Overlap of abnormal photoreceptor development and progressive degeneration in Leber congenital amaurosis caused by NPHP5 mutation

Abstract

Ciliary defects can result in severe disorders called ciliopathies. Mutations in NPHP5 cause a ciliopathy characterized by severe childhood onset retinal blindness, Leber congenital amaurosis (LCA), and renal disease. Using the canine NPHP5-LCA model we compared human and canine retinal phenotypes, and examined the early stages of photoreceptor development and degeneration, the kinetics of photoreceptor loss, the progression of degeneration and the expression profiles of selected genes. NPHP5-mutant dogs recapitulate the human phenotype of very early loss of rods, and relative retention of the central retinal cone photoreceptors that lack function. In mutant dogs, rod and cone photoreceptors have a sensory cilium, but develop and function abnormally and then rapidly degenerate; L/M cones are more severely affected than S-cones. The lack of outer segments in mutant cones indicates a ciliary dysfunction. Genes expressed in mutant rod or both rod and cone photoreceptors show significant downregulation, while those expressed only in cones are unchanged. Many genes in cell-death and -survival pathways also are downregulated. The canine disease is a non-syndromic LCA-ciliopathy, with normal renal structures and no CNS abnormalities. Our results identify the critical time points in the pathogenesis of the photoreceptor disease, and bring us closer to defining a potential time window for testing novel therapies for translation to patients.

Introduction

Primary, non-motile cilia present in most mammalian cells are typically sensory organelles involved in many critical aspects of biology (1,2) including establishment of left-right symmetry, Hedgehog signaling, vision, smell and hearing (2–4). Primary cilia are composed of a ‘9 + 0’ microtubule-based axoneme, which is in turn derived from and anchored to the cell via the basal body (5). In photoreceptors, the sensory cilium forms a narrow isthmus that connects the metabolically active inner segment (IS) to the outer segment (OS). Membrane and soluble proteins essential for phototransduction and critical for maintenance of the OS traffic through the photoreceptor cilium (6,7).

Due to their presence in many cell types, ciliary defects can result in severe developmental and sometimes lethal disorders, collectively called ciliopathies and characterized by one or more overlapping phenotypes including embryonic patterning defects, renal cystic disease, intellectual disability and photoreceptor degeneration. One such ciliopathy, Senior Loken Syndrome (SLSN), includes nephronophthisis (NPHP) and Leber congenital amaurosis (LCA), characterized, respectively, by cystic kidney disease that is the most frequent cause of end-stage renal disease in the first three decades of life, and childhood-onset blindness (8). LCA represents a genetically heterogeneous (https://sph.uth.edu/retnet/; date last accessed July 29, 2016) group of diseases having early onset of profound congenital retinal and visual malfunction that can result from abnormal photoreceptor development, or biochemical defects overlapping with a progressive degenerative time course.

Mutations in NPHP5 (IQCB1) have been identified in patients with SLSN (9), resulting in severe retinal dystrophy and renal disease with a variable age of onset (10). There have been reports of NPHP5 mutations causing LCA without renal symptoms, but the broad range in age of NPHP onset leaves open the possibility that in these cases the renal disease could still develop at a later age (10). NPHP5 patients show early and rapid rod degeneration with unexpected preservation of central cones that function poorly or not at all (11,12). The gene product localizes to renal cilia and to the photoreceptor connecting cilium/transitional zone (9), and is required for early cilia assembly and ciliogenesis (13,14). Additionally, NPHP5 interacts with several proteins responsible for X-linked retinitis pigmentosa (RPGR) (15), LCA, SLSN or Joubert syndrome (CEP290) (10,16,17) and the BBSome subunits that have been implicated in Bardet Biedl Syndrome (18,19). The close interaction of these ciliary proteins, and the overlapping phenotypes observed when defective, suggests that they may function in the same or common pathways. As such, detailed knowledge of the structural and functional consequences resulting from defects in NPHP5 will enhance our understanding of other ciliopathies, and provide insights on therapeutic approaches for disease correction.

A spontaneously occurring canine LCA-ciliopathy model is caused by a single nucleotide insertion in exon 10 of NPHP5, resulting in a 12 amino acid frameshift, and premature stop that truncates the C-terminal 268 aa of the 598 aa native protein (20). Clinical signs include widely dilated pupils under both dim and bright light, severe visual difficulties from an early age, and intermittent nystagmus. In this study, we compared the human and canine retinal phenotypes, and in dogs we examined the early stages of photoreceptor development, the kinetics of photoreceptor loss, the progression of degeneration, and the expression profiles of NPHP5 and selected genes. We now show the canine disease to be a non-syndromic LCA-ciliopathy characterized by abnormal photoreceptor development and concurrent early and progressive degeneration. Our results identify the critical time points in the pathogenesis of the photoreceptor disease, and bring us closer to identifying a potential time window for testing novel therapies for translation to patients.

Results

Human NPHP5-LCA shares a central retinal phenotype with certain ciliopathies but not others

NPHP5-LCA patients had early onset and rapid loss of rod photoreceptor function and structure but retained central retinal photoreceptors, albeit accompanied by reduced cone function (10–12). In contrast to normal retina with preserved photoreceptor nuclear layer (outer nuclear layer, ONL) thickness across the entire cross-sectional image (Fig. 1A), two representative NPHP5-LCA patients (P5, P4; Fig. 1B;Supplementary Table S1 and Fig. S1) showed normal ONL thickness centrally, but there was a decline with eccentricity. The barely detectable ONL beyond 5 degrees showed no obvious distal (deeper in the retina) substructures. A magnified image (right panel, Fig. 1A) showed that normal retina had complex lamination distal to the ONL. There were scattering peaks from the outer limiting membrane (OLM), the inner/outer segments (ISe or EZ, ellipsoid zone), the interface at or near the apical retinal pigment epithelium (RPE) microvilli and cone outer segments (COS), and the interface near the basal RPE and Bruch’s membrane (RPE/BrM). At the NPHP5-LCA fovea, there could be some detectable lamination but it differed from that in the normal (Fig. 1B, magnified images right panels). The discrete reflectivity signals of the normal retina were not seen in NPHP5-LCA; there were less distinct wider bands of high backscatter (termed S⁺) (10–12) and low backscatter (S⁻). P5 had some substructure reminiscent of an OLM, EZ and possibly even a COS lamina, but P4 only showed a wide S⁺band and a narrow S⁻ band.

Figure 1.

Retinal structure and function in human patients with NPHP5-LCA. Cross-sectional OCT scans along the vertical meridian through the fovea, extending 15° into superior (S) and inferior (I) retina in (A) a normal subject (age 30 years) and (B) NPHP5-LCA patients P5 and P4 (ages 15 and 13 years, respectively). Magnified views of the central retina (right; white rectangles) with overlapping longitudinal reflectivity profiles (LRP). Photoreceptor nuclear layer (ONL) is highlighted in blue; EZ (ellipsoid zone) is highlighted in yellow. OLM, outer limiting membrane; COS, cone outer segments. RPE/BrM, retinal pigment epithelium/Bruch’s membrane. In the patients, there are wide hyperscattering bands (S⁺) and more narrow hyposcattering bands (S⁻) distal to the ONL (far right along image). Uncertainty in labeling the abnormal lamination (in P5) is indicated by a ‘?’ near the anatomical labels. Icon (upper left) is location of the scans on a retinal schematic. (C) Comparison of normal LRP through fovea and those of six patients with NPHP5-LCA. Normal LRP (left) is labeled; the six patients (right) have S⁺and S⁻ marked with shades of gray. LRP are aligned by outer edge of EZ (thin horizontal black line). EZ (highlighted yellow) in two of the patients (P6, P5) is definable. (D) Quantitation of thickness of foveal ONL (white bar for normal; n = 11; ages 26–62, error bar is 2 SD). Blue bars represent the patients’ ONL. COS (white bar for normal) or S^- (gray bars for patients) are also plotted. Those patient parameters significantly different from normal are marked by an asterisk. (E) Relationship of foveal cone photoreceptor structure (product of ONL and COS thickness for normal subjects, and patients with BBS1, USH2A and RPGR; product of ONL and S^-for patients with NPHP5 and CEP290) and visual function (cone sensitivity loss, based on dark-adapted 650 nm sensitivity measurements in the foveal region); (24) for five genotypes causing ciliopathies. Normal variability is described by an ellipse encircling the 95% confidence interval of a bivariate Gaussian distribution. Dashed lines indicate region of uncertainty defined as translation of the normal variability along the idealized model of a pure photoreceptor degeneration. Results in patients with BBS1 (n = 6), USH2A (n = 7) and RPGR (n = 16) mutations [modified from (21,24,46,47); Supplementary Table S1] are shown with gray symbols; the values fall inside predicted limits of the model. LCA patients with NPHP5 and CEP290 mutations (red and pink triangles) show better structure than predicted by the model. Cone sensitivity losses in some LCA patients were indeterminate due to severity of dysfunction and are graphed as >30 dB loss.

The complexity of the distal substructures of NPHP5-LCA was further examined using a comparison of the longitudinal reflectivity profiles (LRP) through the normal fovea and those in six patients (Fig. 1C). The LRPs in all patients were divided into S⁺and S⁻ zones. Two patients, P6 and P5, however, had a sufficiently identifiable hyper-reflectivity near the depth of the normal EZ peak to speculate that there was correspondence of these structures. Of interest, these two patients had the best visual acuity of the six patients (Supplementary Table S1). Quantitation of foveal cone ONL in the six NPHP5-LCA patients and comparison with ONL of normal subjects showed that all but two (marked by *) were within normal limits. Assuming that the hyporeflectivity (S⁻) proximal to the RPE/BrM hyper-reflectivity may represent remnant cone OS (10–12), we measured S⁻ thicknesses and found the results to be reduced versus normal cone OS (Fig. 1D).

Severe early onset visual disturbance in NPHP5-LCA patients despite retained central photoreceptor structure suggested a dissociation of function and structure. We quantified the relationship of cone function and structure in these patients and compared the results with those of normal subjects and patients with other ciliopathies to determine if this dissociation is unique to NPHP5 (Fig. 1E;Supplementary Table S1). We defined structure as the product of ONL thickness (a surrogate for photoreceptor numbers) and COS length (surrogate for opsin molecules within each retained photoreceptor) at the fovea. S⁻ was measured in NPHP5-LCA patients. The structure–function data from the different genotypes were plotted, and we applied a simple linear model which has been used to describe this relationship in various retinal degenerations (21–24). The results indicated that central disease due to BBS1, USH2A and RPGR mutations behaved similarly and like pure photoreceptor degenerations. In other words, visual sensitivity was reduced linearly with quantum catch. NPHP5 and NPHP6 (CEP290) patients, however, differed from the others, and fell outside of the 95% confidence interval of the normal variability (Fig. 1E). Taken together, the results quantified the observation that there was a greater degree of dysfunction than could be explained by the loss of cone nuclei, and shortening of cone OS in NPHP5-LCA. The basis for these observations in human disease can then be sought in the canine model.

Natural history of canine NPHP5 retinopathy shows preservation of the cone-rich visual streak

To determine the spatio-temporal disease characteristics and progression of photoreceptor degeneration, ONL thickness measurements were made across the retina of NPHP5-mutant dogs (Supplementary Material, Table S2). An OCT of a representative control dog showed ONL thickness topography that is thickest in the temporal central retina; ONL was relatively thinner in the peripheral and inferior retinal regions (Fig. 2A). A representative NPHP5-mutant dog at 14 weeks shows thinning of ONL indicating that loss of photoreceptors is occurring before this age (Fig. 2B, left). Of note, there was a tendency for a relatively thicker ONL in a horizontal band located immediately above the optic nerve. The same NPHP5-mutant eye imaged 19 weeks later at age 33 weeks demonstrated substantial further degeneration occurring retina-wide; the retention of the horizontal band of ONL superior to the optic nerve had become more distinct (Fig. 2B, right). The band of ONL retention corresponded to the expected location of the cone-rich canine visual streak (25). To quantify the rates of ONL thinning across the retina, 12 NPHP5-mutant eyes were imaged over a range of ages from 7 weeks to 4 years and ONL topographies were sampled at five standardized locations (Fig. 2C). At all locations, progression could be characterized to a first approximation with log-linear curves implying underlying exponential rates of degeneration. At two inferior and two superior loci, the estimated progression rates were −0.73 log₁₀/year and −0.74 log₁₀/year, respectively. At the central visual streak region, the progression rate was −0.37 log₁₀/year, substantially slower than the inferior and superior retinal locations.

Figure 2.

Natural history of disease progression in NPHP5-mutant dogs evaluated with imaging and electrophysiology. (A,B) Pseudocolor maps of ONL thickness topography (upper) in a 35-week-old control dog (A) and a NPHP5-mutant dog (B) at 14 and 33 weeks of age. Insets, near-infrared reflectance images. Arrows on the maps localize the reconstituted OCT scans (lower) along a superior–inferior meridian crossing the central visual streak at the gaps of the lines. ONL on reconstituted scans is highlighted in blue and location of the visual streak is shown with a white wedge. All eyes shown as equivalent right eyes and optic nerve, major blood vessels (black) and tapetum boundary (yellow) are overlaid for ease of orientation. T, temporal; N, nasal retina. (C) ONL thickness quantified as a function of age at five retinal locations (Inset): two inferior loci (left), two superior loci (middle), and a central locus are shown (right). Gray symbols represent control animals; black symbols represent NPHP5-mutant dogs. Regression lines and 95% prediction intervals are shown for the first log unit of ONL loss. (D) Representative ERGs from control (6 and 33 weeks of age), and mutants (6, 14 and 33 weeks of age) in response to mixed rod-cone, or isolated rod or cone (1 Hz or 29 Hz) stimuli. In mutants, dark-adapted rod responses were barely recordable at 6 and 14 weeks, and absent thereafter. They showed an altered waveform, and amplitudes that were <10% of normal using maximal white light stimuli (1.01 log cd·s·m⁻²); cone responses were not recordable. (E) ERG function quantified as a function of age for two cohorts: Group 2 illustrated in D, and Group 1 in Supplementary Material, Figure S2. Both groups show absence of cone function. Whereas Group 1 has better preservation of rod-mediated responses at 6 weeks, these rapidly decay by 14 weeks, and both groups show absent rod and cone ERG function at 33 weeks.

Overlap of abnormal photoreceptor development and progressive degeneration

Functional studies

Retinal function was assessed at 6 weeks of age, near the end of normal retinal maturation, when electroretinography (ERG) responses are approximating adult waveforms and amplitudes (26). All affected dogs had a cone-rod dystrophy phenotype with absolute loss of cone function and variable, but highly abnormal, rod function. In affected dogs, cone responses were not recordable using the highest intensity white light stimuli, either single flashes or 29 Hz flicker stimuli presented on a white background illumination. In contrast, the NPHP5^+/− littermate heterozygous controls had robust cone-mediated responses that were indistinguishable from wild-type dogs. The cone responses from affected animals remained non-recordable throughout the study (Fig. 2D).

In contrast, rod responses were more variably affected. In one group, dark-adapted rod responses were barely recordable, had an altered waveform, a markedly elevated threshold, and amplitudes that were only 10% of normal using maximal white light stimuli (Fig. 2D). The second group also showed elevated thresholds, but better preserved waveforms at 6 weeks of age, and amplitudes that were reduced ∼55% or more depending on the stimuli used (Supplementary Material, Fig. S2). In both groups, rod responses were nearly extinguished by 14 weeks of age, and absent thereafter (Fig. 2E). After 33 weeks, ERG responses were not recordable from any of the older mutant dogs tested (Supplementary Material, Table S2).

Morphologic studies

H&E stained cryosections were used to characterize disease, and to examine for topographic variations within the same eye. Photoreceptor abnormalities were similar throughout the retina [illustrated here for the nasal mid-peripheral (area S2) region (Fig. 3; Supplementary Material, Fig. S3)], although disease progression showed topographic variation in different areas (Nasal > Superior/Inferior > Temporal; data not shown). At 6 weeks, the typical neuroretinal layer organization was preserved in the affected dog, and ONL thickness was similar to control, but photoreceptor IS and OS were irregular and disoriented (Fig. 3A and E). These and subsequent findings (see below) indicated that there was no delay in development in mutants, but that development was abnormal. By 14 weeks, shorter and distorted IS persisted, and there was moderate ONL thinning (Fig. 3B and F). The photoreceptor layer was severely thinned at 33 weeks and contained only remnants of distorted IS and OS, and the RPE was hypertrophied with marked irregularities of the apical surface; the ONL thickness was further reduced (Fig. 3C and G). Finally, at 42 weeks, the latest time point examined, the external limiting membrane was juxtaposed to the RPE, with complete loss of the photoreceptor inner and outer segments, and the ONL thickness was reduced to 1–2 nuclei (Fig. 3D). The outer plexiform (OPL) and inner nuclear (INL) layers also thinned as the disease progressed (Fig. 3D). The rapid and continued loss of ONL over a 36 week time period is illustrated for the mid-peripheral nasal meridian (Fig. 3H).

Figure 3.

Structural changes in the mid-peripheral, nasal region of NPHP5-mutant dog retinas. H&E images of retina from affected and control dogs of different ages. In 6-week mutant (A, control E), IS are irregular in size, disoriented and variation in OS length is apparent from irregular apical RPE contour. (B–D; control F, G) IS irregularities remain, the photoreceptor layer disappears, and the external limiting membrane is opposed to apical RPE at 42 weeks. (H) Graph showing ONL thickness in terms of rows of nuclei at the mid-periphery (S2) of the nasal meridian plotted at each time point. Each data point represents the mean ± SD of three counts. Scale bar = 20 µm; RPE = retinal pigment epithelium, OS = outer segments, IS = inner segments, ONL = outer nuclear layer, OPL = outer plexiform layer, INL = inner nuclear layer.

Photoreceptor cell death

TUNEL assays were done to establish cell death kinetics, and identify the cells and retinal layers involved. In control dogs, the proportion of photoreceptor cells undergoing cell death was low at the earliest time point, and thereafter decreased to background levels. Cell death in mutants was much higher, especially at 6 weeks of age when retinal development was still ongoing (26), an indication of active degeneration superseding developmental death. At all time points, photoreceptor cell death in mutants was greater in the inferior than the superior meridian, and the rate was approximately constant between 14 and 42 weeks of age (Fig. 4A). TUNEL-labeled cells were equally distributed throughout the length of the retina, from the optic disc to the ora serrata, but at earlier ages there were more dying cells located in the vitreal half of the ONL (Fig. 4B and C). Although TUNEL-positive photoreceptors were seen in the outer half of the ONL, it was extremely rare before 33 weeks of age to detect any labeling in the outermost row of ONL nuclei, where cone somas are located (Fig. 4D). TUNEL labeled ONL nuclei located adjacent to the ELM did not co-label with cone arrestin, and were most likely rod cells located adjacent to the external limiting membrane rather than dying cones (Fig. 4F and G).

Figure 4.

Photoreceptor cell death and proliferation in NPHP5-mutant dog retina. (A) Number of TUNEL-labeled photoreceptor cell nuclei per 10⁶ µm² of ONL as a function of age in the superior and inferior meridians of affected and control dogs. Each data point represents the mean ± SD of counts of each animal made on three sections. (B–G): TUNEL labeling (green) and cone arrestin labeling (red; F, G) with a DAPI nuclear counterstain (blue). At 6 weeks a number of nuclei are labeled with TUNEL (arrows) in the vitreal half of the retina in the affected (B) and fewer in the control (C). By 33 weeks TUNEL–labeled nuclei (arrows) continue to be present in the affected (D), but none in the control (E). No cells were co-labeled with TUNEL and cone arrestin at 33 (F) or at 42 weeks (G). At 33 and 42 weeks, TUNEL+ cells were located adjacent to the external limiting membrane. Scale bar = 20 µm; ONL = outer nuclear; INL = inner nuclear layer. (H) Number of PHH3-labeled photoreceptor cell nuclei per 10⁶ µm² of ONL, as a function of age in affected and control dogs. A small number of PHH3-labeled cells were observed in the control dogs at any age. In contrast, far more labeled cells were observed in the affected animals at all ages examined. Each data point represents the mean ± SD of counts of each animal made on three sections taken from the superior meridian.

Photoreceptor proliferation

As the rate of ONL loss appeared less than anticipated for the number of TUNEL-positive ONL cells, particularly at the 6 and 14 week time points, we determined whether mitotic activity in a subpopulation of photoreceptors cells also was occurring. This is a feature of other early onset photoreceptor degenerations (27,28). To this end, we used the mitosis-specific marker PHH3 to identify dividing cells in the ONL, and found background labeling in control (< 3.9 ± 1 cells/10⁶ µm²), but very high levels in the 6 week mutant retina (51.4 ± 7 cells/10⁶ µm²). Mitosis decreased after 6 weeks, but remained constant (∼10 cells/10⁶ µm²) at all subsequent time points examined (Fig. 4H).

Incomplete development of cones, but normal cone insoluble extracellular matrix sheath

Single or double immunolabeling with cone specific antibodies was used to characterize the structural defects in cones. To determine cone density as a function of disease, we plotted the number of cone arrestin labeled cells/1000 μm length of retina in all four meridians. The results showed that the number of labeled cells in the control and mutant retinas in the 6–33 week time period remained relatively constant and comparable (Fig. 5). At the 42 week time point, the number of cones decreased in mutants in accordance with the progressive degeneration present at more advanced disease stages (Fig. 3D).

Figure 5.

Cones develop abnormally, and degenerate early in NPHP5-mutant dog retina. Immunofluorescence labeling of cones with human cone arrestin (red), and DAPI nuclear counterstain (blue), in affected (A-D) and age-matched controls (E-G). Images are from the mid-peripheral (S2) superior meridian. (A) At 6 weeks, the IS were thickened, blunted and distorted, and only a small proportion of shortened OS were present (arrows). (B) By 14 weeks cone IS were contracted, and the cone somas, axons and pedicles remained. (C, D) At later time points, labeled cone somas remained, in some cases (D, arrow) showing labeled axons that were horizontally arranged. (H) Cones are best preserved in the temporal-central meridian where IS and some OS were still present at 42 weeks, but (I) degeneration was present peripherally. (J) Number of arrestin-labeled cones somas and OS per 1000 µm length of retina from affected and control dogs. Each point represents the mean ± SD of counts made in each of the central, mid-peripheral and peripheral regions of the superior, inferior, temporal and nasal meridians. In controls, there is no difference in the counts of cone somas and OS. In mutants, however, very few cone IS had OS at 6 weeks, and the OS were absent at subsequent time points. Scale bar = 20 µm. OS = outer segments, IS = inner segments, ONL = outer nuclear layer, OPL = outer plexiform layer, INL = inner nuclear layer.

Although the mutant retina had a normal number of cones initially, their structure was markedly abnormal, and deteriorated further with age (Fig. 5A–D). In controls, cone arrestin labeled the entire cone, from the outer segment to the synaptic pedicle (Fig. 5E–G). In contrast, the 6-week-old mutant had distorted and condensed IS, and only 5–10% of these had stunted and abnormal OS (7.6 ± 3 OS/1000 µm length of retina; mean ± SD, n = 3; Fig. 5A and J). By 14 weeks, most cone OS were lost, and IS contracted to form an intensely cone arrestin-labeled cytoplasmic ring around the nucleus with a distinct scleral projection. Surprisingly, the cone axons and pedicles remained normal (Fig. 5B and F). At later time points, the cone somas became markedly pleomorphic and cone arrestin labeled the ONL nuclei and processes remaining in the ONL (Fig. 5C, D and G). Morphologically, the temporal retina at 42 weeks of age in the mutant appeared more intact near the area centralis, with a small number of OS and cone pedicles visible between the optic nerve and the area centralis; towards the periphery, the cones were severely disrupted, had tangentially arranged axons and lacked OS, IS or distinct pedicles (Fig. 5H and I).

We then asked whether the cone OS defect is generalized or restricted to one of the cone classes, particularly to the more numerous L/M-cones. To this end, we performed double fluorescence immunolabeling with cone arrestin and S- or L/M-opsin antibodies in sections from the superior, inferior and temporal meridians. The number of S- and L/M-cones, irrespective of presence of OS, was similar between the controls and mutants until 33 weeks, then the number of both cone classes decreased in mutants (Fig. 6A and B). In affected and control dogs there were approximately 6–7 times more L/M-opsin than S-opsin-labeled cones, and these ratios were similar between the two groups; morphologically, however, the retinas were drastically altered in mutants. In the controls, cone opsin labeling was restricted to the OS at all ages (Fig. 6F–H and L–N, respectively). In mutants, S-opsin-label mislocalized to the cone IS, somas, axons and pedicles (Fig. 6C–E), and this finding was present in cones with or without an OS. L/M opsin mislocalization was less prominent in affected retinas. Unless the images were overexposed, the L/M opsin fluorescent labeling was faint or not visible in the cone somas, and rarely in the cone axons or pedicles for all ages (Fig. 6I–K). S- and L/M-opsin-labeled OS remnants were observed in the superior, inferior, and temporal retinas of affected dogs from 6 to 33 weeks of age. OS were present only in the temporal meridian of the 42-week-old affected dog in the area centralis region (data not shown). Overall, there was a considerable reduction in the number of both S- and L/M-opsin-labeled OS in the affected compared with the controls, and this decrease was greater for L/M cones as almost no labeled OS were present after 6 weeks of age (Fig. 6A and B).

Figure 6.

Preferential loss of L/M cone outer segments, but normal formation and preservation of the cone extracellular matrix domain in canine NPHP5 mutant retinas. Changes in the number of (A) S- and (B) L/M- cones with disease in sections double-labeled with cone arrestin and cone-specific antibody. In controls, the ratio of L/M- to S-cones remains constant during development, and the antibodies exclusively label the OS. In mutants, the number of cones based on cone arrestin labeling is comparable to control, but OS are lacking in most; at 6 weeks, approximately 20% of S-cones (∼5 of 21/1000 µm length) have an OS compared with only 14% of L/M-cones (∼20 of 140/1000 µm length). By 14 weeks, all L/M cone OS are lost, but S-cone OS numbers remain constant until 33 weeks; thereafter, degeneration results in loss of cones. Reported values are the mean ± SD of three measurements taken in the central, mid-peripheral and peripheral regions of the superior, inferior and temporal meridians. (C–N) Double fluorescence immunolabeling with cone arrestin (red) and S-opsin or L/M-opsin (green), with a DAPI nuclear counterstain (blue). S-opsin and L/M-opsin are confined to the cone OS in the control samples (F–H and L–N, respectively). In the mutants, S-opsin mislocalizes to the cone IS, somas, axons and pedicles from 6-33 weeks of age (arrows; C–E). Some S- (C) and L/M-opsin-labeled (I) OS remnants were still visible at 6 weeks in the affected dogs. L/M-opsin mislocalization was less apparent in the affected dogs (I–K), but faintly visible in the cone somas. Images are from the mid-peripheral region of the superior meridian. (O–Q) Double fluorescence immunolabeling with cone arrestin (red) and PNA (green) to identify the cones and their surrounding insoluble extracellular matrix domain. In controls (O), PNA label surrounds the cone IS and OS. (P, Q) Even though most mutant cones lack an OS, a distinct PNA labeled matrix domain is present. Scale bar = 20 µm; OS = outer segments, IS = inner segments, ONL = outer nuclear layer; PNA = peanut agglutinin lectin. The dotted line (Q) indicates that the retina and RPE were brought photographically into closer apposition as the two were artifactually separated during sectioning.

To assess the effects of the disease on the extracellular cone matrix domain, we performed double immunolabeling with cone arrestin and PNA in affected and control retinas at 6 and 14 weeks (Fig. 6O–Q). In the controls, PNA labeled the insoluble extracellular matrix domain that enveloped the RPE cone apical sheath, and cone OS and IS (29). In the 6-week-old affected dog, PNA labeling was observed around the distorted cone IS, and also in the subretinal space. Similarly, at 14 weeks, distinct PNA labeling was observed in the interphotoreceptor space even though most cones lacked OS.

Abnormal development of rods

Unlike the very extensive absence of cone OS, a full complement of rod OS was present at 6 weeks of age, but these were irregular and disorganized, and demonstrated extensive rod opsin mislocalization into the IS, ONL and synaptic terminals (Fig. 7A and B). At later time points, the degree of OS disorganization increased, ONL thickness decreased, and the degree of opsin mislocalization intensified, and formed distinct cytoplasmic rings of label around the nuclei (Fig. 7C and D). Rod opsin positive neurites originating from the rod somas extended into the inner retina, and reached as far as the INL in affected dogs at all stages of disease. In contrast, cone neurite sprouting was not observed anytime.

Figure 7.

Rod outer segments form, but are abnormal in canine NPHP5 mutant retinas. Immunofluorescent labeling with rod opsin (green) with a DAPI nuclear counterstain (blue). (A) In the 6 week affected, rod opsin was mislocalized to the IS and diffusely present in the ONL with rod neurite sprouting into the INL (arrow). (B) In the 6-week control, labeling is restricted to the rod OS; two labeled cells are ectopically located at the OPL-INL interface, and represent a normal developmental variation (arrow). (C) The 33 week affected shows increased rod opsin labeling in the ONL, and extensive rod neurite sprouting in the INL (arrows). (D) In the 33 week control, rod opsin labeling was restricted to the OS. Scale bar = 20 µm; OS = outer segments, IS = inner segments, ONL = outer nuclear layer, OPL = outer plexiform layer, INL = inner nuclear layer; * in C indicates a horizontal fold in OS which makes the area out of focus.

Early outer plexiform layer abnormalities

To assess the effects of disease on second order neurons, confocal microscopy was used in sections double labeled with Goα and PKCα which, respectively, label both rod and cone ON bipolars, or only rod bipolars. In mutants, there was considerable retraction of dendritic arborizations in affected retinas as early as 6 weeks of age which continued until 42 weeks; at all stages, the OPL was thinned in comparison to controls (Fig. 8A–G). As well, the PKCα labeled synaptic boutons were lost from the vitreal aspect of the INL at 33 and 42 weeks, and marked reduction of inner plexiform layer thickness was noted (Fig. 8C and D). Additional abnormalities included the presence of PKCα positive rod bipolar cells displaced into the ONL (Fig. 8H), and areas where bipolar cell dendrites extended far into the ONL (Fig. 8I).

Figure 8.

Early abnormalities in bipolar cell dendritic arborizations of the canine NPHP5 mutant retina. Double fluorescence immunolabeling of ON bipolar (Goα-red) and rod bipolar (PKCα-green) cells with a DAPI nuclear counterstain (blue) imaged by confocal (A–G) or standard epifluorescence (H, I) microscopy in control (E–G) and mutant (A–D, H, I) retinas. Rod bipolar cells show co-localization of both proteins and appear orange, while cone ON bipolar cells are labeled only with Goα and appear red. Retraction of the bipolar cell dendrites was present by 6 weeks, and worsened with disease progression. The rod bipolar synaptic boutons that were normal at 6 and 14 weeks (A, B), became attenuated and absent in latter time points (C, D). In spite of the overall bipolar cell dendritic retraction, some dendrites became elongated and extended into the ONL (B- inset, I; arrows), and rod bipolar cells were displaced into the ONL (H; arrow). Images are from the mid-peripheral region of the superior meridian. Scale bar = 50 µm (A–G), 10 µm (A–G inset) and 20 µm (H–I); ONL = outer nuclear layer, OPL = outer plexiform layer, INL = inner nuclear layer, IPL = inner plexiform layer.

Photoreceptor sensory cilium is present in mutant retinas

We used several antibodies that labeled components of the photoreceptor sensory cilium in the canine retina (30). This analysis, however, was only partial as most antibodies tested and used to define the ciliary axoneme were not specific in dog retina; these included antibodies against RPGR, RPGRIP1, RP1 (Supplementary Material, Table S3). As well, antibodies, both commercially available (n = 3) and custom-made (2), against NPHP5 did not label (13). Cognizant of these limitations, we examined by confocal microscopy three antibodies that in dogs labeled components of the photoreceptor sensory cilium: acetylated α tubulin, rootletin (30) and MAP9 (Downs et al., unpublished information). In 6-week mutant retinas, these antibodies labeled different components of the sensory cilium (Fig. 9A–F); because of the IS irregularities in mutants, the rod and cone cilia were readily apparent (Fig. 9A, inset). These preliminary analyses in mutants indicated that photoreceptor sensory cilia formed, and the domains labeled by this limited number of antibodies were present.

Figure 9.

Immunolabeling of the photoreceptor sensory cilium shows the presence of cilia in canine NPHP5 mutant retinas. Confocal images of 6 week affected (A, C, E) and control (B, D, F) retinas labeled with acetylated α tubulin (A, B), rootletin (C, D) and MAP9 (E, F) with a DAPI nuclear counterstain (blue). Different components of the photoreceptor sensory cilium are labeled with the antibodies. In A inset, labeling of the rod (white arrows and cone (yellow arrows) are shown. There are no apparent differences in cilia structure and labeling between affected and control photoreceptors. Scale bar = 20 µm for main figures, and 10 µm for insets; OS = outer segments, IS = inner segments, ONL = outer nuclear layer.

Retinal gene expression

We first examined NPHP5 expression by qRT-PCR in single gene assays using primer sets 5' (NPHP5_1 spanning exons 4–5) and 3' (NPHP5_2 spanning exons 11–12) of the exon 10 deletion (Supplementary Material, Table S4). In two groups of 5 week normal and mutants, n = 3/group, we found an ∼1.5- to 1.9-fold decrease in expression in mutants that was statistically significant, but below the 2.0-fold change criterion threshold for differential expression (Fig. 10A). The decrease was comparable to that found for other genes that were rod-specific, or expressed in rods and cones (Fig. 10A). Surprisingly, in spite of the severe cone structural abnormalities, the expression of a subset of cone-specific genes examined, including L/M- and S-opsin, was not altered (Fig. 10A).

Figure 10.

Expression changes of photoreceptor and pro-death and pro-survival genes at 5 weeks of age in canine mutant retinas. Genes with significant changes (* in A) in expression (measured as fold change by qRT-PCR, with Benjamini–Hochberg adjusted P-values < 0.05) in affected dogs at 5 weeks relative to control dogs at 5 weeks. Only one gene, HSP90, was upregulated in the affected dogs, and all other genes were downregulated. The data for rcd1 and xlpra2 has been published previously (32) and is used to illustrate differences with NPHP5. Only genes with FC < −2 are considered DE. Error bars show SD of biological triplicates; refer to Supplementary Material, Table S5 for summary of results of genes that showed statistically significant expression changes. n/a = not applicable; n.s. = not significant.

The expression analysis was extended to determine the effect of the disease at the 5 week of age time period on additional genes that had a direct role in vision, or that were part of pro-death or pro-survival pathways. To this end, we used a canine-specific qRT-PCR profiling array (31,32), and compared the expression between 5-week-old normal and mutants. This array previously was used in three other non-allelic canine retinal degeneration models (31,32), and informed on the molecular pathways involved in the disease. In total, we examined the expression profiles of 112 selected genes in the retina of 5 week normal and mutant dogs, and found that 33 genes showed statistically significant changes in expression (Fig. 10 A and B, and Supplementary Material, Table S5), of which all but HSP90 were downregulated. Of these 33 genes, 17 were differentially expressed (DE), and downregulated with a fold change <−2. We compared the results with those obtained in two other non-allelic early onset inherited retinal degenerations, rcd1 (PDE6B mutation) and xlpra2 (RPGR mutation) (32), at a comparable disease stage, and using the same profiling array and single gene assays. In 5 week old mutant dogs, only GRK1 was DE and significantly downregulated in NPHP5 as well as in the two other diseases. Interestingly, of the genes involved in autophagy, pro-death and pro-survival, seven DE genes downregulated in NPHP5 mutant retina were upregulated in rcd1 and/or xlpra2: CASP4, FAS and PTPRC are involved in the pro-death, mitochondria-dependent pathway; CASP8 in the pro-death, mitochondria-dependent and -independent pathways; TRADD in the pro-death, mitochondria-dependent and -independent and pro-survival pathways; and TNFSF8 and CD40LD in the pro-death, mitochondria-independent and pro-survival pathways.

Canine NPHP5 disease is non-syndromic

Assessment of renal abnormalities was done using one of several methods. No structural abnormalities were found either with high resolution MRI anatomic scans (Supplementary Material, Fig. S4 and Table S2), or renal ultrasounds (data not shown) in dogs 4.6–9.5 years of age. Renal function studies in mutant dogs 3.7–9.5 years showed normal urine specific gravity, and ruled out tubular (normal BUN and creatinine) or glomerular (normal urinary protein/creatinine ratio) defects (Supplementary Material, Table S6). Although CNS abnormalities are not part of SLSN, these were excluded based on high resolution MRI anatomic scans (Supplementary Material, Fig. S4).

Discussion

Developmental abnormalities of the outer retina and OPL

Mutations in NPHP5 cause an LCA-ciliopathy in man and dogs characterized by early onset, and profound congenital retinal (and visual) malfunction that results from poorly functioning photoreceptors that are abnormally developed and subsequently degenerate. While the disease in both is aggressive, in patients it is not possible to gain insights into the cellular and molecular events that precede the clinical presentation. The principal reason is one of timing in that many of the critical events leading to the photoreceptor abnormalities in man occur pre-term. With the exception of the fovea, the retina of the newborn human is fully developed at birth (33), and foveal development occurs thereafter (34). Analysis of the early disease stages is possible in dogs where retinal differentiation is postnatal.

At birth, the canine outer retina is not developed; photoreceptor differentiation begins at week 1, and the retina is mature after 6–7 weeks of age (26,35). Thus the studies in mutant canines, particularly in the first 6 months of life, are critical to understanding the cellular and molecular characteristics of this LCA-ciliopathy, and providing insight to the orthologous human disease.

Towards the end of postnatal retinal differentiation, the NPHP5 mutant retina shows complete absence of cone ERG function, and rod responses that either are present but abnormal in terms of amplitude and increased dark-adapted thresholds, or are markedly reduced in amplitude. In either case, they represent a cone-rod dystrophy phenotype indicating that normal functional development of photoreceptors does not occur. By 14 weeks, the rod responses are nearly unrecordable, and absent thereafter. The absence of cone-mediated responses is clearly anatomic. As less than 10% of the cones have an OS, it is not surprising that cones are unable to generate signals to photic illumination. On the other hand, as rods have a full complement of OS before the retina fully degenerates, yet loses rod function much earlier it is, an indication that there is a dissociation of structure and function. This likely results from impaired trafficking of key phototransduction proteins through the photoreceptor sensory cilium, and is reflected by the extensive rod opsin, as well as S-cone opsin, mislocalization into the inner segment and ONL (36,37). This abnormality may not be restricted solely to trafficking between IS to OS, as in the bbs4 null mouse there is impaired synaptic transmission as well which contributes to abnormalities in ERG rod b-wave (37). A similar situation is likely present in canine NPHP5, and may influence the dramatic retraction of the cone ON bipolar and rod bipolar cell dendrites that results in compression and narrowing of the OPL during postnatal development.

A feature of the cone disease is the differential effect on the two cone classes, and the topographic distribution of disease. While the absence of cone OS is widespread, L/M-cones are more severely affected. In comparison to S-cones of which 20% have OS at 6 weeks of age, only 14% of L/M-cones do so. In the next 8 weeks, all L/M-cone OS are lost, but S-cones retain this structure, albeit abnormal, until cone cell loss begins after 33 weeks. Although widespread, these abnormalities were less severe in the area centralis region where better preserved cones, some with OS, were present [present study and (20)].

In spite of the lack of cone OS, it was important to find that the cone extracellular matrix domain labeled with PNA was present and intact. This complex, insoluble structure of poorly understood origin (29), forms an 'exoskeleton' that surrounds the RPE cone sheath, and cone IS and OS, and in photoreceptor degenerations it adjusts to the contours of the diseased cells (38). Preservation of this matrix domain will be critical for future studies with corrective gene therapy in order to provide the appropriate extracellular environment that would allow treated cones to regrow their OS.

Photoreceptor sensory cilium

In mutant rods and cones, the photoreceptor sensory cilium, as determined by immunolabeling of a limited subset of its domains, is present, yet the majority of NPHP5-mutant cones are unable to form an OS during development. At this time, it is not possible to better characterize the organization of the cilium, and its different structural domains, as most of the antibodies tested, including commercial and custom made NPHP5 antibodies, did not result in specific labeling. In vitro studies using a variety of cell lines, e.g. RPE-1, HEK293 and others, have shown that NPHP5 localizes to the centrosome and cilia, and, together with CEP290 is required for ciliogenesis. In serum deprived, quiescent cell lines, either depletion of NPHP5 by siRNA, or introduction of pathogenic mutations, caused NPHP5 and/or CEP290 mislocalization, and were ineffective in binding of CEP290; both abnormalities impair cilia assembly (13).

Unlike cultured cells, the photoreceptor is post-mitotically differentiated, albeit abnormal in mutants, and, with limited exceptions (see below), is non-dividing. Based on the slight reduction of NPHP5 mRNA expression, and lack of nonsense-mediated decay, we posit that a truncated NPHP5 is expressed, and is partly functional. Although the mutation in dogs (20) truncates the C-terminal 268 amino acids and eliminates the second of two BBS binding domains, and the CEP290 binding domain (18), the truncated protein appears to be sufficient to drive differentiation and cilia formation. The canine mutation, presumably, also would delete the second calmodulin-binding (CaM), domain and 5 of the 29 amino acid of the first domain (20); in cultured cells, mutations in this domain do not impair cilia formation (13,39). Regardless, it is surprising that ciliogenesis is not fully impaired given the extent of the canine NPHP5 mutation.

The findings in the dog, especially for rods, are different from NPHP5 knockout mice in which there is proper anchoring of the basal body to the cell cortex, but there is impaired formation of the ciliary transition zone, and failure of rod and cone outer segments to form (40). The differences in phenotypes resulting from the mutations in the dog and mice suggests that different domains of the protein may be necessary for ciliogenesis, and outer segment formation, and these differ between rods and cones. These issues are best addressed in photoreceptor cells rather than with in vitro approaches using cells with primary cilia as the latter will not form an outer segment. Examining these questions in dogs will require a complete suite of antibodies that identify the different domains of the sensory cilium to better assess the interactions between NPHP5, CEP290, members of the BBSome, as well as other functional domains that may be adversely impacted by the mutation.

Photoreceptor cell death

Apoptotic cell death is the final common pathway in most retinal degenerative diseases (41), but the rate of degeneration, and cell death pathway utilized varies in a disease, mutation and species-specific manner [see (42) for review]. In NPHP5 mutant retinas, there was a high number of TUNEL positive cells at 6 weeks of age, the earliest age examined, and indicate that degeneration and cell death overlap abnormal photoreceptor development.

Apoptotic cells were more numerous in the inferior than the superior retinas, and this difference was maintained during the subsequent 36-week study period. From 14 to 42 weeks of age, the rate of cell death, based on the number of remaining photoreceptor cells, was rather constant. As previously reported for the RPGR-X-linked RP canine model (43), the kinetics of photoreceptor cell loss is best described by a model of constant risk of cell death after an initial burst of cell death that occurs at approximately 6 weeks. At all the time points examined, the apoptotic cells were exclusively rods. Given that cones also disappear from the ONL, a slower rate of cone degeneration also occurs, but the mechanism by which this occurs is unknown presently.

Photoreceptor cell proliferation

A feature of early onset inherited retinal degenerations, at least in canine models, is the concomitant mitosis of a subgroup of rod photoreceptors that continue to proliferate independent of the apoptotic ones. Rod photoreceptor origin of these dividing cells has been established by showing that the dividing cells are surrounded by a rim of delocalized rod-opsin immunolabeling, and by excluding any contribution of Müller glia, microglia or nestin positive stem cells to the dividing cell population. This process was first identified in the STKL38 (erd) mutant retina, and now extended to RPGR (xlpra2) and PDE6B (rcd1) models (27,28). In NPHP5 mutants, there is a burst of photoreceptor mitosis that coincides with the peak of cell death established by TUNEL labeling. Thereafter, mitosis decreases to a lower rate (∼20% of peak levels) that remains constant until 42 weeks, the last point analyzed. Although rod proliferation may delay slightly the degeneration in the NPHP5 mutant, the outer nuclear layer continues to thin as more and more photoreceptors die and are lost.

Retinal gene expression

Using a custom-made qRT-PCR profiling array and single gene analyses, we interrogated the expression of NPHP5 and a subset of photoreceptor-specific/enriched genes, as well as genes involved in cell survival or cell death pathways. For this analysis, we selected the 5-week time period as the retina is near the end of postnatal differentiation period, and degeneration has started. Furthermore, the presence of an almost full complement of photoreceptors, even though markedly diseased, suggested that gene expression information at this age would be important, as this is a likely time point for therapeutic intervention by gene augmentation.

NPHP5 expression at this age was decreased ∼1.5- to 1.9-fold in mutants using primer sets located 5' and 3' of the exon 10 deletion. This decrease was comparable to that found for other genes that were rod-specific, or expressed in rods and cones, and indicated that the deletion mutation in NPHP5 did not result in nonsense-mediated decay. Our previous study using northern analysis showed very low levels of expression of NPHP5 in control tissues, and only after exposing the film for 11 days. In contrast, the mutants showed almost no message, results that were interpreted as representing nonsense-mediated decay (20). The present results using two different primer sets for qRT-PCR refute this conclusion, and, indeed, show expression of mutant message that is appropriate to the disease state of the retina.

The expression profiles for genes expressed in rods and in both rods and cones showed uniform and significant down regulation except for Recoverin and RPGRORF15. At this age the rods, although diseased, showed distinct structural preservation. In contrast, expression of a subset of cone-specific genes was unchanged. For example expression of both cone opsins, and both subunits of the cone CNG channel was normal, or slightly elevated, even though the great majority of cones lacked an OS. In regards to the cell death and survival pathway analysis, down-regulation of all tested genes except HSP90 was the rule. In a previous study, we proposed that activation of the TNF gene superfamily was a feature of 3 early-onset photoreceptor degenerations in dogs, and might represent a common cell death pathway that could be amenable to therapeutic intervention (32). In this case, NPHP5, which is comparable in disease time course to the others, is a notable exception.

Canine NPHP5 is non-syndromic

NPHP5 mutations in patients commonly cause SLSN which includes nephronophthisis and LCA, with end-stage renal disease occurring in the first three decades of life or earlier in some individuals (8). In terms of the renal disease, there can be considerable phenotypic variability with some patients developing renal disease later in life, and some not at all (10–12). We used high resolution 9.4 Tesla MRI imaging, and renal ultrasounds, to establish normal renal structures in mutant dogs. Moreover, renal function studies ruled out glomerular or tubular defects in dogs ranging in age from 3.7 to 9.5 years. A similar lack of kidney phenotype has been observed in NPHP5 knockout mice (40). Although CNS abnormalities are not considered part of SLSN, we were able to rule out such defects in the dog.

Human and canine disease phenotypes and translational applications

Models of human diseases are key for understanding the pathophysiology of disease en route to developing and evaluating potential treatments. The current work showed that NPHP5-mutant dogs recapitulate the human retinal but not the renal phenotypes. In patients, the renal disease can be mild or severe and renal transplantation, if necessary, is an available and successful therapy, although obviously a major procedure (10–12). The retinal disease, however, has no therapy. A model that isolates the retinal disease from the renal disease allows experimental work to be performed to understand and someday propose therapy for the blindness.

In both patients and dogs, there is very early loss of rods and relative retention of the central retinal cone photoreceptors which have severely impaired function. Taken together with the importance of central macular vision for humans and our recent finding of the existence of a fovea-like region in canine eyes (25), it will be of substantial translational significance to evaluate whether gene augmentation therapy can improve central cone photoreceptor function by correcting the ciliopathy, forming outer segment structures and restoring trafficking of relevant proteins to these newly formed outer segments. Also important will be evaluation of whether reconstituted retinal signals activate higher visual centers in a blindness that overlaps with development. Success in treating NPHP5-associated LCA-ciliopathy will also be highly relevant to the most common molecular form of human LCA caused by CEP290 mutations which is the only retinopathy that shares the distinct phenotype of NPHP5-LCA (11).

Materials and Methods

Patient studies

Patients (ages 5–81 years) with mutations in the following genes were included: NPHP5 (n = 6), CEP290 (n = 5), BBS1 (n = 6), USH2A (n = 7) and RPGR (n = 16) (Supplementary Material, Table S1). Informed consent or assent was obtained; procedures followed the Declaration of Helsinki and had institutional review board approval. All patients underwent a complete eye examination, genetic analyses, as well as specialized tests of retinal cross-sectional structure and vision.

Animal studies

The NPHP5 dogs used originated from a clinically affected male American pit bull terrier, and have the same mutation (Supplementary Material, Table S2). The disease, originally referred to as cone-rod dystrophy 2 (20,44), is a severe early onset retinal blindness more appropriately considered LCA. All dogs were bred at the Retinal Disease Studies Facility (RDSF, Kennett Square, PA), and represent outcrosses having a common, but heterogeneous genetic background (45). Clinical status was determined phenotypically by ERG in animals euthanatized prior to the identification of the causal mutation, and/or genetically by genotyping. For terminal procedures, the dogs were deeply anesthetized by intravenous injection of pentobarbital sodium, the eyes enucleated and then immediately euthanatized. The research was conducted in full compliance with the University of Pennsylvania Institutional Animal Care and Use Committee (IACUC) approval, adhered to the Association for Research in Vision and Ophthalmology (ARVO) Resolution for the Use of Animals in Ophthalmic and Vision Research, and followed the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health.

Optical coherence tomography (OCT) imaging and analyses

Patients. Cross-sectional images of the retina were obtained with a spectral-domain optical coherence tomography (OCT) instrument (RTVue-100, Optovue Inc., Fremont, CA). Data from a limited number of ciliopathy patients were collected using time-domain (TD) OCT instruments (OCT1, OCT3; Carl Zeiss Meditec, Dublin, CA). Overlapping, non-averaged, 9 mm-length scans through the fovea were used to study the retina, and were analyzed with custom programs (MatLab 7.5; MathWorks, Natick, MA). For the outer retinal sublaminae, signal peak assignments were based on our previously published work (21,24,46,47). Foveal ONL and distal substructure thicknesses were measured from three OCT scans through the fovea obtained at the same visit in the NPHP5-LCA patients, and compared with measurements in groups of patients with four other genotypes. The structural data were compared with foveal function (absolute threshold at fixation, 650-nm target (48,49); a normal data set with both measurements was used (n = 16, ages 22–62).

Dogs. En face and retinal cross-sectional imaging was performed with the dogs under isofluorane inhalation general anesthesia. Overlapping en face images of reflectivity with near-infrared illumination (820 nm) were obtained (Spectralis HRA + OCT, Heidelberg, Germany) with 30° and 55° diameter lenses as previously described (30). All normal ONL maps were registered by the centers of the optic nerve head and rotated to bring the area centralis regions in congruence and a map of mean normal ONL thickness was derived (30). The area centralis region of NPHP5-mutant dogs was determined by superimposing a normal template onto mutant eyes by alignment of the optic nerve head, major superior blood vessels and the boundary of the tapetum. Next, NPHP5-mutant ONL maps were registered to the normal map by the center of the optic nerve and estimated area centralis region, and an ONL fraction was derived by dividing each ONL thickness sample by the corresponding mean normal value. This derived ONL fraction map was sampled at five locations: two loci were in superior retina, two loci in the inferior retina and one locus was the central visual streak.

Retinal function studies

Full-field flash ERGs were recorded from both eyes of littermate mutant and heterozygous control dogs under general anesthesia (induction with IV propofol; maintenance with isofluorane) using a custom-built Ganzfeld dome fitted with the LED stimuli of a ColorDome stimulator (Diagnosys) and methods previously described (30). Waveforms were processed with a digital low-pass (50 Hz) filter to reduce recording noise if necessary. After 20 min of dark adaptation, rod and mixed rod-cone–mediated responses (averaged four times) to single 4 ms white flash stimuli of increasing intensities (from −3.24 to 1.01 log cd·s·m⁻²) were recorded. Following 5 min of white light adaptation (1.53 log cd·m⁻²), cone-mediated signals (averaged 10 times) to a series of single flashes (from −2.24 to 1.01 log cd·s·m⁻²) and to a 29.4-Hz flicker (averaged 20 times; from −2.24 to 0.76 log cd·s·m⁻²) stimuli were recorded. The protocols used separately assessed rod- and cone-mediated responses.

Anatomic and immunohistochemical studies

The eyes of age-matched affected and littermate heterozygous control dogs were used for morphologic examination (Supplementary Material, Table S2); additional wild-type control dogs were used as well. Eyes were fixed, cryoprotected, and embedded in optimal cutting temperature (OCT) medium (43). Cryosections 7 µm or 10 µm in thickness were cut from the four meridians (Supplementary Material, Fig. S3), air-dried at room-temperature and used for structural, immunohistochemical (IHC), cell-counting, TUNEL and cell-proliferation studies. Hematoxylin and eosin (H&E) stained sections were examined by light microscopy (Axioplan, Carl Zeiss Meditec GmbH, Oberkochen, Germany) in contiguous fields extending from the optic disc to the ora serrata, and quantitative evaluation of the ONL thickness was done on sections from the four meridians of the left eyes at three specific locations (Supplementary Material, Fig. S3). At each of these sites, the number of rows of nuclei in the ONL were counted in at least three representative areas of a 40× field and averaged. Images were captured digitally (Spot 4.0 camera, Diagnostic Instruments, Inc., Sterling Heights, MI).

For some antibodies, optimal labeling was achieved by performing either high or low pH antigen retrieval prior to immunolabeling (Supplementary Material, Table S3; Supplementary Materials and Methods). Immunolabeling was performed by standard IHC methods previously described (27,43). Slides were incubated overnight with one primary antibody, and subsequently incubated with a fluorochrome-labeled secondary antibody (Alexa Fluor 568 or 488, Invitrogen, Carlsbad, CA). This process was repeated with a second primary and secondary antibody for dual-labeling. Mitotic cells were visualized by Phospho-histone H3 (PHH3) labeling (50,51), and apoptotic nuclei by TUNEL (terminal deoxynucleotidyl transferase-mediated biotinylated UTP nick end labeling) labeling performed according to the manufacturer’s instructions (Roche, Indianapolis, IN). Peanut agglutinin (PNA) labeling was used to label the cone insoluble extracellular matrix domain (29), and done by incubating tissue sections in fluorescently-tagged PNA for 1 hour. Cell nuclei were stained with DAPI (49,69-diamino-2-phenylindole). Slides were mounted with a medium composed of polyvinyl alcohol and DABCO (1,4 diazabicyclo-[2.2.2]octane) (Gelvatol, Sigma Aldrich, St. Louis, MO), and examined with either an epifluorescence microscope (Axioplan, Carl Zeiss Meditec) or a confocal microscope (Leica SP5 confocal laser scanning spectral imaging system on a DM6000 microscope).

Cell counting. Details of the methods used to quantify cell numbers are presented in Supplementary Materials and Methods (Cell Counting). TUNEL- and PHH3-labeled cells in the ONL were counted throughout the entire length of the sections taken from the superior (TUNEL and PHH3) and inferior (TUNEL only) meridians. Cone numbers/unit length of photoreceptor layer were determined by counting the number of cone arrestin-labeled OS or somas, in control or affected samples. A similar method was used to count the number of S- and L/M-cone subpopulations, labeled with the specific cone opsin antibodies.

Gene expression quantification

Three affected and three homozygous normal age-matched control dogs at 5 weeks of age were used for gene expression analysis using a custom-made canine-specific qRT-PCR profiling array (Applied Biosystems, Foster City, CA) (31,32). Additional genes were analyzed in single qRT-PCR assays, and all genes are listed in Supplementary Material, Table S4. Two different primer sets were used to analyze NPHP5 expression, and were located 5' (NPHP5_1 spanning exons 4–5) and 3' (NPHP5_2 spanning exons 11–12) of the exon 10 deletion. qRT-PCR was performed with strict adherence to the guideline for minimum information for publication of quantitative real-time PCR experiments (MIQE) (52) using methods published previously (32,53). Canine-specific genes (Supplementary Material, Table S4) were amplified with SYBR green, using primers designed with Primer Express Software v3.0, or using Taqman assays (Applied Biosystems). C_T values of each gene were normalized to GAPDH, and the ΔΔC_T method was used to compare affected with control samples (54). An unpaired t-test was performed and the p-values adjusted with the Benjamini–Hochberg (BH) step-up false discovery rate (FDR) controlling procedure (55) to determine statistical significance (BH-adjusted P < 0.05). Differentially expressed (DE) genes were defined as those with a statistically significant change in expression levels, measured as fold change (FC), where FC > +2 or <−2.

Analysis of brain and renal structure and renal function

Different studies were carried out to determine the presence of central nervous system (CNS) and/or renal abnormalities in NPHP5 mutant dogs. As controls, we used two different strains of retinal degeneration affected dogs: xlpra2 caused by a non-LCA ciliopathy gene mutation in RPGR (56), and rcd1 caused by a PDE6B mutation (57) (Supplementary Material, Table S2). Details are presented in Supplementary Materials and Methods.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

The authors are grateful to Dr. C. Craft (University of Southern California, CA) and Dr. William Tsang (Institut de recherches cliniques de Montréal) for providing, respectively, the human cone arrestin and NPHP5 antibodies, Svetlana Savina (University of Pennsylvania, PA) for assistance with laboratory techniques, Inna Martynyuk for assistance with confocal microscopy, and the staff of the Retinal Disease Studies facility for animal care and technical assistance for in vivo studies. We are grateful to Dr. Ariel Mosenco, University of Pennsylvania School of Veterinary Medicine for advice in selecting tests for renal function, and interpretation of results, to Mr. Joe McGrane for collecting and fixing the brains and kidneys, Dr. Stephen Pickup of the Department of Radiology, University of Pennsylvania Perelman School of Medicine for MRI studies, and Dr. Wil Mai, Section of Radiology, University of Pennsylvania School of Veterinary Medicine for renal ultrasound analyses.

Conflict of Interest statement. None declared.

Funding

National Institutes of Health (EY-06855, EY-17549, EY-020516, P30EY-001583, R24EY-022012, R24EY-023937, 2PNEY-018241), Foundation Fighting Blindness Center grants, NIH-Merck/Merial Summer Research Fellowship NIH T35 RR07065, Hope for Vision, Alcon Research Institute and Research to Prevent Blindness.

Author notes

†

The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors.

‡

Present address: College of Veterinary Medicine and Biomedical Sciences, Texas A&M University, College Station, TX 77845, USA.