AAV-Mediated Expression Targeting of Rod and Cone Photoreceptors with a Human Rhodopsin Kinase Promoter

Abstract

purpose. Gene therapy for retinal degeneration requires well-defined promoters that drive expression in rod and cone photoreceptors. This study was undertaken to develop short, active derivatives of the human rhodopsin kinase (RK) gene promoter for targeting transgene expression in rods and cones. RK, also known as G protein–coupled receptor kinase 1 (GRK1), is a component of the light adaptation pathway expressed in rods and cones.

methods. Human RK (hRK) promoter and its concatemers or derivatives extending into the conserved 5′ untranslated region (5′-UTR) were assayed for promoter activity in WERI retinoblastoma or Crx/Sp1-supplemented HEK-293 cells. The derivative displaying the highest activity was linked to a GFP reporter and packaged in a pseudotyped adenoassociated viral vector (AAV2/5). The AAV vector was tested in vivo by subretinal injections in wild-type mice, in the all-cone Nrl ^−/− mice, and in the cone-rich diurnal Nile grass rat (Arvicanthis niloticus). Control eyes received a similar AAV2/5 vector carrying a mouse rod opsin (mOps) promoter-controlled GFP reporter.

results. The hRK promoter with the full 5′ untranslated sequence (–112 to +180) was the most active in cell culture. Delivered by the AAV2/5 vector, RK promoter drove GFP expression specifically in photoreceptors. In rods, hRK promoter-mediated expression was as efficient as, but appeared more uniform than, mOps promoter-mediated expression. In cones, the hRK promoter drove expression, whereas the mOps promoter did not.

conclusions. The hRK promoter is active and specific for rod and cone photoreceptors. Because of its small size and proven activity in cones, it is a promoter of choice for somatic gene transfer and gene therapy targeting rods and cones.

Hereditary photoreceptor degeneration, clinically known as retinitis pigmentosa (RP), encompasses a group of blinding retinal diseases with diverse genetic origins and clinical manifestations. Typical RP leads to visual deficit in middle age and progresses to severe visual loss in later life. Leber congenital amaurosis (LCA) describes a more severe form of photoreceptor degeneration with visual deficit in early childhood. Genetic defects in more than 100 genes are known to underlie the etiology of photoreceptor degeneration (http://www.sph.uth.tmc.edu/retnet/). These genes perform essential functions in rods, cones, and retinal pigment epithelium (RPE). Gene-based therapies offer a potential treatment for photoreceptor degeneration, as illustrated in a number of studies in model systems. In these studies, replacement gene constructs were delivered to the retina through gene delivery vectors, usually those derived from adenoassociated virus (AAV), which proved effective as a vehicle for gene delivery to retinal photoreceptors. ^{1 2 3 4 5 6 7 8 9 10 11 12} Recombinant AAV-2 (rAAV-2) vectors were among the first to be developed. Pseudotyped AAV vectors are AAV2 vector sequence packaged into viral capsids derived from different serotypes. ¹³ Pseudotyped AAV2/5 vectors are more efficient than AAV-2 vectors in mediating gene delivery to photoreceptors and RPE cells. ^{13 14 15}  

Apart from efficient gene delivery vectors, further progress toward clinical application of gene-based therapies requires well-defined promoter/enhancer elements with uniform, photoreceptor-restricted transcriptional activity. AAV vectors have a relatively small carrying capacity for foreign DNA that typically does not exceed 4.5 kb. The limited carrying capacity of AAV vectors dictates that such promoters be short and, ideally, no more than several hundred bases in length. A number of promoters have been studied with variable success in driving expression of foreign genes in photoreceptors. ^{11 16 17 18 19 20 21} Short derivatives of rod and cone opsin promoters have been validated as efficient and specific for driving expression in rods and in cones, respectively. However, many severe forms of photoreceptor degeneration, such as LCA, have primary gene defects in rods and cones requiring gene therapy so that all photoreceptors can be targeted simultaneously. At present, no well-characterized compact promoter targets rods and cones. In several studies, rod opsin (rhodopsin) gene promoters were shown to be “leaky,” allowing transgene expression in rods and in cones. ^{21 22 23} Among other potential rod–cone-specific genes characterized to date, interphotoreceptor retinoid-binding protein (IRBP), ^{24 25 26} CRX, ²⁷ and RK ^{28 29} gene promoters were reported to have activity in vivo. Multiz genomic alignments (http://genome.ucsc.edu/cgi-bin/hgGateway) show a diffuse IRBP promoter structure with multiple conserved elements scattered over a broad region upstream of the transcription start site. Similarly, the CRX gene promoter, ²⁷ 2 kb in length, is much too long to be accommodated in an AAV vector for most applications. RK is expressed in rod and cone photoreceptors. A single conserved 0.2-kb enhancer/promoter (hereafter referred to as promoter) resides near the transcriptional start sites of the RK gene. Cross-species comparisons show the 0.2-kb TATA-less region, together with its extension into the 5′ untranslated region, to be the sole conserved segments in the entire 5′ flanking sequence. ³⁰ This is in contrast to the diffusely organized promoters of other photoreceptor-specific genes extending upstream for several kilobases. ²⁹ Functional studies of the RK promoter in transiently transfected retinoblastoma cells and in transgenic mice revealed this segment as a key active enhancer within the 5′ flanking region. ²⁹  

In this study, we examined the compact human RK promoter and its derivatives for their potential usefulness in the context of AAV-mediated gene targeting of rod and cone photoreceptors. Compared with an opsin promoter construct in parallel, in vivo analysis revealed an important difference between the two promoters. Although both targeted expression in rods with high efficiency, the RK, but not the opsin, promoter drove expression in cones. 

Materials and Methods

Materials

WERI-RB1 and HEK-293 cell lines and pXP1 plasmid were from American Tissue Culture Collection. Cell culture supplies including media were purchased from Invitrogen. Transfection reagent (GenePORTER) was from Genlantis (San Diego, CA), and pRSV-EGFP plasmid was from Stratagene (La Jolla, CA). Other expression plasmids were kindly provided by the following investigators: pCMV-LacF (β-galactosidase) from R. Palmiter (University of Washington, Seattle, WA), pCDNA3.1HBC (bovine Crx) from Shiming Chen (Washington University, St. Louis, MO), and pEVR2/Sp1 (human Sp1) from Guntram Suske (Philipps University, Marburg, Germany). Luciferase and β-galactosidase assay kits were purchased from Tropix (Bedford, MA). All other chemicals were from Sigma (St. Louis, MO). 

Luciferase Reporter Plasmids and In Vitro Transient Transfection Assays

RK promoter-reporter constructs were based on previously described pRK-112/+87-Luc and p-1904/+87-GFP parent plasmids. Numbers in the plasmid designate the 5′/3′ extent of the sequence included in the RK promoter DNA fragment relative to the transcription start sites. ²⁸ All clones were engineered according to standard protocols, and all final constructs were verified by DNA sequencing. Constructs carrying 5′ UTR extensions were derived by incorporation of oligonucleotides adjacent to the promoter. pRK-112/+133-Luc was the product of ligation of the synthetic double-stranded AegI-XmaI oligomer (sense, 5′-CCGGTCTCCCAGGGGCTTCCCAGTGGTCCCCAGGAACCCTCGACAGGGC-3′; antisense, 5′-CCGGGCCCTGTCGAGGGTTCCTGGGGACCACTGGGAAGCCCCTGGGAGA-3′) in sense orientation into the XmaI site of the previously described pRK-112/+87-Luc. ²⁸ Further ligation of an additional synthetic fragment (sense, 5′-CCGGTCTCTCTCGTCCAGCAAGGGCAGGGACGGGCCACAGGCCAAGGGC-3′; antisense, 5′-CCGGGCCCTTGGCCTGTGGCCCGTCCCTGCCCTTGCTGGACGAGAGAGA-3′) into the regenerated XmaI site in pRK-112/+133-Luc led to pRK-112/+180-Luc. Additional constructs, pRK(-112/+87)₂-Luc, pRK(-112/+87)₃-Luc, and pRK(–112/+87)₆-Luc, containing variable repeats of the RK promoter upstream of the luciferase gene, were developed on a modified version of pRK-112/+87. After conversion of a single HindIII site in the pRK-112/+87-Luc to the NheI site by Klenow fill-in and religation, additional copies of the RK enhancer/promoter (BglII-BamHI, –118 to +87) were sequentially introduced into the BglII site of the modified plasmid in the sense direction upstream of the luciferase gene. Fragment orientations were verified by restriction digestion and sequence analysis. 

Transient transfection of luciferase reporter constructs into WERI-RB1 cells were carried out as previously described. ²⁸ Briefly, 2 to 5 million suspension-grown WERI cells per milliliter serum/antibiotic–free medium were exposed to 2 μg reporter plasmid with 100 ng pCMV-LacF, together with transfection reagent. Luciferase and β-galactosidase activities were measured in cell lysates 48 hours after transfection (Lumat LB 9507; Berthold Technologies, Bad Wildbad, Germany) using substrates provided in the kits (Tropix). Measured luciferase activities were normalized against the β-galactosidase activities from each well then divided by the normalized activities for promoterless plasmid to determine relative luciferase activities (RLUs). Select luciferase constructs were cotransfected with transcription factor plasmids encoding Crx ^{31 32} and Sp1 ³³ into HEK293 cells. Crx and Sp1 have putative recognition sites on the RK promoter and potentially on the 5′ UTR. 

Recombinant AAV Construct and AAV Vector Production

To construct the AAV vector, a pRK-112/+180-GFP plasmid was digested with EcoRI, blunt ended with Klenow treatment, and digested again with HindIII. The 1.2-kb insert containing the RK P/E linked to a downstream GFP reporter and ending with an SV40 polyA signal sequence was isolated. The backbone parental plasmid, pAAV-MCS6-IZsGreen vector, provided the AAV ITR packaging sequences. This vector was cut with SpeI, blunt ended with Klenow, and digested with HindIII again so that it was compatible for ligation with the insert. The ligation reaction mixture was transformed into DH5α-competent bacterial host, and the new plasmid pAAV-RK-GFP was purified from the bacteria. The pD10/mOps-GFP vector, in which GFP was driven by a minimal mouse opsin promoter (nucleotides –218 to +17 upstream from the transcription start site), was described previously. ¹¹ AAV2/5 pseudotyped vectors (AAV-hRK-GFP and AAV-mOps-GFP) were generated by tripartite transfection (AAV vector plasmid encoding the gene of interest, AAV helper plasmid pLT-RC03 encoding AAV Rep proteins from serotype 2 and Cap proteins from serotype 5, and adenovirus helper miniplasmid pHGTI-Adeno1) into 293A cells. Transfection was performed using a protocol developed by Xiao et al. ³⁴ Two days after transfection, cells were lysed by repeated freeze and thaw cycles. After initial clearing of cell debris, the nucleic acid component of the virus producer cells was removed by nuclease (Benzonase; Merck KGaA, Darmstadt, Germany) treatment. Recombinant AAV vector particles were purified by iodixanol density gradient. Purified vector particles were dialyzed extensively against PBS and titered by dot blot hybridization. A pRK-112/+87-GFP plasmid was similarly packaged into the AAV vector. 

Animals and Vector Delivery

At postnatal day (P)30 to P60, 16 wild-type (WT) mice were placed under general anesthesia with intraperitoneal injection of ketamine (90 mg/kg)/xylazine (9 mg/kg). A 0.5% proparacaine solution was applied to the cornea as a topical anesthetic. Pupils were dilated with topical application of cyclopentolate and phenylephrine hydrochloride. Under an ophthalmic surgery microscope, a small incision was made through the cornea adjacent to the limbus with an 18-gauge needle. A 33-gauge blunt needle fitted to a Hamilton syringe was inserted through the incision, with care taken to avoid the lens, and was pushed through the retina. All injections were made subretinally within the nasal quadrant of the retina. Each animal received 2 μL AAV-hRK-GFP at 1 × 10¹² particles per milliliter in the left eye (OS) and the equivalent volume of the AAV-mOps-GFP at the same concentration in the right eye (OD). Visualization during injection was aided by the addition of fluorescein (100 mg/mL AK-Fluor; Alcon, Inc., Fort Worth, TX) to the vector suspensions at 0.1% by volume. Fundus examination during the injection found the entire retina detached, confirming successful subretinal delivery. In addition to WT mice, Nrl ^−/− mice ³⁵ (n = 14) and Nile grass rats (Arvicanthis niloticus; n=13) were injected with AAV-hRK-GFP (OS) and AAV-mOps-GFP (OD). The Nile grass rat is a diurnal animal, ³⁶ and another member of this genus (Arvicanthis ansorgei) has been shown to have a cone-rich retina. ³⁷ All experiments involving animals were approved by the Institutional Animal Care and Use Committee and were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Immunofluorescence Analysis

Eyes were enucleated, placed in 2% formaldehyde and 0.1% glutaraldehyde/PBS, and their anterior segments and lens were removed. Fixation continued in this fixative for 2 hours. The fixed tissues were soaked in 30% sucrose/PBS for 3 hours or overnight, shock frozen, and sectioned along the superior-inferior meridian at 10-μm thickness. Immunolabeling on frozen sections was performed as described. ³⁸ Rhodamine-labeled peanut agglutinin (PNA) was purchased from Molecular Probes (Eugene, OR). Chicken anti–mouse cone opsin antibodies, ¹¹ rabbit anti–mouse cone arrestin antibody, ³⁹ and monoclonal anti–rhodopsin antibody (rho1D4) ⁴⁰ were previously described. Stained sections were photographed on a confocal laser scanning microscope (model TCS SP2; Leica, Wetzlar, Germany). For quantitative comparison, the excitation energy levels and duration of exposure were kept the same for paired samples, and 12-bit images were taken to preserve the dynamic range of the original signal intensities. Image data were analyzed quantitatively using the ImageJ 1.34s software (National Institutes of Health, Bethesda, MD). 

Results

Analysis of the RK Promoters in Heterologous Cell Systems

To optimize the human RK promoter for photoreceptor-directed transgene expression, we began with a transient transfection approach to select maximally active derivatives of the promoter core. Transient transfection assays in WERI retinoblastoma cells have previously been found to be informative. Previous studies have also suggested that, in general, inclusion of nonconserved sequences further upstream has an inhibitory effect on RK promoter activity in vivo and in vitro. ²⁸ We, therefore, focused on comparing derivatives carrying multiple copies of the promoter or 3′ extension into the conserved 5′ UTR. As shown in Figure 1A , the extended RK promoter constructs were progressively more active, with more of the 5′ UTR included (filled bars). The construct with the full conserved 5′ UTR, pRK-112/+180-Luc, was 2 to 3 times as active as the basal construct pRK-112/+87-Luc. Inclusion of an additional copy of promoter region inserted in tandem (pRK(-112/+87)₂-Luc) also led to enhanced activity. Inclusion of more copies, as in pRK(-112/+87)₃-Luc and pRK(-112/+87)₆-Luc, did not lead to enhancement of activity. The pRK-112/+87-Luc and pRK-112/+180-Luc constructs were further tested in HEK293 cells cotransfected with Sp1+ and Crx expression plasmids. We found approximately threefold higher activation of pRK-112/+180-Luc than of p-112/+87-Luc (Fig. 1A ; open bars). Thus, pRK-112/+180 was the strongest RK promoter derivative among those tested. 

RK Promoter Drives Robust and Specific Transgene Expression in Retinal Photoreceptors

To determine the promoter activity of pRK-112/+180 sequence (Fig. 1B)in vivo, we placed a GFP reporter under its transcriptional control and packaged the resultant expression construct into an AAV2/5 vector to generate AAV-hRK-GFP (Fig. 1C) . For comparison, we also packaged a GFP reporter construct driven by a mouse opsin promoter into AAV2/5 (AAV-mOps-GFP). The latter promoter is a minimal mouse opsin promoter composed of nucleotide sequence –218 to +17 that was previously shown to drive robust expression in photoreceptors. ¹¹ Both vectors were delivered into adult animals by subretinal injections. We evaluated the animals for GFP expression at three time points after injection: 2 to 3 weeks, 6 to 8 weeks, and 16 weeks. Some GFP-positive photoreceptors were detectable at 2 weeks. The signals appeared stronger and more widespread, covering nearly the full expanse of the retina at 6 to 8 weeks, and remained so at 16 weeks (Fig. 2A) . No expression in the RPE or in the inner retina was ever observed (Fig. 2B) , confirming the photoreceptor specificity of the two promoters. Both promoters appeared similarly effective in driving expression in photoreceptors. RK promoter-mediated expression appeared relatively uniform, whereas the opsin promoter resulted in many cells that appeared extraordinarily brighter compared with the surrounding cells (Fig. 2B) . The visual impression that the mOps promoter led to greater cell-to-cell variability is further illustrated by surface plots of the GFP signals (Fig. 2C) . 

RK promoter, but not opsin promoter, drives expression in cone photoreceptors.

We next asked whether either promoter or both promoters could drive GFP reporter expression in cone and rod photoreceptors. We first examined whether GFP-positive photoreceptor cells were confluent, which would be indicative of a promoter being active in both types of photoreceptors. As shown in Figure 3A , wherever transduction was maximal in the retina, all photoreceptors were GFP positive in eyes that received AAV-hRK-GFP. In contrast, periodic “blank spots” were seen along the outer margin of the outer nuclear layer, where cone nuclei are typically located, in eyes injected with the mOps vector. Given the thin optical sectioning (at 3 μm), it is unlikely that cells at different depths could have accounted for the apparent GFP confluence. This observation suggested that rods and cones expressed GFP reporter in the hRK vector–injected retinas. Injected retinas were also examined by double labeling with cone photoreceptor markers, including cone opsin, peanut agglutinin, and cone arrestin. Similar results were obtained with all three markers, but cone arrestin labeling provided the better marker because its distribution largely overlapped with that of GFP within the photoreceptor cell bodies. Cone photoreceptors in the RK vector–injected retina appeared positive for GFP (Fig. 3B) . By comparison, mOps vector–injected retina showed blank areas where cone arrestin staining was positive (Fig. 3B) . 

Although our data suggested that the RK promoter was active in cones and that the opsin promoter was not, this interpretation remained tentative. In the mouse retina, the dense packing of photoreceptors and the small number of cones surrounded by rods constituted a technical hurdle to an unambiguous interpretation about the cellular origins of GFP. Hence, we sought additional evidence in retinas in which cones were more abundant. Because of a developmental defect, the Nrl ^−/− mouse retina has an all-cone photoreceptor population, with a gene expression profile resembling that of blue cones. ³⁵ We reasoned that if the RK promoter was active in cones, it should drive GFP expression in a larger number of photoreceptors in the Nrl ^−/− retinas. This proved to be the case. As shown in Figure 4A , hRK promoter drove GFP expression in the photoreceptor layer. In contrast, the opsin promoter exhibited no detectable activity in the Nrl ^−/− mouse retina. As a control experiment, we performed immunolabeling of the same set of experimental retinas with cone or rod opsin antibodies to verify that they were indeed cone dominant. The results confirmed that most Nrl ^−/− photoreceptors expressed blue cone opsin, with a few positive for the green cone opsin and none for rhodopsin (Fig. 4B) . Nrl ^−/− retinas had many photoreceptor rosette formations in the central and midperipheral retinas. Their photoreceptor inner and outer segments also appeared much shorter than normal. Interestingly, hRK-driven GFP expression was more apparent in the superior retinas. Because the GFP-positive cells were almost confluent in this region, they did not appear to correspond to the few scattered green cones. Nevertheless, analyses of the AAV-injected Nrl ^−/− retinas illustrated a differential transcriptional activity of the two promoters in cones. 

An additional approach for demonstrating promoter activity in cones took advantage of the cone-rich retina in Nile grass rats (A. niloticus). ³⁷ In the retina of the closely related species A. ansorgei, more than 30% of photoreceptors are cones, and green cones outnumber blue cones by a factor of 10³⁷. By immunostaining for cone opsins (not shown) and cone arrestin (Fig. 5) , we confirmed that the A. niloticus retina was similarly cone rich. As illustrated by cone arrestin immunostaining, rods and cones are stratified in such a way that cones occupy the outer half of the outer nuclear layer that lies close to the inner segments (Fig. 5 , arrow). RK and opsin promoters drove robust reporter gene expression in rods, as evidenced by the strong GFP signal in the inner portion of the outer nuclear layer. The hRK promoter led to expression in cones (Fig. 5 , upper panels). In contrast, in the mOps vector–injected retinas, the outer half of the outer nuclear layer was completely negative for GFP (Fig. 5 , lower panels). These observations support the conclusion that the RK promoter was active in rods and cones, whereas the opsin promoter was measurably active in rods only. 

Discussion

A limiting factor in the development of gene therapy for retinal degenerative disorders has been the scarcity of well-characterized, dually active promoters capable of driving gene expression specifically in rods and cones. Many of the promoters characterized to date are diffusely organized with crucial cis-acting elements distributed over a wide expanse of genome extending in each direction from the transcription start sites. Resultant lengths of the functional promoters frequently surpass the size limits for incorporation into AAV vector constructs. Such scattered organization is highlighted by the recent availability of multi-genome alignments (http://genome.ucsc.edu/cgi-bin/hgGateway) showing that even the well-characterized photoreceptor-specific promoters to date extend well beyond the boundaries defined by in vivo functional studies ^{19 41} (and our unpublished data, 2006). Rhodopsin kinase, with a crucial role in light adaptation ^{42 43} and abundant expression in rods and cones, thus far appears to have the most favorable promoter configuration. It has a single, highly active 0.2-kb enhancer/promoter near the transcriptional start sites. ^{28 29} The RK promoter is unique in that its functionally defined enhancer/promoter, together with a segment of the 5′ untranslated segment, is the only segment in the entire 5′ flanking sequences that is conserved. As such, this segment is likely to have all the functionally active elements. ^{28 29}  

The robust RK promoter activity observed in this study is probably a function of the compact, self-contained configuration and the strength of the native promoter in rods and cones. This may in part explain the disproportionate levels of RK promoter activity compared with the short opsin promoter, given the fact that the native opsin gene is expressed at a much higher level than rhodopsin kinase. A deficiency in the short opsin promoter may result from exclusion of much of the potentially important 5′ flanking sequences scattered upstream and perhaps other locations in the gene. In cell culture studies, the conserved segment of 5′ UTR, next to the TATA-less promoter, further enhanced activity, putatively through promoting transcriptional complex formation, stabilizing mRNA structure, or recruiting translational apparatus. In vivo, the basal RK promoter (pRK-112/+87) delivered in an AAV vector also drove robust expression in rods and cones (our unpublished observations, 2006). Quantitative analysis of the relative strengths of the two promoters and efforts to engineer less active promoter derivatives are under way. It is likely that a range of promoter strengths will be needed for optimal expression of different replacement genes. 

An important question for this study was whether the RK promoter was active in cone and rod photoreceptors. Because it was technically difficult to determine with certainty whether a cone was GFP positive when it was densely packed among rods that were also GFP positive, we expanded our studies to include rodent retinas that were enriched for cones. The Nrl mutant mice have a defect in the specification of photoreceptors such that their retinas contain a nearly pure blue cone population. ³⁵ The Nile rat is a diurnal animal with a retina that has at least 30% cone photoreceptors, most of which express the middle wavelength (green) opsin. ³⁷ Thus, we were able to confirm in the cone-rich retinas that the RK promoter drove expression in rods and cones. Given that the Nrl mutant mice and the Nile rats are enriched for blue and green cones, respectively, our data appear to suggest that the RK promoter is active in green and blue cones and indeed in all photoreceptors. In all our in vivo assays, the GFP reporter signals were always stronger in rods than in cones. It remains unclear whether this reflected a real difference in the promoter activity in rods and cones or whether this resulted from different posttranscriptional processing of the GFP reporter in these two cell types. 

Under the experimental conditions and with the specific short mouse opsin promoter we used in this study, we were unable to confirm the opsin promoter as having transcriptional activity in cones. Various forms of the short mouse opsin promoter have been found “leaky” in cones, with detectable transgene expression in cones in several transgenic and somatic gene transfer studies. ^{21 22 23} The reason for the different outcomes remains unclear. Other than subtle differences in the promoter sequences used, it is possible that the mOps-mediated cone expression level might have been below the detection limit in this study. Another interesting feature of the RK promoter is the comparatively uniform expression pattern. This may be an important potential advantage for gene therapy applications because variable expression from cell to cell may lead to insufficient expression or toxicity. We conclude that the short RK promoter, as described in this article, is a promoter of choice for mediating expression in rods and cones. As such, it will be a valuable tool for vector-mediated human ocular gene therapies. 

 Supported by the National Eye Institute (Grants EY13600 [SCK] and EY10581 [TL]), EU EVI-GENORET (RRA), the Foundation Fighting Blindness, the Foundation for Retina Research, and Research to Prevent Blindness (challenge grant [Department of Ophthalmology, State University of New York, Buffalo, New York]).

 Submitted for publication February 28, 2007; revised April 2, 2007; accepted June 6, 2007.

 Disclosure: S.C. Khani, None; B.S. Pawlyk, None; O.V. Bulgakov, None; E. Kasperek, None; J.E. Young, None; M. Adamian, None; X. Sun, None; A.J. Smith, None; R.R. Ali, None; T. Li, 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.

 

Figure 1.

 

Activities of human RK promoter derivatives in vitro. (A) Relative luciferase activity in WERI-RB (filled bars) or HEK-293 (open bars) after transfection with RK promoter-luciferase plasmids. Plasmids carrying unmodified basal promoter (pRK-112/+87-Luc), promoter with added segments conserved 5′ UTR (pRK-112/+133-Luc, pRK-112/+180-Luc), or promoter concatemers were transfected into WERI-RB cells, and resultant luciferase activities were recorded in relative luciferase units. Bold arrows: core promoter sequence (–112/+87). Extensions into the 5′ UTR are represented by striped blocks. P-Luc (promoterless control), pRK-112/+87-Luc and p-112/+180-Luc were also assayed by cotransfection with Crx and Sp1 into HEK-293 cells. The β-galactosidase normalized activity of the constructs was represented as a percentage of the p-112/+87 activity. (B) Sequence and known regulatory elements in the RK promoter region. H1/HDRE indicates potential homeodomain (Crx) binding site. Solid underline: GC-rich regions presumably interacting with Sp proteins are indicated by (GC/SP). Dashed underline: putative Nrl and Nr2e3 binding sites. (C) Schematic diagram of the AAV-hRK-GFP construct. ITR, inverted terminal repeat; SV40SD/SA, splice donor/acceptor sequences derived from the SV40 virus; SV40 polyA, the polyadenylation site from SV40 virus.

Figure 2.

 

GFP reporter gene expression in WT mouse retinas after subretinal delivery of the AAV constructs. (A) Six weeks after injection, the retina shows widespread GFP expression. Areas of coverage and percentages of GFP-positive cells were comparable with those of opsin and RK promoters. (B) Expression was confined to the photoreceptor cell layer only. No expression was observed in the RPE or inner retinal neurons. Upper panels: GFP fluorescence images. Lower panels: GFP fluorescence images superimposed on Nomarski images to illustrate all layers of the retina. RPE, retinal pigment epithelium; OS, outer segment; IS, inner segment; ONL, outer nuclear layer; INL, inner nuclear layer; IPL, inner plexiform layer; GC, ganglion cell layer. (C) RK-mediated GFP expression in photoreceptors appeared to be more uniform than that of the opsin promoter, as illustrated in the surface plots from the corresponding GFP fluorescence in (B). To increase the dynamic range, 12-bit images were acquired and analyzed with ImageJ image analysis software.

Figure 3.

 

Analyses of reporter expression in cone photoreceptors in WT mice. (A) Higher magnification views of GFP confocal fluorescence images photographed at 3-μm thickness. RK-AAV–injected retinas showed 100% of the cells positive for GFP; the relatively thin optical section of the image makes it unlikely that GFP-negative cells could be masked by positive cells from a different depth. In contrast, AAV-mOps–GFP injected retinas showed the characteristic blank spots at the outer margin of the ONL, where cone nuclei were concentrated. (B) Double labeling with a cone arrestin antibody showed that most cones in the AAV-hRK–GFP injected retinas (left panels) were positive for GFP (arrows). AAV-mOps-GFP injected retinas (right panels) were blank where cone marker staining was positive (arrows). Upper panels: GFP fluorescence. Lower panels: GFP fluorescence (green) superimposed on cone arrestin immunofluorescence (red). RPE, retinal pigment epithelium; OS, outer segment; IS, inner segment; ONL, outer nuclear layer; arr, arrestin.

Figure 4.

 

Analysis of GFP reporter expression in the Nrl ^−/− mutant mouse retinas. (A) Three weeks after subretinal injection, eyes receiving AAV-hRK-GFP had many photoreceptors positive for GFP (left). No GFP-positive photoreceptors were found in the contralateral eyes of the same animals injected with AAV-mOps-GFP (right). Superior portions of the retinas are shown. Cell nuclei stained blue with Hoechst dye 33342 illustrate the different retinal layers. (B) The Nrl ^−/− mutant mouse has an all-cone outer retina, as was previously reported. The retinas shown in (A) were labeled with different opsin antibodies to determine their cell types. Photoreceptors were overwhelmingly positive for the blue cone opsin (left panel). Occasional cells were seen to express the green cone opsin (middle panel; arrows). One of the green cones was located inside a photoreceptor rosette (arrowhead). No evidence of rhodopsin was observed, as illustrated by the monoclonal antibody rho1D4 (right panel). The few bright red spots in the inner retina were blood vessel profiles illustrated by the anti–mouse IgG secondary antibody. Cell nuclei were stained blue with Hoechst dye 33342. OS, outer segment; IS, inner segment; ONL, outer (photoreceptor) nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer.

Figure 5.

 

Analysis of GFP reporter expression in the cone-rich A. niloticus retina. Cone and rod photoreceptors in this retina are stratified in the photoreceptor nuclear layer (ONL). As indicated by the cone marker cone arrestin immunostaining (red), cone nuclei are packed in the outer rows in the ONL (closer to the inner segments; arrowhead), whereas rod nuclei are in the inner rows of the ONL close to the outer plexiform layer (synapses). After injection with AAV-hRK-GFP, GFP expression was found in rod and cone photoreceptors (upper panels). In the retinas injected with AAV-mOps-GFP (lower panels), cones were completely blank whereas rods were strongly positive for GFP. RPE, retinal pigment epithelium; OS, outer segment; IS, inner segment; ONL, outer nuclear layer; OPL, outer plexiform layer.

The authors thank Agnieszka Lis for assistance with transient transfection assays, Cheryl Craft and Xuemei Zhu (the Mary D. Allen Laboratory for Vision Research, Doheny Eye Institute) for the cone arrestin antibody, Anand Swaroop (University of Michigan) for the Nrl ^−/− mice, Laura Smale (Michigan State University) for the breeding colony of Nile grass rat (A. niloticus), the Research Vector Core at Harvard Medical School for packaging AAV vectors, and David Hicks (Université Louis Pasteur) for helpful suggestions on Nile rats.