Different tropism of adenoviruses and adeno-associated viruses to corneal cells: implications for corneal gene therapy

PURPOSE: Diseased corneas are potential targets for viral-based gene therapy to normalize (stimulate or inhibit) the expression of specific proteins. The choice of viral vectors is important to achieve optimal effect. The purpose of this study was to compare the tropism to different corneal cells of recombinant adenovirus (rAV) and recombinant adeno-associated virus (rAAV) constructs using live rabbit and organ-cultured human corneas. METHODS: rAV constructs harbored the green fluorescent protein (GFP) gene under the control of major immediate early cytomegalovirus (CMV) promoter. rAAV constructs from virus serotypes 1, 2 5, 7, and 8 had GFP under the chicken β-actin promoter and CMV enhancer. For organ culture, 16 healthy and diabetic postmortem human corneas were used. Five or fifteen μl rAV at 10(7) plaque forming units per 1 μl were added for 2 days to culture medium of uninjured corneas that were further cultured for 5–32 days. rAAV were added at 1.2–7.8×10(10) vector genomes per cornea for 3 days to each cornea; the culture then continued for another 14–23 days. Corneal cryostat sections were examined by immunohistochemistry. Live rabbit corneas were used following excimer laser ablation of the corneal epithelium with preservation of the basal cell layer. Equal numbers of rAAV particles (2x10(11) vector genomes) were applied to the cornea for 10 min. After seven days to allow for corneal healing and gene expression the animals were euthanized, the corneas were excised, and sections analyzed by immunohistochemistry. RESULTS: By direct fluorescence microscopy of live organ-cultured human corneas GFP signal after rAV transduction was strong in the epithelium with dose-dependent intensity. On corneal sections, GFP was seen in all epithelial layers and some endothelial cells but most keratocytes were negative. In rAAV-transduced organ-cultured human corneas GFP signal could only be detected with anti-GFP antibody immunohistochemistry. GFP was observed in the epithelium, keratocytes, and endothelium, with more pronounced basal epithelial cell staining with rAAV1 than with other serotypes. No difference in the GFP expression patterns or levels between normal and diabetic corneas was noted. The rabbit corneas showed very similar patterns of GFP distribution to human corneas. With all rAAV serotype vectors, GFP staining in the epithelium was significantly (p=0.007) higher than the background staining in non-transduced corneas, with a trend for rAAV1 and rAAV8 to produce higher staining intensities than for rAAV2, rAAV5 (p=0.03; rAAV5 versus rAAV1), and rAAV7. rAAV serotype vectors also transduced stromal and endothelial cells in rabbit corneas to a different extent. CONCLUSIONS: rAAV appears to reach many more corneal cells than rAV, especially keratocytes, although GFP expression levels were lower compared to rAV. rAV may be more useful than rAAV for gene therapy applications requiring high protein expression levels, but rAAV may be superior for keratocyte targeting.