A Common Outer Retinal Change in Retinal Degeneration by Optical Coherence Tomography Can Be Used to Assess Outcomes of Gene Therapy

Abstract

Identifying early disease hallmarks in animal models with slow disease progression may expedite disease detection and assessment of treatment outcomes. Using optical coherence tomography, a widely applied noninvasive method for monitoring retinal structure changes, we analyzed retinal optical sections from six mouse lines with retinal degeneration caused by mutations in different disease-causing genes. While images from wild-type mice revealed four well-separated hyper-reflective bands in the outer retina (designated as outer retina reflective bands, ORRBs) at all ages, the second band (ORRB2) and the third band (ORRB3) were merged in retinas of five mutant mouse lines at early ages, suggesting the pathological nature of this alteration. This ORRB change appeared to be degenerating photoreceptor related, and occurred before obvious morphological changes that can be identified on both hematoxylin and eosin-stained sections and electron microscopic sections. Importantly, the merging of ORRB2 and ORRB3 was reversed by treatment with adeno-associated viral vector-mediated gene replacement therapies, and this restoration occurred much earlier than measurable functional or structural improvement. Our data suggest that the ORRB change could be a common hallmark of early retinal degeneration and its restoration could be used for rapid and noninvasive assessment of therapeutic effects following gene therapy or other treatment interventions.

Introduction

Inherited retinal dystrophies (IRDs) are caused by gene mutations that compromise the function and viability of retinal cells, leading to vision loss or blindness. They affect millions of people worldwide and no effective treatment is available.¹ However, gene therapy has emerged as a promising treatment option for some forms of IRDs. In the past two decades, the effectiveness of gene therapy has been proven in a number of animal models simulating human retinal disorders.^2,3 More than thirty gene therapy protocols have been or are being tested in patients with IRDs, including retinitis pigmentosa (RP), Leber congenital amaurosis (LCA), X-linked retinoschisis, choroideremia, and Stargardt disease (www.clinicaltrials.gov). Importantly, an adeno-associated viral (AAV) vector-mediated RPE65 gene transfer aimed at treating type 2 LCA has been approved by the U.S. Food and Drug Administration as the first gene therapy product for a retinal disease.⁴ In the coming years, more IRDs will foreseeably become treatable by gene therapy.

Preclinical efficacy studies of gene therapy for an IRD are usually conducted in animal models of the human disease, and therapeutic effects in treated eyes are evaluated at the time point when disease phenotypes have become observable in untreated control eyes. If disease progression is slow, a long waiting period following vector administration is necessary. Such a situation has been exemplified by our previous study involving Rpgr knockout mice, in which retinal degeneration only became obvious after 1 year of age.⁵ Identifying early disease signs emerging before prominent phenotype change may help shorten the waiting period, which is crucial for preclinical development of a treatment.

Optical coherence tomography (OCT) provides retinal cross-sectional images of humans and live animals with resolutions close to those of histological sections.⁶ OCT has become the gold standard for retinal imaging in eye clinics for a large number of retinal pathologies, including age-related macular degeneration⁷ and diabetic retinopathy.⁸ In addition, this technology has been applied extensively to small mammals⁹ such as mice and rats or larger animals^10,11 such as pigs, dogs, and primates. The lamina pattern of OCT in normal subjects is well described. For both human and mouse retina, there are four distinguishable hyper-reflective bands of the outer retina,¹² namely outer retina reflective bands (ORRBs) 1–4,¹³ which correspond to the outer limiting membrane, ellipsoid zone (EZ), interdigitation zone (IZ), and retinal pigment epithelium (RPE)/Bruch's membrane complex, respectively.^12,14 In this study, we investigated whether subtle pathological changes in the outer retina at the early phase of retinal degeneration could be captured by OCT imaging. To that end, we conducted OCT on several retinal degenerative mouse models with mutations in distinct disease-causing genes. An early optical band change, the merged ORRB2 and ORRB3, was commonly observed in five mouse lines with mutations that directly cause progressive photoreceptor degeneration. This alteration was reversed in mice receiving gene replacement therapy. We believe that this OCT change is a common hallmark of retinal degeneration, which could be utilized for early disease detection, monitoring of disease progression, and assessment of therapeutic responses to treatment interventions.

Materials and Methods

Mouse lines and husbandry

The following mouse lines were used in this study: C57BL/6J (Jackson Laboratories, Bar Harbor, ME), Rpgr^−/Y,¹⁵ Ttll5^−/−,¹⁶ Rpgr^−/Y;Cep290^rd16/+, Reep6^−/−,¹⁷ Rp2^−/Y,¹⁸ and Tg RHO P347S transgenic mice.¹⁹ All mice were maintained in National Institutes of Health animal care facilities in controlled ambient illumination on a 12-h light/12-h dark cycle. Studies conform to the ARVO statement for the Use of Animals in Ophthalmic and Vision Research. Animal protocols were approved by the National Eye Institute Animal Care and Use Committee.

OCT imaging

OCT images were collected using the Bioptigen spectral domain ophthalmic imaging system (Envisu UHR2200; Bioptigen, Durham, NC) equipped with an animal imaging mount and a rodent alignment stage. Before imaging, pupils of mice were first dilated with 1% tropicamide and 2.5% phenylephrine. Then, mice were anesthetized by an intraperitoneal injection of ketamine (80 mg/kg) and xylazine (8 mg/kg) or by inhalation of isoflurane/air mixture (at concentrations of 4% in the induction phase and at 1.5% during imaging procedures). The anesthetized mice were placed into the animal cassette of the rodent alignment stage and their positions were adjusted so that the optic nerve heads (ONHs) of their tested eyes were positioned at the center of the viewing pane. For image acquisition, rectangular scans (1.4 × 1.4 mm at 1000 A-scan × 100 B-scan × 5) and radial scans (1.4 mm radial diameter at 1000 A-scan × 4 B-scan × 40) were performed. The radial scanned images were carefully reviewed before the analysis to ensure that the images were flat to exclude optical artifacts caused by altered band reflectivity as reported.²⁰

OCT image analysis

Repeated B-scan images (5 for volume scan or 40 for radial scan) were averaged to reduce speckle noise. The averaged images were used for thickness measurement and intensity profile plotting. The outer nuclear layer (ONL) or ORRB thickness was measured at two points (600 μm from the ONH center in both nasal and temporal directions) using InVivoVue 2.4. Reader software (Bioptigen), and the average value of the two measurements for each eye was used. The brightness-based intensity profile was calculated using ImageJ software (National Institutes of Health, Bethesda, MD) in the outer retinal area (25 × 50 μm) containing all ORRBs of the magnified images.

AAV vectors and subretinal injections

An AAV vector plasmid carrying the human RPGR-ORF15 expression cassette driven by a human rhodopsin kinase promoter was made and described previously.⁵ To make a self-complementary (sc) AAV vector carrying the human REEP6 cassette, the human REEP6.1 coding sequence (NCBI Reference Sequence: NM_001329556.3) was placed downstream of a human rhodopsin promoter and a human beta-globin intron and upstream of a human β-globin polyadenylation site. The full sequences of the vector plasmids are available upon request. Production, purification, and quantitation of AAV vectors were described previously.²¹ The human RPGR-ORF15 vector was packaged into AAV9, and the human REEP6.1 vector was packaged into AAV8 as an sc form. The vectors were injected subretinally into mouse eyes, as previously described.⁵ The AAV9 human RPGR vector was injected at a dose of 1 × 10⁹ vector genomes (vg)/eye, while the AAV8 human REEP6.1 vector was injected at 3 × 10⁸ vg/eye.

Histology

Mice were euthanized by carbon dioxide asphyxia, followed by cervical dislocation. To prepare semithin plastic sections, eyeballs were enucleated and fixed in a mixture of 1% glutaraldehyde and 4% paraformaldehyde for 72 h. Tissues were embedded in methyl methacrylate, sectioned at 1 μm, and then stained with hematoxylin and eosin (H&E). H&E-stained slides were observed under an Axio Imager Z1 microscope (Zeiss, Jena, Germany).

Electron microscopy

Enucleated mouse eyes were fixed in 2.5% glutaraldehyde and 0.5% osmium tetroxide in phosphate-buffered saline, dehydrated, and embedded in Spurr's epoxy resin. Ultrathin sections (100 nm) of mouse eyes were prepared, double-stained with uranyl acetate and lead citrate, and viewed with a JEOL JEM-1010 transmission electron microscope (Tokyo, Japan) equipped with a digital imaging camera.

Statistical analyses

Quantitative data are presented as mean ± standard error of the mean. A two-tailed paired t-test was used to determine statistically significant changes between vector-injected and vehicle-injected eyes. In all statistical analyses, significance was defined as p < 0.05.

Results

Identification of a common OCT change of the outer retina in retinal degenerative mouse models

We conducted OCT imaging on six mouse lines, all of which have mutations in disease-causing genes for retinal degeneration, at ages ranging from postnatal day 22 (P22) to 6 months. These mouse lines included Rpgr^−/Y (38 mice), Rpgr^−/Y;Cep290^rd16/+ (37 mice), Ttll5^−/− (5 mice), Reep6^−/− (6 mice), Rp2^−/Y (9 mice), and one human mutant rhodopsin transgenic line Tg RHO P347S (4 mice). Retinal degeneration with different progression rates in these mouse lines was well documented.²² Wild-type (WT) C57BL/6J mice were used as controls.

At the time of imaging, most of these mouse lines displayed relatively intact retinal layers, although ONL thinning was noticeable in some. Four hyper-reflective bands (ORRB1–4) were clearly distinguishable in the outer retinas of WT mice. However, a common band change was observed in all the mutant mouse lines except for Rp2^−/Y mice (Fig. 1A, B). The boundary between ORRB2 and ORRB3 was almost invisible, with the two bands merged in Rpgr^−/Y, Rpgr^−/Y;Cep290^rd16/+, Ttll5^−/−, Reep6^−/−, and Tg RHO P347S mice.

Figure 1.

Representative OCT retinal images of the C57BL/6J and six different mouse lines with retinal degeneration. (A) OCT images collected from C57BL/6J (WT, 6 months old, 3 mice), Rpgr^−/Y (6 months old, 38 mice), Ttll5^−/− (3 months old, 5 mice), Rpgr^−/Y;Cep290^rd16/+ (3 months old, 37 mice), Reep6^−/− (2 months old, 6 mice), human-mutant (P347S) rhodopsin transgenic (P22, 4 mice), and Rp2^−/Y (4 months old, 9 mice) mice. Red arrowheads point to the merged bands of ORRBs 2 and 3. Scale bar, 100 μm. (B) Magnified images selected from (A). The four ORRBs are labeled 1, 2, 3, and 4 on the corresponding layers in the image of the WT mouse. Red arrowheads point to the merged bands of ORRB2 and ORRB3. Scale bar, 50 μm. OCT, optical coherence tomography; ORRBs, outer retina reflective bands; P22, postnatal day 22; WT, wild-type.

Both RPGR and TTLL5 proteins are localized at the connecting cilia (CCs) of photoreceptors. TTLL5 glutamylates RPGR-ORF15 in its Glu-Gly-rich repetitive region, and this post-translation modification of RPGR is critical for its function in photoreceptors. Mutations in these two genes lead to similar phenotypes with late onset and slow progression of photoreceptor degeneration.¹⁶ Therefore, it was not surprising that the OCT sections from Rpgr^−/Y and Ttll5^−/− mice exhibited similar patterns. ORRB2 and ORRB3 were apparently merged, but other layers were very similar to those of WT mice (Fig. 1A, B).

Rpgr^−/Y;Cep290^rd16/+ mice, a hybrid Rpgr^−/Y mouse line carrying a heterozygous hypomorphic allele of Cep290 (Cep290^rd16/+), exhibited an earlier onset and faster progression of retinal degeneration than the Rpgr^−/Y mice.²³ OCT sections from Rpgr^−/Y;Cep290^rd16/+ mice revealed more advanced retinal degeneration than in Rpgr^−/Y or Ttll5^−/− mice. Besides the merged ORRB2 and ORRB3, moderate thinning of the ONL was also observed (Fig. 1B).

REEP6 is a protein localized at the rod inner segments (ISs) and outer plexiform layer of the retina and its deficiency leads to retinal degeneration through disruption of endoplasmic reticulum (ER) homeostasis and protein trafficking.^17,24 Similar to the patterns of the Rpgr^−/Y;Cep290^rd16/+ mice, Reep6^−/− mice exhibited the merged band of ORRB2 and ORRB3 and moderate thinning of the ONL (Fig. 1B).

Rhodopsin is localized at the outer segments (OSs) of rod photoreceptors. Tg RHO P347S mice displayed the fastest retinal degeneration among the mouse lines we examined. At P22, not only was the boundary between ORRB2 and ORRB3 missing, but the other two ORRB boundaries (ORRB1/ORRB2 and ORRB3/ORRB4) were also destroyed, and significant thinning of the photoreceptor layer had occurred (Fig. 1B).

RP2 is localized at the ISs of photoreceptors, serving as a GTPase activation protein of Arf-like protein 3 to facilitate the trafficking of prenylated proteins such as GRK1 from the ER to the OS. In contrast to the other five mutant lines, but similar to WT mice, ORRB2 and ORRB3 were clearly separated (Fig. 1A, B), and merging of the two bands was observed neither in 4-month-old nor in 18-month-old Rp2^−/Y mice (Fig. 1B; Supplementary Fig. S1). Previous studies by our group¹⁸ and others²⁵ have shown that RP2 deficiency reduces both rod and cone function at an early age in mice. We also observed that while cone photoreceptors progressively degenerate, the rod function mainly reduces within 4 months after birth and then remains stable. The progressive degeneration of cone photoreceptors, which account for a small proportion (<5% of total photoreceptors) of the mouse retina, may not be sufficient for detection on OCT images. Therefore, merging of ORRB2 and ORRB3 observed in other mutant mouse lines may represent an abnormality in the progressively degenerating photoreceptors, including both rods and cones. As retinal degeneration in the five mutant mouse lines is caused by distinct pathological mechanisms and the afflicted proteins are localized at different subcellular locations, merging of ORRB2 and ORRB3 could be a common sign of photoreceptor abnormality regardless of disease mechanisms.

Monitoring the outer retina band change in Rpgr-deficient mice

As shown in Fig. 1, merged ORRB2 and ORRB3 was observed in two slow retinal degenerative mouse models, Rpgr^−/Y and Ttll5^−/−. We were curious about whether this band change is an early sign that occurs before significant loss of photoreceptors. Thus, we collected daily optical sectional images from Rpgr^−/Y mice and age-matched WT mice from the youngest age available for OCT imaging. We observed that ORRB2 and ORRB3 were still separated in Rpgr^−/Y retina on P25 and that the retinal structure was indistinguishable from that of the age-matched WT mouse (Fig. 2A, B). From P26 to P29, ORRB2 and ORRB3 in Rpgr^−/Y retina gradually merged together, without obvious changes in ORRB1, ORRB4, and ONL (Fig. 2A, B). The intensity profile of the outer retina, when plotted according to brightness, confirmed the dynamic changes observed for ORRB2 and ORRB3 between P26 and P29 in Rpgr^−/Y retina (Fig. 2B, right lower panel). The peak (between the dotted lines) corresponding to the dark layer between ORRB2 and ORRB3 shifted daily in the direction of brightness between P26 and P29. It nearly disappeared on P29, indicating that ORRB2 and ORRB3 became merged. However, no significant change was observed in the thickness of ORRB 1–4 (P26: 53.25 ± 1.26 vs. P29: 54 ± 0.82, p = 0.44) or ORRB 2–4 (P26: 38 ± 1.41 vs. P29: 39 ± 2, p = 0.35), while significant band pattern changes occurred. This indicates that the merging of ORRB2 and ORRB3 may be due to a change in reflectivity, rather than structural alteration. Although timing of this change varied slightly between the mouse individuals, the band change occurred in all tested 38 Rpgr-deficient mice without exception. This alteration was mutant related as it was not detected in any WT mice over the same period (Fig. 2A, B). It is noteworthy to mention that at P25, merging of ORRB2 and ORRB3 was already observed in the retina of the Rpgr^−/Y;Cep290^rd16/+ mouse line with a relatively faster degeneration (Supplementary Fig. S2).

Figure 2.

Monitoring of dynamic band changes in the OCT retinal images of young Rpgr^−/Y mice. (A) OCT images collected from C57BL/6J and Rpgr^−/Y for five consecutive days from P25 to P29. Scale bar, 100 μm. (B) Magnified images (left panels) selected from (A) and intensity profiles (right panels) for outer retinal regions, marked by red bars. Arrowhead indicates the boundaries between ORRB2 and ORRB3. Dark regions divided by two dotted lines correspond to the boundaries between ORRB2 and ORRB3. Scale bar, 50 μm. a.u., arbitrary units; EZ, ellipsoid zone; ONL, outer nuclear layer; RPE, retinal pigment epithelium.

We further collected OCT images from Rpgr^−/Y mice from P30 to 15 months. Merged ORRB2 and ORRB3 was observed at all time points of imaging during the entire period, suggesting that once it is formed, it cannot be reversed spontaneously (Supplementary Fig. S3). ONL thinning progressed at a very slow rate, which became apparent in 15-month-old mice, consistent with previous reports.^5,26 At 15 months, boundary cracks between ORRB1 and ORRB2 were also observed. In contrast, all retinal layers, including four ORRBs, were well preserved even in 24-month-old WT mice (Supplementary Fig. S3).

Analyses of retinal histological sections in Rpgr-deficient mice

Our collective results showed that ORRB2 and ORRB3 were completely merged at P30 in most tested Rpgr^−/Y mice. Therefore, we compared retinal histological sections from P30 Rpgr^−/Y mice with those from age-matched WT mice to identify structural changes corresponding to the optical change. H&E-stained histological sections were prepared from the same eyes of the mice right after the OCT scan (Fig. 3A). Although the OCT change in Rpgr^−/Y mice was clearly distinguishable from that of WT mice, no corresponding change was found in the histological sections. Overall, the thickness and structural integrity of the retina in histological sections of Rpgr^−/Y mice were indistinguishable from those of WT mice (Fig. 3A).

Figure 3.

Comparison of retinal optical sections and retinal histological sections of C57BL/6J and Rpgr^−/Y mice. (A) OCT optical sections (left) and H&E-stained histological sections (right) taken from the same eyes of C57BL/6J (WT) and Rpgr^−/Y mice on P30. The four ORRBs are labeled 1, 2, 3, and 4 on the corresponding layers in the image of the WT mouse. Scale bar, 50 μm. (B) Representative electron micrographs of inner and outer segments in the longitudinal retinal sections from WT and Rpgr^−/Y mice on P30. Scale bar, 1 μm. CC, connecting cilium; H&E, hematoxylin and eosin; INL, inner nuclear layer; IPL, inner plexiform layer; IS, inner segment; Mt, mitochondria; NFL, nerve fiber layer; OLM, outer limiting membrane; OPL, outer plexiform layer; OS, outer segment.

To identify more subtle structural changes, we compared the ultrastructure of retinal sections from the same WT and Rpgr^−/Y mice using electron microscopy (EM). It was suggested that ORRB2 corresponds to mitochondria in the EZ of ISs, while ORRB3 corresponds to the IZ containing OS and microvillus processes of the RPE.²⁷ We carefully examined the ultrastructure in the vicinity of ISs and OSs and did not find any obvious differences between WT and Rpgr^−/Y mice. The OSs, ISs, and CCs were normally organized, and the OS membranous discs were tightly packed and flattened in both mouse lines (Fig. 3B).

Based on electroretinography (ERG) measurements, no functional impairments have been well documented in Rpgr-deficient mice at early ages, such as at P30, in studies performed by us and other independent groups.^15,28 Therefore, the merging of ORRB2 and ORRB3 may occur before retinal functional impairment and noticeable histological changes in Rpgr^−/Y mice.

Restoration of optical bands following gene replacement therapy

Our end goal is to determine whether this OCT change can be utilized as a parameter to evaluate the outcome of treatment. To test this, gene replacement therapy was given to Reep6^−/− and Rpgr^−/Y;Cep290^rd16/+ mice. Reep6^−/− mice were treated with AAV8-human-REEP6, while Rpgr^−/Y;Cep290^rd16/+ mice were treated with AAV9-human-RPGR. Each mouse received a subretinal injection of the vector in one eye and vehicle injection in the other eye.

Before treatment, OCT was conducted to confirm that ORRB2 and ORRB3 were completely merged and that there was no difference between the two eyes in each 2-month-old Reep6^−/− mouse (Fig. 4A, B). In vehicle-injected control eyes, the merged ORRB2 and ORRB3 were not separated during the entire monitoring period. In addition, significant ONL thinning was observed at 3 months postinjection (PI) (Fig. 4B, C). In contrast, ORRB2 and ORRB3 were clearly separated in vector-treated eyes at 1.5 months PI. These two bands were almost completely separated at 3 months PI, similar to those in WT retina (Fig. 4A, B). In addition, the intensity profile of the outer retina exhibited a peak corresponding to the boundary between the two bands, in the vector-treated eye at 1.5 months PI, and the peak shifted further toward the dark direction at 3 months PI (Fig. 4B, right lower panel). The ONL thickness was reduced by 12.4% in control eyes between 1.5 months and 3 months PI, while remaining unchanged in vector-treated eyes over the same period (Fig. 4C). At 3 months PI, the ONL thickness was 23.6% greater in vector-treated eyes than in control eyes (Fig. 4D).

Figure 4.

Restoration of ORRB pattern following gene therapy in Reep6^−/− mice. (A) OCT images collected from vehicle-injected and vector-injected eyes of Reep6^−/− mice at Pre-Inj and two time points (1.5 and 3 months) of PI. Scale bar, 100 μm. (B) Magnified images (left panels) selected from (A) and intensity profiles (right panels) for outer retinal regions, marked by red bars. Arrowhead indicates the restored boundaries between ORRB2 and ORRB3. Dark regions divided by two dotted lines correspond to the boundaries between ORRB2 and ORRB3. Scale bar, 50 μm. (C) Quantitative analysis of ONL thickness in vehicle-injected and vector-injected eyes of Reep6^−/− mice (n = 3) at Pre-Inj and two time points (1.5 and 3 months) of PI. *p < 0.05. Error bars, SEM. PI, postinjection; Pre-Inj, preinjection; SEM, standard error of the mean.

Similar results were obtained when using the Rpgr^−/Y;Cep290^rd16/+ mouse line. Without vector treatment, ONL thickness was reduced by 17% from 3 to 6 months PI, in addition to the merging of ORRB2 and ORRB3. The boundary between ORRB1 and ORRB2 was also partially destroyed (Fig. 5A–C). In contrast, ONL thickness was well preserved in vector-treated eyes. At 6 months PI, a 22% greater thickness in the ONL was observed in vector-treated eyes than in control eyes (Fig. 5D). In addition, the ORRB2 and ORRB3 band pattern was partially restored in vector-treated eyes (Fig. 5A, B). Moreover, the boundary between ORRB1 and ORRB2 was well preserved in vector-treated eyes (Fig. 5A, B). These results collectively indicate that merging of ORRB2 and ORRB3 is reversible by gene replacement therapy, and the band pattern correlates well with photoreceptor healthiness.

Figure 5.

Restoration of ORRB pattern following gene therapy in Rpgr^−/Y;Cep290^rd16/+ mice. (A) OCT images collected from vehicle-injected and vector-injected eyes at 3 and 6 months PI. Scale bar, 100 μm. (B) Magnified images selected from (A). Black arrowhead and open arrowhead indicate the restored boundary between ORRB2 and ORRB3 and the boundary between ORRB1 and ORRB2, respectively. Scale bar, 50 μm. (C) Quantitative analysis of ONL thickness in vehicle-injected and vector-injected eyes (n = 4) at 3 and 6 months PI. ***p < 0.001, **p < 0.01. Error bars, SEM.

Next, we tested whether the ORRB2 and ORRB3 band pattern could be used to evaluate treatment effects in mice with very slow retinal degeneration. To that end, Rpgr^−/Y mice were treated with AAV9-human-RPGR vector at 1 month after birth. As shown earlier (Fig. 2), ORRB2 and ORRB3 are completely merged at this age. Following vector treatment, separation of ORRB2 and ORRB3 was first seen at 3 months PI and became even clearer at 6 months PI (Fig. 6A, B). At 3 months PI, there was no difference between vector-treated and control eyes in ONL thickness. Although the difference was statistically significant at 6 months PI, it was only ∼5%, mainly due to the slow rate of photoreceptor death (Fig. 6C, D). Therefore, restoration of this band pattern occurred earlier than distinguishable preservation of ONL thickness. Restoration of the band pattern was not observed in vehicle-injected eyes at any time point (Fig. 6A, B). In our previous efficacy study, neither functional nor structural rescue was detected in Rpgr^−/Y mice before 12 months PI by ERG or histological analyses.⁵ Therefore, restoration of the ORRB2 and ORRB3 band pattern at 3 to 6 months PI may represent a more rapid indicator of therapeutic effects.

Figure 6.

Restoration of ORRB pattern following gene therapy in Rpgr^−/Y mice. (A) OCT images collected from vehicle-injected and vector-injected eyes at 3 and 6 months PI. Scale bar, 100 μm. (B) Magnified images selected from (A). Arrowhead indicates the reconstructed boundaries between ORRB2 and ORRB3. Scale bar, 50 μm. (C) Quantitative analysis of ONL thickness in vehicle-injected and vector-injected eyes of Rpgr^−/Y mice (n = 4) at 3 and 6 months PI. **p < 0.01. Error bars, SEM.

Discussion

Technological innovations in OCT allow us to distinguish detailed retinal layer structures and certain cellular compartments by micrometer-scale resolution imaging. In the present study, we identified the merging of ORRB2 and ORRB3 as a common and early OCT change in the outer retina of five mouse lines with distinct gene mutations leading to retinal degeneration. In retrospect, we actually observed merged ORRB2 and ORRB3 in 1 month and older retinoschisin-deficient mice.¹³ More recently, this phenomenon was also revealed in P22 and older Rpe65^−/− mice.²⁹ All these findings strongly suggest that merging of ORRB2 and ORRB3 could be a common hallmark of early retinal degeneration, regardless of the disease-causing genes and underlying disease mechanisms.

In our current study, the Rp2^−/Y mouse line is the only mutant line in which merging of ORRB2 and ORRB3 was not observed. Although reduction of rod and cone function happens early in this mouse line, rod function becomes stable after 4 months, while cone degeneration continues.¹⁸ In addition, we did not find obvious ONL thinning either in young or old Rp2^−/Y mice, indicating that rod degeneration is very mild. As rods account for >95% of total retinal photoreceptors in mice,³⁰ ISs and OSs of rods could be the major sources of ORRB2 and ORRB3. Therefore, formation of merged ORRB2 and ORRB3 may require progressive degeneration of rods, and cone degeneration alone may not be able to lead to the change. Further work is needed to confirm this hypothesis.

The structural basis leading to this optical change before prominent retinal degeneration remains unknown. In the normal retina, ORRB2 and ORRB3 are supposed to correspond to the mitochondrion-containing ellipsoid portion of the IS and distal portion of the OS, respectively.^12,14 Therefore, abnormality in either ISs or OSs is likely to contribute to merging of ORRB2 and ORRB3, although other factors cannot be ruled out. In Rs1-KO retina, it was suggested that lack of the regular array of densely packed, long thin mitochondria adjacent to the internal plasma membrane may account for the merged ORRB2 and ORRB3.¹³ In Rpe65^−/− retina, disarrangement and vacuolation of the OS discs and the variable sizes of OSs were supposed to be responsible for the diffused ORRB2 and ORRB3 in younger and older mice, respectively.²⁹ To explore the structural basis of the merged ORRB2 and ORRB3 in Rpgr^−/Y mice, we analyzed histological sections from a P30 retina when merging had just completed and no OCT changes in other retinal layers had occurred. However, we did not see any abnormalities in either ISs or OSs of Rpgr^−/Y retina at P30 (Fig. 4), either by HE staining or by EM analyses. In theory, any dynamic factor leading to lower transparency or higher reflectivity at the vicinity of the ellipsoid portion of the IS and distal portion of the OS could be the cause of this optical change, without resulting in structural changes that are detectable using traditional histological methods. A previous study revealed merged ORRB2 and ORRB3 in WT retina after light challenge, which was restored to normal when the light challenge was removed.⁹ Although we cannot rule out the possibility that merging of ORRB2 and ORRB3 may have different structural bases caused by different disease mechanisms, it is also likely that it is a common reaction in response to all sorts of stresses, either endogenous or exogenous. It is noteworthy that in previously reported OCT images, the boundary between ORRB2 and ORRB3 disappeared in conditional knockout mice with higher levels of reactive oxygen species by RPE-specific deletion of superoxide dismutase 2,³¹ but the boundary was restored in the mice after dietary antioxidant supplementation.³² Further study of the oxidative stress-induced response in photoreceptors may provide a clue to aid in understanding the mechanisms underlying the observed OCT changes.

It was well documented that progression of retinal degeneration in Rpgr^−/Y mice is very slow. Photoreceptor loss and ERG amplitude reduction become obvious only when mice are older than 6 months or even 1 year.^5,15 However, merging of ORRB2 and ORRB3 occurs between P25 and P30 (Fig. 2), much earlier than other measurable morphological and functional changes in the retina. In Ttll5^−/− mice, the same thing may hold true, although the exact timing of the merging was not examined. Therefore, merging of ORRB2 and ORRB3 appears to be an early sign of retinal degeneration in mice, which may be used to help early diagnoses in patients with retinal diseases.

Merging of ORRB2 and ORRB3 could be reversed by gene therapy. In our previous study, the normal pattern of ORRB2 and ORRB3 was restored in Rs1-KO retina after the AAV8-hRs1 vector was provided.¹³ In the present study, the merged ORRB2 and ORRB3 were separated in Reep6^−/− and Rpgr^−/Y;Cep290^rd16/+ mice after gene therapy, in concomitance with the preservation of ONL thickness. It is noteworthy that in Rpgr^−/Y mice, restoration of the normal pattern of ORRB2 and ORRB3 began within 3 months post-treatment. In our earlier study, it took at least 12 months after treatment to observe improved ERG amplitudes in the treated eyes and 18 months to conclusively identify therapeutic effects.⁵ Using restoration of the normal reflective band pattern as a treatment end point, the length of the efficacy study may be shortened significantly. Therefore, in addition to the intrinsic advantages of OCT, including its noninvasive nature and the ability to conduct studies in live animals with repeated longitudinal monitoring, restoration of the normal ORRB pattern can be used as an early indicator of treatment effects in mouse models with slow retinal degeneration. In the present study, gene therapy was conducted at an early phase of retinal degeneration. Studies by us and others have shown that photoreceptor death can be halted or delayed even when gene therapy is given at the mid to late stage of retinal degeneration.^5,33 It remains to be seen if the normal pattern of ORRB can be restored after delayed treatment.

Taken together, our data strongly suggest that the change that occurs between ORRB2 and ORRB3 in retinal OCT images could be a useful indicator to predict retinal degeneration and to monitor therapeutic effects following treatment interventions. Our OCT findings provide a far earlier detectable parameter than other functional and morphological assessments in retinal degenerative mouse models. If the same optical change could be identified in human patients with retinal degenerative diseases, we expect that the parameter would be broadly used for early diagnosis and assessment of therapeutic effects.

Author Disclosure

All authors declare that they have no competing interests.

Funding Information

This work was supported by the Intramural Research Program of National Eye Institute (1ZIAEY000443-12).

Supplementary Material

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3