ mutations prevent the formation of higher order complexes, some pathogenic mutations of *Prph2* lead to the formation of an abnormal high molecular weight aggregates that include both Prph2 and Rom1 [81,82]. A patient mutation causing the exchange of the tyrosine at position 141 for a cysteine (Y141C), e ffectively adding another cysteine in the D2 loop, causes mostly defects in the macula but some cases of RP were also reported [30,82,83]. The Prph2 and Rom1 complexes in Y141C knockin retinas were significantly altered, showing the formation of abnormal high molecular weight aggregates containing both Prph2 and Rom1. This is a remarkable finding considering that Rom1 is normally excluded from higher order complexes [82]. These mice display structural and functional defects in ROS and COS. The formation of abnormal high molecular weight aggregates could also be found in another patient mutation in which tryptophan in Prph2 is substituted by arginine at position 172 (R172W) [81]. Transgenic mice carrying this mutation displayed the formation of high molecular weight aggregates of Rom1 while Prph2 complexes were not a ffected. Additionally, these high molecular weight Rom1 aggregates were more abundant in cones, a result that is in line with the decline of cone function in this mouse model [81].

Taken together, these results demonstrate that the exact regulation of the Prph2/Rom1 complex formation is critical for proper OS development, maintenance and disc size in both rods and cones as well as for their function. For OS initiation, proper complex formation seems to be irrelevant. It is worth noting that these studies point to a striking di fference between rods and cones with regard to how they are a ffected by the di fferent *Prph2* mutations. Mutations, which primarily a ffect the rods, seem to be resulting from haploinsu fficiency or loss-of-function e ffects [78,84,85], while those e ffecting cones seem to be more susceptible to changes in the complex formation, likely due to gain-of-function defects [78,79,81,82,86–88]. Another di fferentiation between rods and cones is how they process Rom1. While ablation of Prph2 results in the complete loss of ROS and severely disorganized COS, elimination of Rom1 (*Rom1*−/−) is less severe, indicating that Rom1's role is more in fine tuning of disc sizing [27]. It is worth noting that most studies considering these functions were performed in WT murine rods. A recent study performed on a *Nrl*−/− background using the K153 Δ mutation of Prph2 demonstrated a loss of Prph2/Rom1 interaction specifically in cones [79]. In order to unravel the precise di fferences in the roles of Rom1 in rods versus cones, further studies are needed.

The D2 loop of Prph2 contains a motif for N-linked glycosylation, a motif which is absent in the D2 loop of Rom1 [24]. In rods, the N-glycosylation on Prph2 was found to be expendable for ROS [89]. In transgenic mice expressing the unglycosylated form of Prph2 specifically in rods, the interaction with Rom1 or the formation of complexes was not a ffected. Furthermore, expressing unglycosylated Prph2 on a *Prph2*−/− background achieved a full rescue of the phenotype, further proving that the N-glycosylation of Prph2 is unnecessary for its function in rods [89]. A knockin mouse model expressing unglycosylated Prph2 [75] recapitulated the rod results observed by Kedzierski et al. [89]. However, while rods were una ffected, cone function was found to decrease significantly in these mice. Furthermore, a significant decrease in Prph2 and Rom1 levels in the cones of the knockin mice was also observed. These results further highlight the di fferential functional roles of Prph2 in rods versus cones.

Apart from its interaction with Rom1, Prph2 was found to interact with additional OS specific proteins. One of those interactors is with cyclic nucleotide gated channel B1a (CNGB1a) [90,91]. CNGB1a was shown to be relevant for the correct sizing and alignment of the rod discs, supporting the hypothesis that the interaction between Prph2 and CNGB1a plays a role in the proper shaping of the discs [92,93]. Additionally, Prph2 was also found to directly interact with rhodopsin (Rho), thus forming a complex of Prph2, CNGB1a and Rho to anchor the rim of the discs to the OS plasma membrane and the rims of adjacent discs. It has been hypothesized that these interactions are responsible for correct alignment and stacking of the discs [94]. This was supported by findings that co-depletion of these three proteins exacerbated both functional and structural defects in rod photoreceptors [95].

### **3. Insight from Mouse Models into the Pathophysiology of** *Prph2* **Mutations**

### *3.1. Transgenic Mouse Models of Prph2 Mutations*

Genetically engineered mouse models expressing pathogenic mutations of Prph2 have proven to be valuable tools in studying the pathophysiology of Prph2 related blinding diseases. These models have been generated as transgenic expressing *Prph2* under the control of heterologous promoters that in some cases are specific to rods, cones or both.
