|Home | About | Journals | Submit | Contact Us | Français|
To evaluate retinal structure and photoreceptor mosaic integrity in subjects with OPN1LW and OPN1MW mutations.
Eleven subjects were recruited, eight of whom have been previously described. Cone and rod density was measured using images of the photoreceptor mosaic obtained from an adaptive optics scanning light ophthalmoscope (AOSLO). Total retinal thickness, inner retinal thickness, and outer nuclear layer plus Henle fiber layer (ONL+HFL) thickness were measured using cross-sectional spectral-domain optical coherence tomography (SD-OCT) images. Molecular genetic analyses were performed to characterize the OPN1LW/OPN1MW gene array.
While disruptions in retinal lamination and cone mosaic structure were observed in all subjects, genotype-specific differences were also observed. For example, subjects with “L/M interchange” mutations resulting from intermixing of ancestral OPN1LW and OPN1MW genes had significant residual cone structure in the parafovea (~25% of normal), despite widespread retinal disruption that included a large foveal lesion and thinning of the parafoveal inner retina. These subjects also reported a later-onset, progressive loss of visual function. In contrast, subjects with the C203R missense mutation presented with congenital blue cone monochromacy, with retinal lamination defects being restricted to the ONL+HFL and the degree of residual cone structure (8% of normal) being consistent with that expected for the S-cone submosaic.
The photoreceptor phenotype associated with OPN1LW and OPN1MW mutations is highly variable. These findings have implications for the potential restoration of visual function in subjects with opsin mutations. Our study highlights the importance of high-resolution phenotyping to characterize cellular structure in inherited retinal disease; such information will be critical for selecting patients most likely to respond to therapeutic intervention and for establishing a baseline for evaluating treatment efficacy.
Mutations in the long-wavelength (L) and middle-wavelength (M) cone opsin genes (designated OPN1LW and OPN1MW, respectively) have been associated with a wide range of visual defects including red-green color vision deficiency, blue cone monochromacy (BCM), X-linked cone dystrophy, X-linked cone dysfunction, and high myopia with abnormal cone function.1–16 While characterization of visual function in these individuals is relatively straightforward, less is known about how the presence of OPN1LW and OPN1MW mutations affects retinal structure. Such information will be of paramount importance for advancing efforts to restore cone function in individuals with OPN1LW and OPN1MW mutations.
Recent studies have shown that OPN1LW and OPN1MW mutations resulting in congenital red-green color vision defects are associated with a variable retinal phenotype, with some individuals showing disrupted cone structure and/or thinning of the outer nuclear layer (ONL).8,14,17,18 It is difficult to draw definite conclusions about the pathogenicity of a specific mutant from comparisons of these individuals, as there may be other factors influencing the retinal phenotype. For example, during development, there is competition between the first two genes in the X-chromosome opsin array in the nascent L/M cones that ends with only one of the two genes being expressed in each cell.19 It has been shown that the relative proportion of cones expressing each of the two genes in the L/M array varies widely (over 40-fold).20,21 Thus previously observed differences in retinal phenotype may be confounded by differences in the relative expression of the mutant opsin with respect to the normal opsin. As the degree of retained cone photoreceptor structure is related to the therapeutic potential of a given retina,22 elucidation of genotype-specific retinal phenotypes is essential.
In one of the more serious vision disorders associated with OPN1LW and OPN1MW mutations, a single type of mutant opsin is expressed in all the cones that would have been L or M in a normal eye. In these subjects, rods and short-wavelength (S) cones are the only photoreceptors expressing normal photopigments. These individuals offer the opportunity to directly evaluate the effect of different OPN1LW and OPN1MW mutations. These mutations can be placed into one of three categories: (1) mutations that produced random nonhomologous missense substitutions at single amino acid positions1,3,12,16; (2) partial or complete deletion of an exon15,23; and (3) a recently identified category involving intermixing of ancestral OPN1LW and OPN1MW genes to produce “L/M interchange” mutations with deleterious combinations of nucleotides at normal polymorphic positions.7,8,10,13 While at least one L/M interchange mutation has been shown to directly cause cone malfunction (Greenwald SH, et al. IOVS 2012;53:ARVO E-Abstract 4643), it was recently shown that in addition to any functional changes in the photopigment caused by the mutations, many of the L/M interchange mutations also interfere with recognition of exon 3 by the splicing mechanism.24 Some of the variants incompletely interfere with splicing, so full-length mRNA is produced as well as the inappropriately spliced transcript. Whether there are structural differences between the mutation categories, or for different mutations within a category, has been unknown.
Here we used adaptive optics scanning laser ophthalmoscopy (AOSLO) and spectral-domain optical coherence tomography (SD-OCT) to examine 11 subjects for whom all cones except the S cones express one of six mutant opsins. There were differences in the anatomy and in the course and severity of vision loss across mutation categories. The subjects with L/M interchange mutations reported a later-onset progressive loss of visual function, while those with the C203R mutation showed a typical congenital BCM phenotype. We observed significant disruption of retinal lamination and of cone mosaic topography in all subjects, though the degree of disruption was generally greater for subjects with L/M interchange mutations than for those with random mutations. These differences provide insight into the underlying mechanisms responsible for loss of structure and function in these subjects. Furthermore, while the cone loss observed may limit success of any efforts to restore L/M cone function using gene therapy in any of these subjects, it may be possible to develop strategies to slow or halt the degenerative changes in people harboring L/M interchange mutations.
Written informed consent was obtained after the nature and possible consequences of the study were explained. This study followed the tenets of the Declaration of Helsinki and was approved by all local ethics committees. We examined eight subjects for whom the clinical phenotype has been previously described and three new subjects (see Table and Supplementary Material, http://www.iovs.org/content/53/13/8006/suppl/DC1). Two male subjects (JC_0826, 22 years; JC_0847, 23 years) with normal color vision were included for comparison, and data from two previously published normative databases were used for comparison of the SD-OCT studies. The data used for comparison against the horizontal line scans consisted of 93 subjects with an average age of 25.7 ± 8.2 years,14 and the data used for comparison against the topographical thickness maps consisted of 60 subjects with an average age of 29 ± 8.42 years.25 Axial length measurements were obtained on all subjects (Zeiss IOL Master; Carl Zeiss Meditec, Dublin, CA) in order to determine the scale (in microns per pixel) of each retinal image. Prior to all retinal imaging, each eye was dilated and cycloplegia was induced through topical application of a combination of phenylephrine hydrochloride (2.5%) and tropicamide (1%).
For the three newly recruited subjects, DNA was isolated from whole blood, and the opsin genes were amplified and sequenced using previously described methods.7
Volumetric images of the macula were obtained using high-definition OCT (Cirrus HD-OCT; Carl Zeiss Meditec). Volumes were nominally 6 mm × 6 mm and consisted of 128 B-scans (512 A-scans/B-scan). Retinal thickness was calculated using the Macular Analysis software on the Cirrus (software version 5.0; Carl Zeiss Meditec), which is automatically determined via measurement of the difference between the inner limiting membrane (ILM) and retinal pigment epithelium (RPE) boundaries. High-resolution SD-OCT images of the macula were acquired (Bioptigen, Research Triangle Park, NC). High-density line scans (1000 A-scans/B-scan, 100 repeated B-scans) were acquired through the foveal center, then registered and averaged as previously described.26 No transformation (e.g., flattening), filtering, or other postprocessing was applied to these images.
For retinal sublayer analysis, we manually segmented the ILM, the outer plexiform layer (OPL), the external limiting membrane (ELM), and the RPE on the high-resolution line scans as previously described using ImageJ (National Institute of Mental Health, Bethesda, MD).14 The ILM–RPE distance provides total retinal thickness, the ILM–OPL distance provides the inner retinal thickness, and the OPL–ELM distance provides the ONL plus Henle fiber layer thickness (ONL+HFL).
An adaptive optics scanning light ophthalmoscope (AOSLO) was used to obtain images of the photoreceptor mosaic at various retinal locations. JC_0183, JC_0184, JC_0440, JC_0441, JC_0355, JC_0356, JC_0118, and JC_0430 were imaged at the University of Rochester; details of this imaging system have been previously published.27 The remaining three subjects and the normal controls were imaged at the Medical College of Wisconsin, and details of this system have also been published.28 The systems derive from the basic design described by Gray et al.,29 with the afocal telescopes folded to varying degrees.27 Both systems utilized a 97-channel deformable mirror (ALPAO, Biviers, France) as the wavefront corrector, with a Shack-Hartman wavefront sensor used to measure the wavefront. The imaging source was either a 775 nm or a 796 nm super-luminescent diode (Inphenix Inc., Livermore, CA), and the light exposure was kept below the safe maximum permissible exposure set forth by ANSI.30,31 Images were processed to remove distortions induced by the sinusoidal motion of the resonant scanner by estimating the distortion from images of a calibrated Ronchi ruling and then resampling the images over a grid of equally spaced pixels. A separate calibration was done for each subject. Images were registered to improve signal-to-noise as previously described.32
The registered images from each subject were combined into a single montage (Adobe Photoshop; Adobe Systems, Inc., San Jose, CA). This montage was scaled and aligned to the LSO image from the HD-OCT, in which a crosshair was placed at the center of the foveal pit. The scaling was based on theoretical magnification of each system and the alignment performed using blood vessel patterns. The locations to be analyzed were determined based on the distance from the center of the foveal pit, and the original image comprising that portion of the montage was set aside for subsequent density analyses. Cone density was measured at selected retinal locations (0.4, 0.8, 1.2, 1.6, 2.0 mm) using manual identification of cone structures (80 μm × 80 μm sampling window). For rod analysis, we utilized a smaller sampling window (55 μm × 55 μm) than we did for the cone density analysis, as the cones in these subjects were more sparse and irregularly spaced. Rod density was calculated at various locations between 0.5 and 3 mm using a previously described semiautomated algorithm in which the user could add/subtract missed or erroneous cell identifications.33
The Table provides a summary of subject demographics, with the clinical phenotype on 8 of the 11 having been reported previously. All four subjects with C203R mutations and the subject with the exon 2 deletion presented a classical BCM phenotype, with vision based solely on S cones and rods.12,15 The two brothers with W177R mutations (JC_0355 and JC_0356) had an onset in the first decade with no history of nystagmus; both showed subsequent deterioration of visual acuity and color vision, macular pigment epithelial disturbance, and severe generalized cone dysfunction on ERG.16 Both had good S-cone function with absent L/M cone function on psychophysical testing, though the data could not entirely exclude some residual L/M cone function in JC_0355.16
The remaining four subjects (JC_0347, JC_0564, JC_0118, KS_0577) had OPN1LW and OPN1MW mutations that fell into the newly discovered category involving intermixing of ancestral genes to produce L/M interchange mutations with deleterious combinations of nucleotides at normal polymorphic positions in exon 3.7,8,10,13,19,24 All four of these subjects had a late-onset, progressive phenotype that was in stark contrast to that of the subjects with a C203R mutation or the exon 2 deletion (see Supplementary Material, http://www.iovs.org/content/53/13/8006/suppl/DC1). While the subjects with W177R mutations also had a progressive phenotype, it was earlier in onset than observed for the L/M interchange mutations and resulted in more complete loss of L/M cone function.
Topographical maps of retinal thickness show variable but significant macular thinning in individuals with all three classes of mutation (see Supplementary Material and Supplementary Fig. S1, http://www.iovs.org/content/53/13/8006/suppl/DC1). The average (±SD) central subfield (CSF) thickness (central 1 mm) for the 11 subjects examined here was 194 ± 34 μm, compared to an average (±SD) value for 60 normal subjects from our lab of 266 ± 19 μm; P < 0.0001, Mann-Whitney test.25 The subjects with random point mutations had more normal CSF thickness (216 ± 20 μm) than the individuals with L/M interchange mutations (174 ± 28 μm); P = 0.0381, Mann-Whitney test. In addition, the retinal thinning in the subjects with L/M interchange mutations was more widespread (see Supplementary Material and Supplementary Fig. S1, http://www.iovs.org/content/53/13/8006/suppl/DC1).
To investigate the reduction in retinal thickness in more detail, we examined the thickness of the inner retina and the ONL+HFL from high-resolution horizontal cross sections (Fig. 1), and compared it to previously reported normative data.14 Consistent with the topographical thickness analysis, we observed that subjects with random point mutations had more normal total retinal thickness than subjects in the other two genetic categories (Fig. 2A). However, subjects with random point mutations had thinning restricted to the ONL+HFL, while subjects with L/M interchange mutations (LVAVA, LIAVS, and LVVVA) or the exon 2 deletion showed parafoveal thinning of the inner retina in addition to reduced ONL+HFL thickness (Figs. 2B, B,22C).
In the normal retina, at least four hyperreflective bands comprise the outer photoreceptor complex (see JC_0847 in Fig. 1). The innermost peak of the outer photoreceptor complex is attributed to the ELM, while the second layer is now thought to derive from the ellipsoid portion of the inner segment (ISe).34 The third layer is attributed to the RPE contact cylinder, and the fourth band is attributed to the RPE. As shown in Figure 1, we observed disruption of the ISe in all subjects; however, there were differences between the mutation classes that paralleled the differences in retinal thinning. Five of the six individuals with random point mutations (JC_0183, JC_0184, JC_0440, JC_0441, and JC_0355) had a focal disruption of the ISe near the foveal center. In contrast, subjects with L/M interchange mutations had generally greater disruption of the ISe: JC_0118 (LIAVS) and KS_0577 (LVVVA) had a large area of ISe loss (as did the subject with the exon 2 deletion, JC_0430), and JC_0564 (LVAVA) showed diffuse mottling of the ISe. However, these differences did not segregate perfectly with genotype category, as JC_0347 (LVAVA) had a well-defined focal lesion and JC_0356 (W177R) had diffuse mottling of the ISe.
In the six individuals with the focal ISe disruption, the boundaries of the disruption were marked using ImageJ35; the average (±SD) width was 99.2 ± 43.5 μm, consistent with previous estimates of the size of the S-cone free zone in humans.36–38
Foveal montages created by stitching together multiple overlapping images are shown in Figure 3. Differences between the genotypic categories parallel those observed in the SD-OCT images. We observed a hyporeflective area at or near the foveal center in the six aforementioned subjects with a focal ISe disruption on SD-OCT, consistent with this small area lacking healthy, waveguiding cones. A sparse population of hyperreflective cones, presumably S cones, surrounded this hyporeflective area. The three subjects with a larger area of ISe loss on SD-OCT had irregular areas of hyperreflectivity in the foveal montages with minimal cone structure present, consistent with disruption of L, M, and S foveal cones in their macula. The remaining two subjects had more diffuse photoreceptor mosaic disruption across the foveal montage with only sporadic hyperreflective cones, in keeping with the irregularly disrupted ISe.
Shown in Figure 4 are plots of the putative S cones observed in the foveal montages for the six subjects with a discrete disruption of the ISe, compared to a plot of similar data from Curcio et al.38 While the analysis may not capture every S cone, it does provide a robust way to visualize the relative absence of reflective cones in a particular region. The size of the presumed S-cone free zone, determined by finding the largest circle that could be placed within the cone mosaic without encroaching on any cones, ranged from 50 to 120 μm, with an average of 72 μm for the six subjects. This is consistent with previous estimates of the size of the S-cone free zone,36–38 providing further support that the focal ISe represents complete loss of L/M cone structure in the foveola. Interestingly, the center of the presumed S-cone free zone did not always align to the center of the foveal pit.
The parafoveal cone mosaic in subjects with OPN1LW and OPN1MW mutations is severely disrupted compared to normal (Fig. 5). However, there are important differences between the genetic categories. Between L/M interchange mutations and those with C203R mutations and W177R mutations, the severity of the losses was reversed compared to that in the foveal region. While the macula was generally thinner and the foveal region more disrupted for the interchange mutations as a group, their parafoveal cone numbers were better preserved in the parafovea (Fig. 6). The difference was largest at 2 mm from the fovea, the most eccentric location measured. We were not able to resolve the rod mosaic in all subjects, but measurements made outside the central area of ISe disruption (visible on OCT) showed rod density consistent with that measured previously in normals with the same technique (Fig. 7).39
Assuming a starting cone density equal to that of an average normal,40 the residual density for the subjects with the random point mutations was 8.5% of normal (SD = 3.0%), consistent with what would be expected if they had only S cones remaining.37,38 In contrast, the subjects with L/M interchange mutations had residual densities that were on average 23% of normal (SD = 10.7%). Factoring in the expected population of S cones, we estimate that this corresponds to a loss of 86% of the L/M cones (SD = 9%). As normal cone density is highly variable, it is impossible to determine the exact degree of cone loss (or the degree of cone retention); however, it was clear that the retinas harboring C203R or W177R had very few, if any, residual L/M cones. Likewise, there was certainly residual L/M cone structure in subjects with L/M interchange mutations, consistent with the residual L/M cone function measured with the ERG in KS_0577, JC_0347, and JC_0564 (Kuchenbecker J, et al. IOVS 2012; 53:ARVO E-Abstract 6400). Structure/function agreement was also seen between the two subjects with the W77R mutation; JC_0355 had higher cone density and less retinal thinning than his brother (JC_0356), consistent with his slightly better vision in one eye and better L/M cone function on psychophysical testing.16
While all subjects had disrupted photoreceptor mosaics and reduced retinal thickness, there were significant differences between the mutation classes. The imaging results provide a number of insights into the basis for the observed phenotypic differences between the mutation categories. In general, the six subjects with random missense mutations (C203R and W177R) had the healthiest-appearing retinas on SD-OCT. The small disruption of the ISe that was present was shown to correspond to the S-cone free zone,36–38 with the focal loss of the ISe being the result of the absence of healthy L or M cones, S cones, or rods to provide structure at the foveal center.
In contrast to the C203R and W177R mutations, the L/M interchange mutations (four subjects) and exon 2 deletion (one subject) appear to be more disruptive to the overall foveal architecture. Only one of five of these subjects (JC_0347, LVAVA) had a small focal disruption of the ISe; the others had diffuse mottling of the ISe or a much larger absence of the ISe (extending into the parafovea). Compared to the subjects with the random point mutations, these subjects had greater retinal thinning that involved the inner retina as well as the ONL. This suggests that these L/M interchange mutations cause, or are associated with, degenerative changes that result in damage to neighboring cells (S cones and rods) in addition to those expressing mutant L/M opsin.
The specificity of the photoreceptor damage to the macular region in the subjects with L/M interchange mutations is striking when images of the fovea are compared to more peripheral ones. Retinoid by-products of the visual cycle and all-trans-retinal (atRAL) are particularly toxic.41 It is possible that the cones expressing the L/M interchange mutants remain viable through young adulthood but are defective in their ability to participate normally in the visual cycle. This could lead to a buildup of atRAL or other toxic retinoids, affecting not only the cones expressing the mutant opsin, but also parafoveal S cones and rods. If these toxic by-products are concentrated in the fovea where the cone density is 20 times higher than in the peripheral retina,40 it could explain why the collateral damage is higher at the central retina compared to more eccentric locations (Fig. 1).
Advances in gene therapy have generated a great deal of excitement regarding the restoration of cone function in a variety of retinal diseases.42–48 While subjects with L/M interchange mutations had the greatest degree of residual parafoveal L/M cone structure, the presence of macular atrophy and inner retinal thinning in most of these subjects would limit the therapeutic opportunity in these individuals at later stages of the disease. However, strategies may be developed to slow the degenerative effects of these mutations. In contrast, the subjects with C203R or W177R mutations generally had more preserved retinal lamination, but adaptive optics imaging revealed no evidence for retained L/M cone structure. It is unclear whether any cone cell bodies remain, though given that rods appear to have expanded to fill in the space occupied by the cones, this would imply degeneration of at least the inner and outer segments.
We did not examine subjects with deletions involving the locus control region (LCR), so it remains to be seen how complete absence of L/M opsin affects cone photoreceptor integrity compared to the expression of mutant opsin. Furthermore, we are unable to say if there are structural differences between individuals harboring different L/M interchange mutations. Previous evidence showed that an individual with the LIAVA mutant expressed by one of the two genes in the L/M array had random, non-waveguiding cells throughout the cone mosaic, normal ISe integrity, normal inner retinal thickness, and no change in cone structure measured over a span of 8 years.8,18 This suggests that the LIAVA mutant may not result in the progressive loss of L/M cones (or does so on a much slower time scale), and it will be important to longitudinally assess the progressive nature of the various L/M interchange mutations to better determine the therapeutic potential in these individuals.
In addition to using the imaging tools described here to prioritize potential subjects most suitable for intervention, by characterizing the degree of residual cone structure in subjects with OPN1LW and OPN1MW mutations, it will be valuable to employ the same techniques when evaluating the safety and efficacy of any future therapeutic intervention. Such an approach has already been demonstrated in patients with retinitis pigmentosa receiving ciliary neurotrophic factor,49 where preservation of cone structure was observed in the absence of a significant functional improvement in vision. It is entirely plausible that structural recovery or preservation precedes functional changes; however, this will remain unclear until high-resolution imaging metrics become a routine part of the outcome measures used in clinical trials.
The authors thank Pooja Godara, Zach Harvey, Jennifer Norris, Jungtae Rha, Netta Smith, and Phyllis Summerfelt for assistance with the study.
Supported by NEI Grants P30 EY001931 (JC), R01 EY017607 (JC), T32 EY014537 (AMD), UL1 RR031973 (KES, JC), R01 EY014375 (DRW), P30 EY001319 (DRW), P30 EY001730 (MN, JN), R01 EY009303 (MN), and R01 EY009620 (JN); supported by grants from the Gene and Ruth Posner Foundation (JC), Foundation Fighting Blindness (JC, MM), The E. Matilda Ziegler Foundation for the Blind (JC), the RD & Linda Peters Foundation (JC), unrestricted departmental grants from Research to Prevent Blindness (Medical College of Wisconsin, University of Washington, University of Illinois-Chicago), the Juvenile Diabetes Research Foundation (8-2002-130), Fight for Sight UK (AJH, MM), Moorfields Eye Hospital Special Trustees (AJH, MM), and the National Institute for Health Research Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology (AJH, ATM, MM); and supported in part by Grant No. 3000003241 from the Chief Scientist Office of the Ministry of Health, Israel, and the Yedidut Research Grant (DS, EB). J. Carroll and A. Dubra are recipients of Career Development Awards from Research to Prevent Blindness. A. Dubra is the recipient of a Career Award at the Scientific Interface from the Burroughs Wellcome Fund. M. Michaelides is a recipient of a Career Development Award from the Foundation Fighting Blindness. This investigation was conducted in part in a facility constructed with support from the Research Facilities Improvement Program; Grant C06 RR016511 from the National Center for Research Resources, National Institutes of Health. DRW holds patents on adaptive optics imaging.
Disclosure: J. Carroll, Imagine Eyes, Inc. (S); A. Dubra, None; J.C. Gardner, None; L. Mizrahi-Meissonnier, None; R.F. Cooper, None; A.M. Dubis, None; R. Nordgren, None; M. Genead, None; T.B. Connor Jr, None; K.E. Stepien, None; D. Sharon, None; D.M. Hunt, None; E. Banin, None; A.J. Hardcastle, None; A.T. Moore, None; D.R. Williams, Canon (F), Polgenix (F), Pfizer (C), P, GlaxoSmithKline (C), P, Pfizer (R); G. Fishman, None; J. Neitz, None; M. Neitz, None; M. Michaelides, None