Efficient Gene Transfer into Retinal Cells Using Adenoviral Vectors: Dependence on Receptor Expression

Abstract

purpose. A number of ocular diseases are potentially amenable to gene therapy interventions if appropriate vectors for the targeted administration of therapeutic genes can be identified. In vitro and in vivo transduction efficiency of a Group C serotype 5 adenoviral vector containing the fiber domain derived from a Group B serotype 35 adenovirus and the gene encoding green fluorescent protein (AdV5/F35-GFP) was compared to an AdV5-GFP vector for transgene delivery to human retinoblastoma and to human and murine retinas.

methods. The distribution of the adenoviral receptors CAR and CD46 on normal and malignant retinal tissues was determined using immunohistochemistry. Human retinoblastoma cells were incubated with either AdV5-GFP or AdV5/F35-GFP, and the expression of the reporter protein was compared using quantitative fluorescence and fluorescent-activated cell sorting. Mice were given a single subretinal injection of either viral vector, and eyes were enucleated at specified times after injection for histopathologic examination. Human cadaver eyes were similarly examined ex vivo.

results. CAR was expressed in retina except in photoreceptor outer segments. CD46 was expressed in photoreceptor inner and outer segments. Both vectors efficiently transduced the human retinoblastoma cells in vitro. However, the amount of the transgene expressed using AdV5/F35-GFP was more than sixfold greater than that when AdV5-GFP was used. In vivo, AdV5/F35-GFP at doses as low as 10⁵ infectious units (IU) transduced cells in all layers of the retina especially photoreceptors and occasional neuronal cells, and Müller cells as well as retinal pigment epithelial cells, whereas AdV5-GFP transduced only retinal pigment epithelial cells and occasional photoreceptors and Müller cells.

conclusions. AdV5/F35 chimeric vectors may be superior to AdV5 for gene therapy applications targeting the photoreceptor.

The majority of the adenoviral vectors currently used have been derived from Group C serotype 5 (AdV5) viruses by deletion of portions or all of the viral coding sequences and replacement of those genes with the therapeutic or reporter gene of interest. The Coxsackie–adenovirus receptor (CAR) has been identified as a cellular receptor for adenovirus Group C serotypes 2 and 5 (AdV2, AdV5) fibers and for Coxsackie B viruses. ^{1 2} Subsequent studies have demonstrated that, with the exception of Group B adenoviruses, representative serotypes from Groups A, C, D, E, and F all use CAR as a cellular fiber receptor. ³ Adenoviral vectors based on the serotype 5 virus also require the presence of α_v-integrins for internalization. ^{4 5 6} Insufficiency or lack of the adenoviral receptors on the target cells may therefore be one of the major limitations associated with the use of recombinant adenoviruses for gene therapy. 

Adenoviral vectors use the same fiber/knob structures that protrude from the viral capsid of the wild-type adenovirus to interact with the CAR receptor for binding to susceptible cells. However, not all target cells that express CAR at sufficient levels may allow efficient viral transduction and high-level transgene expression. Recently, an AdV5/F35 chimeric vector in which the fiber protein from Group B serotype 35 virus has been substituted for the original AdV5 fiber was developed. ⁷ Binding of this chimeric vector to the target cells is CAR-independent and appears to use CD46 as its receptor. ⁸ The AdV5/F35 chimeric vector effectively transduces AdV5-resistant cells such as hematopoietic cells. ⁷  

AdV5 vectors efficiently transduce the retina pigmented epithelium (RPE), however they poorly transduce most other retinal cells including photoreceptors. ^{9 10 11} The aim of this study was to compare the in vitro and in vivo transduction efficiency and transgene expression levels in the human retinoblastoma-derived cell line Y79 and in the various retina layers using the chimeric AdV5/F35 vector or its parent adenovirus serotype 5 vector (AdV5). 

Methods

Cells and Viruses

The Y79 and Weri human retinoblastoma cell lines and the HEK293 human embryonic kidney cell line (American Type Culture Collection, Rockville, MD) were grown in Dulbecco’s modified eagle medium (Gibco/BRL, Life Technologies, Grand Island, NY), supplemented with 5% fetal bovine serum (HyClone, Logan, UT), 100 units/mL penicillin and 100 μg/mL streptomycin (Gibco/BRL, Life Technologies) at 37°C in 5% carbon dioxide-supplemented humidified air. The Y79 and Weri human retinoblastoma cells were maintained in suspension at a concentration of 10⁵ to 10⁶ cells/mL. Passages two through eight were used for the experiments in this study. The HEK293 cells were grown as an attached monolayer. Adenoviral 5 (AdV5) constructs containing a green fluorescent protein (AdV5-GFP, 145 viral particles/infectious unit [vp/IU]) reporter gene or expressing fiber domains from adenovirus 35 and containing the GFP reporter gene (AdV5/F35-GFP, 40 vp/IU) were provided by the Center for Cell and Gene Therapy, Baylor College of Medicine. These constructs were derived from the first-generation E1/E3 deleted adenovirus vector by inserting the GFP expression cassette under the control of the cytomegalovirus (CMV) promoter into the E3 deleted region. The AdV5-GFP and AdV5/F35-GFP vectors were maintained and propagated separately in HEK293 cells. At maximum virus infection when the cytopathic effect (CPE) can be seen, cells were harvested and viruses were extracted from the cells by three consecutive freeze/thaw cycles and purified as previously described. ¹² Purified viral stocks were stored at −80°C in 4% sucrose (10 mM Tris [pH 8.0], 2 mM MgCl₂) for in vitro and in vivo experiments. 

In Vitro Transduction and Analysis of GFP

To study the transduction efficiency of AdV5-GFP and AdV5/F35-GFP, Y79 cells were seeded in 6-well plates (1 × 10⁵ cells/mL). Cells were transduced with AdV5-GFP or AdV5/F35-GFP at the indicated viral concentrations (IU/mL) and incubated at 37°C in 5% CO₂ humidified air. At 48 hours post-transduction, the relative fluorescence intensity as a measure of the amount of the transgene (GFP) expressed and the percentage of GFP-expressing cells were determined by fluorescence activated cell scanning (FACS). 

In Vivo Transduction

All animal studies were performed in accordance with the policies stated in the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the guidelines approved by the Baylor College of Medicine IACUC. Two groups of 2-month-old Rag-2 immune deficient mice or C57BL/6 immune competent mice were given single subretinal injections of AdV5-GFP or AdV5/F35-GFP at 10⁷ IU in 1 μL of viral dilution buffer (10 mM Tris, 4% sucrose, and 2 mM MgCl₂, pH 8.0 [HyClone]). As control, the contralateral eyes were injected with an equal volume of viral dilution buffer. 

Immunohistochemistry

At the indicated times after viral injection, the eyes were harvested and immediately fixed in 10% formalin. The tissues were processed and embedded in paraffin. The blocks were sectioned (4 to 5 μm thickness) and then incubated at 60°C for 1 hour to deparaffinize. After incubation in xylene for 3 minutes, tissues were washed with ethanol (100%, then 95%), rinsed with deionized water, and then treated with proteinase K (Dako Corporation, Carpinteria, CA). Because autofluoresence in the retina interferes with direct observation of GFP expression, immunocytochemistry using an antibody to GFP was used. Samples were blocked in hydrogen peroxide, followed by incubation with a rabbit anti-GFP antisera (BD Clontech, Palo Alto, CA), affinity purified rabbit anti-CAR antisera, ¹ rabbit anti-α_v-integrin antisera (BD Pharmingin, San Diego, CA), or mouse anti-CD46 (BD Clontech) for 30 minutes and then an anti-rabbit or anti-mouse second antibody conjugated with horseradish peroxidase using Dako Envision+ Systems (Dako Corporation) for 30 minutes After incubation with the peroxidase substrate diaminobenzidine (DAB) solution for 8 minutes, samples were washed in PBS and then counterstained with Meyer’s hematoxylin (Poly Scientific, Bay Shore, NY). The hematoxylin solution was washed off with deionized water followed by ethanol (100% and then 95%) and then incubated with xylene. Tissue samples were observed microscopically for a brown precipitate indicating the expression of the antigen. In some experiments, double labeling with a mouse anti-GFAP antibody (Dako Corporation) was performed to identify cells in the retina of glial origin transduced by the AdV5/F35 vector. For the double labeling experiments the Double Staining Kit (Dako Corporation) was used in conjunction with a Mouse on Mouse kit (Vector Laboratories Inc., Berlingeme, CA). The GFAP antibody was applied first with an appropriate second antibody conjugated to peroxidase. The chromogen DAB was applied to stain the GFAP immune complexes brown. After proteinase K digestion, the anti-GFP antibody was applied with an alkaline phosphatase conjugated second antibody. The chromogen Fast-red (Dako Corporation) was applied staining the GFP immune complexes red. The slides were then counterstained with hematoxylin. Antibodies when used at high concentrations can result in nonspecific staining. All antibodies were used at dilutions where the tissues known not to express the antigens did not react with the chromogens. Reaction for GFP was further examined by Western blot analysis in transduced and nontransduced tissues. Although faint bands could be identified in overdeveloped blots in nontransduced retina, the only unique band found in transduced retina had the molecular weight of GFP and this signal identified at least 90% of the total signal as determined by densitometry. Human retina and retinoblastoma tissues were obtained as discarded pathologic specimens and processed as above for immunohistochemical analysis. 

Results

Weri and Y79 Retinoblastoma Cells Express CAR and CD46

To confirm the expression of the AdV5 receptor CAR and the AdV35 receptor CD46, 2 × 10⁵ cultured Weri human retinoblastoma cells were harvested and stained with the anti-CAR monoclonal antibody RmcB (American Type Culture Collection) or the monoclonal antibody anti-CD46 and goat-anti-mouse fluorescence conjugated second antibody or only the second antibody (control). CAR- and CD46-bearing cells were quantitated using fluorescence activated flow cytometry (FACS). More than 90% of the cells specifically reacted with the monoclonal antibody to human CAR and CD46. Less than 5% of the Weri cells reacted with the FITC conjugated second antibody alone (Fig. 1) . These results indicate that the Weri retinoblastoma cells bear both AdV5 (CAR) and AdV35 (CD46) receptors. Similar results were found when the Y79 retinoblastoma cell line was analyzed. A similar analysis of α_v-integrin expression on Y79 cells using a rabbit anti-α_v-integrin showed only low-level expression of this protein (10 to 20% of cells) as has been previously described. ⁷  

Expression of CAR, CD46, and α_v-Integrin in Retinoblastoma

Tumors from children with retinoblastoma were formalin-fixed and embedded in paraffin. Sections were prepared for immunocytochemistry as described in the Methods section and incubated with either affinity purified rabbit anti-CAR, monoclonal anti-CD46, or rabbit anti-α_v-integrin and peroxidase-conjugated goat anti-rabbit or rabbit anti-mouse second antibody and treated with DAB. After counter staining with hematoxylin, the slides were observed under phase microscopy (Fig. 2) . Retinoblastoma expressed CAR and CD46 but failed to express α_v-integrin. Rosettes found in retinoblastoma tumors strongly expressed CD46. 

Expression of CAR, CD46, and α_v-Integrin in the Retina

Formalin-fixed mouse or human retina were embedded in paraffin, cut into 4–5 μm sections and prepared for immunocytochemistry as described in the Methods section. The sections were incubated with affinity purified rabbit-anti-CAR, anti-α_v-integrin, or monoclonal anti-CD46 and peroxidase-conjugated goat anti-rabbit or rabbit anti-mouse second antibody and treated with DAB. After counter staining with hematoxylin, the slides were observed under light microscopy. The results in the mouse (detailed in Fig. 3 ) and human retina (not shown) were similar for expression of all proteins. CAR expression was noted in the ganglion cell layer, inner and outer plexiform layers, inner and outer nuclear layers, and Müller cells. Whereas the photoreceptor inner segments were positive, the photoreceptor outer segments were distinctly negative for staining. RPE cells and the optic nerve were positive for CAR. CD46 expression was found in photoreceptor inner and outer segments. α_v-Integrin expression was found in the ganglion cell layer, inner and outer plexiform layers, inner nuclear layer, and in photoreceptor inner segments. The outer nuclear layer showed only a few cells staining positive. Photoreceptor outer segments were negative for α_v-integrin expression. RPE cells were positive for α_v-integrin. 

AdV5/F35-GFP Results in More Transgene Expression in Y79 Cells than AdV5-GFP

Transduction efficiency of AdV5/F35-GFP and AdV5-GFP in Y79 cells was evaluated based both on the percentage and on the relative fluorescence intensity of GFP-positive cells. Human retinoblastoma Y79 cells (1 × 10⁵ cells/mL) were incubated with either virus at the indicated concentrations of viral infectious units. After 48 hours, cells were harvested and analyzed by FACS for the percentage of the GFP-expressing cells. There was no significant difference in the percentage of Y79 cells transduced by either AdV5/F35-GFP or AdV5-GFP at similar viral concentrations (Fig. 4A) . At a concentration of 1 × 10⁷ IU/mL the relative fluorescence intensity as a measure of the amount of the transgene (GFP) expressed by AdV5/F35-GFP vector was sixfold greater than the AdV5-GFP vector (Fig. 4B) . Greater concentrations of viral vector were cytotoxic to the target cells (as measured by trypan blue exclusion) without increasing the amount of transgene expressed (data not shown). ¹³  

AdV5/F35-GFP Transduces Both Retinal Cells and RPE

AdV5 vectors are relatively ineffective at transducing most retinal cells including photoreceptors in vivo. To determine whether changing the fiber protein from F5 to F35 would improve adenoviral transduction of retinal cells, mice were given single subretinal injections of AdV5-GFP or AdV5/F35-GFP at a dose of 10⁷ IU in 1 μL of viral dilution buffer. Two weeks after injection the eyes were harvested, fixed, sectioned, and examined for GFP expression using single- and double-labeling immunohistochemical methods as described. GFAP expression was examined using double-labeling immunohistochemistry to distinguish cells of glial origin from neuronal cells. When compared to the control (Fig. 5A) , both viral vectors transduced RPE cells (Figs. 5B 5J) . Although some transduction of photoreceptor and Müller cells was observed close to the injection site with AdV5 (Fig. 5C) , the AdV5/F35 chimera transduced all layers of the retina, especially photoreceptors (Figs. 5F 5G 5H 5I) in both immune competent and immune deficient mice (Figs. 5D 5E) . The GFP was observed in the photoreceptors, synaptic regions and could be observed in nerve fibers in the optic nerve (Fig. 5F) . The ciliary body, lens, and corneal epithelium also had evidence of transduction by AdV5/F35 (not shown). As expected, transduction was greatest around the injection sites (Figs. 5D 5E) . At viral vector doses of 10⁵ IU/μL, the AdV5 vector transduced only the RPE whereas the AdV5/F35 transduced photoreceptors and RPE. The other layers of the retina were not effectively transduced at this dose by either viral vector (data not shown). 

Expression of GFP Is Stable in both Immune Competent and Immune Deficient Mice

Efficiency of expression of GFP was approximated by counting the percentage of cells that were positive in the region immediately adjacent to the injection site at both 2–4 weeks and 7–8 months after injection of the AdV5/F35 viral vector in immune competent (C57BL/6) and immune deficient (Rag-2) mice (Table 1) . Animals that had been injected with the AdV5 vector had a comparable level of RPE cells transduced as the animals with the AdV5/F35 vector but had a very low percentage (<5%) of transduced cells in all retinal layers. Variability in the percent of cells transduced within similar groups of animals is due to the injection technique. Although the percentage of RPE cells decreased over time, the percentage of photoreceptors containing GFP was maintained in the immune competent mice even after 8 months. Thirty-five Rag-2 and 35 C57BL/6 mice were injected with the AdV5/F35-GFP vector and their eyes were examined at intermediate time points. Similar patterns of expression were observed. 

Discussion

AdV5 vectors have been modified to contain the shorter fiber/knob structures of the serotype AdV3 (AdV5/F3) ^{14 15} or AdV35 (AdV5/F35) ¹⁶ Group B viruses. This substitution was sufficient to transfer all infectious properties from AdV3 or AdV35 to the chimeric vector. Recent studies with the modified fiber chimeras AdV5/F3 in ovarian cancer cells ¹⁷ and AdV5/F35 in hematopoietic stem cells ¹⁸ as well as in primitive stem cell subsets ¹⁹ have demonstrated a higher level of transgene expression than with the parent AdV5 vector. These chimeric vectors bind to receptors distinct from CAR and enter the cells by an α_v integrin-independent pathway. ^{3 20} The AdV5/F35 chimera has been shown to transduce both CAR-positive and CAR-negative cells efficiently and to mediate high transgene expression. ¹⁹  

The hypothesis that the AdV5/F35 chimera vector may more efficiently transduce retinal cells and result in higher levels of transgene expression than the parent AdV5 vector was examined. The apparent tropism of different serotypes in different tissues results from virus interaction with distinct cellular receptors. ²¹ The Group B serotype AdV35 virus recognizes a receptor distinct from CAR ³ that has been identified as CD46. ^{8 22} The AdV5/F35 chimeric vector used in the experiments detailed in this report is identical with the AdV5 parent vector except the AdVF35 fiber is expressed instead of the native AdV5 fiber. Both vectors contain a GFP reporter transgene driven by a cytomegalovirus (CMV) promoter. Therefore, differences in transduction efficiency and transgene expression should be limited to the properties construed on the viral vector by this change in fiber protein. 

The transduction efficiency and transgene expression levels of AdV5-GFP and AdV5/F35-GFP vectors were first examined in cultured retinoblastoma Y79 cells, a cell line derived from a human retinal tumor. At identical viral infectious unit concentrations, both viruses efficiently transduced the Y79 cells. Nearly 100% of the Y79 cells expressed GFP when the cells were incubated with either adenoviral vector; however, the amount of GFP expressed in the cells transduced by AdV5/F35 was significantly greater (sixfold) than in cells transduced by AdV5. AdV5 vector requires a high density of CAR for fiber binding ^{1 2} and α_v-integrins for internalization ⁵ whereas the AdV35 binds and enters the cells using CAR and α_v-integrin independent pathways. ²⁰ One possible explanation for the data presented in this report is that there are more AdV35 fiber receptors (CD46) on the retinoblastoma cells than CAR, thereby allowing more AdV5/F35 viral particles to bind to the cell. Another possible explanation is that there are fewer α_v-integrins expressed on the cell than the AdV35 internalization protein. Y79 cells have been demonstrated to express low levels of α_v-integrin (data not shown). ⁷ Wickham and coworkers have demonstrated that AdV5 vectors can transduce cells even when the vectors have been modified not to express fiber or penton base, ²³ although the magnitude of transgene expression was reduced. The number of infectious units of AdV5-GFP required to express transgene in retinoblastoma cells in vitro was two orders of magnitude greater than that required in HEK293, a cell line that expresses both CAR and α_v-integrins (data not shown). However, when transduction was assessed using adenoviral vector to cell ratios less than one, the difference in GFP expression mediated by AdV5/F35 compared with that mediated by AdV5 was still evident (data not shown). Since the statistical probability of more than one copy of virus entering a given cell under these conditions is low, the increased transduction observed with the AdV5/F35 vector is probably not due to increased copy number of active viral particles in the cell but may be due to a CD46-linked biochemical pathway that allows more efficient viral processing and transgene expression. 

The utility of the AdV5/F35 vector to express the GFP transgene in the retina in vivo was investigated in mice using subretinal injections of AdV5-GFP or AdV5/F35-GFP. The eyes were examined immunohistochemically for the presence of GFP to distinguish GFP expression from the high autofluorescent background of the retina. Both viral vectors efficiently transduced the RPE. Some transduction of photoreceptor and Müller cells was observed around the AdV5 vector injection site; however, the AdV5/F35 chimera vector more efficiently transduced photoreceptors. The outer and inner plexiform layers clearly expressed GFP. The neuronal axons could be visualized entering the optic nerve. Lower doses of the AdV5 viral vector (10⁵ or 10⁶ IU) transduced only the RPE cells whereas the AdV5/F35 vector at the same doses clearly resulted in transgene expression in the photoreceptor layer. The other layers of the retina did not express the transgene with either vector at these lower doses. A human eye donated at time of autopsy was injected with AdV5/F35-GFP. GFP transgene expression could be detected in photoreceptors and RPE after subretinal injection of AdV5/F35 in situ (data not shown). 

A recent report demonstrated that intravitreous injections of vector derived from a Group D AdV37 virus efficiently transduce retinal cells including photoreceptors whereas a Group B AdV3 vector and a Group C AdV5 vector did not transduce retinal cells. ¹¹ AdV37 had originally been thought to use CAR as its receptor; however, more recent evidence has suggested that AdV37 may use an alternative but not yet characterized receptor. ²⁴ In our studies, AdV5 vectors transduced photoreceptors when injected subretinally but with limited efficiency compared with the AdV5/F35 chimera vector. CAR and α_v-integrin are not expressed on photoreceptor outer segments. Thus, the reduced transduction efficiency by AdV5 vectors might be explained by steric hindrance caused by the tight packing of photoreceptors that could interfere with effective binding of the large AdV5 fiber to photoreceptor inner segment CAR. Although AdV3 and AdV35 both are Group B adenoviruses, the homology between the amino acid sequences of their fibers is only approximately 60%. ¹⁸ That AdV35 binds to different receptor(s) than AdV3 but appears to use common structural elements for internalization (but not α_v-integrins) has been demonstrated. ²²  

CD46, a cell surface protein used by the measles virus as its receptor, ²⁵ is also the receptor for AdV35 and AdV11. ^{8 22} Using immunohistochemical analysis, CD46 is found to be expressed on photoreceptor inner and outer segments and the RPE in both human ²⁶ and murine retinas and on retinoblastoma cells. While CD46 is found on numerous tissues in primates, this complement regulating protein appears to have expression limited to the testes and eye in rodents. ^{27 28 29 30}  

AdV37 receptors may be expressed in the retina while the AdV3 receptor is not. Therefore AdV5/F35 vector transduction in Y79 cells can be explained by the presence of CD46 on the cell surface. Although viral receptor and integrins appear to account for adenoviral transduction in vitro, in vivo observations suggest that additional factors in the local environment play a significant role in adenoviral transduction. CAR expression does not strictly correlate with adenoviral-mediated transgene expression in vivo. ³¹ CAR ablation does not affect AdV5-mediated transgene expression in vivo. ^{23 32 33 34} Hyaluronic acid appears to enhance adenoviral transduction through interactions with CD44 (Chaudhuri et al. IOVS 2003;43:ARVO E-Abstract 4620). Even though retinoblastoma tumors do not express α_v-integrins, these tumors are effectively transduced by AdV5 vectors in vitro and in vivo. ¹³ The current understanding of adenoviral transduction is not complete and cannot be predicted from knowledge of receptor and integrin expression alone. 

Transgene expression in photoreceptors as measured by the percentage of GFP positive cells remained stable for at least 8 months in both immune competent and immune deficient mice. This is in marked contrast to transgene expression delivered by adenoviral vectors to the systemic environment where expression is measured for only a few weeks. The short-term expression in the nonocular environment is thought to be due to immune responses directed against expressed adenoviral antigens in early generation adenoviral vectors. ^{35 36 37} The eye is immune tolerant of antigens including first generation adenoviral vectors injected into the anterior and vitreous chambers. ^{38 39} The ability to achieve long-term transgene expression in the ocular environment suggests that adenoviral vectors could play a role in gene replacement therapy for photoreceptor degenerative diseases. The decrease in the percentage of RPE cells expressing GFP may be explained by the ability of RPE cells to proliferate in contrast to retinal cells. ⁴⁰ Adenoviral vectors maintain DNA separate from the host genome and the viral DNA is therefore not replicated and passed to the daughter cells. ⁴¹  

High in vitro and in vivo transduction efficiencies as well as high levels of transgene expression indicate that the AdV5/F35 chimeric vector may be superior to the parent AdV5 vector for ocular gene therapy applications involving cells of retinal origin. Depending on the target cell, there are now several adenoviruses that can be used selectively in the ocular environment. AdV5-mediated gene therapy is useful for treating retinoblastoma. ¹³ AdV5 may be particularly useful for this disease since the virus targets the tumor but not the retina. Chimeric viruses such as AdV5/F35 vectors may be useful for treating retinal diseases and diseases of the RPE. Careful selection of adenoviral constructs for targeted delivery of therapeutic transgenes to the ocular environment may offer opportunities for the specific treatment of a wide variety of ocular diseases. 

 Supported by the National Institutes of Health (CA97762), the Foundation for Research, the Retina Research Foundation, and the Knights Templar.

 Submitted for publication July 13, 2003; revised December 29, 2003 and February 5, 2004; accepted February 13, 2004.

 Disclosure: J.N. Mallam, None; M.Y. Hurwitz, None; T. Mahoney, None; P. Chévez-Barrios, None; R.L. Hurwitz, 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.

 

Expression of CAR and CD46 on Weri retinoblastoma cells. Cultured Weri human retinoblastoma cells (1 × 10⁶) were harvested and incubated with the second antibody only (Control) or with the anti-human CAR mouse monoclonal primary antibody RmcB or mouse monoclonal antibody to CD46 followed by a goat anti-mouse fluorescein-conjugated second antibody. CAR and CD46 were confirmed to be present on the Weri cells by FACS analysis.

Figure 2.

 

Immunocytochemical analysis of CAR, CD46, and α_v integrin expression in retinoblastoma. Human retinoblastoma tumors were examined by immunohistochemistry using antisera against CAR (A), mouse monoclonal antibody against CD46 (B), or antisera against α_v integrin (C). The tumors stained positive for CAR and CD46 (especially in rosettes) expression but negative for the α_v integrin. Magnification, ×40.

Figure 3.

 

Immunocytochemical analysis of CAR, CD46, and α_v integrin expression in murine retina. Murine retina was analyzed by immunohistochemistry using anti-CAR (A), anti-CD46 (B), anti-α_v-integrin (C), or second antibody alone (D). CAR is expressed in ganglion cells (GC), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL) especially at the synaptic region, inner segments of photoreceptors (IS), and retinal pigment epithelium (RPE). The outer segments of photoreceptors (OS) do not appear to express CAR. There are positive cytoplasmic processes from Müller cells traversing through the outer nuclear layer (ONL). CD46 expression was limited to IS, OS, and RPE. α_v-Integrin expression was detected in the GC, IPL, INL, OPL, IS, and RPE and only few cells in the ONL but not in the OS. Magnification, ×40.

Figure 4.

 

Comparison of transduction efficiency of AdV5-GFP to AdV5/F35-GFP in Y79 retinoblastoma cells. (A) The Y79 human retinoblastoma cells were transduced by AdV5-GFP (○) or AdV5/F35-GFP (•) at the indicated concentration of infectious units. After 48 hours, cells were harvested and the percentage of cells expressing GFP was quantitated by FACS analysis. There was no significant difference in the percentage of GFP-positive cells transduced by either vector. (B) Y79 human retinoblastoma cells transduced by AdV5-GFP (○) or AdV5/F35-GFP (•) at the indicated concentration of infectious units were harvested 48 hours post-transduction. The relative fluorescence intensity was used to indicate the amount of the reporter gene (GFP) expressed in the cells transduced by either virus. The amount of GFP expressed by the AdV5/F35-GFP vector at 1 × 10⁷ IU/mL was sixfold greater than the amount of GFP expressed by the AdV5-GFP vector.

Figure 5.

 

Comparison of transduction efficiency of AdV5-GFP to AdV5/F35-GFP in murine retina. Mice were given subretinal injections of either AdV5-GFP or AdV5/F35-GFP (10⁷ IU). Enucleated eyes were examined 2 weeks after injection by immunohistochemistry using anti-GFP and peroxidase-conjugated second antibody as described in the Methods section. A sham-injected eye shows no staining of any retinal cell layers with anti-GFP (A). Retina pigment epithelium (RPE) stains reddish brown near the site of AdV5-GFP injection. These RPE cells appear enlarged and hyperplastic (B). GFP is also detected in Müller cells (MC) and occasional full-length cell bodies and nuclei of photoreceptors at the site of injection (C). Rag-2 and C57BL/6 mice examined 2 weeks after injection with AdV5/F35-GFP show reddish-brown staining in the photoreceptors and RPE beyond the injection sites (arrows). The retinal detachment is an artifact of processing as is the light brown staining in the lens (L) (D, E). Few nerve fibers in the optic nerve also stain (arrow) (F). Higher magnification indicates labeling of the entire photoreceptor including the outer segments (OS), inner segments (IS), nucleus (ONL), and synaptic zone in the outer plexiform layer (OPL) (G). Sections were also double labeled with anti-GFAP (stained brown), a glial cell marker, and anti-GFP (stained red). A Müller cell can be identified in the inner retinal layer that stains for both GFAP (brown) and GFP (red). The arrow points to a Müller cell foot process that stains positive for both chromogens. Cells containing large nuclei (ganglion cells) and the cells in the inner nuclear layer are negative for both GFAP and GFP (H). The green arrows point to brown-labeled glial cells in the ganglion cell layer (GC) that weakly stain red, suggesting the presence of GFP. Black arrows point to two cells in the inner nuclear layer (INL) showing the presence of GFP but not GFAP (I). These cells are possibly horizontal cells because they contain dendrites directed toward photoreceptors. The RPE shows transduced cells staining bright red and nontransduced cells with football-shaped brown melanin granules. The underlying choroid (C) contains dark brown-pigmented melanocytes (J). Magnification: (A, B) ×20; (C, G–J) ×40; (D, E) ×4; (F) ×10.

Table 1.

 

Analysis of AdV5/F35-GFP Transgene Expression in Immune Competent and Immune Deficient Mice

Strain	n	Time (months)	RPE	Nuclear Layers	Müller Cells	Photoreceptors	Neuronal Cells
C57BL/6	2	0.5–1	80% (r: 70%–99%)	5% (r: 1%–10%)	15% (r: 1%–30%)	25% (r: 5%–40%)	1% (r: 0%–1%)
Rag-2	2	0.5–1	90% (r: 80%–99%)	5% (r: 5%)	5% (r: 5%)	60% (r: 30%–90%)	1% (r: 1%)
C57BL/6	4	7–8	30% (r: 10%–40%)	5% (r: 1%–10%)	5% (r: 1%–5%)	20% (r: 10%–30%)	5% (r: 1%–10%)
Rag-2	4	7	35% (r: 10%–40%)	10% (r: 1%–10%)	5% (r: 1%–10%)	20% (r: 5%–25%)	10% (r: 1%–10%)

 Percentage of cells was determined by counting the number of cells labeled by a GFP antibody (as described in Methods) in the vicinity of the injection site. The percentages are averages of the number (n) of animals evaluated. The range of observations (r) is given.

The authors thank Juan-Ru Lin for her expert technical assistance with the subretinal injections.