Long-Term Transgene Expression in the RPE after Gene Transfer with a High-Capacity Adenoviral Vector

Abstract

purpose. To analyze duration of gene expression in the retinal pigment epithelium (RPE) in immunocompetent animals after gene transfer with a high-capacity adenoviral (HC-Ad) vector.

methods. An HC-Ad vector was constructed to express the enhanced green fluorescence protein (EGFP) from the human CMV promoter. This vector (HC-AdFK7) was used to transduce rat RPE cells in cell culture and after subretinal injection in vivo in adult immunocompetent Wistar rats. In cell culture, expression of EGFP was analyzed by fluorescence microscopy. In vivo expression was monitored by scanning laser ophthalmoscopy and stereo fluorescence microscopy. After enucleation of the eyes, immunohistochemical and morphologic analyses by fluorescence light microscopy and electron microscopy were performed.

results. In vitro, RPE cells were efficiently transduced with HC-AdFK7. Expression persisted for the observation time of 8 weeks. In vivo, the RPE was efficiently transduced with low doses of HC-AdFK7. EGFP synthesis was confirmed by antibody staining and found to be stable for the complete study period of 6 months. The neuroretina was well preserved over areas of subretinal vector administration, and significant morphologic changes were not detected. There was no accumulation of inflammatory T cells or macrophages.

conclusions. In contrast to previous results with earlier generation adenoviral vectors, subretinal injection of an HC-Ad vector in immunocompetent rats resulted in long-term transgene expression without evidence of adverse immune reactions or significant toxicity, probably because of the absence of expression of the viral gene from this vector. Thus, HC-Ad vectors are suitable for the treatment of eye disorders that require durable gene expression in the RPE.

Several hereditary and acquired retinal diseases are known to primarily affect the retinal pigment epithelium (RPE). The RPE is involved in retinol metabolism and in photoreceptor outer segment phagocytosis. In addition, in RPE cells several trophic and antiangiogenic factors are produced that are essential for the function and the integrity of the retina (for review, see Ref. ¹ ). Several types of retinitis pigmentosa are caused by mutations in genes that are exclusively expressed in the RPE and not in the photoreceptor cells. ^{2 3 4} Best disease is an example of a primary disorder of the RPE. ⁵ Age-related macular degeneration (AMD) is a very common multifactorial disorder that affects the RPE. In industrialized countries AMD is the leading cause of blindness in individuals more than 65 years of age. ⁶ Visual impairment is probably caused by macular photoreceptor degeneration due to progressive changes in the RPE and the choriocapillaris. ⁷ Formation of fibrovascular membranes in the macular region complicates the natural course of the disease and leads to sudden and severe visual loss. ⁸ Treatment options for this complication, known as choroidal neovascularization (CNV), include laser photocoagulation, photodynamic therapy, and submacular surgery. ⁹ After surgical excision of areas of CNV, invariably there is additional damage to the RPE. ¹⁰  

The disappointing long-term results of current therapies for AMD or genetic RPE disorders indicate that there is an urgent need for the development of new treatment strategies. Gene therapy has a significant potential for the treatment of retinal disorders. Viral vectors, derived from adenovirus, adenoassociated virus (AAV) or from lentivirus have been successfully used in preclinical models to deliver reporter or therapeutic genes into the different cell layers of the retina. ¹¹ The specific clinical disorder, the cell type to be transduced, the size of the transgene, the need for regulated gene expression, and the turnover of the target cell type may dictate the choice and the design of the vector that has the greatest chance of success in treatment of the particular disease. Although approaches involving gene therapy show promising results in long-term transduction of photoreceptor cells using lentiviral and AAV systems, ^{12 13 14} stable RPE transduction using adenoviral vectors has not been achieved until recently. 

Adenovirus serotype 5–based vectors have been demonstrated to efficiently and selectively transduce the RPE after subretinal vector injection. ^{15 16} However, in all experiments in which either reporter genes or therapeutic genes have been delivered into this tissue using E1-deleted (first-generation) adenoviral vectors, expression has been transient. Without exception, duration of expression has been less than 4 months. This has been interpreted to be the consequence of an immune response directed against either viral proteins or transgenic proteins expressed from the vector resulting in the removal of transduced RPE cells by effector cells of the immune system. HC-Ad vectors are characterized by the absence of viral coding sequences. Therefore, they feature reduced toxicity and immunogenicity and are suitable for delivery of large DNA fragments of up to 36 kb into target cells (for review, see Ref. ¹⁷ ). In the present study, stable gene expression of a neoantigen in the RPE was observed in adult immunocompetent rats after gene delivery by subretinal injection of an HC-Ad vector. This was achieved in the absence of adverse immune reactions or toxicity. This result contrasts favorably with previous results of only transient gene expression that have been observed with the use of earlier-generation adenoviral vectors. Therefore, HC-Ad vectors are very promising for long-term treatment of inherited and acquired disorders that require gene expression in the RPE. 

Materials and Methods

The experiments in this study were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the German Law for the Protection of Animals. 

Construction of the HC-Ad Vector HC-AdFK7

Plasmid pFK7 was generated by inserting the AflIII-AflII fragment (blunt-ended with Klenow) from pEGFP-N1 into the blunt-ended NotI site of pSTK129 (Kochanek S, paper in preparation). pSTK129 is a shuttle plasmid that can be used to generate high-capacity adenoviral vectors and consists of the left terminus of the adenovirus type 5 (nucleotide [nt] 1–440), a 20-kb DNA fragment derived from the human HPRT locus ¹⁸ HUMHPRTB (gene map positions, 1777–21729), a NotI cloning site, a 6.5-kb human fragment of C346 (locus, HUMDXS455A; cosmid map positions, 10205–16750), and the right terminus of adenovirus type 5 (nt 35818–35935). The insert of pFK7 is flanked by PmeI restriction sites. 

Rescue and Propagation of HC-AdFK7

To rescue the HC-AdFK7 vector (Fig. 1) , 5 μg pFK7 was cleaved with PmeI and transfected into 293Cre66 cells (G. Schiedner, paper in preparation) that were subsequently infected with helper virus AdLC8cLuc ¹⁹ at a multiplicity of infection (MOI) of five. After complete cytopathic effect (CPE) was observed, the medium and the infected cells were harvested and freeze thawed to release the virus. Aliquots of the crude vector lysate were serially passaged through 293Cre66 cells, as described. ^{19 20} The final vector stock was purified by CsCl equilibrium density gradient centrifugation. 

Titration and Quality Control

The yield of HC-AdFK7 after CsCl equilibrium density centrifugation was 2 × 10¹² particles, as determined by optical density of 260 nm (OD260). The infectious titer (1 × 10⁷ green-forming units per microliter) was determined in triplicate experiments by infecting HeLa cells with different numbers of particles and counting the resultant green cells with a fluorescence microscope. 

RPE Preparation and Transduction

The bulbi of adult pigmented Long-Evans rats were opened with a 360° incision at the ora serrata. The neural retina was removed. The exposed RPE was immersed in 0.25% trypsin solution containing 0.02% EDTA (Roche Molecular Biochemicals, Mannheim, Germany) for 20 minutes at 37°C before the RPE cells were carefully detached from the stroma with the sterilized fire-polished tip of a Pasteur pipette. The isolated cells were cultured in DMEM nutrient mixture (Gibco, Grand Island, NY) to 90% confluence. Increasing doses of HC-AdFK7 were used to transduce RPE cells to determine the optimal dose for 100% transduction of a given number of RPE cells. GFP expression was monitored by fluorescence microscopy with an FITC filter set (AF Analysentechnik, Tübingen, Germany). 

Surgical Procedures

Twelve-month-old Wistar rats were anesthetized by CO₂ inhalation and intraperitoneal injection of Ketanest (Parke-Davis, Morris Plains, NJ) for approximately 40 minutes. After additional topical anesthesia with proparacaine eye drops, a lateral canthotomy was performed. The conjunctiva was opened at the limbus, and the anterior chamber was decompressed by corneal puncture. With a 27-gauge needle, 5E+04 to 5E+06 infectious particles of HC-AdFK7 in a final volume of 0.5 μL was injected into the subretinal space of the central part of the superior hemisphere of each eye. 

Fluorescence Imaging

Rats were anesthetized by Ketanest injection. After application of mydriatic eye drops the animals were examined by scanning laser ophthalmoscopy (SLO; Rodenstock, Munich, Germany). The SLO provided the following excitation wavelengths: 780 nm (infrared), 514 nm (argon green), and 488 nm (argon blue). By using the fluorescein barrier filter and the latter excitation wavelength, GFP fluorescence was monitored. Infrared imaging served as a test for specificity of fluorescence. The fundus images were stored on S-VHS media. Rats were examined 1 week after the initial procedure. Thereafter, monthly SLO measurements were performed. 

Harvest of Transduced Eyes

After the study period of 6 months, one half of the eyes were enucleated and fixed overnight with 4% paraformaldehyde. The anterior segment and the lens were removed, and the posterior part of the eyecup was prepared as a wholemount by radial incisions toward the optic nerve. The neurosensory retina was gently removed, and the RPE layer was examined with a microscope (Axiovert; Carl Zeiss, Wetzlar, Germany) with an FITC filter set (AF Analysentechnik) and a digital camera (Orca ER; Hamamatsu, Hamamatsu City, Japan) in combination with image-aquisition software (Openlab; Improvision, Inc., Heidelberg, Germany) acquisition software. 

Light and Electron Microscopy

For electron microscopy, eyes were processed as follows. The corneae were removed and the eyes were fixed overnight at 4°C in 4% glutaraldehyde and 0.1 M cacodylate buffer (pH 7.4) containing 100 mM sucrose. Fluorescent areas in flatmount preparations were excised and postfixed with 1% OsO₄ at room temperature in 0.1 M cacodylate buffer for 3 hours, stained en bloc with uranyl acetate, and embedded in Spurr resin after dehydration in a graded series of acetone. The blocks were sectioned semiserially in 0.7-μm-thick sections. Ultrathin sections were stained with uranyl acetate and lead citrate and observed under an electron microscope (model 902 A; Carl Zeiss). For light microscopy, eyes were fixed as described earlier and incubated in 18% sucrose overnight before freezing at −80°C. Some fluorescent areas were excised and were processed for paraffin embedding. 

Immunologic Studies

The wholemount retinas after fixation with 4% paraformaldehyde were washed 4 times for 10 minutes with 0.05 M TBS solution. For blocking, the retinas were incubated for 10 minutes with 0.5 M NH₄Cl (Sigma, Diesenhofen, Germany) and 0.25% Triton (Serva, Heidelberg, Germany) and then two times for 60 minutes with 0.5 M TBS containing 5% bovine albumin fraction V (Sigma). For the detection of macrophages, monocytes or dendritic cells the antibody T3003 and for the detection of T cells the pan–T-cell antibody KiTiR (both from BMA, Augst, Switzerland) were used, both at a dilution of 1:50 in an overnight incubation at 4°C. After a wash in TBS, the sections were developed for 1 hour at room temperature with a 1:800 dilution of a Cy3-labeled goat anti-mouse antibody (Dianova, Hamburg, Germany). The samples were examined by microscope (Axiovert; Carl Zeiss) with a rhodamine filter set (AF Analysentechnik). 

Immunolabeling of GFP

The eyes were enucleated and frozen in isopentane. For immunocytochemical staining, frozen sections were air dried on glass slides coated with 0.05% poly-l-lysine. Sections were incubated in 1% goat serum in 0.05 M PBS (pH 7.4) for 15 minutes. After three washes with PBS, the sections were incubated in 0.3% H₂O₂ in methanol for 30 minutes and washed with PBS. Subsequently, the sections were incubated at 4°C overnight with a polyclonal antiserum against GFP from rabbit (Dianova) diluted 1:200 in PBS. After three washes with PBS, the sections were covered with goat serum for 15 minutes. The secondary antibody (anti-rabbit IgG coupled to Cy3) was diluted 1:800 in PBS, and the sections were incubated for 1 hour at room temperature. After three washes with PBS, the sections were dehydrated, embedded in Entellan (EM Science, Darmstadt, Germany) and investigated under a fluorescence microscope (Axioplan; Carl Zeiss, Oberkochen, Germany), using Cy3 (excitation 550 nm, emission 570 nm)-specific or EGFP (excitation 471 nm, emission 503 nm)-specific filter sets. Photographs from the same area were taken with a digital camera (Orca; Hamamatsu) with both filter sets separately. Overlays were made with image-aquisition software (Openlab; Improvision). As a control, the anti-GFP antibody was omitted. 

Measurement of Lipofuscin-like Autofluorescence

Flatmount preparations containing fluorescent RPE cells were observed with a filter set with excitation and emission maxima at both 405 nm and 460 nm. 

Results

RPE Cell Transduction In Vitro

Fluorescence light microscopy revealed EGFP-specific signals 24 hours after transduction (Fig. 2) . A vector dose of 20 MOI resulted in 80% EGFP-positive cells 48 hours after infection. At a vector dose of 50 MOI, 100% of the RPE cells were EGFP positive 48 hours after transduction. Three to 7 days after gene transfer, the intensity of the EGFP fluorescence reached its maximum and persisted for the observation time of 8 weeks. The morphology of the HC-AdFK7-transduced RPE cells did not differ from nontransduced cells. 

Long-Term EGFP Expression In Vivo

One month after subretinal injection of 5E+05 infectious units of HC-AdFK7 areas of patchy and/or continuous fluorescence were observed by SLO at the site of injection (Fig. 3a) . EGFP signals were specific, and infrared filter application did not show any background fluorescence. Six months after the initial treatment, at the end of the study period, there were still areas of bright EGFP fluorescence that did not differ significantly in extent and intensity compared with the results at months 1 through 5. Figures 3b and 3c show the same eye as in Figure 3a , 3 months and 6 months, respectively, after transduction. 

Dose-Dependent EGFP Expression In Vivo

The injection of 5E+05 infectious units led to virtually complete transduction of the RPE cell layer(Figs. 4a 4b) . The administration of more than 5E+05 infectious units did not further improve transduction. Three days after injection into the subretinal space of 5E+04 infectious units of HC-AdFK7, by fluorescence microscopy of cryosections, approximately 50% of the RPE cells in the area of the subretinal injection were EGFP positive (Figs. 4c 4d) . 

Specificity of EGFP Fluorescence

To evaluate the specificity of EGFP fluorescence at the protein level, we performed antibody labeling on cryosections using anti-EGFP antibodies. Figure 4e shows colocalization of the Cy3-labeled anti-EGFP antibody and sites of EGFP fluorescence. 

Light Microscopy and Electron Microscopy

Photoreceptor outer segments were slightly shorter in areas of the subretinal injection. There was no apparent infiltration by inflammatory cells at the site of the former injection. We did not observe signs of neuroretinal damage due to intermittent retinal detachment, possibly because of the small total volume injected. The retinas showed full thickness of all layers 6 months after injection. Their histologic appearance was normal (Fig. 5) . The images obtained with the SLO system in vivo were comparable to those that were acquired by fluorescence microscopy of the wholemounts of the enucleated eyes (Fig. 6) . By electron microscopy, an intact photoreceptor-RPE interface was observed in the areas of former subretinal injection of the vector (Fig. 7) . There were mild changes in the architecture of the rod outer segments, which most likely were due to the former temporal retinal detachment at the sites of injection. Accumulation of lipofuscin in the RPE cells, appearing as multiple dark vesicles surrounded by a single membrane, reflected the advanced age of the treated animal. 

Fluorescence Microscopy in Flatmount Preparations

Six months after subretinal injection EGFP expression was detected in the areas of subretinal injection of the HC-AdFK7 vector. The green fluorescence signals were observed only in RPE cells and not in any other cell types. A clear border of transduced and nontransduced cells (Fig. 6a) was demonstrated. The specific GFP fluorescence (Fig. 6b) was not visible when a lipofuscin filter set was used (Fig. 6c) . 

At the site of the fluorescent RPE cells, there was no gross change in architecture or density of the tissue layer. An invasion of inflammatory cells such as macrophages or T cells was not observed. 

Immunologic Analysis

Using specific antibodies, inflammatory cellular infiltrates consisting of macrophages, monocytes, or T cells were not detected 6 months after the initial injection. The pattern of immunostaining in treated eyes using the antibodies T3003 and KiTiR was not different from that in untreated control samples (data not shown). 

Discussion

In this study, efficient gene transfer into the RPE of adult and immunocompetent rats was achieved by subretinal injection of low doses of an HC-Ad vector. Injection resulted in stable expression of a reporter gene for at least 6 months in the absence of a local immune reaction or of toxicity. This finding was unexpected, because in all RPE gene transfer studies reported to date involving the use of E1-deleted adenoviral vectors, gene expression has been transient, in most cases lasting for less than 12 weeks. ^{21 22 23 24} This has been explained by immune responses directed against viral or transgenic proteins resulting in the immune-mediated removal of transduced RPE cells, despite the immune-privileged status of the subretinal space. ^{21 24 25 26 27}  

In our study, we detected only minor ultrastructural changes at the sites of vector administration that were similar to those after temporary retinal detachment alone. EGFP toxicity ^{28 29} and glycerol toxicity may have contributed to this minor effect. The favorable expression and toxicity profiles of the HC-Ad vector used in this study are in agreement with those in skeletal muscle, ^{30 31 32} hepatocytes, ^{20 33} and the central nervous system. ^{34 35} However, expression of a foreign antigen in skeletal muscle results in immunorecognition and in the loss of gene expression, even if the protein is expressed from an HC-Ad vector. ³⁰ Adenoviral-mediated delivery of the Escherichia coli lacZ gene into the striatum of immunocompetent rats results in persistent expression when the transgene is expressed both from a first-generation and from an HC-Ad vector. However, peripheral subcutaneous challenge with an adenoviral vector expressing an unrelated transgene results in the loss of expression in animals transduced with the first-generation but not with the HC-Ad vector. ³⁵  

These results indicate that in the brain, an immune-privileged organ as is the eye, expression of viral functions is required for the loss of expression on peripheral challenge. The collective experience in many studies in which different transgenes were delivered to the RPE using first-generation adenoviral vectors suggests that expression of viral or transgenic proteins alone or in combination are responsible for the observed loss of expression, probably caused by immune- or toxicity-related mechanisms. Thus, although the subretinal space is immune privileged, protection appears to be not sufficient to prevent immunorecognition of transgenic proteins if they are expressed from first-generation vectors. The results of the present study indicate that the shortcomings of the earlier generation of adenoviral vectors can be overcome by the use of HC-Ad vectors. This new vector combines the advantages of adenoviral vectors (high titer, efficient RPE transduction, strong transgene expression, efficient transduction of quiescent cells) with durable gene expression that so far has been achieved only with lentiviral or AAV vector systems. In contrast to these vectors HC-Ad vectors only rarely integrate into the host cell’s genome, minimizing the risk of insertional mutagenesis. A loss of HC-Ad vector genomes during cell division does not play a significant role in the RPE, because the RPE is a postmitotic cell layer. Furthermore, HC-Ad vectors can carry large DNA fragments with sizes of up to 36 kb, thus allowing the delivery of several transgenes at the same time or the inclusion of regulatable gene expression systems. The use of low doses of this vector type in the RPE cells further minimizes the risk of negative side effects. 

In conclusion, HC-Ad vectors are very promising vectors for the treatment of eye disorders that require stable gene expression in the postmitotic RPE cell population. 

²  Contributed equally to this work.

 Supported by Grant FKZ01KS9502 from the Federal Ministry of Education and Research and grants from the Center for Molecular Medicine Cologne (SK), the Köln Fortune Program and the Retinovit Foundation (US), and the Boehringer Ingelheim Foundation (FK).

 Submitted for publication July 24, 2001; revised December 20, 2001; accepted January 25, 2002.

 Commercial relationships policy: I (TTL, SK); N (FK, IS, US).

 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.

 

Genomic structure of HC-AdFK7. HC-AdFK7 has the following features from left to right: the left terminus of adenovirus type 5 (nt 1–440), a 20-kb DNA fragment derived from the human HPRT locus ¹⁸ (locus HUMHPRTB; gene map positions, 1777–21729), the EGFP expression cassette derived from pEGFP-N1, a 6.5-kb human fragment of C346 (locus HUMDXS455A; cosmid map positions, 10205–16750), and the right terminus of adenovirus type 5 (nt 35818–35935). The locations of the inverted terminal repeats (ITRs) and the packaging signal (ψ) are indicated.

Figure 2.

 

In vitro gene transfer into RPE cells. Cultured rat RPE cells expressed EGFP 24 hours after transduction with 50 MOI of HC-AdFK7.

Figure 3.

 

SLO fundus images after subretinal injection of HC-AdFK7. Arrow: Group of transduced RPE cells. (a) EGFP-expressing cells 1 month, (b) 3 months, and (c) 6 months after transduction in the same eye of the same rat. This group of cells is shown as a flatmount preparation in Figure 6a and 6a fraction of it is shown in Figure 7 in an electron micrograph.

Figure 4.

 

In vivo gene transfer into the RPE. An unstained cryosection (a) through a rat eye after subretinal injection of HC-AdFK7 in a bright-field image. Arrows: RPE cell layer. (b) Same section under fluorescent light. The RPE specifically expressed EGFP after gene transfer with 5E+05 infectious units. Other retinal cells are not transduced. (c) An unstained cryosection is shown as a bright-field image after injection of HC-AdFK7. (d) Same section as in (c) under fluorescent light. Every other RPE cell was EGFP positive after gene transfer with 5E+04 infectious units. In the overlay (e) yellow or orange indicates specific fluorescence in both wavelengths. Green fluorescent RPE cells were labeled with anti-EGFP antibodies coupled to Cy3 in a cryosection. Micrographs were taken from the same area with both filter sets separately (Cy3: excitation 550 nm, emission 570 nm; EGFP: excitation 471 nm, emission 503 nm).

Figure 5.

 

Architecture of the retina after subretinal injection of HC-AdFK7. A paraffin-embedded section through the transfected RPE cell layer 6 months after the initial application. Arrows: RPE. The morphology of all retinal layers was not altered. Infiltrating mononuclear cells were not visible.

Figure 6.

 

(a) Flatmount preparation of the RPE layer expressing EGFP 6 months after subretinal injection of HC-AdFK7. Individual RPE cells are recognizable by their hexagonal shape (arrow). This group of transduced cells is shown under the laser scanning microscope at two earlier time points in Figures 3a and 3b . This group of transfected cells is shown under the laser scanning microscope in Figure 3 . A flatmount preparation containing fluorescent RPE cells (b) is shown, by using a filter set with excitation and emission maxima at 405 nm and 475 nm (c). Therefore, the green fluorescence is different from lipofuscin.

Figure 7.

 

Electron micrograph in the area of the group of RPE cells in Figure 3 . An intact photoreceptor–RPE interface is shown at the ultrastructural level in an area of former subretinal injection of HC-AdFK7. There were mild changes in rod outer segment (ROS) architecture appearing as swollen disc membranes ( ), as typically occur in areas of temporal retinal detachment at the sites of injection. Accumulation of lipofuscin in the RPE cells, appearing as multiple dark vesicles (small arrows), reflect the advanced age of the treated animal. Large arrow: Bruch membrane. Arrowheads: microvilli. N, nucleus of an RPE cell.

The authors thank Hanna Janicki for excellent technical assistance and Frank Graham for the generous gift of AdLC8cLuc.