Photoreceptor outer segment as a sink for membrane proteins: hypothesis and implications in retinal ciliopathies

Abstract

The photoreceptor outer segment (OS) is a unique modification of the primary cilium, specialized for light perception. Being homologous organelles, the primary cilium and the OS share common building blocks and molecular machinery to construct and maintain them. The OS, however, has several unique structural features that are not seen in primary cilia. Although these unique features of the OS have been well documented, their implications in protein localization have been under-appreciated. In this review, we compare the structural properties of the primary cilium and the OS, and propose a hypothesis that the OS can act as a sink for membrane proteins. We further discuss the implications of this hypothesis in polarized protein localization in photoreceptors and mechanisms of photoreceptor degeneration in retinal ciliopathies.

Introduction

The photoreceptor outer segment (OS) is a primary cilium-derived subcellular organelle. This structure is filled with proteins necessary for the phototransduction cascade. As expected from its ciliary origin, the OS shares its basic structural elements with primary cilia, and common mechanisms are used to build and maintain these two organelles (see review articles (1,2) for more comprehensive discussion on this topic). The parallel between the OS and the primary cilium culminates in the co-existence of pathologic conditions in both photoreceptors and other cells with primary cilia in human genetic diseases associated with ciliary defects. Furthermore, cilia-related molecular defects are one of the major causes of inherited retinal degenerations (3–5). For more comprehensive reviews on ciliopathies, we recommend recent publications (6,7).

Although the OS shares many of its features with primary cilia, there are several features that are unique to the OS. These include the size, the high membrane content, and the renewal process of the OS. These unique features have an important impact on the localization of membrane proteins in both normal and diseased photoreceptors. Here, we overview the structural properties of the OS, propose a hypothesis about the role of the OS, and discuss its implications in membrane protein localization and retinal ciliopathies.

Primary Cilia and Outer Segments: Homologous but Distinct Organelles

The ciliary compartment of the photoreceptor is composed of three main sub-compartments: the basal body, the connecting cilium (CC), and the OS. The basal body is at the base of the ciliary compartment and composed of nine triplet microtubules derived from the mother centriole. It functions as a microtubule-organizing center from which the nine axonemal microtubule doublets grow. Basal bodies are anchored to the periciliary membrane, the border between the ciliary membrane and the cellular plasma membrane, by transition fibers. The periciliary membrane and the transition fibers act as docking stations for proteins bound for the ciliary compartment (8–15). In frog photoreceptors, a prominent structure (named the periciliary ridge complex (PRC)) is observed in this area and rhodopsin-containing vesicles fuse to the PRC for rhodopsin trafficking to the OS (16,17). The periciliary membrane is also where endocytosis of ciliary membrane proteins occurs (18–21). In addition to acting as a platform for ciliary protein delivery and internalization, the periciliary membrane and the transition fibers constitute part of the ciliary gate that prevent random passage of macromolecules into and out of the ciliary compartment (15,22–25).

Distal to the basal body is the photoreceptor CC, equivalent to the transition zone (TZ) in primary cilia. Significant progress has been made during the past decade in our understanding of the constituents and architecture of the CC/TZ, and functions of the proteins that build the CC/TZ. For more in-depth discussions on this topic, we recommend recently published review articles (15,25–27). Proteins that localize to the CC/TZ are involved in ciliogenesis and regulating protein trafficking into and out of the ciliary compartment. Since the OS and the ciliary proper lack the protein synthesis machinery (e.g. ribosomes) and the CC/TZ is the only physical conduit between the OS/ciliary proper and the cell body, all ciliary proteins synthesized in the cell body must pass through the CC/TZ to reach their destination. This makes the CC/TZ an ideal place to control the trafficking of proteins in and out of cilia, and many proteins that localize to the CC/TZ are thought to be a part of the ciliary gate. Indeed, mutations disrupting CC/TZ proteins alter the protein composition of the ciliary compartment and cause various ciliopathies with retinal involvement (28–39). While the size of the OS in photoreceptors and the ciliary proper in primary cilia varies considerably depending on the species and the cell types (see below), the diameter of the TZ and the CC is remarkably invariant (250–300 nm) in all cells investigated (40–43). In contrast, the length of the photoreceptor CC is significantly longer than that of the TZ in primary cilia: 1–1.5 μm long in mouse photoreceptors and 0.25–0.3 μm in typical primary cilia (40–42).

Further distal to the CC is the OS, where considerable deviations from the primary cilium are observed. Furthermore, there are several notable differences between the two major photoreceptor cell types, rod and cone photoreceptors. One of the major differences between the OS and the primary cilium is the size. Typical primary cilia in mammalian cells are 0.25–0.3 μm in diameter and 1–9 μm (mostly 3–6 μm) long (volume = 0.05–0.6 μm³), which corresponds to <0.05% of the total cell volume (44,45). In contrast, human and mouse rod OSs are 1.5–2 μm in diameter and 25–30 μm long (volume = 45–94 μm³) (Fig. 1), corresponding to 30–40% of the total cell volume (41,46,47). Frog rod OSs are even larger: ∼5 μm wide and 50–60 μm long (i.e. ∼1,000 μm³ in volume) (48,49) (Fig. 1). Although cone OSs are smaller than rod OSs, they are still significantly larger than typical primary cilia. Another major difference between the OS and the primary cilium is the membrane content. The OS is highly membrane-enriched compared to primary cilia. In primary cilia, ciliary plasma membrane is the only membranous structure and other vesicular/membranous structures are normally not observed inside the cilium. In contrast, the OS is filled with membranous structures called discs. The vast majority of discs in rod OSs are essentially large vesicles separate from the OS plasma membrane and called ‘closed’ discs (50–52). Approximately 1% of discs most proximal to the CC are topologically continuous with the OS plasma membrane (therefore, essentially part of the OS plasma membrane) and called ‘open’ discs (50–52). Discs in cone OSs are mostly or entirely ‘open’, depending on the species (53–57). Finally, the OS is continuously and rapidly renewed and, particularly in rod OSs, newly delivered membrane proteins are isolated in closed discs at the base of the OS (58–60). Although continuous delivery of proteins and lipids to the ciliary compartment and shedding at the ciliary tip may not be unique to the OS (61,62), the rate, scale, and mechanisms of the renewal are unique to the OS. In mammalian rods, 8∼10% of discs are shed at the distal end every day and engulfed by retinal pigment epithelial cells (58–60,63). To compensate for this loss and maintain the OS, new OS components are continuously delivered to the base of the OS and form new open discs. Previously formed open discs become closed discs as the newly generated open discs displace them. One notable change that occurs to membrane proteins during this open-to-closed disc transition is their separation from the OS plasma membrane. Although cone OSs are thought to be renewed similarly as in rods, our understanding of the cone OS renewal is limited because of the patency of the cone OS.

Figure 1.

Schematic representation of mouse and frog photoreceptors drawn roughly to scale. (From Pearring et al., 2013 (1); adapted from Elsevier Ltd. with permission).

Outer Segment as a Sink for Membrane Proteins

The large size, the high membrane content, and the unique renewal process of the OS predicts that the OS would have different properties than primary cilia with respect to the distribution of membrane proteins in a cell. In typical cells with primary cilia, membrane proteins with no specific targeting signals localize mostly to the plasma membrane (Fig. 2A). In a hypothetical situation in which the ciliary gate is absent or defective, a fraction of such membrane proteins would be found within the cilium, assuming even distribution along the entire plasma membrane. However, the quantity in the ciliary membrane would be negligible compared to that of the rest of the plasma membrane because of the small volume of the primary cilium. Similarly, if the ciliary gate were permissive (not defective) to certain membrane proteins (again with no ciliary targeting or retention signals), a small fraction of such proteins would be found within the cilium. In photoreceptors, however, the large volume and the high membrane content of the OS provides an ample space to accommodate membrane proteins. Therefore, when the ciliary gate is absent or defective, a large fraction of membrane proteins with no targeting signals are expected to localize to the OS (Fig. 2B). If the ciliary gate is permissive to certain membrane proteins, a considerable amount of such proteins are expected to be found within the OS. Besides, the conversion of open discs to closed discs during the renewal process limits further diffusion of membrane proteins out of the OS by trapping them into topologically isolated closed discs. The continuous renewal of the OS (including the removal of older components at the distal tip) allows continuous absorption of membrane proteins from the cell body. Based on this inference, we propose that the OS can act as a sink for membrane proteins, which absorbs and disposes membrane proteins that have access to the OS.

Figure 2.

Photoreceptor outer segments as a membrane protein sink. (A) Localization of a hypothetical membrane protein (green) with no targeting signals in typical cells with primary cilia. Left; when the ciliary gate is intact, right; when the ciliary gate is defective or absent. (B) Localization of a hypothetical membrane protein (green) with no targeting signals in rod photoreceptors. Left; when the ciliary gate is intact and the protein is actively excluded from the OS, middle; soon after the ciliary gate is disrupted, right; after the entire outer segment is renewed. Schematic cells are not drawn to scale.

The membrane protein sink hypothesis for the OS can explain the propensity of membrane protein localization to the OS in photoreceptors. The OS was previously shown to be a default destination for membrane proteins with no targeting signals in frog photoreceptors (64,65). In addition, certain inner segment (IS) membrane proteins require active exclusion from the OS for their IS localization (66). Although the propensity is less prominent (presumably because of the relatively smaller size of the mouse OS), similar phenomena were also observed in mouse photoreceptors (67). The OS being a default destination for membrane proteins is consistent with the membrane protein sink hypothesis. Interestingly, the ciliary gate in photoreceptors appears to be permissive to the recombinant proteins used in the aforementioned studies despite their diverse origins, suggesting that it may not be highly selective. In this regard, it is noteworthy that the access of soluble proteins to the ciliary compartment is largely dependent on the size of the protein and diffusion (24,68–72).

One prediction from the membrane protein sink hypothesis is that a considerable amount of OS targeted membrane proteins would still localize to the OS (but not exclusively), even when mechanisms for the OS trafficking are impaired. This in fact has been observed in multiple animal models, in which protein trafficking to the OS is disrupted. For example, although the rate of rhodopsin trafficking to the OS decreases in KIF3A-deficient photoreceptors and a fraction of rhodopsin mislocalizes to the IS in various intraflagellar transport (IFT) mutants, a significant subset of rhodopsin still localizes to the OS, obscuring the requirement for IFT proteins in rhodopsin trafficking (73–86). Similar partial mislocalization of presumed cargos (e.g. rhodopsin, GC1, GRK1, and PDE6) was observed in animal models deficient for ARL3 (or overexpressing a constitutively active form of ARL3), PDE6D, RAB8A, TULP1, and UNC119A, as long as the OS forms (87–94). Finally, rhodopsin mutants that misfold or lack the OS targeting signals show partial but not complete mislocalization (see (95) for a recent review on this topic). Although the presence of alternative transport mechanisms and targeting signals cannot be ruled out, these observations are consistent with the membrane protein sink hypothesis and suggest that lateral diffusion along the plasma membrane may partly contribute to the OS localization of the OS-resident membrane proteins.

The membrane protein sink hypothesis further predicts that, to achieve and maintain the polarized protein distribution between the IS and the OS, photoreceptors would need efficient mechanisms to exclude non-OS proteins from the OS. This also suggests that photoreceptors would be prone to the accumulation of membrane proteins in the OS when these mechanisms fail. As mentioned earlier, certain IS proteins need to be actively excluded from the OS for their IS localization (64,66,67). In these studies, removal of short peptides which encompass potential IS targeting signals was sufficient to induce OS accumulation of the protein. In a more recent study, a plethora of non-OS proteins were found to accumulate in the OS when Bardet-Biedl syndrome (BBS) proteins were lost (96). The vast majority of the abnormally accumulated proteins were membrane-associated proteins. These findings support the membrane protein sink hypothesis and suggest that BBS proteins are part of the mechanisms to prevent the encroachment of IS proteins on the OS. Finally, these findings suggest that aberrant accumulation of IS proteins in the OS may be causally involved in the pathogenesis of photoreceptor degeneration in certain retinal ciliopathies.

Relevance to Retinal Ciliopathies and Prospects

One can envisage two main mechanisms to exclude proteins from the OS: 1) preventing OS entry at the ciliary gate and 2) exporting from the OS. As discussed earlier, proteins that localize to the CC/TZ and the transition fibers constitute the ciliary gate. Proteins that belong to this group include (but are not limited to) AHI1, CC2D2A, CEP164, CEP290, FAM161A, MKS1, NPHP5, RPGRIP1, RPGRIP1L, SDCCAG8, SPATA7, TCTN1, and TMEM67 ((15) and references therein). Studies in animal models and human patients with mutations in the genes encoding these proteins showed that they are essential for the OS development and/or homeostasis, and eventually for the photoreceptor survival (97–111) (also see (4,5) for more comprehensive review of retinal ciliopathies). Although most of these proteins are required for the OS development and the trafficking of specific OS-resident proteins, their roles in preventing OS accumulation of IS proteins have not been investigated. Some of them may be necessary mainly or additionally to prevent unauthorized entry of IS proteins into the OS. Future studies should address this question.

With respect to the export of non-OS proteins, two groups of proteins may be involved: i) IFT proteins and ii) BBS proteins. IFT is a mechanism that moves proteins along the ciliary axoneme in both anterograde (from the basal body to the ciliary tip) and retrograde (from the ciliary tip to the basal body) directions. IFT proteins are essential for the OS development at least partly due to their roles in protein trafficking to the OS (73–86,112). Although the precise molecular functions of the BBS proteins in photoreceptors remain to be determined, they associate with IFT particles and are implicated in ciliary trafficking, particularly in the retrograde direction (113–119). These two groups of proteins are thus strong candidates for the export machinery, and retinal ciliopathies caused by mutations in relevant genes may be associated with aberrant accumulation of IS proteins in the OS. However, it should be noted that active export of IS proteins from the OS has not been demonstrated for any proteins. Therefore, the presence of active export mechanisms awaits experimental validation.

Consideration of the OS as a membrane protein sink provides novel insights into the mechanisms by which photoreceptors establish and maintain polarized protein distribution between the IS and the OS, as well as the mechanisms of photoreceptor degeneration in retinal ciliopathies. Mis-trafficking of OS proteins is often suspected as a primary or early defect in retinal ciliopathies. This is probably true in many diseases, considering the high demand for protein trafficking to the OS. However, the membrane protein sink hypothesis based on the structural properties of the OS predicts that aberrant accumulation of non-OS proteins in the OS may be common and causally involved in certain retinal ciliopathies. Future studies should address the prevalence and the significance of this pathologic condition in retinal ciliopathies.