Abstract
Retinitis pigmentosa (RP), the most common form of inherited retinal degeneration, is associated with different groups of genes, including those encoding proteins involved in centriole and cilium biogenesis. Exome sequencing revealed a homozygous nonsense mutation [c.304_305delGA (p. D102*)] in POC5, encoding the Proteome Of Centriole 5 protein, in a patient with RP, short stature, microcephaly and recurrent glomerulonephritis. The POC5 gene is ubiquitously expressed, and immunohistochemistry revealed a distinct POC5 localization at the photoreceptor connecting cilium. Morpholino-oligonucleotide-induced knockdown of poc5 translation in zebrafish resulted in decreased length of photoreceptor outer segments and a decreased visual motor response, a measurement of retinal function. These phenotypes could be rescued by wild-type human POC5 mRNA. These findings demonstrate that Poc5 is important for normal retinal development and function. Altogether, this study presents POC5 as a novel gene involved autosomal recessively inherited RP, and strengthens the hypothesis that mutations in centriolar proteins are important cause of retinal dystrophies.
Introduction
Retinitis pigmentosa [(RP (MIM 268000)] is the most prevalent form of inherited retinal degeneration, affecting between 1: 2000 and 1: 7000 persons worldwide (1–3). RP is characterized by initial loss of rod function, which causes defective dark adaptation, and progressive involvement of both rods and cones. The fundus appearance of RP depends on the stage of retinal degeneration. In the earliest stages when electroretinography (ERG) reveals only rod dysfunction, the fundus is usually normal in its appearance. Arterial narrowing, fine dust-like intra-retinal pigmentation and loss of pigmentation in the retinal pigment epithelium are the earliest changes observed. With disease progression, the appearance of the characteristic ‘bone spicules’ can be observed, derived from the intra-retinal clumping of melanin (4,5).
RP is a genetically heterogeneous disease and can be inherited in an X-linked, autosomal-dominant or autosomal-recessive pattern. So far, mutations in at least 60 different genes have been identified to cause RP (RetNet, http://www.sph.uth.tmc.edu/RetNet/; date last accessed September 29, 2017) (6). In most cases, the resulting phenotype is restricted to the retina (non-syndromic RP). However, around 20–30% of patients suffer from additional, non-ocular symptoms (syndromic RP). It is well-established that different (types of) mutations in a single gene can result in a different phenotypic outcome. For example, mutations in USHRN (previously USH2A) can result in either non-syndromic RP or Usher syndrome (deafblindness) (7,8). Currently, over 30 syndromes with RP as a phenotypic hallmark are recognized (2).
Photoreceptors are organized into several compartments. Photoconversion occurs in the outer segment, metabolic and regulatory processes take place in the cell body (inner segment). The synapse relays the electrical signals resulting from photoconversion to other retinal neurons (9). The outer and inner segment are linked by a specialized primary cilium (connecting cilium), which act as a transport highway for proteins involved in phototransduction, or essential for outer segment generation and maintenance. The known RP-associated genes encode proteins with variable function, such as rod-specific phototransduction (RHO; MIM *180380), transcriptional regulation (e.g. NRL; MIM *162080) pre-mRNA splicing (e.g. PRPF31; MIM *606419) and ciliary function (e.g. C2orf71; MIM * 613428). The molecular mechanisms through which loss of function of photoreceptor proteins with different functions result in retinal degeneration remains elusive. Mutations in genes encoding components of the connecting cilium and associated transport machinery are also an important cause of syndromic retinal degeneration such as Joubert Syndrome and Bardet-Biedl Syndrome (10,11).
We report here the identification of a homozygous protein-truncating mutation in POC5, encoding Protein Of Centriole 5 (POC5), as the underlying cause of syndromic RP in a family of Moroccan/Yemenite Jewish origin. Centrosomes are comprised of two orthogonally arranged centrioles that are surrounded by pericentriolar material. In ciliated cells, the mother centriole will form the basal body from which the ciliary axoneme extends (12). The importance of centriolar and centrosomal proteins in retinal development and function is well-recognized. Mutations in several of these proteins are known to cause inherited retinal degeneration (13,14). Previously, missense single nucleotide variants in the C-terminus of POC5 were found to contribute to the occurrence of idiopathic scoliosis (15). We used a comprehensive zebrafish knockdown approach to demonstrate that Poc5 is essential for correct photoreceptor outer segment formation, and for photoreceptor function in zebrafish. These findings support the causality of the mutation in POC5 for recessively inherited RP.
Results
Clinical features
The proband in this study was a 19-year-old girl, descent of healthy Moroccan/Yemenite Jewish parents. She was born at full term after an uneventful pregnancy. Birth weight was 2400 grams, which is small for the gestational age. She had no dysmorphic features on physical exam, but she had short stature and microcephaly with a head circumference of −3.8SD (father’s circumference −2SD, mother’s head circumference within normal range). Her final height is 143 cm (−3SD for age). The height of her mother, father and grandmother are 156cm, 165cm, and 141 cm, respectively. She was diagnosed with early, fast puberty. However, endocrine evaluation revealed normal hormonal profiles.
From the age of 6 years, she had glasses due to myopia, and later mild central posterior subcapsular cataract/polar cataract was identified (no intervention was required). At age 14, she was diagnosed with RP. Full-field electroretinography (ffERG) revealed a rod > cone pattern of injury: non-detectable rod responses, practically non-detectable scotopic mixed rod-cone responses, but still a sizable photopic cone response: 30 Hz cone flicker amplitudes of 20 microvolts (our lower limit of normal being 60 microvolts) were recorded, with a delayed implicit time of 38–40 ms (upper limit of normal being 33 ms). Visual acuity at age 19 (on her last visit in 2016) was 20/40. Goldmann visual fields were constricted at that time to 10–20 degrees from fixation with the IV4e target. Fundus findings included salt and pepper-like hyperpigmentation in the midperiphery accompanied by early bone-spicule-like pigmentary changes, mild optic disc pallor, and mild attenuation of retinal vessels, findings that fall within the spectrum typically associated with RP (Fig. 1A).
Fundus imaging in patient with POC5-associated retinal degeneration and Family Pedigree. (A) Wide field (OPTOS) pseudocolor imaging of the right and left eyes at 17 years of age (left, top row) shows salt and pepper changes and early bone spicule-like pigmentary changes in the retinal midperiphery with relative preservation of the posterior poles. Corresponding Fundus Autofluorescence (FAF) images (second row) show hyperfluorescence in the macular region with hypofluorescent spots throughout the more peripheral retina. A ring of hyperfluorescence surrounding the maculas can be appreciated on Heidelberg FAF images of the post poles taken at age 19 (third row). Optical coherence tomography (OCT) scans at the same age show presence of the photoreceptor layer in the area of the foveas without cystoid changes, but thinning and loss of the photoreceptor outer nuclear layer in extrafoveal regions (right, top two rows). The findings are largely symmetric between the two eyes. (B) Family pedigree.
Systemically, she suffered from recurrent glomerulonephritis episodes with focal segmental glomerulonephritis on one kidney biopsy and immunoglobulin A (IgA) nephropathy on another biopsy. Due to hematuria and proteinuria she was treated with angiotensin-converting-enzyme inhibitors. Her blood pressure and kidney ultrasound are in normal range.
Recurrent episodes of elevated creatine phosphokinase (CPK) and muscle pain/cramps, increased in winter time. Electromyogram and CPK levels were normal between episodes. Metabolic evaluation showed no abnormalities. Echocardiogram and electrocardiogram were performed without any findings.
Genetic evaluation
Genetic evaluation was done in tiers. First a chromosomal microarray (CMA) was performed in order to rule out chromosomal microdeletions and duplications that could cause RP and other systemic disease. No deletions/duplications were found. Also known Moroccan and Yemenite Jewish RP founder mutations were tested: FAM161A [c.1355delCA] and CERKL [c.238 + 1G > A]. The tested mutations were not found. Since multiple genes are known to cause both non-syndromic and syndromic RP, a Clinical Trio Whole exome sequencing was performed. Variants in three genes previously associated with retinal degeneration were found.
The clinical lab reported two variants in ABCA4 (NM_000350), a gene previously associated with retinal degeneration. One variant in ABCA4 [c.5882G > A; p.G1961E] was inherited from the mother, one from the father [c.466A > G; p.I156V]. The p.G1961E variant is one of the most frequent ABCA4 variants that were in the past related to Stargardt disease. Based on its high prevalence in control individuals (minor allele frequency 0.0047 in non-Finnish Europeans), and its underrepresentation in homozygous STGD1 cases, it was deemed to have a mild effect on ABCA4 function (16) and it is often associated with late-onset Stargardt disease with foveal sparing [Zernant et al. (2017) (17)—and refs therein]. This variant has not been found thus far in early-onset STGD1 cases that sometimes resemble cone-rod dystrophy or RP. The p.I156V variant shows similar allele frequencies in all 3, 270 reported European STGD1 cases and 33, 231 non-Finnish European controls (0.00168 vs. 0.00172, respectively) rendering it very likely a benign variant (16).
In-house bioinformatic re-analysis of the complete exome data (Supplementary Material, Tables S1 and S2) revealed a heterozygous variant in AIPL1 [NM_001033054; c.G716T (p.R239L)]. This variant is worth mentioning, as there is one report that describes autosomal dominant retinal degeneration due a 12bp deletion in AIPL1 (18). The proband inherited the variant in AIPL1 from her mother, who appeared asymptomatic upon clinical and retinal examination. Furthermore, the variant has a high frequency in in-house control exomes and in the gnomAD database (MAF 0.2369% http://gnomad.broadinstitute.org, November 28, 2017) (Supplementary Material, Table S1). Therefore, the c.G716T (p.R239L) variant in AIPL1 was regarded as benign.
Third, a homozygous variant of unknown significance was reported in POC5 (NM_0010992271; c.304–305delGA; p.D102*). This variant has not been published or associated with RP before. Validation and segregation analysis identified both parents and the brother as carriers. As it was a rare nonsense mutation, it was considered as the primary cause underlying the observed phenotype. The homozygous mutation is meaningful despite the non-consanguineous parents as it is known the different branches of the Jewish diaspora share a genetic background shown by identical-by-descent (IBD) segment (19,20).
Expression and localization of POC5
Public databases of protein and mRNA expression show ubiquitous expression of POC5. The FANTOM5 RNA expression dataset, part of the human protein atlas, is the only dataset that includes retinal tissue (https://www.proteinatlas.org; date last accessed November 29, 2017) (21). It shows that POC5 expression is particularly high in neural and retinal tissue, and in the testis (Fig. 2A). Using reverse transcriptase (RT)-PCR analysis in 21 different tissue samples, we validated the tissue distribution of POC5 in our own lab, and found similar results, including robust POC5 expression in the retina (Supplementary Material, Fig. S1A). The zebrafish genome contains one POC5 orthologue (poc5, XM_685988.8), of which the encoded protein has 42% amino acid identity compared with human POC5. The amino acid similarity is even stronger, particularly at the centrin-binding domains (Supplementary Material, Fig. S2A). To evaluate whether zebrafish is suitable to study the role of POC5 in the retina, we first investigated the expression of poc5 in zebrafish larvae. We could identify poc5 transcripts throughout development and amongst the maternally contributed mRNA of oocytes (Supplementary Material, Fig. S1B).
POC5 expression studies (A) POC5 mRNA expression. Dataset is obtained from the protein atlas, FANTOM5 database. POC5 is ubiquitously expressed. Highest expression is found the testis, neural and retinal tissue. See Supplementary Material, Fig. S1A for RT-PCR validation. (B) Poc5 Immunohistochemistry. Mouse retinal sections (upper panel), incubated with antibodies recognizing Poc5 and Centrin, revealed distinct immunoreactivity of Poc5 at the connecting cilium of photoreceptors. None of the other retinal cells showed evidence of Poc5 expression. Detailed magnification (right-most panels), revealed Poc5 immunoreactivity along the entire connecting cilium (CC; equivalent of the transition zone in primary cilia). Similarly, the larval zebrafish retina (lower panel) shows Poc5 localization (red puncta) at the base of the connecting cilium of photoreceptor cells, visualized with anti-centrin antibodies (green puncta). Scale bar = 10 µm. Corresponding negative controls are shown in Supplementary Material, Figure S5A.
POC5 is well established as a centriolar protein. To investigate if POC5 also localizes to the centriolar/basal body in photoreceptors, we performed immunohistochemistry in mouse and zebrafish retina (Fig. 2B). Immunohistochemistry of the mouse retina, using an antibody directed at the C-terminal region of POC5, revealed distinct localization of Poc5 at the connecting cilium of photoreceptors. Upon higher magnification, it becomes clear that the Poc5 immunoreactivity overlaps with centrin immunoreactivity. Multiple centrin proteins exist, and are found in different parts of the connecting cilium, basal body and centrioles (22). The antibody used here recognizes all centrins, and therefore serves as a marker for the entire connecting cilium, (counterpart of the transition zone of primary cilia), the basal body and the daughter centriole. Poc5 localized predominantly to the connecting cilium. We observed a similar ciliary localization of Poc5 in the retina of zebrafish larvae, as there is a clear overlap in localization between Poc5 and centrin immunoreactivity. In the larval photoreceptor, the connecting cilium is very short, and therefore immunohistochemistry with the antibody directed against centrin results a punctate staining (Fig. 2B).
Knockdown of zebrafish poc5 results in decreased retinal function
The distinct ciliary localization of Poc5 in the zebrafish retina allows us to further study the effect of mutations in POC5 in zebrafish larvae. Morpholino (MO) knockdown has been previously conducted for poc5 in zebrafish in relation to idiopathic scoliosis (15). The authors showed that poc5 knockdown, as well as the introduction of human missense variants in POC5, results in the development of scoliosis-like curvatures in the body axis of zebrafish. Zebrafish morphants did not survive past 72 hours post-fertilization (hpf). Although poc5 morphants showed additional phenotypes such as small eyes and head, these phenotypes were not further investigated. Investigation of Poc5 in the retina requires the morphants to survive up to 4dpf. Therefore, we first evaluated the effect of different doses of poc5 splice-blocking morpholino on pre-mRNA splicing, survival and larval morphology.
Injection of the splice-blocking MO resulted in two alternatively splice transcripts, including either the intron upstream of exon 5, or both introns flanking exon 5. In both transcripts, the incorporation of intronic sequence results in a premature translation terminal codon (Supplementary Material, Fig. S3). Near-complete alternative splicing of poc5 was observed when 8ng of splice-blocking MO was injected. At this dose, survival of poc5 morphants up to 4dpf was similar to that of wildtypes and injection controls. A second MO, blocking the translation initiation site (ATG) of poc5, was employed to further establish the specificity of poc5 knockdown. For this MO, injection of up to 8ng did not affect survival, and therefore allowed for the investigation of retinal function. Both MOs only sporadically resulted in the previously published body-axis curvature, other severe developmental defects at a dose of 8ng (Fig. 3A and B;Supplementary Material, Fig. S4A). The tested doses furthermore did not affect the size of the morphants eyes (Supplementary Material, Fig. S4B and C). We visually examined the pronephros of poc5 morphants, but did not observe any obvious defects. We selected a dose of 8ng for both MOs to assess the effect of poc5 knockdown on retinal function, as this did not result in any of the typical phenotypes that are associated with MO toxicity and off-target effects.
Effect of poc5 knockdown on zebrafish development and retinal morphology. (A) Overview of larval morphology after injection with the Poc5 ATG MO. The curved tail is the only morphological phenotype of the curved morphants. The severely affect morphants also suffer from poor eye development, pericardial edema and a truncated or curved tail. (B) Phenotype distribution. Results of two biological replicates of n = 50–70 each (one biological replicate for 2ng and 4ng MO). (C) Poc5 morphants show decreased outer segment (OS) length. The white brackets show examples of the way outer segments were measured. The specificity of this phenotype was confirmed by co-injection of wildtype POC5 mRNA, which could rescue the phenotype. Retinal lamination is normal in poc5 morphants. scale bar = 0.5 mm. (D) Quantification of photoreceptor outer segment lengths (**P < 0.01; ***P < 0.001, Kruskal Wallis multiple comparison test; average of 240–280 OS measurements per group, n = 12–14 eyes).
poc5 knockdown affects photoreceptor development and function
The effect of poc5 knockdown was assessed in the retina of 4dpf morphants for both MOs. Immunoreactivity of Poc5 was significantly decreased for both morpholinos (Supplementary Material, Fig. S5A and B). However, residual Poc5 immunoreactivity could be detected in morphants treated with the splice-blocking MO. This could be caused by either Poc5 translated from maternally contributed mRNA, the MO being metabolized or the incomplete alternative splicing that we observed upon RT-PCR (Supplementary Material, Figs S1B and S3).
Retinal lamination and morphology was normal in morphants and controls. Loss of Poc5 function did result in defective outer segment formation, as the average length of photoreceptor outer segments was significantly decreased in both poc5 morphants (Fig. 3C and D;Supplementary Material, Fig. S6). This shows that also for the splice-blocking MO, knockdown was sufficient during retinal development to establish a phenotype in photoreceptors. Co-injection of full-length wildtype human POC5 mRNA together with either MO rescued the outer segment phenotype completely, confirming the specificity of this phenotype. As the first nucleotides following the translation initiation codon are highly conserved between human and zebrafish (Supplementary Material, Fig. S2B), the first 22 nucleotides of the POC5 construct were saturated with synonymous substitutions to prevent binding of the ATG morpholino.
Retinal function was assessed through measurements of larval movement in response to a light stimulus, the visual motor response (VMR). As previously published, the dark-to-light VMR is initiated through retinal signaling, while the dark-to-light VMR is independent of the retina (23). The results of a typical experiment show a decreased VMR response in poc5 morphants compared with control injected siblings (Fig. 4A). The VMR response occurs immediately after the light is switched on. For both MOs, the combined results of multiple experiments show a larval response in the first second after the light stimulus (Fig. 4B). Co-injection of either MO with wildtype human POC5 mRNA could rescue the phenotype, and restore the VMR to normal levels.
poc5 knockdown results in a decreased visual motor response. (A) typical result of a single VMR experiment. The distance moved is plotted in the 20 s before light-on, and 30 s after (n = 21–47). The black line represents the control injected larvae, the grey line with open circles the poc5 morphants, and the dotted line with open circle the larvae co-injected with poc5 MO and wildtype POC5 mRNA. (B) Average VMR of all experiments using the translation-blocking or splice-blocking MO. The VMR is plotted as the average velocity of the larvae immediately after light-on. The average VMR is reduced in poc5 morphants, but not controls. Co-injection of POC5 mRNA with the either poc5 MO restore the VMR. (*P < 0, 05; **P < 0, 01, Mann–Whitney U test; translation blocking MO n = 30–75, splice-blocking MO n = 47–71).
In line with the patient’s short stature, microcephaly and the association of POC5 with idiopathic scoliosis, we investigated skeletal development in 4dpf zebrafish larvae. No changes were found between poc5 morphants and controls in the morphology of the cartilaginous structures of the jaws, gill arches and vertebrae. The only difference was found in the notochord tip, which was more frequently mineralized in poc5 morphants (41%, n = 17) compared with control injected larvae (11%, n = 18). However, since notochord tip is mineralized by a process unique for teleosts (24), and this phenotype could not be rescued by co-injection of wildtype human mRNA, this phenotype is difficult to interpret and might not be specific for poc5 knockdown (Supplementary Material, Fig. S7).
Discussion
Here, we report a homozygous nonsense mutation in POC5 in a patient with a syndromic form of RP that includes kidney disease and microcephaly. Our functional studies demonstrate that loss of POC5 function causes retinal degeneration in a zebrafish model, confirming POC5 as the causative gene for the patient’s retinal disease. We can however not exclude that the ABCA4 variants have a modifying effect on the phenotypic outcome of the retinal disease in this patient.
POC5 is an evolutionarily conserved protein that is localized at the distal part of the centrioles (25). The protein contains Sfi1p-like centrin-binding repeats (CBR), through which it binds to centriolar proteins centrin 2 and centrin 3. POC5 is recruited to the procentrioles during the G2/M phase, but is not essential for procentriole assembly. It is however required for assembly of the distal half of the centriole, centriole elongation and cell cycle progression (25).
In light of the patient’s retinal phenotype, we investigated the expression of Poc5 in the retina of mouse and zebrafish. In both species, Poc5 could only be detected at the connecting cilium photoreceptor cells, and not in any of the other retinal cells. The centrin binding repeats of POC5 are known to bind centrin 2 and 3, both of which are found in the photoreceptor connecting cilium, the basal body and the daughter centriole (22,25). Based on the role of POC5 in centrioles, we expected to find Poc5 at the basal body and daughter centriole of the cilium. However, the immunofluorescent signal resulting from the Poc5 antibody is strongest in the connecting cilium, the equivalent of the transition zone in primary cilia, where it can interact with centrin 2 and 3. While Poc5 is known to be involved in centriole assembly and elongation and cell cycle progression, the localization of Poc5 in the connection cilium suggests an additional function for Poc5, possibly unique for the post-mitotic photoreceptors.
MO-induced blocking of poc5 translation in zebrafish demonstrates that the protein is required for normal photoreceptor development and function. The immunofluorescent signal of centrin was still present, suggesting that proper localization of centrin in the connecting cilium is not dependent on Poc5. While retinal lamination and morphology were still normal, the length of photoreceptor outer segments was significantly decreased in poc5 morphants as compared with controls. We previously observed similar defects in photoreceptor outer segment formation in zebrafish morphants for Poc1b and other ciliary components (14,26). POC1B, like POC5, is a centriolar protein involved in syndromic and non-syndromic retinal degeneration (14,27). Our histological findings correlate with the results of the VMR, which was decreased in poc5 morphants, but not absent. The VMR, the locomotor response directly after a dark-to-light transition, is well established as a readout of retinal function (23,28). Co-injection of wildtype human POC5 mRNA could rescue the observed retinal phenotype, confirming specificity of our knockdown approach. Together, these data show that loss of Poc5 function affects retinal function, and present mutations POC5 as a novel cause of RP.
We further examined the poc5 morphants for additional defects in relation to the patient’s syndromic phenotype, but found no obvious defects in the kidney, skeleton and length of the morphants. By itself, these results are inconclusive for the patient’s renal phenotype, as to the best of our knowledge, it cannot be studied in zebrafish as it is inflammatory in nature.
Following notion that heterozygous variants in the POC5 gene were found to contribute to the occurrence of idiopathic scoliosis (15), it is striking that the patient does not display any defects in her axial skeleton. Furthermore, the scoliosis phenotype was not observed in either of our zebrafish poc5 morphants. In the study of Patten et al.,poc5 morphants die at 72 h post fertilization (hpf). To measure retinal function, larvae need to survive until 96hpf, which may suggest that incomplete knockdown is required for extended survival. However, in our poc5 ATG morphants, knockdown was complete up to 4dfp, as no Poc5 could be detected in the retina upon immunohistochemistry. Body axis curvatures in zebrafish morphants in often associated with MO toxicity or off-target effects. Rescue experiments, combined with a second MO targeting the same gene, are required to establish specificity of the phenotype. While we could rescue the phenotype induced by both MOs with human POC5 mRNA, the scoliosis study lacks certain validations. In contrast with our approach, there is a strong possibility that the ATG MO binds directly to the human POC5 mRNA that was used in the scoliosis study. As such, the POC5 mRNA interferes with directly with the MO, rather than rescuing the phenotype through production of human POC5. Furthermore, rescue experiments were not published to confirm the specificity of splice-blocking MO. Based on our phenotypic data of the poc5 morphants and the lack of the proper controls in the study of Patten et al., we have no indications that loss of Poc5 function result in deformities of the axial skeleton in zebrafish.
For years, our patient’s glomerulonephritis was presumed to be ‘post-streptococcal’. Once the retinal degeneration was identified, the suspicion that the two conditions are linked was raised. Mutations in genes involved in centriolar proteins give rise to numerous form of syndromic and non-syndromic retinal degeneration. In most cases, these mutations disrupt key components of cilium formation, signaling or transport. These disorders are collectively called ciliopathies, and share several hallmark features including retinal degeneration, kidney cysts and intellectual disabilities (10,29). Two centriolar proteins that are implicated in the development of ciliopathies are POC1A and POC1B. A protein-truncating mutation in POC1A was shown to cause a ciliopathy by Shaheen et al. 2012 in three families with primordial dwarfism (30). Another mutation in POC1A was described as a cause of SOFT syndrome, an osteocutaneous syndrome, which includes short stature, onychodysplasia, facial dysmorphism and hypotrichosis (31). Mutations in POC1B are described to cause either non-syndromic autosomal recessive cone dystrophy or cone-rod dystrophy (14) or a severe syndromic retinal ciliopathy (27), underlying the variation in clinical representation that can exist within a single gene.
The patient’s RP could be a clinical representation of ciliary defects, but most of her other symptoms are not typical ciliopathy features. We investigated the presence of additional variants that may explain the non-retinal phenotypes. Bioinformatic reanalysis of the clinical exome revealed a compound heterozygous variant in FAT1 was also identified (see Supplementary Material, Table S1). The missense variant was inherited from the mother and the frameshift insertion was found the novo in the proband. Mutations in FAT1 are reported as a potential cause of glomerulotubular nephropathy in humans (32). The clinical findings associated with FAT1 mutations included a combination of steroid resistant nephrotic syndrome (SRNS), tubular ectasia, haematuria, but not retinal degeneration. Our proband did not have nephrotic syndrome or histologic findings on kidney biopsy, but we cannot rule the involvement of the FAT1 variants out.
Despite the clear localization in the photoreceptor connecting cilium, POC5 was not identified as a ciliary protein in the large-scale experiments that defined the ciliary protein landscape (33–36). On the other hand, there are reports of mutations in genes encoding proteins involved in centriole biogenesis that result in syndromic disorders without the hallmark features of ciliopathies (37,38). For example, mutations in PLK4 (Polo-like kinase 4), also encoding an important centriolar protein, cause syndromic microcephaly with or without ciliopathy features (35). PLK4 is a protein kinase involved in centriole duplication and as such, functions in the same pathway as POC5 (39). There is a significant phenotypic overlap between the proband and the published PLK4 cases (microcephaly, cataract, retinal degeneration). However, the type of retinal degeneration differs between cases. It is interesting to speculate on the presence of a pathophysiological relationship between POC5 and PLK4 mutations. A recent paper shows that human microcephaly protein RTTN directly interacts with STIL in centriole assembly (40). Down regulation of RTTN blocked the loading of the centriolar proteins POC1B and POC5 to the distal-portion centrioles. RTTN is furthermore suggested to act as an upstream regulator of CEP295, which was recently reported to be essential the recruitment of POC5 and POC1B to the distal portion of centrioles (41). Furthermore, a protein module of PLK4, STIL and SAS6 was shown to function at the core of centriole duplication (42,43). A functional relationship between POC5 and PLK4 might exist through RTTN and STIL. This is further substantiated by the identification of mutations in RTTN in patients with primary microcephaly and primordial dwarfism (44). It can therefore be extrapolated that the severe microcephaly of our proband, and maybe to some degree her height, could be related to the POC5 mutation. Further investigation is required to elucidate the phenotypic relation between POC5, RTTN and PLK4. While our zebrafish knockdown model is highly suitable to investigate the retinal function of Poc5 (14,45), additional models are required to scrutinize the full extent of POC5 pathophysiology.
In conclusion, whole exome sequencing led to the identification of a homozygous POC5 protein-truncating mutation in a patient with syndromic RP. Loss of Poc5 function in zebrafish results in early-onset retinal dysfunction, confirming the causality of the mutations in POC5 for retinal degeneration in the proband.
Materials and Methods
Subjects and clinical evaluation
All biological samples were collected after written informed consent of the patient, in agreement with the regulations of the institutional review boards (Beilinson Medical Center, Petah Tikva Israel and Hadasah Ein Karem, Jerusalem, Israel) and the Helsinki guidelines.
Genetic analysis and variant identification
Trio clinical exome sequencing including mitochondrial DNA (including parents and proband samples) was outsourced to Ambry Genetics (Aliso Viejo, CA, USA). Genomic DNA was isolated from blood samples, and prepared using the SeqCap EZ VCRome 2.0 (Roche NimbleGen) or the IDT xGen Exome Research Panel V1.0. Each DNA sample is sheared, adaptor ligated, PCR-amplified and incubated with the exome baits. Captured DNA is eluted and PCR amplified. Final quantified libraries are seeded onto an Illumina flow cell and sequenced using paired-end, 100 cycle chemistry on the Illumina HiSeq 2500. Initial data processing, base calling, alignments and variant calls are generated by various bioinformatics tools. Data are annotated with the Ambry Variant Analyzer tool (AVA), including: nucleotide and amino acid conservation, biochemical nature of amino acid substitutions, population frequency, and predicted functional impact. The data were filtered for small insertions and deletions, canonical splice site alterations, and non-synonymous alterations. The following sites are used to search for previously described gene mutations and polymorphisms: the Human Gene Mutation Database (HGMD), the Single Nucleotide Polymorphism database (dbSNP), 1000 genomes, and online search engines (e.g. PubMed). Variants are then filtered further based on the recessive inheritance pattern of this family. All candidate alterations underwent confirmation by automated fluorescence dideoxy (aka ‘Sanger’) sequencing.
The exome raw data were reanalysed in-house. The BWA-MEM tool (http://bio-bwa.sourceforge.net; date last accessed November 25, 2017) was used to map the reads, and a combination of several variants callers (Samtools, http://samtools.sourceforge.net; Freebayes, https://github.com/ekg/freebayes; GATK, https://software.broadinstitute.org/gatk and Pindel, http://gmt.genome.wustl.edu/packages/pindel/; date last accessed November 25, 2017) were used to call the variants and their genotypes. Allele frequency threshold for filtering the polymorphic variants was set to 1% in any population from public domain databases (ExAC, gnomAD, GME) and in local database of diverse Israeli populations (that contains about 1K exomes). Variants with total homozygote count greater than 2 were filtered out. The resulting candidates were validated manually, and subsequently segregation analysis was performed with Sanger sequencing.
No additional mutations could be identified within the exome data of approximately 500 unsolved cases of inherited retinal degeneration.
PCR amplification and direct sequencing of POC5
Genomic DNA was isolated from venous blood, collected in EDTA 5% tubes with a Gentra Puregene Blood Kit (Qiagen). The POC5 c.304_305delGA variant was tested by Sanger sequencing using primers 5′-CCCCAGCATGTTGTATTTTG-3′ (forward), and 5′-TGCTGAAATTATTCCTCTTACTACATA-3′ (reverse) [NM_001099271.1, ENST00000428202.6 according to the Recommendations for the Description of DNA Sequence Variants (Ensembl Browser)]. Both strands of the polymerase chain reaction (PCR) products were sequenced with BigDye Terminators on an ABI 3100 sequencer (Applied Biosystems, Foster City, CA, USA). Sequence chromatograms were analysed using SeqScape software version 1.1 (Applied Biosystems).
Tissue distribution
The expression of POC5 in human tissues was assessed on cDNA generated from a Human Total RNA Master Panel (cat.no. 636643, lot.no. 8101369A; Mountain View, CA, USA), using primers located in exons 1 and 4. PGM1 (MIM 171900) was used as a positive control. Primers are listed in Supplementary Material, Table S3.
Zebrafish breeding and egg collection
Tupfel long fin zebrafish were bred and raised under standard conditions (46). Both adult and larval zebrafish were kept in a light-dark regime of 14 h light: 10 h darkness. All experiments were carried out in accordance with European guidelines on animal experiments (2010/63/EU). Zebrafish eggs were obtained from natural spawning and reared at 28.5 °C in E3 embryo medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, and 0.33 mM MgSO4), supplemented with 0.1% methylene blue.
Morpholino knockdown
We searched the zebrafish genome for orthologues of the human POC5 gene, and identified poc5 (XM_685988.8) as the only orthologue. Antisense morpholino oligonucleotides (MO), obtained from GeneTools (Philomath, OR, USA), were designed to block the translation-initiation (ATG) site of poc5 (5’-TCTCCCCTTCATCTGATGACATCTT-3’) and the splice donor site of poc5 exon 5 (5’-AACCGCAAGTGCAATACAAACCTTA-3’). Exon 5 of poc5 is one of few exons result in an altered reading frame when excluded from the transcript. Furthermore, the splice donor site of poc5 exon 5 is only target for against which a splice-blocking MO can be direct for knockdown purposes based on the design criteria of the manufacturer. A morpholino directed against a human β-globin intronic mutation (5’-CCTCTTACCTCAGTTACAATTTATA-3’) was used as a standard negative control. Morpholinos were diluted and injected as previously described (14). A minimum sample size of 60 embryos per condition was used in each experiment. At least two biological replicates were performed per knockdown experiment. All experiments included a group of control injected embryos in which the negative control MO was delivered at a dose matching the highest dose of poc5 MO.
To determine the efficiency of splice-blocking, RNA was isolated from 25 uninjected, 25 control MO-injected and 25 splice MO-injected embryos (2 dpf) using Trizol reagent (Thermo Fisher Scientific, Waltham, MA, USA) according to manufacturer’s instructions. Here, 250 ng of total RNA was used to produce first-strand cDNA. Reverse transcription was performed using the Superscript II cDNA synthesis kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. Subsequently, PCR analysis was performed using primers on poc5 exon 4 and exon 6 (Supplementary Material, Table S3). The resultant amplicons were analysed using Sanger sequencing.
The complete coding sequence of human POC5 isoform 1 (encoding 575 amino acids; NCBI reference sequence NP_001092741), was amplified from Marathon-ready retina cDNA (Thermo Fisher Scientific), and cloned in the pCR4_TOPO vector (Thermo Fisher Scientific). mRNAs were transcribed with the T3 mMESSAGE mMACHINE Kit (Thermo Fisher Scientific) according to manufacturer’s instructions and co-injected with MOs as described above. For the rescue experiment using a translation blocking MO, synonymous substitutions were introduced in the first 22 nucleotides to prevent binding of the MO to the synthetic POC5 mRNA.
At 4 dpf, embryos from each group, including controls and wildtypes, were analysed for overall body morphology, renal cysts, eye size, retinal histology and visual motor response (VMR). Poc5 knockdown was validated with anti-Poc5 immunostaining in 4-day-old morphants.
Photoreceptor outer segment staining and measurement
Photoreceptor outer segments of larvae of 4 dpf were visualized with boron-dipyrromethene (1: 5000), as described by Dona et al. (26). Outer segments of both eyes were measured with Adobe Photoshop CC 2015 (Adobe, San Jose, CA, USA). For each knockdown experiment, eyes of 4 to 7 fish were measured.
Immunostaining and microscopy
For histological analysis, zebrafish larvae were rinsed in 30% sucrose in PBS and directly frozen in Tissue Tek (without fixation). Immunohistochemistry was performed on coronal sections of four-day-old morphants and age-matched controls. For histological analysis of the mouse retina, eyes of 2-month-old mice were frozen in Tissue Tek [without fixation. Sections were incubated with antibodies directed against Poc5 (1: 500; cat.no. A303341A, Bethyl Laboratories, Montgomery, TX, USA) and pan-centrin 20H5 (1: 500; cat.no. 04–1624, Merck Millipore, Darmstadt, Germany)]. Sections were washed with PBS, permeabilized for 20 min in 0.01% (v/v) Tween-20 in PBS, and washed again with PBS. Next, sections incubated in blocking buffer (10% normal goat serum, 2% BSA in PBS), followed by overnight incubation with primary antibodies in blocking buffer at 4 °C. After washing with PBS, secondary antibodies were incubated in blocking buffer for 60 min at room temperature. Samples were counterstained with DAPI and mounted with Prolong Gold (Life Technologies). For all sections, Alexa 488 and 568-conjugated goat anti-mouse and goat anti-rabbit secondary antibody were used (1: 500; Thermo Fisher Scientific).
Quantification of Poc5 immunofluorescence
The intensity of Poc5 immunofluorescence was measured using FIJI software version 1.47v (47). First, the outer segment layer was selected and isolated from the picture based on the centrin immunofluorescence signal. Subsequently, a mask was made based on the centrin staining using the ‘Find Maxima’ option (noise = 50), and dilated five times. To find the exact location of Poc5 immunofluorescence, the centrin mask and Poc5 layer were combined. Find Maxima (noise = 10) was used to identify the Poc5 immunofluorescence within the centrin mask. The resulting mask was dilated three times and touching objects were separated using the watershed option. Subsequently, the maximum gray value of the identified regions was measured on the original image of Poc5 immunofluorescence (Analyse Particles option; size = 0–50, pixel circularity = 0.00–1.00).
Zebrafish retinal function measurements
The visual motor response (VMR) was measured through live video tracking of zebrafish larvae using the DanioVision system and the EthoVision XT11 software from Noldus Information Technology (Wageningen, the Netherlands). Zebrafish larvae of 4 dpf were placed individually in wells of a 48-well plate containing 0.25 ml E3 medium. Prior to video tracking, larvae were placed in the dark in the DanioVision system for 20 min to acclimate. The VMR protocol consisted of a 10-min period in the dark, followed by 10 min white light, 10 min darkness and 5 min light, to measure the locomotor behavior during light-to-dark transitions.
Supplementary Material
Supplementary Material is available at HMG online.
Acknowledgements
The authors gratefully acknowledge Dr Miriam Schmidts for fruitful discussion on the ciliary nature of the patient’s phenotypes, and Tom Spanings for excellent fish husbandry. The authors are grateful for the support of The Milken Family Foundation & The Lowell Milken Family Foundation.
Conflict of Interest statement. None declared.
Funding
Dutch patient foundations ‘Stichting Blindenhulp’, ‘Rotterdamse Stichting Blindenbelangen’, ‘Stichting tot Verbetering van het Lot der Blinden’ and ‘Stichting voor Ooglijders’ (awarded to EvW, FC, LH and EdV), the Foundation Fighting Blindness (PPA-0517–0717-RAD to EvW and FC), Israelian Ministry of Science, Technology & Space (grant number 3–13545 to P-SY), and the Yedidut Research Fund (to EB).
References
Author notes
The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors.
The authors wish it to be known that, in their opinion, the last two authors should be regarded as joint Last Authors.
