RDH12 and RPE65, Visual Cycle Genes Causing Leber Congenital Amaurosis, Differ in Disease Expression

Abstract

purpose. Human blindness caused by mutation of visual cycle genes has been discussed as potentially treatable by retinoid replacement either through gene transfer or pharmacological bypass. Mutations in the RDH12 gene disrupt the visual cycle in vitro, but little is known of the in vivo effects of mutant RDH12, other than the association with severe early-onset autosomal recessive retinal disease. The relationship of retinal organization and visual function in patients with RDH12 mutations was determined and comparisons made with the disease from mutations in another visual cycle gene, RPE65.

methods. Young patients with RDH12 mutations were studied with optical coherence tomography (OCT) and colocalized measures of vision with dark-adapted absolute thresholds. Results were compared to those in patients with RPE65 mutations.

results. Retinal architecture of patients with RDH12 mutations was appreciably distorted, precluding identification of the normal laminae. Some RDH12-mutant retinas were remarkably thick and others were thin, but all had the same dysplastic pattern. A comparison with the structural and functional consequences in patients with mutations in RPE65 indicated that the pathogenesis of retinal degeneration in RDH12 mutations was distinctly different.

conclusions. The results demand critical consideration of the human disease mechanism and the therapeutic approach in patients with mutations in the putative visual cycle gene RDH12.

The human visual (or retinoid) cycle is composed of biochemical reactions and transport mechanisms that maintain the response of retinal photoreceptors to light and are thereby critical to vision. ^{1 2} The complex pathways, some essential and some redundant, involve metabolism of vitamin A and retinoid cycling between the retinal pigment epithelium (RPE) and the photoreceptors. When light is absorbed in photoreceptors by photopigment molecules, 11-cis-retinal is isomerized to all-trans-retinal, and the visual cycle acts to restore 11-cis-retinal for further light absorption. A high level of scientific interest in the participating cells and reactions of the visual cycle dates back more than a century and continues to the present day. ^{1 2 3 4 5 6} As genes responsible for visual cycle molecules have been identified, some have been associated with human retinal diseases. ^{1 2}  

A putative visual cycle disease gene recently identified is RDH12 (retinal dehydrogenase 12), which encodes a member of the short-chain dehydrogenase/reductase superfamily of proteins. ^{7 8 9 10 11 12 13} RDH12 has its highest expression in the retina ¹³ and has been localized to photoreceptors. ⁷ The autosomal recessive human retinopathy associated with RDH12 mutations is blinding at young ages, ^{14 15 16} described clinically as a form of early-onset retinal degeneration or Leber congenital amaurosis (LCA). Clinical testing has prompted speculation about the similarities or differences between affected individuals with RDH12 mutations and those with mutations in RPE65 (Gal A, et al. IOVS 2006;47:ARVO E-Abstract 2975), ^{14 15 16} a critical visual cycle gene encoding the isomerohydrolase in RPE. ^{6 17 18} Comparison of these two severe human eye diseases associated with visual cycle gene mutations is of interest because of success at gene transfer and retinoid replacement therapies in RPE65-deficient animals ^{19 20 21 22 23 24 25} and the potential relevance of such work in humans. ²⁵  

Is the human disease due to RDH12 mutations simply another early-onset retinal degeneration caused by an abnormal visual cycle, and is it ready to be positioned in the queue for retinoid replacement therapy, either by gene transfer or pharmacological bypass? ^{1 26 27} With high-resolution microscopy of the living eye, we studied retinas from compound heterozygotes with RDH12 mutations. In vitro function analyses of these specific alleles have demonstrated reduced expression and profound loss of catalytic activity when assayed for ability to convert all-trans retinal to all-trans-retinol. ¹⁶ Our in vivo studies found an unexpected and dramatic retinal laminopathy detectable in the first decade of life. This is in striking contrast to the disease expression leading from RPE65 mutations. ²⁵ The pathogenesis of human RDH12 mutations is thus not a slowly progressive retinal degeneration superimposed on disruption of 11-cis-retinal synthesis. ²⁸  

Materials and Methods

Subjects

There were four patients with RDH12 mutations, representing three families (Table 1) . Molecular diagnostics and in vitro studies of the mutations have been reported. ¹⁶ Other patients included four with RPE65 mutations, ^{25 29 30} four with CRB1 mutations, ³¹ and four with retinal degenerations screened as negative for RPE65 and RDH12 mutations. Normal subjects (n = 30; age range, 8–56 years) were also studied. Informed consent and assent were obtained; procedures adhered to the Declaration of Helsinki and were approved by the institutional review board. 

Optical Coherence Tomography

In vivo microscopy of the retinal cross section was obtained with optical coherence tomography (OCT; Carl Zeiss Meditec, Dublin, CA). The principles of the method and our recording and analysis techniques have been published. ^{25 31 32 33 34 35 36} Data were acquired with an OCT3 instrument with a theoretical axial resolution in retinal tissue of ∼8 μm. In all subjects, overlapping OCT scans of 4.5 mm length were used to cover horizontal and vertical meridians up to 9 mm eccentricity from the fovea. In a subset of patients, dense raster scans were performed to sample a 18 × 12-mm² region of the retina centered on the fovea. ^{25 33} At least three OCT scans were performed at each retinal location. A video fundus image was acquired and saved with each OCT scan by the commercial software. In addition, the fundus image visible during the complete session was recorded continuously on a video cassette recorder. 

Postacquisition processing of OCT data was performed with custom programs (MatLab ver. 6.5; The MathWorks, Natick, MA). Longitudinal reflectivity profiles (LRPs) making up the OCT scans were aligned by using a dynamic cross-correlation algorithm ³² with a manual override when crossing structures (for example intraretinal pigment), which interrupted local lateral isotropy of signals. Repeated scans were laterally aligned and averaged to increase the signal-to-noise ratio and allow better definition of retinal laminae. ²⁵  

Retinal thickness was defined as the distance between the signal transition at the vitreoretinal interface (labeled T1 in Ref. ³² ), and major signal peak corresponding to the RPE. ³¹ In normal subjects, the RPE peak was assumed to be the last peak within the two- or three-peaked scattering signal complex (labeled ORCC in Ref. ³² ) deep in the retina. In patients, the presumed RPE peak was sometimes the only signal peak deep in the retina; at other times, it was apposed by other major peaks (see 1 Figs. 2 3 ). In the latter case, the RPE peak was specified manually by considering the properties of the backscattering signal originating from layers vitread and sclerad to it. Outer photoreceptor nuclear layer (ONL) thickness was defined as the major intraretinal signal trough delimited by the signal slope maxima and measured as previously described. ^{25 31 33} In some patients and at some retinal locations, ONL thickness was not measurable due to laminopathy. 

For topographic analysis, the precise location and orientation of each scan relative to retinal features (blood vessels, intraretinal pigment, and optic nerve head) were determined in the video images of the fundus. LRPs were allotted to regularly spaced bins in a rectangular coordinate system centered at the fovea; the waveforms in each bin were aligned and averaged. For two-dimensional maps (see Figs. 1A 1B 1C 1D ), 0.3 × 0.3-mm² bins were used for sampling whereas 0.15 × 0.15-mm² bins were used for analysis along the horizontal and vertical meridians (see Figs. 2A 2B 2C 2D 2E 2F ). The retinal thickness was measured as just described, missing data were interpolated bilinearly, thicknesses were mapped to a pseudocolor scale, and location of blood vessels and the optic nerve head was overlaid for reference. ²⁵ Lower and upper limits of normal retinal thickness were specified, and the patients’ results were subtracted to determine the loci of significant thickening or thinning. 

Quantitative analysis of LRP morphology at individual retinal loci was performed by using templates produced from groups of individuals from each genotype and normal subjects (see Fig. 2G ). Individual waveforms were first scaled along the axial direction to the average retinal thickness of the respective group and then averaged to result in templates. ³¹ To compare LRP morphology, RDH12, and CRB1 templates were further scaled to mean normal thickness. 

Psychophysics

Psychophysical thresholds were measured (1.7° diameter, 200-ms duration stimuli) at fixation in the dark-adapted state. Long/middle wavelength cone function was determined with 650-nm stimuli and compared with normal data determined during the cone plateau phase of dark adaptation. Details of the visual function techniques and analysis methods have been published. ^{25 31 33 36}  

Results

Topographical Maps of Retinal Thickness in RDH12-Mutant Retina

Normal human retina (Fig. 1A)has a central depression or foveal pit surrounded by an annulus of increased thickness, representing displaced inner retinal layers from foveal formation. Retinal thickness declines with distance from the parafoveal peak, except for the crescent-shaped thickening at superior and inferior poles of the optic nerve, representing converging axons from ganglion cells. Retinal topography in three patients with RDH12 mutations showed a range of results. Patient (P)3, at age 13, had extreme retinal thickening except at the fovea (Fig. 1B) . A difference map comparing P3 with normal (n = 5, ages 21 to 26; normal limits defined as ±2 SD from mean) indicates that P3 had subnormal thickness at the fovea, but supernormal thickness throughout the sampled region (Fig. 1B , inset). P4, at age 21, had thickness topography that appeared normal; a difference map showed foveal thinning and scattered islands of supernormal thickness (Fig. 1C) . The fovea in P2 (age 11), younger than P4 by a decade, was thinned centrally but attained normal thickness with increasing eccentricity (Fig. 1D) . Further quantitation of thickness results are provided as profiles across the horizontal and vertical meridians (Fig. 1E) . Data from all four patients are plotted in relation to normal (n = 25, age range, 8 to 56 years; limits defined as ±2 SD from the mean). Siblings P1 and P2, representing the youngest ages studied at 8 and 11 years, were at the lower limit of normal retinal thickness except in the fovea, which was subnormal. The oldest patient, P4, was at the upper limit of normal; P3, at an age similar to that of the young siblings, had remarkable thickening with the exception again of the abnormally thinned fovea. 

Markedly Distorted Laminar Architecture in RDH12- but Not in RPE65-Mutant Retina

Histology-like cross-sectional images of RDH12-mutant retina along the vertical meridian through the fovea revealed dramatic differences from normal laminar architecture, independent of total retinal thickness (Figs. 2A 2B 2C 2D) . Normal retina has a foveal depression and the surrounding retina is organized into laminae. Cellular layers generally correspond to low-reflectivity bands, and synaptic layers show higher reflectivity. ³¹ Other sources of high signal reflectivity are at the interface between retina and vitreous, a deep complex composed of photoreceptor inner and outer segments, RPE, and anterior choroid (termed the outer retinal choroidal complex, ORCC) and the nerve fiber layer. ^{25 31 32 33} In a vertical section, the nerve fiber layer, the unmyelinated axons of the ganglion cells, arched above and below the central retina and was very pronounced (Fig. 2A) . In contrast to the representative normal (age 19), the RDH12-mutant retina of P3 (age 13) was thin at the fovea and grossly thickened in adjacent surrounding retina. Normal laminae were not discernible. There was high reflectivity in a band representing almost half the retinal thickness, and deeper in the retina was a less reflective layer. High reflectivity was also seen near the depth of the normal ORCC (Fig. 2B) . A similar dysmorphic pattern was present in P4 (age 21), but the section was not as thick as that of P3, and there seemed to be a decline in thickness at greater distances from the fovea (Fig. 2C) . P2 (age 11 years) had an even thinner fovea, and the adjacent retina was dysmorphic (Fig. 2D)but within normal limits for total retinal thickness (Fig. 1E) . 

Comparisons of retinal organization in two other forms of LCA are shown (Figs. 2E 2F) . An RPE65-mutant retina (P6, age 20 years), in contrast to RDH12, retained normal laminar architecture (Fig. 2E) . The main abnormality was in the ONL, which was evident at and around the fovea and in the superior retina, but was reduced in other regions. A CRB1-mutant retina (P11, age 24 years) showed laminar disorganization and thickening that were highly reminiscent of those seen in RDH12. Reflectivity waveforms from a region in the temporal retina illustrate the differences in retinal layering between normal and these three different molecular causes of LCA (Fig. 2G) . The representative normal waveform has peaks and troughs that correspond to the low and high reflectivities in cross-sectional images. RPE65-mutant retina is thinner than normal, and this is mainly attributable to reduced ONL. RDH12- and CRB1-mutant reflectivity waveforms each represent an average of four individuals and are scaled to the size of the normal for easy comparison. Individual thin lines are the waveforms that form the average. The two groups of waveforms are alike and severely abnormal. We asked whether there was thickened nerve fiber layer in RDH12-mutant retinas as we found in CRB1, ³¹ but no such thickening was detected (data not shown). 

Foveal Cone Vision Predictably Related to Photoreceptor Layer Thickness in RDH12-Mutant Retina, Unlike the Relationship inRPE65-Mutant Retina

A relationship between photoreceptor layer thickness and visual function has recently been established in normal human retinas and in retinal degenerations. ²⁵ Human RPE65-mutant retinas tend to behave as if there is a visual cycle defect: there is a dissociation of retinal function and structure. In other words, there is more visual impairment than predicted from loss of photoreceptors. ²⁵ We have defined the relationship of structure and function for RPE65-mutant retinas in comparison to a group of non-RPE65 retinal degenerations at the cone-rich fovea and in rod-rich retinal regions. ²⁵ To test the hypothesis that RDH12-mutant retina was behaving like a disease with a defective visual cycle, we collected data at the fovea for photoreceptor function and structure and the relationship examined (Fig. 3) . The disorganization of the surrounding retina precluded other loci from being examined in this way. 

Superimposed on en face infrared fundus images of the RDH12-mutant retina of P4 (Fig. 3A)and the RPE65-mutant retina of P6 (Fig. 3B)are examples of the function and structure data sources for determining the relationship of the parameters (Fig. 3C) . Whereas foveal nuclear layer thickness was reduced by approximately 20 μm in RDH12-P4 compared with that in RPE65-P6, cone-mediated visual sensitivity was approximately 1.5 log units (30 times) better in the RDH12 patient than in the patient with RPE65 mutations. A graph of all data shows that the four patients with RDH12 mutations had abnormally reduced foveal cone vision and abnormally thinned ONL. RDH12 data were consistent with the theoretical expectations of vision loss due to photoreceptor loss, and they were not distinguishable from a group of patients in whom mutations in RPE65 or RDH12 genes were not found (Fig. 3C) . Specifically, in RDH12 mutations there was no greater structure than expected for the level of dysfunction. In contrast, three of four patients with RPE65 mutations had ONL thickness that was greater than expected for the level of dysfunction (Fig. 3C) , similar to what we had reported. ²⁵  

Discussion

Photon capture by the 11-cis-retinal chromophore in rhodopsin, the visual pigment of rod photoreceptors, leads to an all-trans isomer. The visual cycle serves to restore 11-cis isomer by enzymatic and transport mechanisms. ^{1 2 3 4 5} Mutations in visual cycle genes cause different effects on the vertebrate retina and some of these lead to human eye disease. ^{1 2 5} Enzymatic visual cycle pathway molecules studied to date fall into two categories: (1) those with a proven but nonessential or redundant role in the cycle; and (2) those that cause a major visual cycle blockade with little or no visual pigment production and/or devastating human phenotype. In the first category are the RPE components RDH5, RDH10, RDH11 ^{9 10 11 37 38} and the photoreceptor components RDH8, RDH11, and RDH14. ^{8 10 12} Deficiency states (by genetic engineering or naturally occurring) lead to delays in (but sufficient) production of visual pigment and a relatively minor phenotype, suggesting redundancy in the pathway. ¹⁶  

In the second category are two RPE molecules with enzymatic roles in the visual cycle: RPE65 and LRAT. Both cause profoundly defective production of 11-cis retinal in animals with genetically engineered or naturally occurring deficiency states, ^{19 20 27 39} and humans with RPE65 or LRAT gene mutations have early-onset autosomal recessive blindness. ^{25 29 40 41 42 43} In animals with RPE65 or LRAT deficiency, retinoid replacement by gene transfer or pharmacological bypass has restored vision, ^{19 20 21 22 23 24 25 27 44} leading the way to consideration of human trials for these potentially reversible biochemical defects. 

RDH12 is a putative visual cycle enzyme localized to photoreceptor inner segments. Mutations in RDH12 are associated with early-onset human blindness. ^{13 14 15 16} The human phenotype in RDH12 mutations, assessed by clinical eye examinations, has been described as severe (Gal A, et al. IOVS 2006;47:ARVO E-Abstract 2975). ^{14 15} There are no reports of animal models to date. Our detailed structural and functional studies in young patients with RDH12 mutations led to the unexpected result that RDH12-mutant retinas were remarkably different from those with RPE65 mutations. The rarity of LRAT-associated disease precluded that comparison. Whereas RPE65-mutant human retina behaved like a visual cycle enzymatic defect with relatively preserved retinal structure, RDH12-mutant retina was grossly dysplastic at comparable or younger ages. In the only retinal region with structure that permitted comparison with function, there was no evidence of an RPE65-like disproportionate loss of photoreceptor function compared with structure. ²⁵ Although exact localization of RDH12 is required, our finding of relative preservation of foveal cone function and structure in the associated human diseases does not support the notion that RDH12 has a key role in the cone visual cycle. 

What could be the basis of the dysplastic retinal response (DRR) that we observed in RDH12 mutations? Other human and murine observations in recent years provide clues that lead to a hypothesis about the pathobiology. We noted similarly coarse and delaminated retina in the early-onset recessive retinal disease caused by CRB1 mutations (documented from age 2). ³¹ Subsequently, two murine models of human CRB1-associated disease (rd8 and Crb1 ^{− / −}) were found to have patchy retinal disorganization that may be the equivalent of the human manifestation. ^{45 46} In another early-onset human disease of photoreceptor development caused by NR2E3 mutations, we provided evidence of a transformation from laminar organization to disorganization at localized retinal regions. ³³ Two animal models with Nr2e3 deficiency (rd7 and Nrl ^{− / −}) also had disturbed retinal organization with whorls and folds. ^{47 48 49} A similar sequence from normal lamination to dysplasia occurs in the murine model of the autosomal dominant polyglutamine expansion disease, spinocerebellar ataxia type 7. ^{50 51}  

A parsimonious explanation is that the DRR is a pathway triggered by some forms of photoreceptor damage and is independent of the primary cause. For example, light exacerbates and genetic background ameliorates the retinal dysplasia in Crb1 ^{− / −} mutants. ^{45 46} The DRR pathway probably triggers Müller glial cell activation, hypertrophy, and possibly proliferation, leading to disorganization of the laminar organization. A recent analysis of the genome response to retinal damage indicates that there is a photoreceptor-to-Müller glial signaling pathway involving photoreceptor-derived endothelin-2, EDN2. ⁵² Several murine genetic and nongenetic retinal diseases, including the rd7 (Nr2e3-deficient) mouse, showed induction of EDN2 as a response to photoreceptor damage. Such a pathway, presumably also present in human disease, may lead from primary or secondary photoreceptor damage to Müller cell activation and to a DRR. Human DRR may involve discernible stages by in vivo microscopy. Late DRR may be associated with retinal thinning with extensive gliosis and retinal remodeling, as commonly observed in human retinal degenerations ⁵³ and in animal models. ^{54 55} The present study and our recent work ^{31 33 56} point to early DRR associated with disorganized and thick retina that may be detectable when the local rate of Müller cell hypertrophy or proliferation exceeds the rate of apoptotic photoreceptor cell loss. The role of genetic ⁴⁵ and environmental factors ⁴⁶ in this complex response to photoreceptor disease requires further study. 

Anticipation of DRR and greater understanding of the pathway could lead to specific therapy to prevent this process. Without such prevention, any treatment attempts aimed at a monogenic cause would not be worth attempting. The natural history of onset of DRR could make the difference between early intervention or timed intervention after monitoring. Our data in RDH12-mutant retinas show a disease phenotype distinctly different from the prototypical visual cycle phenotype in RPE65-mutant retinas. Thoughts of therapy in RDH12 disease with DRR should involve nongenetic or nonspecific therapies to preserve central retinal function or, if this is lost, intervention with retinal prostheses. ⁵⁷  

 Supported by the National Institutes of Health, Foundation Fighting Blindness, the Macula Vision Research Foundation, the F. M. Kirby Foundation, the Macular Disease Foundation, Ruth and Milton Steinbach Fund, Alcon Research Institute, and Mackall Trust.

 Submitted for publication June 2, 2006; revised July 21, 2006; accepted November 14, 2006.

 Disclosure: S.G. Jacobson, None; A.V. Cideciyan, None; T.S. Aleman, None; A. Sumaroka, None; S. B. Schwartz, None; E.A.M. Windsor, None; A.J. Roman, None; E. Heon, None; E.M. Stone, None; D.A. Thompson, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Characteristics of Patients with LCA

Gene and Patient	Age at Visit (y)	Mutations	Reference*
RDH12
P1, †	8	Y194X/A206D	16
P2, †	11	Y194X/A206D	16
P3	13	A47T/L99I	16
P4	21	T55M/R295X	16
RPE65
P5	17	97del20bp/97del20bp	25 29 30
P6	19	G40S/R91Q	This study
P7	19	R91W/R91W	This study
P8	23	R91W/1059insG	This study
CRB1
P9	12	C948Y	31
P10	14	C948Y/C1218F	31
P11	24	R764C/V13361	31
P12	28	749del3bp/749del3bp	31

*  Previous reports of genotype and/or phenotype by the authors.

†  From the same pedigree.

Figure 1.

 

Retinal thickness topography of RDH12-mutant retina. (A– D) Topographical maps of retinal thickness mapped to a pseudocolor scale in a normal subject (A) and patients with RDH12 mutations (B–D). Difference maps (insets) show regions of retina that are abnormally thickened (pink), abnormally thinned (blue), or within normal limits (white; defined as ±2 SD). Traces of major blood vessels and location of optic nerve head are overlaid on each map depicted as right eyes; fovea and optic nerve are shown in (A). (E) Retinal thickness along the horizontal and vertical meridians in the patients. Gray region represents normal limits (mean ±2 SD) of retinal thickness. F, fovea. N, nasal, T, temporal, S, superior, I, inferior.

Figure 2.

 

Retinal laminar architecture in RDH12 mutations compared with other molecular subtypes of Leber congenital amaurosis. (A– F) Cross-sectional scans along the vertical meridian (inset, A) in a normal subject (A) and in patients with mutations in RDH12 (B–D), RPE65 (E), and CRB1 (F). Lamina representing INL and ONL are labeled in the normal subject (A); a retinal region outside the fovea with retained ONL is also shown in the RPE65 patient (E, arrow). (G) Normalized reflectivity profiles from a retinal locus 1.3 mm in the temporal retina (inset) are shown for a normal subject and for RDH12-mutant retinas compared with RPE65- and CRB1-mutant retinas at the same locus. Axial length of RDH12 and CRB1 reflectivity profiles are normalized to normal retinal thickness. Signal features representing INL and ONL (dark-gray regions) are marked on normal and RPE65 profiles.

Figure 3.

 

Foveal structure as a function of foveal sensitivity in patients with RDH12 or RPE65 mutations. (A, B) Infrared reflectance images from representative patients with RDH12 or RPE65 mutations. Overlaid are foveal dark-adapted absolute sensitivity to a red (650 nm) stimulus (horizontal bar; lower limit of normal range is 29 dB). Also overlaid are averaged OCT scans along the horizontal meridian passing through the fovea (white rectangular region corresponds to the size of the stimulus used for measuring sensitivity). (C) Foveal outer nuclear layer (ONL) thickness as a function of foveal sensitivity redrawn from a previous report with the addition of new data from patients with RDH12 or RPE65 mutations, as well as patients (Pts) screened negative for RDH12 or RPE65 mutations. Normal variability is described by an ellipse encircling the 95% confidence interval of a bivariate Gaussian distribution. Dotted lines: the idealized model of the relationship between structure and function in pure photoreceptor degenerations and the region of uncertainty that results by translating the normal variability along the idealized model. Redrawn from Jacobson SG, Aleman TS, Cideciyan AV, et al. Identifying photoreceptors in blind eyes caused by RPE65 mutations: prerequisite for human gene therapy success. Proc Natl Acad Sci USA. 2005;102:6177–6182.

The authors thank Elaine Smilko, Andy Cheung, Michelle Doobrajh, Marisa Roman, Alexandra Windsor, and Malgorzata Swider for critical help.