Retinitis pigmentosa (RP) is the leading cause of inherited blindness. RP is a genetically heterogeneous disorder, with more than 100 different causal genes identified in patients. Central to disease pathogenesis is the progressive loss of retinal photoreceptors. Photoreceptors are specialised sensory neurons that exhibit a complex and highly dynamic morphology. The highly polarised and elaborated architecture of photoreceptors requires precise regulation of numerous cytoskeletal elements. In recent years, significant work has been placed on investigating the role of microtubules (specifically, the acetylated microtubular axoneme of the photoreceptor connecting cilium) and their role in normal photoreceptor function. This has been driven by the emerging field of ciliopathies, human diseases arising from mutations in genes required for cilia formation or function, of which RP is a frequently reported phenotype. Recent studies have highlighted an intimate relationship between cilia and the actin cystoskeleton. This review will focus on the role of actin in photoreceptors, examining the connection between actin dysregulation in RP.
The tri-laminar retina has evolved to process light waves from our environment and deliver them to the occipital lobe for visual processing. It is arguably our most sensitive sensory organ. Indeed, evidence suggests that its evolution predates the brain, which subsequently evolved to deal with the information gathered by primitive photoreceptors [1,2]. One hundred and twenty million rods and six million cones begin this process, and they are able to do so thanks to their exquisitely modified primary cilia. Cilia dysfunction leads to photoreceptor degeneration and visual loss in a group of untreatable genetic diseases known as the retinal dystrophies, of which RP is the major disease.
The initial rod photoreceptor loss in RP results in night blindness (nyctalopia) and progressive constriction of the patient's visual field. Eventually, cone photoreceptors degenerate, with loss of central vision . Despite enormous genetic heterogeneity, as well as functional heterogeneity of mutated genes, most cases of RP exhibit a classic phenotypic appearance of peripheral bone spicule formation, blood vessel attenuation, and optic disc pallor. Although the phenotypic presentation of RP is due to photoreceptor loss, photoreceptor loss can be triggered by defects in the photoreceptors themselves, or by defects in other retinal cells, especially the overlying retinal pigmented epithelium (RPE) . Importantly, RP is often associated with pathologies in other tissues in syndromic disease. Common examples of these are Usher syndrome (RP and deafness), Alport syndrome (RP, deafness, and renal glomerular dysfunction), and Bardet–Biedl syndrome (RP, obesity, polydactyly, and developmental delay) [5–7].
While advances over the last decade in sequencing technologies have greatly enhanced our ability to identify genes that, when mutated, give rise to both syndromic and non-syndromic RP, our understanding of the pathophysiology of photoreceptor loss is still incomplete. Recently, it has become evident that correct regulation of the cytoskeletal protein actin, which is essential for normal cellular physiology, is perturbed in many cases of RP and this perturbation influences photoreceptor survival [8–10]. This review will document the evidence of actin's role specifically in photoreceptor health and disease. To do so, we must first discuss how actin influences key processes in primary cilia.
The primary cilium
Primary cilia are signalling organelles present on vertebrate cells and essential for tissue homeostasis. Cilia can be motile (e.g. sperm) or immotile (e.g. photoreceptors) and can exist as single or multiple protrusions on a single eukaryotic cell. Primary cilia comprise, in the main, a basal body (BB) and an axoneme. The BB, which describes the mother centriole when it is associated with a cilium, is a barrel of nine triplet microtubules and appendages (sub-distal and distal) that anchor to the membrane at the cilia base . The axoneme comprises nine microtubule doublets which aid cilia transport. Proximally, the axoneme is known as the transition zone, featuring characteristic Y-links that connect the axoneme to the ciliary membrane, that functions to maintain a ciliary protein composition distinct from the rest of the cell . The organelle-specific make-up allows non-motile cilia to function as sensory antennae, responding to their environment by enriching signal-transducing components through highly regulated trafficking pathways . Cilia dysfunction is associated with a wide spectrum of devastating, systemic human disorders known as the Ciliopathies.
Ciliopathies can result in abnormalities in human development (birth defects) or post-natal health. They can cause defects in multiple organs with features including mental disability, polydactyly, cystic kidneys, obesity, and hearing loss . Importantly, a key feature of many systemic ciliopathies is the presence of a retinal dystrophy. Thus, insights into basic cilia processes can inform us as to basic photoreceptor physiology. The cilia field is an emerging but rapidly expanding one, with advances in research tools now allowing us to probe their biology in health and disease. With these advances, and the novel experiments that they have subsequently allowed, we are becoming increasingly aware of the crucial role that regulation of the actin cytoskeleton plays in cilia function.
The role of actin in cilia physiology
Actin is a globular protein that forms microfilaments, playing a role in cell motility, vesicle transport, and cell signalling. Both ciliogenesis and ciliary signalling are directly influenced by actin. Actin nucleation (branching) inhibits ciliogenesis, while actin-severing factors, such as cofilin and gelsolin-family proteins, enhance their formation [14–17]. Furthermore, actin-binding proteins are present in the cilia and are dynamically regulated with changes in actin polymerisation . Recent work has shed light on the mechanisms through which actin dynamics may help in cilia formation, especially following the observation that increased RAB8-positive transport vesicles are present at the base of lengthened cilia when actin is depolymerised [15,17]. Thus, actin may create a mechanical barrier to the vesicle movement or membrane remodelling required for the process .
A further role for actin in cilia function in mammalian cells has been demonstrated with the recently described process of ectocytosis. Ectocytosis is a process through which primary cilia regulate sensory signalling. While some content is trafficked out of cilia by intraflagellar transport (IFT), other proteins, including G-protein coupled receptors (GPCRs), are eliminated by shedding membrane vesicles (ectosomes) . Ectocytosis is driven by dynamic, localised actin turnover. Actin perturbation mislocalises GPCRs to the cell body via ‘retrieval-mediated ciliary-removal’, an inefficient, IFT-dependent process . Pharmacologically destabilising actin leads to the accumulation of ciliary tip buds unable to ectocystose . These two actin-dependent processes, vesicle trafficking and membrane turnover (ectocytosis), are key to the cilium's function as a signalling organelle and are of huge importance in photoreceptor turnover and maintenance.
The role of actin in photoreceptor physiology
Owing to the extraordinary amount of protein trafficking and membrane turnover required to maintain normal vision, the photoreceptor is the most metabolically active cell in the body . Photoreceptor vulnerability to cilia dysfunction probably reflects the sheer scale of cilia-based trafficking and membrane turnover required for vision. As highly specialised primary cilia, photoreceptor outer segments (OS) — where vision is initiated — comprise a stack of membranous discs filled with the GPCR rhodopsin, extending from the cell's transition zone, known as the ‘connecting cilium’ (CC). OS discs are continually turned over. Furthermore, to maintain photoreceptor health, thousands of rhodopsin molecules/minute must traffic from their site-of-synthesis in the inner segment through the CC . Rhodopsin mis-trafficking or mislocalisation results in photoreceptor degeneration, the hallmark of retinitis pigmentosa (RP) .
But despite its unique position in initiating the visual cascade, the photoreceptor shares much in common with other sensory cilia, including the actin-dependent processes outlined above. Rhodopsin traffics to the base of the photoreceptor connecting the cilium in Rab8-positive vesicles (rhodopsin transport carriers) . In the photoreceptor, this process is actin-mediated . Furthermore, ectocytosis is remarkably similar to the process by which photoreceptors shed GPCR (rhodopsin)-containing discs, with the ectosome proteome overlapping that of photoreceptor discs [26,27]. Moreover, photoreceptors have an innate ability to form ectosomes, defaulting to ectocytosis when the actin-dependent process of disc formation is perturbed [28–31]. Thus, dysregulated actin turnover may affect rhodopsin targeting and disc formation, leading to photoreceptor stress. Indeed, recent work in functionally dissecting the genetics of RP has shed light on how the photoreceptor regulates actin and how its dysregulation leads to disease. Here, we will discuss key genes involved in the process.
Retinitis pigmentosa 2
Mutations in the RP2 gene account for ∼15% of cases of X-linked RP . Despite being ubiquitously expressed, patients with RP2 mutations exhibit no extra-retinal phenotypes. The RP2 protein contains two distinct domains. The N-terminal region was first predicted to contain a cyclase-associated protein domain (Figure 1), a protein domain involved in binding actin monomers, based on sequence homology . However, the crystal structure and subsequent biochemical studies revealed that this region of RP2 functions as a GTPase-activating protein for the small GTPase, ARL3 . The C-terminal domain of RP2 shares homology with nucleotide diphosphate kinases (Figure 1), but lacks key catalytic residues that would confer kinase activity . Instead, this domain is thought to serve as a protein–protein interaction domain. Supporting this, the C-terminus of RP2 was shown to form a complex with two proteins involved in actin dynamics, osteoclast-stimulating factor 1 (OSTF1) and Myosin IE (Myo1e), with a human pathogenic mutation disrupting the complex formation . In vitro, loss of RP2 results in perturbed cellular migration, probably through cytoskeletal defects. How OSTF1 and Myo1e may contribute to the pathogenesis in vivo remains to be investigated, but provides evidence that it may, in part, involve actin dysregulation.
Domain structure of actin-associated photoreceptor proteins mutated in inherited retinal degenerative disease.
Schematic diagram illustrating localisation of retinitis pigmentosa proteins that have been associated with actin dysregulation.
The principal role of RP2 is regulating the trafficking of acylated proteins from their site of synthesis in the photoreceptor inner segment across the connecting cilium and into the outer segment . This is achieved via ARL3, which, when in its active GTP-bound state, triggers the release of acylated cargo in the outer segment. However, Drosophila studies demonstrated that active ARL3 resulted in the disruption of actin filaments phenocopying cytochalasin D treatment . Since ARL3 localises to both photoreceptor inner and outer segments, enhanced activity of ARL3 due to mutation of RP2 could profoundly disrupt the actin cytoskeleton, contributing to photoreceptor trafficking defects and morphology.
In humans, the Crumbs family of paralogous proteins is encoded by the CRB1, CRB2, and CRB3 genes. Mutation of CRB1 is a frequent cause of Leber Congenital Amaurosis, cone-rod dystrophy, and RP . The major CRB1 transcript in retina encodes a large (1406 amino acids) single-pass transmembrane protein (Figure 1). The first 25 amino acids encode a signal peptide that is subsequently cleaved, leaving a large extracellular amino-terminus consisting of multiple EGF (epidermal growth factor)-like and laminin AG-like extracellular domains, followed by a single transmembrane domain and a short (38 amino acids) intracellular domain comprising a juxta-membrane FERM (4.1 Ezrin Radixin Moesin) and a carboxy-terminal PDZ domain-binding motif.
More than 200 distinct disease-causing mutations have been identified in CRB1, with the vast majority affecting the extracellular domain . Within the retina, CRB1 is expressed in both the photoreceptors and Muller glia, localising to the subapical region of photoreceptor inner segments, and is thought to participate in cell–cell adhesion at the retinal outer limiting membrane (OLM) via the extracellular domain. Disease-causing mutations are thought to disrupt cell–cell adhesions, disrupting the OLM leading to defects in retinal lamination. The precise nature of this function is not fully understood, especially as CRB1 can be alternatively spliced to encode a secreted protein lacking the transmembrane and intracellular domains.
Most of our understanding of Crumbs protein function has come from studies of the Drosophila homologue (crumbs, of which there is only one) and the mammalian CRB3 paralogue, with the majority of research aimed at the function of the intracellular domain. In Drosophila epithelia, crumbs interacts with components of the actin cytoskeleton including spectrin and moesin via the FERM-binding motif, stabilising the apical actin cytoskeleton . Additionally, Crumbs is required for rhodopsin transport in Drosophila photoreceptors via the actin motor Myosin 5, although given the architectural differences between Drosophila and mammalian photoreceptors, it is not clear if mammalian CRB1 performs a similar role . However, similar to Drosophila crumbs, mammalian CRB3 does interact with both moesin and ezrin to regulate actin dynamics in different cell types. Additionally, CRB3 has been shown to interact with actin-related protein 2 (Arp2) and epidermal growth factor receptor kinase substrate 8 (Eps8), two critical regulators of actin polymerisation and bundling, respectively .
Crumbs proteins are well-known regulators of cell polarity, specifically apico-basal polarity. Cell polarity is regulated, in part, by two evolutionarily conserved protein complexes, the PALS1–PATJ–Crumbs and PAR3–PAR6–aPKC complexes. CRB3 can directly interact with both PALS1 and PAR6 via its C-terminal PDZ-binding motif, and these interactions are essential for its function in controlling apical membrane identity [41,42]. In addition to this role in regulating cell polarity, CRB3 is also critical for the formation of primary cilia in epithelial cells through Par6 and aPKC . Very recently, aPKC was shown to regulate actin dynamics at the BB and regulate cilia length, suggesting that CRB3 could regulate ciliogenesis via modulation of actin dynamics in an aPKC-dependent fashion . While CRB3 appears to be predominantly expressed in epithelia, CRB3 also localises to the connecting cilium in photoreceptors , suggesting that it may be a critical regulator of actin in these cells. Unlike CRB1, no mutations in either CRB2 or CRB3 have been identified in RP patients. However, given the almost identical intracellular domains of the three paralogues, CRB1 is likely to share many of the same protein partners as CRB3, warranting further investigation into photoreceptor (and Muller cell) actin dysregulation as a pathogenic mechanism in patients with CRB1-associated retinal dystrophies.
The Fascin family of proteins comprise of three members in mammals, FSCN1, FSCN2, and FSCN3. All three are 55 kDa proteins containing four β-trefoil domains (Figure 1) and function to cross-link actin filaments to form actin bundles critical for the formation of multiple actin-based cellular structures including lamellipodia and stress fibres . Mutation of FSCN2, specifically the 208delG mutation, has been shown to cause both RP and age-related macular degeneration in Japanese patients in an autosomal dominant manner [47,48]. More recently, a point mutation, P406L, in the FASCIN2 gene has been associated with cone dystrophy . The pathogenicity of the 208delG mutation was called into question with the finding that this mutation did not correctly segregate in a Chinese cohort of families with RP . Why this mutation did not cause RP in some Chinese individuals remains unknown, especially as follow-up studies in mice demonstrated that the 208delG mutation in mice resulted in a classical RP phenotype of progressive retinal degeneration .
The FSCN2 protein is principally expressed in rod photoreceptors and in the cochlea. In the photoreceptors, FSCN2 localises to inner segments and to the calycal processes . The exact function of FSCN2 in photoreceptors is not clear, but accompanying progressive degeneration, EM studies revealed misalignment of outer segment membrane discs at the distal end of the connecting cilium . Given its localisation in photoreceptor calycal processes and cochlea stereocilia, FSCN2 may co-operate with other Usher proteins to mediate disc morphogenesis through actin regulation.
Mutations of the PROM1 gene, which encodes the protein prominin (also known as CD133), cause a wide range of retinal dystrophies including early onset of severe autosomal recessive RP, macular dystrophy, and cone-rod dystrophy [53–55]. The encoded protein is a five-span transmembrane protein that is widely expressed (Figure 1), with much research focussed on its role as a progenitor and cancer stem cell marker. It is expressed on the plasma membrane in most cells, localising to cellular protrusions including cilia, microvilli, and lamellipodia.
In both rod and cone photoreceptors, prominin localises to the base of the outer segment where, together with peripherin, it is thought to regulate disc morphogenesis [54,56]. Prominin is a key component of ciliary ectosomes (see above). Animal models have shown that loss of prominin results in disorganised outer segment discs, although the underlying mechanisms are not fully understood, but binding assays have shown that prominin can directly interact with actin filaments . It is not clear if mutation of prominin directly affects actin dynamics in the outer segment, but evidence from Drosophila points to the involvement of prominin with a network of proteins including crumbs in maintaining the outer segment architecture .
Retinitis Pigmentosa GTPase Regulator
Mutations in the Retinitis Pigmentosa GTPase Regulator (RPGR) gene cause 70–80% of X-linked RP and result in a severe phenotype with early visual loss . RPGR is an alternatively spliced protein with a constitutive variant containing all 19 exons and a retinal-enriched isoform with 14 exons followed by a disordered, glutamic acid/glycine-rich open reading frame (Figure 1). RPGR localises to the connecting cilium, where there is growing evidence it acts to regulate actin turnover. Increased actin polymerisation is seen in the connecting cilia of the Rpgr knock-out mouse as well as RPGR-mutant, iPSC-derived photoreceptor cultures [9,23,59]. Furthermore, siRNA-mediated knock-down of RPGR leads to increased actin stress fibres in hTERT-RPE cells and increased F-actin is seen in a different Rpgr knock-out mouse model [10,60]. While the molecular mechanism remains to be fully elucidated, RPGR mutations perturb RPGR's interaction with the actin-severer gelsolin, reducing its activation .
Quite how this influences photoreceptor health is unknown. The resulting rhodopsin mislocalisation seen in Rpgr−/− photoreceptors is reminiscent of the ‘retrieval-mediated ciliary-removal’ seen when ectocytosis is perturbed, pointing to a role for RPGR in disc formation. Indeed, common proteins like Prominin, required for both ectosome and disc formation, are reduced in Rpgr-null mouse photoreceptors and abnormal disc formation is seen in Rpgr−/− mice [26,27,61,62]. Alternatively, both RPGR and gelsolin are known to interact with the scaffold protein Whirlin, a key component of the periciliary membrane complex (PMC) . The PMC is thought to serve as a gate to regulate protein entry to the cilia. Thus, control of actin at the PMC by RPGR may regulate rhodopsin trafficking. Further work is required to pick apart these mechanisms.
The alternatively spliced, putative PDZ scaffold protein Whirlin is expressed in cochlear hair cells and the CC of photoreceptors [64,65]. It is encoded by the DFNB31 gene, and N-terminus mutations in Whirlin cause Usher's Syndrome Type 2, a recessive syndromic form of RP and non-congenital sensori-neural deafness . C-terminal mutations, however, segregate to a deafness-only phenotype . Whirlin has a role in cytoskeletal assembly both in inner ear stereocilia and in photoreceptors, where its interaction with the actin cross-linking protein espin is important in regulating the actin filament network in the PMC, defined by the presence of the proteins usherin, whirlin, or ADGRV1 [64–70,65]. Whirlin interacts with RPGR in the photoreceptor CC, and this link between RPGR and the Usher protein network is of particular relevance given the network's clear links to the actin cytoskeleton [63,71]. Though in one study, it was claimed that rhodopsin targeting to the outer segment is abrogated in whirlin-mutant mice, earlier studies revealed normal rhodopsin localisation [72,73]. At the PMC, post-Golgi vesicles are thought to dock and sort their cargoes [71,74]. Thus, it is possible that Whirlin, as part of the Usher protein network at the PMC, may facilitate targeting rhodopsin or other molecules to the CC. Additionally, whirlin is also found at the outer limiting membrane, where it is thought to interact with the polarity protein membrane-palmitoylated protein 1 (MPP1) via a PDZ domain interaction . As MPP1 is associated with the PALS1 and CRB1 polarity complex, this suggests that whirlin and other Usher proteins may also regulate actin to maintain apico-basal cell polarity .
Myosin VIIa is a member of the unconventional myosin superfamily of proteins and thus is an actin-binding molecular motor . Mutations in Myosin VIIa cause USH1B, the commonest form of Usher's Syndrome type 1 . Myosin VIIa localises to the CC in photoreceptors and appears necessary for rhodopsin transport through the CC . Indeed, abnormally high levels of rhodopsin accumulate in the CC in the absence of myosin VIIa, with normal transport appearing to be via actin filaments at the ciliary membrane [79,80]. However, a lack of retinal phenotype seen in Myosin VIIa mutant mice has compromised the field's attempts to determine the protein's precise role in the photoreceptor .
In humans, the full-length Espin protein contains an actin-bundling module and an extra actin-binding motif, both involved in binding actin filaments, in addition to a WH2 motif thought to bind monomeric actin (Figure 1) . Recently, a consanguineous Pakistani family segregating Usher's Syndrome Type 1 were shown to harbour an in-frame deletion in ESPN . In vitro, the mutation was shown to compromise microvillar elongation ability, which is believed to stem directly from Espin's ability to cross-link actin. Espin colocalises with Whirlin in the photoreceptor, and its expression pattern in the retina is altered when Whirlin is perturbed . While further work is required, this provides further evidence linking the Usher complex to actin regulation in the photoreceptor CC/PMC.
The mammalian photoreceptor outer segment is arguably the most specialised and architecturally complex cilium in the body, and as described above, disruption of actin-associated processes probably underlies the pathophysiology of many inherited retinal degenerative diseases. This complexity presents numerous challenges in understanding the precise role actin and its modulators play in normal photoreceptor function. First and foremost, much of our understanding of actin dynamics has arisen from elegant in vitro studies utilising cultured cells and live-cell imaging. However, unlike many other neuronal cell types for which there are passable immortalised cell lines, there are currently no photoreceptor cell lines that produce elaborated primary cilia approximating the photoreceptor outer segment, presenting a challenge for imaging outer-segment actin dynamics. A solution to this problem lies in the recent advances in the development of retinal organoids . Retinal organoids can now be grown in vitro from both embryonic and induced pluripotent stem cells, containing all major retinal cell types organised in distinct layers recapitulating in vivo retinal morphology. These models will provide excellent models for investigating actin dynamics, especially in retinal development. However, the utility of these models in studying retinal degeneration, especially late onset degeneration or those involving outer segment disc morphology defects, will be more limited, as these in vitro models still need to be refined to faithfully produce photoreceptors with outer segments as seen in vivo.
In vivo models including Drosophila, Xenopus, zebrafish, and mice have been invaluable in the study of retinal biology, including understanding disease mechanisms underlying RP. Advances in genome editing technologies, such as CRISPR, have allowed not only the generation of animals with human disease-causing mutations, but also generation of animals with endogenously epitope-tagged retinal proteins. Going forward, these animals will prove invaluable for investigating photoreceptor biochemistry through the use of proteomics. Additionally, when combined with advance imaging techniques, such as light-sheet microscopy and correlative light electron microscopy, these animal models will enable improved imaging within the retina. These types of studies will prove essential to unravel the complexities of actin regulation in the photoreceptor, in both normal and disease states.
This review has focussed on those genes linked either directly or indirectly to actin regulation in photoreceptors, with a specific emphasis on those linked to cilia and outer segment biology. Of course, actin-dependent processes regulate many aspects of normal physiology in all cells, including photoreceptors. Therein lies the challenge in dissecting which processes, when disrupted, lead to disease. Undoubtedly, disruption of actin will lead to defects in cell–cell contacts, synaptic transmission, trafficking, and intracellular signalling, all of which may contribute to pathogenesis in inherited retinal degenerations. Additionally, diseases such as RP may not originate due to defects in the photoreceptors themselves and also due to mutations in genes in other retinal cell types. The best example of this is mutations that lead to retinal pigmented epithelial (RPE) dysfunction. Many functions of the RPE depend on tight control of the actin cytoskeleton, such as the formation and maintenance of cell–cell contacts, secretion of neurotrophic factors, and phagocytosis of photoreceptor outer segments. Disruption of any of these functions leads to retinal degeneration. In summary, as in many tissues, actin is critical to normal physiology of the retina.
Future studies will better inform on how actin is regulated in different cellular compartments, in different cells, potentially leading to novel approaches to halt the progression of vision loss in inherited retinal degenerative disease.
epidermal growth factor
4.1 Ezrin Radixin Moesin
osteoclast-stimulating factor 1
periciliary membrane complex
retinal pigmented epithelium
retinitis pigmentosa GTPase regulator
R.M. is funded by the Wellcome trust and the Academy of Medical Sciences. T.W.H is funded by the Medical Research Council (MRC) Human Genetics Unit and a University of Edinburgh Chancellor's fellowship. This work was supported by grant funding from RP Fighting Blindness (GR588 — www.rpfightingblindness.org.uk) and a Million Dollar Bike Ride pilot grant from The Orphan Disease Center (MDBR-18-110-CRB1 — http://orphandiseasecenter.med.upenn.edu).
The Authors declare that there are no competing interests associated with the manuscript.