|Home | About | Journals | Submit | Contact Us | Français|
Autosomal recessive retinitis pigmentosa (RP), a heterogeneous group of degenerations of the retina, can be due to mutations in the MFRP (membrane-type frizzled-related protein) gene. A patient with RP with MFRP mutations, one of which is novel and the first splice site mutation reported, was characterized by noninvasive retinal and visual studies. The phenotype, albeit complex, suggested that this retinal degeneration may be a candidate for gene-based therapy. Proof-of-concept studies were performed in the rd6 Mfrp mutant mouse model. The fast-acting tyrosine-capsid mutant AAV8 (Y733F) vector containing the small chicken β-actin promoter driving the wild-type mouse Mfrp gene was used. Subretinal vector delivery on postnatal day 14 prevented retinal degeneration. Treatment rescued rod and cone photoreceptors, as assessed by electroretinography and retinal histology at 2 months of age. This AAV-mediated gene delivery also resulted in robust MFRP expression predominantly in its normal location within the retinal pigment epithelium apical membrane and its microvilli. The clinical features of MFRP-RP and our preliminary data indicating a response to gene therapy in the rd6 mouse suggest that this form of RP is a potential target for gene-based therapy.
The retinal pigment epithelium (RPE) plays a critical role in vision, maintaining the structural integrity, function, and survival of photoreceptor cells (Sparrow et al., 2010). On its apical side, the RPE extends numerous microvilli around the light-sensitive photoreceptor outer segments and into the interphotoreceptor matrix. Microvilli considerably increase the surface area of the RPE apical membrane, and establish a critical interface for many RPE functions including phagocytosis of shed outer segments, visual chromophore transport and regeneration, production of trophic and antiangiogenic factors, directional transport of oxygen and nutrients from the choroid to sustain the high metabolic rate of photoreceptors, and removal of water and aqueous waste products from the subretinal space (Strauss, 2005; Bonilha et al., 2006). Mutations in genes expressed in RPE cells can result in retinal degeneration and loss of vision (Sparrow et al., 2010). Some of these include RPE65 and LRAT (Leber congenital amaurosis, LCA), MERTK (early-onset retinitis pigmentosa, RP), BEST1 (Best disease), TIMP3 (Sorsby fundus dystrophy), EFEMP1 (malattia leventinese), RDH5 (fundus albipunctatus), and RLBP1 (retinitis punctata albescens).
A relative newcomer to this group is autosomal recessively inherited RP caused by mutations in the human MFRP (membrane-type frizzled related protein) gene, located on chromosome 11q23 (Ayala-Ramirez et al., 2006; Crespí et al., 2008; Zenteno et al., 2009; Mukhopadhyay et al., 2010). The MFRP gene encodes a type II transmembrane protein of 584 amino acid residues, which consists of an N-terminal cytoplasmic region, a transmembrane domain, and an extracellular region with a C-terminal cysteine-rich domain similar to that observed in Wnt-binding frizzled proteins (Katoh, 2001; Kameya et al., 2002). Mfrp is expressed as one element of a dicistronic transcript (Kameya et al., 2002; Hayward et al., 2003; Mandal et al., 2006), which also encodes the complement C1q tumor necrosis factor-related protein-5 (C1QTNF5/CTRP5). CTRP5 also has a human disease association; specifically, a Ser163Arg mutation causes an autosomal dominant late-onset retina-wide degeneration, which can feature neovascular macular degeneration (Milam et al., 2000; Jacobson et al., 2001; Hayward et al., 2003). The functional relationship between the two proteins remains under investigation (Fogerty and Besharse, 2011).
With clinical trials of gene therapy ongoing for another autosomal recessive RPE disease leading to retinal degeneration, RPE65-LCA (reviewed in Cideciyan, 2010), we inquired whether MFRP-RP was a potential candidate for this form of treatment. A patient with MFRP-RP was examined in detail by noninvasive studies and the results were compared with those of patients with other molecularly proven MFRP-RP in the literature. Like RPE65-LCA, there is an animal model of the MFRP-RP disease for proof-of-concept studies. The rd6 mouse has a naturally occurring autosomal recessive retinal degeneration associated with a 4-bp deletion in a splice donor site in the Mfrp gene (Hawes et al., 2000; Kameya et al., 2002). This results in the skipping of exon 4 and deletion of 58 amino acids from the Mfrp protein (Kameya et al., 2002). The present study uses subretinal delivery of a self-complementary tyrosine-capsid mutant adeno-associated virus serotype 8 (AAV8) (Y733F) vector carrying the mouse Mfrp gene to determine whether photoreceptor degeneration can be prevented in rd6 mice, thereby paving the way for further proof-of-concept, dosing, and safety studies in gene therapy en route to a clinical trial in patients with MFRP-RP.
A patient with RP underwent a complete eye examination and detailed studies of retinal phenotype. Genomic DNA from peripheral blood lymphocytes was obtained from the patient and family members and molecular genetic analyses led to the identification of MFRP mutations (Ayala-Ramirez et al., 2006; Crespí et al., 2008; Zenteno et al., 2009). Patients with retinal degenerations, including other forms of RP (n=6; age, 7–78 years) and choroideremia (n=3; age, 13, 34, and 38 years), as well as normal subjects (n=34; age, 5–58), were included for comparison of phenotype. Informed consent was obtained for all subjects; procedures were in accordance with the Declaration of Helsinki and were approved by the institutional review board.
Full-field electroretinograms (ERGs) were performed with bipolar Burian-Allen contact lens electrodes and a standard protocol using an Espion system (Diagnosys, Lowell, MA) with methodology previously described (Jacobson et al., 1989; Aleman et al., 2002). Kinetic visual fields and dark- and light-adapted chromatic static threshold perimetry (200-msec duration, 650- and 500-nm stimuli, dark-adapted, and 600nm, light-adapted; 1.7° diameter target) were performed with a modified HFA-750i analyzer (Zeiss-Humphrey, Dublin, CA). Our methods for data collection and analyses have been published (Jacobson et al., 1986, 2010; Roman et al., 2005). Reflectance imaging and reduced-illuminance autofluorescence imaging (RAFI) were performed with near-infrared (NIR) and short-wavelength (SW) lights with a confocal scanning laser ophthalmoscope (SLO) (HRA2; Heidelberg Engineering, Heidelberg, Germany) as described (Cideciyan et al., 2007, 2011; Jacobson et al., 2011). Retinal cross-sectional imaging used a spectral-domain optical coherence tomography (SD-OCT) unit (RTVue-100; Optovue, Fremont, CA) with published recording and analysis techniques (Cideciyan et al., 2011; Jacobson et al., 2011). The three-dimensional SD-OCT raster scans were performed for topographic analysis. Postacquisition processing of OCT data was performed with custom programs (MATLAB 6.5; MathWorks, Natick, MA). Further methodological details are provided in the online supplement (supplementary data are available online at www.liebertonline.com/hum).
rd6 mice (originally provided by B. Chang, Jackson Laboratory, Bar Harbor, ME) were bred and maintained in the University of Florida Health Science Center Animal Care Services Facility (Gainesville, FL) under 12-hr-on/12-hr-off cyclic lighting. Wild-type C57BL/6 mice served as controls and were from the University of Florida or University of Pennsylvania (Philadelphia, PA) Animal Care Services Facility. All experiments were approved by University of Florida and University of Pennsylvania Institutional Animal Care and Use Committees and conducted in accordance with the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research.
A self-complementary AAV8 vector containing a Y733F point mutation in a highly conserved surface-exposed capsid tyrosine residue was used for packaging the wild-type murine Mfrp cDNA under the control of the ubiquitous, constitutive smCBA (small chicken β-actin) promoter. AAV8 (Y733F) vector was produced by the two-plasmid cotransfection method in HEK 293 cells and purified according to previously reported methods (Petrs-Silva et al., 2009). Viral titer was determined by real-time PCR. Subretinal injections were performed on postnatal day 14 (P14) under anesthesia as described (Pang et al., 2008). In brief, the nasal cornea was penetrated with a 30.5-gauge disposable needle; and a 33-gauge unbeveled, blunt-tip needle on a Hamilton syringe was introduced into the subretinal space. Each eye received 1μl of vector at a titer of 1×1012 genome copies/ml, leaving the left eye as an untreated control. Preliminary experiments were also performed with buffer-injected rd6 eyes as possible controls, and it was observed that there was rapid degeneration in these eyes, likely secondary to poor reattachment after retinal detachment. To avoid bias that would lead to an apparent treatment effect due to surgery-related damage to control rd6 eyes, we chose to use an untreated control eye for comparison with treatment. Eleven rd6 mice had subretinal injections of vector in the right eye. At 6 weeks postinjection, mice with no signs of injection-related trauma to the anterior segment were included in further studies.
Full-field ERGs were elicited in 2-month-old (P14-treated) rd6 mice (n=5) and wild-type C57BL/6 controls (n=20; mean age, 3.4 months; age range, 1 to 10 months), using methods previously reported (Roman et al., 2007; Pang et al., 2008; Caruso et al., 2010). A series of stimuli were generated with a UTAS SunBurst system (LKC Technologies, Gaithersburg, MD) or a custom-built ganzfeld stimulator with a computer-based system (EPIC-XL; LKC Technologies). Mice were dark-adapted (>12hr) and anesthetized by injection of a mixture of ketamine-HCl (72mg/kg) and xylazine (5mg/kg). Pupils were dilated with topical agents (phenylephrine hydrochloride, 2.5%; tropicamide, 1%) under dim red light. Dark-adapted ERGs were elicited with 0.02 and 2 scot-cd·sec·m–2 stimuli. In wild-type mice, the dimmer flash produces a b-wave driven by rod postreceptoral activity whereas the brighter flash produces an a-wave dominated by rod photoreceptor activity and a b-wave that is driven by both rod and cone postreceptoral neurons (Weymouth and Vingrys, 2008). In addition, cone-isolated ERGs were elicited with 25 phot-cd·sec·m–2 stimuli on a rod-suppressing background light after a preadaptation period. In wild-type mice, this flash produces a b-wave driven by cone postreceptoral neurons (Weymouth and Vingrys, 2008). Efficacy of the uniocular treatment in rd6 eyes was determined by evaluating interocular differences (IODs). IODs for all ERG parameters were expressed as the amplitude difference between the two eyes divided by the mean value. t tests were used to determine the statistical significance of differences between IODs of rd6 mice and wild-type animals.
For morphological analysis, treated rd6 mice (n=4) had both eyes fixed in 10% formalin solution, processed for paraffin embedding, sectioned at a thickness of 4μm, and stained with hematoxylin and eosin. The sections were visualized by light microscopy. Comparisons were made with adult C57BL/6 mice (n=4). The outer nuclear layer (ONL) thickness was measured at three locations from within the mid-superior retina of wild-type, treated, and untreated rd6 eyes (four animals per group). The differences between the ONL thickness of AAV-treated and the uninjected contralateral left eyes were analyzed by Student t test for paired samples. A difference was considered significant at p<0.05.
For immunostaining, deparaffinized tissue sections were dewaxed in xylene and rehydrated in a graded series of ethanol, and then incubated with 0.5% Triton X-100 for 15min, followed by blocking with a solution of 2% albumin for 30min. The sections were incubated with mouse MFRP affinity-purified polyclonal antibody (AF3445, R&D Systems, Minneapolis, MN) or anti-ezrin (Sigma-Aldrich, St. Louis, MO). Secondary antibodies were Alexa-594 or Alexa-488 fluorophore (Molecular Probes/Invitrogen, Eugene, OR) diluted 1:500 in 1× phosphate-buffered saline (PBS). All sections were examined by fluorescence microscopy, using a Leica TCS SP2 laser scanning confocal microscope (Leica, Heidelberg, Germany). Albino control mouse retinas were used as immunostaining controls to prevent the melanin in the RPE from interfering with the signal.
AAV8 (Y733F)-smCBA-MFRP-treated and untreated rd6 eyes were carefully dissected, and the MFRP protein was detected by Western blotting. The eyecups were homogenized by sonication in 1× PBS containing complete protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany). After centrifugation, each pellet was resuspended in a buffer containing 50mM Tris-HCl (pH 7.4), 1% Triton X-100, 2% sodium dodecyl sulfate (SDS), and 10% glycerol, and used for Western blot analysis. Aliquot extracts containing equal amounts of protein were subjected to SDS–polyacrylamide gel electrophoresis (PAGE), using a 4–12% gradient gel, and transferred to polyvinylidene difluoride (PVDF) membranes (Millipore, Bedford, MA). After incubation for 1hr in Odyssey blocking buffer (Li-Cor, Lincoln, NE), the membranes were probed either with a primary antibody against MFRP (mouse MFRP AF3445; R&D Systems) or an antibody against α-tubulin (rabbit polyclonal ab4074; Abcam, Cambridge, MA) as an internal control. The signals were detected with an infrared IRDye 800 dye-conjugated secondary antibody (Rockland Immunochemicals, Gilbertsville, PA). Visualization of specific bands was performed with the Odyssey infrared fluorescence imaging system (Odyssey; Li-Cor).
A 19-year-old female patient had spectacle correction in childhood, There were some complaints of night and peripheral vision disturbances as well as difficulty with reading; these visual symptoms were noted mainly from the second decade of life. Ancestry was German/English/French with no known parental consanguinity; parents and siblings had no visual complaints. Visual acuity was 20/30 with a refractive error of +11.00 sphere in each eye. Corneal curvatures, ultrasound A-scan, and anterior chamber depth were 52.00D spherical equivalent (normal mean±SD, 43.95±1.470) (AlMahmoud et al., 2011), 16.4mm (normal, 23.67±0.9) (Oliveira et al., 2007), and 2.2mm (normal, 2.9±0.3) (Fontana and Brubaker, 1980), respectively. Corneal diameters and intraocular pressures were normal. Near-infrared (NIR) reflectance images of the fundus showed irregular margins of the optic nerve head, and peripheral retinal regions with chorioretinal atrophy and bone spicule-like pigment (Fig. 1A). Autofluorescence imaging with SW and NIR excitation were consistent with a ~20° diameter central region of retained RPE with normal or nearly normal signals originating from lipofuscin and melanin fluorophores. Surrounding the central region was an annulus of hyperautofluorescence apparent in both SW- and NIR-RAFI. This annulus could originate from unmasking of existing fluorophores, accumulation of additional fluorophores, or chemical changes affecting their fluorescence efficiency. Surrounding the hyperautofluorescent ring was an intermediate level of autofluorescence like that previously described in other forms of RP (Cideciyan et al., 2011). Optic disc drusen were detectable as small hyperautofluorescent dots on SW-RAFI.
The high hyperopia with pigmentary retinopathy prompted questions concerning whether the patient had MFRP-RP (Ayala-Ramirez et al., 2006; Crespí et al., 2008; Zenteno et al., 2009; Mukhopadhyay et al., 2010). Partial nucleotide sequencing of MFRP exon 5 from the proband showed a heterozygous 1-bp deletion (Supplementary Fig. S1A; c.498delC, arrow) (supplementary data are available online at www.liebertonline.com/hum), which predicts a prematurely truncated protein (p.Asn167ThrfsX25). The normal exon 5 sequence is shown (Supplementary Fig. S1B). In the second allele a novel heterozygous G-to-T mutation (Supplementary Fig. S1C, arrow) at the conserved 5′ donor splice site was demonstrated at exon/intron 9. The normal exon/intron 9 sequence is also illustrated (Supplementary Fig. S1D). Parental DNA analysis showed that the father carried one allele with the one base deletion and the mother carried one allele with the splice site mutation.
Full-field ERGs were abnormal (Fig. 1B). Rod b-waves were barely detectable (11μV; normal, 299±52μV); a mixed cone–rod ERG had reduced a- and b-wave amplitudes (a-wave, 33μV [normal, 297±65μV]; b-wave, 35μV [normal, 497±111μV]); and cone ERGs were reduced in amplitude (for single flash, 61μV [normal, 173±32μV]; and for flicker, 17μV [normal, 172±35μV]). A rod sensitivity loss map by psychophysics showed no detectable rod function in the far peripheral field, but there was some retained rod sensitivity (between ~1.5 and 2.5 log10 units reduced) in a wide region of the central field (Fig. 1C, left). Cone sensitivity loss was most evident in the nasal field, but there was detectable cone function elsewhere in the field. At fixation, cone sensitivity was within normal limits but was reduced by 0.5–1.5 log10 units with increasing eccentricity into the temporal peripheral field (Fig. 1C, right). A kinetic visual field showed slight generalized constriction (more evident in the nasal field) to the V-4e target (70% of normal extent; 90% is 2 SD less than the normal mean) (Jacobson et al., 1989); the I-4e target was detected only centrally (10% of normal extent; 90% is 2 SD less than the normal mean) (Fig. 1C, inset).
Topographical maps using OCT of retinal thickness and maps of inner retina and photoreceptor outer nuclear layer (ONL) are shown for normal subjects compared with the patient with MFRP-RP (Fig. 1D and E). Whereas retinal thickness was within normal limits across most of the retina except in the very central macula, the inner retina was significantly thicker than normal across the retinal area studied. The ONL, on the other hand, was thicker than normal at the fovea, within normal limits in the parafoveal region, but thinner than normal across the remainder of the retina sampled (Fig. 1D and E, insets, are comparison maps to normal limits). Cross-sectional images revealed abnormal retinal laminar architecture in the patient with MFRP-RP. A normal fovea has a structural pit due to the concentric displacement of an inner nuclear layer (INL) and limited hyperreflective tissue vitreal to the cone outer nuclear layer (Fig. 1F). In contrast, the MFRP-RP fovea was devoid of a normal central depression and had substantial inner retinal lamination with a discernable INL, microcystic changes anterior to the ONL, and considerable tissue vitreal to the INL (Fig. 1G). The deep hyporeflective layer had no cystic structures and measured 182.8μm (normal foveal ONL in the central 0.6mm, 90.1±10.5μm; n=16; age, 11–28 years) (Jacobson et al., 2007).
We then asked whether the inner and outer segment (IS/OS) layer thickness was also different from normal in the patient with MFRP-RP and found the thickness across the central 4mm to be within normal limits (data not shown). This would suggest that this hyperthick layer may be complicated by Muller cell swelling, such as we suggested may be the cause of a similar-appearing effect in patients with choroideremia (CHM) (Jacobson et al., 2006). Thickness at the fovea with concomitant loss of the foveal pit at certain stages of CHM could reach 150–190μm, similar to that seen in this patient with MFRP-RP. A more common cause of thickened central structure in RP is cystoid macular edema (CME), but no cysts were visible within the ONL layer scans of the patient with MFRP-RP. The thick hyperreflective inner retinal layer across the foveal region in this patient measured 160μm, which is in dramatic contrast to the relatively thin foveal hyperreflectivity at the vitreoretinal interface in normal subjects (9.8±1.5μm; n=7; age, 9–45 years). Patients with CHM with hyperthick foveal hyporeflectivity and no cystic changes (n=3; age, 13, 34, and 38 years) and patients with RP with CME (n=6; age, 7–78 years) showed thicker than normal hyperreflectivity at the vitreoretinal interface, but the values were small (average, RP with CME, 34.8μm; CHM, 30.7μm) by comparison with that in the patient with MFRP-RP (Fig. 1H).
The natural history of retinal function in the rd6 mouse has been reported to show a decrease in ERG amplitude detectable by P25 followed by a progressive decline (Hawes et al., 2000; Won et al., 2008). Dark-adapted ERGs, dominated by rod function, show greater reduction than light-adapted ERGs, mediated by cone function, at early disease stages (Won et al., 2008). Representative ERG waveforms driven by the activity of rods, both rods and cones (mixed rod–cone), and cones are shown for a wild-type mouse and for the untreated and treated eyes of a 2-month-old rd6 mouse (Fig. 2A). At the ages studied, for rod b-wave amplitudes, the mean of untreated rd6 eyes was 58% of mean wild-type, whereas, for cone-isolated b-waves, the mean of untreated rd6 eyes was 89% of mean wild-type. Despite the tendency for lower values compared with wild-type, all rd6 eyes were within normal limits for these parameters. For mixed rod–cone ERGs, the majority (60%) of the untreated rd6 eyes had abnormally reduced amplitudes with mean values being reduced to 45 and 56% of mean wild-type for a- and b-waves, respectively.
ERGs performed 6 weeks after the uniocular subretinal injection of 1μl (109 total vector genomes) of AAV8 (Y733F) vector carrying wild-type mouse Mfrp cDNA showed there were higher amplitudes in treated versus untreated rd6 eyes. ERGs for all treated rd6 eyes were within normal limits. For the representative rd6 mouse (Fig. 2A), the treated eye had a rod-isolated b-wave amplitude of 462μV as compared with 225μV in the untreated eye (interocular difference, 69%). Across all rd6 animals, treated eyes had larger amplitudes (range, 316–462μV) compared with untreated eyes (160–238μV) with interocular differences (treated minus untreated) averaging 54% (range, 34–69%) as compared with 3% (range, −21 to +31%) in wild-type mice (Fig. 2B). With the mixed rod–cone ERGs, the treated rd6 eyes had larger amplitudes (a-wave range, 132–265μV; b-wave range, 326–604μV) as compared with untreated eyes (a-wave range, 47–147μV; b-wave range, 153–369μV) with interocular differences averaging 83% (range, 35–99%) for a-waves and 57% (range, 40–72%) for b-waves. Wild-type mice for the mixed rod–cone ERGs had a mean interocular difference near zero (range, −24 to +32%) (Fig. 2B). For the cone-isolated ERG, the treated rd6 eyes also had larger b-wave amplitudes (range, 129–175μV) as compared with untreated eyes (range, 84–128μV), with interocular differences averaging 35% (range, 22–50%). Interocular differences in wild-type mice for the cone-isolated ERG had a mean near zero (range, −35 to +38%) (Fig. 2B). In summary, interocular differences for all four ERG parameters in rd6 mice were significantly larger than in wild-type mice. Taken together with the fact that untreated eyes at this age had reduced retinal function compared with wild-type, the results suggest that the vector–gene injection rescued rod- and cone-mediated ERG function in the rd6 mice.
Light microscopy was used to evaluate the effect of the gene therapy on rd6 mice (Fig. 2C). Treated eyes had greater numbers of photoreceptor nuclei compared with the contralateral untreated eyes. The ONL of untreated rd6 eyes was about 5–7 nuclei thick (Fig. 2C, middle), whereas the vector-injected retina had 9 or 10 nuclei (Fig. 2C, right), which is similar to that in the wild-type eyes (Fig. 2C, left). ONL thickness (mid-superior retina) was 47.8±4.9μm in treated retinas compared with 29.9±1.7μm in untreated rd6 retinas (p<0.05) (Supplementary Fig. S2A). Treated rd6 retinas also had well-organized, normal length outer segments, in contrast to the untreated eye, in which outer segments were both shortened and disorganized. Abnormal cells, likely of phagocytic nature (Hawes et al., 2000), were visible in the subretinal space of untreated rd6 mice (Fig. 2C, arrows, middle) but absent in treated eyes.
MFRP expression was evaluated by immunohistochemistry in retinal sections (Fig. 3). Previous studies have shown that MFRP in wild-type mice is predominantly expressed in the RPE and ciliary epithelium whereas MFRP protein is not detected in rd6 eyes (Kameya et al., 2002; Mandal et al., 2006a; Won et al., 2008). This is confirmed in the present study (Fig. 3A and B). After gene therapy, intense labeling was present in the RPE (Fig. 3C), and it was similar to the pattern of MFRP in wild-type retina (Fig. 3A). When viewed at low magnification, the treated rd6 eye displayed widespread and nearly continuous expression of MFRP in the RPE (Fig. 3C, inset). This indicates that a wide expanse of retina was included in the subretinal injection. In addition, MFRP expression, as driven by the ubiquitous smCBA promoter, was detectable in photoreceptor inner segments and ONL when the images were purposely overexposed to enhance the low-level non-RPE transgene expression pattern throughout the retina (Supplementary Fig. S2B).
Western blot analysis confirmed the presence of an immunoreactive MFRP band in treated eyes only (Fig. 3F). The MFRP immunoreactive band migrated at about 120kDa, a much larger size than the predicted molecular mass of 65kDa of wild-type protein, but consistent with results of previous studies (Won et al., 2008; Fogerty and Besharse, 2011). This high molecular weight band suggests that MFRP may exist as a dimer in RPE cells. Previous studies have shown that some of the frizzled proteins are capable of forming homodimers, and a region containing the cysteine-rich domain is implicated in this process (Carron et al., 2003). Cysteine-rich domains have also been shown to form a conserved dimer interface within crystal structures, suggesting that dimerization may have a biological function in frizzled receptor signaling (Dann et al., 2001).
In earlier studies MFRP expression was found to be restricted to the base of the RPE apical membrane (Mandal et al., 2006a), but we observed intense labeling of RPE microvilli in both wild-type and treated rd6 mice (Fig. 3D and E). We further confirmed this finding by staining with an anti-ezrin antibody, a marker of RPE apical processes (Supplementary Fig. S3).
Before relatively routine molecular analysis, clinical reports noted the rare association of high hyperopia, microphthalmos, and retinal degeneration with or without other ocular abnormalities (e.g., Buys and Pavlin, 1999; Ghose et al., 1985; MacKay et al., 1987). The MFRP gene, first identified more than a decade ago (Katoh, 2001), was initially associated only with humans having high hyperopia, reduced axial length, but no retinal degeneration (Sundin et al., 2005). Screening for MFRP mutations in a wide spectrum of retinal degenerations was negative (Pauer et al., 2005) and species differences were used to explain the lack of retinal degeneration in humans (Sundin, 2005). More recently, however, autosomal recessive families with MFRP mutations were identified with RP, posterior microphthalmos, foveal abnormalities, and optic disc drusen (Ayala-Ramirez et al., 2006; Crespí et al., 2008; Zenteno et al., 2009; Mukhopadhyay et al., 2010). A list of clinical and molecular features of the 16 published patients (age range, 16–60 years), representing 7 families, is provided (Supplementary Table S1). Visual acuities varied from 20/25 to no light perception (NLP). All had high hyperopia (range, +13.50 to +29.00) and abnormal ERGs. Corneal diameters were normal. Data from the MFRP-RP patient in the present study are consistent with the clinical phenotype in these other reports (Supplementary Table S1).
The convexity forming a domelike configuration in the central retina with persistent inner retinal structure across the presumptive fovea in human MFRP disease suggests a disturbance in the complex early cell migrations that lead to normal foveal development. Nine of the previously published patients with MFRP-RP had OCT scans and all showed the lack of a foveal pit and thickening of inner retinal tissue across the foveal region (Ayala-Ramirez et al., 2006; Crespí et al., 2008; Zenteno et al., 2009; Mukhopadhyay et al., 2010). We postulate that this is the result of foveal maldevelopment (Sundin et al., 2008), likely complicated by effects of degenerative retinal disease. Formation of the normal foveal pit occurs during the second half of gestation and is due to progressive thinning of ganglion cell and inner nuclear cell layers that overlie the foveal cone cells due to their centrifugal displacement (reviewed in Provis et al., 1998). The process is complete at about 1 year of life. Another process, but a centripetal one, leads to increasing cone cell density in the foveal pit and this continues for years after birth. Taken together, the good visual acuity in many patients with MFRP-RP (Supplementary Table S1), normal inner and outer segment thickness measurements in our patient, and thicker than normal foveal ONL suggest that the cone cell centripetal movement has occurred. It is also conceivable that Muller cell swelling secondary to noncystic macular edema is the basis of the thickened hyporeflective layer (Jacobson et al., 2006). The excessively thickened superficial hyperreflective layer above the cone cells suggests persistence of tissue that should have moved centrifugally if there was normal development. A comparison with different retinopathies indicated that even the thickest of other pre–cone cell layers we examined are far thinner than the MFRP-RP layer. The OCT appearance without a foveal depression has some similarities but also differences from other maldevelopment retinopathies such as albinism (Marmor et al., 2008; McAllister et al., 2010) and nanophthalmic eyes without retinal degeneration (Bijlsma et al., 2008). The findings in the extracentral retina of photoreceptor degeneration and inner retinal thickening point to retinal remodeling in MFRP-RP, as we have noted in many retinal degenerative diseases (e.g., Jacobson et al., 2006; Aleman et al., 2009; Cideciyan et al., 2011).
The AAV8 (Y733F) tyrosine capsid mutant has emerged as a promising vector for the treatment of rapidly degenerating animal models of RP caused by mutations in photoreceptor cells (Pang et al., 2011). Here, we demonstrate the ability of a subretinally delivered AAV8 (Y733F) vector containing a smCBA promoter driving expression of the murine Mfrp gene to rescue retinal function and prevent photoreceptor cell death in the rd6 mouse, the early-onset RPE-based model of MFRP-RP. To minimize the surgical trauma associated with subretinal injections and to still be able to provide rapid and widespread transgene expression before the initiation of cell death, treatment was provided on P14 in rd6 mice, coincident with eye opening when the photoreceptor cell layer was still fully intact. AAV8 (Y733F)-smCBA-MFRP vector delivery at this age led to significantly higher ERG amplitudes relative to untreated eyes when assessed 6 weeks later. Histology of treated retinas revealed preserved photoreceptor ONL with better organized inner and outer segments than control untreated rd6 retinas. Moreover, accumulation of abnormal, phagocytic-like cells in the subretinal space was not observed in vector-treated eyes. Treatment also led to intense and widespread expression of the MFRP transgene, predominantly localized to the apical RPE membrane and its microvilli, where the wild-type protein normally resides in unaffected controls.
The function of the MFRP protein is not clear at this time, and the mechanism of photoreceptor degeneration in rd6 mice could be multifactorial. The presence of MFRP throughout the apical membrane and its actin-rich microvilli, both in wild-type and treated rd6 mice, suggests that it may play a structural role by maintaining the normal morphology of the RPE apical processes. MFRP expression normally increases markedly after birth, coincident with the development of microvilli and outer segments in mice (Won et al., 2008). Thus, rd6 retinal degeneration may be caused by compromised RPE microvilli leading to adhesion defects between RPE and photoreceptor outer segments. In addition, experiments suggest that MFRP and CTRP5 physically interact, and could therefore be both functionally and transcriptionally dicistronic (Mandal et al., 2006a; Shu et al., 2006). CTRP5 contains a short-chain collagen sequence and a C-terminal globular complement 1q (C1q) domain that appears to bind the extracellular MFRP region containing tandem repeats of two cubilin (CUB) domains and a low-density lipoprotein receptor-related sequence (Shu et al., 2006). C1q is a key component of the classical pathway of complement activation, and is known to be involved in many critical processes including innate and adaptive immunity, inflammation, apoptosis, cell adhesion, and monocyte chemotaxis (Kishore et al., 2004; Lu et al., 2008). Interestingly, a study describing a murine Mfrp null mutant has shown that CTRP5 is upregulated in both rd6 and Mfrp174delG (Fogerty and Besharse, 2011). Thus, the lack of MFRP protein might lead to uncontrolled signaling by CTRP5, triggering the migration of phagocytic cells from the choroid into the subretinal space, as seen in rd6 mice. Morphological characterization of an Mfrp/Ctrp5 double-knockout mouse could prove useful for testing this hypothesis in future studies.
Is MFRP-RP a target for future gene therapy trials? Although a limited number of patients have been reported to date, there is a recognizable human phenotype with a dramatic refractive error that should make clinical and genetic screening more productive in the future than previously (Pauer et al., 2005). Evidence of central retinal maldevelopment and early peripheral retinal degeneration with loss of normal retinal architecture indicating remodeling makes a complex target for gene therapy as it is currently performed. Despite reports of safety and efficacy of subretinal gene therapy as currently performed in the RPE65 form of LCA (reviewed in Cideciyan, 2010), a surgically induced subfoveal retinal detachment in a well-functioning but maldeveloped central retina may present more risk than benefit. The patient with MFRP-RP whom we examined, however, had preserved although abnormal peripheral retinal function (at the end of the second decade of life) by ERG and psychophysics, both for rod and cone photoreceptor systems. This suggests the nonfoveal retina is a potential therapeutic target. In terms of relevant rd6 experiments to perform, our relatively short-term and preliminary studies with one vector serotype demand to be confirmed and extended to longer term efficacy and safety studies, dose–response experiments, and future work to address the potentially safer approach of intravitreal delivery of Mfrp cDNA driven by RPE cell-specific promoters (e.g., tyrosine capsid mutant AAV vectors; Petrs-Silva et al., 2011).
Supported in part by NIH grants R01EY11123 and P30EY021721, and by grants from the Macula Vision Research Foundation, Foundation Fighting Blindness, and Research to Prevent Blindness. A.V.C. is an RPB Senior Scientific Investigator.
W.W.H. and the University of Florida have a financial interest in the use of AAV therapies, and own equity in a company (AGTC Inc.) that might, in the future, commercialize some aspects of this work.