Long-Term, Targeted Genetic Modification of the Aqueous Humor Outflow Tract Coupled with Noninvasive Imaging of Gene Expression In Vivo

Abstract

purpose. To address a problem impeding research into glaucoma-associated genetic mutations and glaucoma gene therapy and achieve permanent, targeted transgene expression in the trabecular meshwork (TM). Lentiviral vectors are known to transduce human donor eye TM ex vivo, but efficacy in vivo has not been shown. More generally in the field of gene therapy, the authors hypothesized that distinctive properties of the intraocular aqueous circulation could facilitate solving problems of accessibility, targeting, and scale that have hindered realization of gene therapy in other settings.

methods. A domestic cat model was developed in which long-term in vivo studies were performed. After dose-response studies in primary human TM cells, 19 cats received anterior chamber (AC) injections of stepped doses (10⁶–10⁸ transduction units) of lentiviral vectors encoding different marker transgenes (β-galactosidase, Aequorea victoria green fluorescent protein [GFP], or Renilla reniformis GFP). Animals were monitored serially for transgene expression and IOP.

results. High-grade, stable transgene expression in the TM was achieved and monitored noninvasively over time in living animals. Extensive expression resulted after a single transcorneal injection, persisted for at least 10 months (time of death in the present studies), and was targeted to the TM. The initial IOP did not differ significantly from the IOP at the end of the study (P = 0.4). Aequorea GFP was superior to Renilla GFP. Vectors were effective enough to cause GFP-specific overexpression cytotoxicity at the highest dose, which was solved by dose reduction.

conclusions. High-grade transgene expression in this large-animal model persisted stably for at least 10 months after a single transcorneal lentiviral vector injection, was highly targeted, and could be monitored serially and noninvasively in living animals. These studies provide a basis for developing realistic disease models and administering glaucoma gene therapy.

Glaucoma, a group of optic neuropathies that afflict 70 million people, is a leading cause of irreversible blindness. ¹ Elevated intraocular pressure (IOP) is the major causal risk factor. Risk of blindness remains high because of the partial efficacy and compliance burden of existing therapies. ² A therapy that could correct glaucoma pathophysiology permanently would be desirable. Permanent genetic reprogramming of anterior chamber outflow tract physiology by gene therapy has attracted attention as an ideal solution in theory because of the disease’s life-long chronicity and the emerging understanding of its genetic basis. ^{3 4 5 6} The trabecular meshwork (TM) is a key intraocular structure to target, because it controls IOP by controlling outflow of aqueous humor. ^{7 8}  

Additional features make glaucoma an intriguing potential proof-of-concept disease for gene therapy. The small amount of tissue that would require targeting, and its accessibility, could enhance the feasibility of adequate levels of corrective gene transfer. In the absence of a selective growth advantage for gene-altered cells, and/or a tissue that supports proliferation in vivo after ex vivo gene transfer, ⁹ achieving permanent transduction of most of a relevant tissue has remained a major hurdle in most gene therapy situations. Specific tissue targeting is problematic if the target cells cannot be isolated ex vivo (e.g., Ref. ⁹ ). The anterior chamber can also be visualized through clinically feasible imaging methods, suggesting a means of monitoring the in vivo expression of an integrated transgene over time during gene therapy developmental studies. 

Long-term, stable, high-grade, and properly targeted transgene expression in the TM has not been achieved. Liposomes, ¹⁰ adenovirus, ^{11 12} adenoassociated virus, ¹³ and herpes simplex virus vectors ¹⁴ have been limited by short duration, inflammation, or lack of sufficient targeted transduction. In contrast to DNA virus- or plasmid-based vectors, lentiretroviral vectors integrate permanently into the genomes of transduced cells as an obligate part of the life cycle. ¹⁵ The advantage over conventional oncoretroviral vectors is their ability to integrate in nondividing cells. Consistent with the metabolically active yet mitotically quiescent phenotype of the endothelial-like cells that the TM comprises, lentiviral vectors have been shown to transduce the TM of explanted, perfused human anterior chambers without detectable toxicity, while oncoretroviral vectors do not. ¹⁶ Lentiviral vectors derived from human or feline immunodeficiency viruses (HIV-1 or FIV) are equally efficacious. ^{16 17} However, the organ explant model is artificial in many respects, such as short viability with consequent inability to assess long-term outcomes, substitution of the complex aqueous humor with synthetic medium, and the complete lack of a systemic immune response. In rodents, the outflow tract is too rudimentary to permit robust comparisons. We therefore proceeded to assess the possibility of long-term targeted gene expression with the required characteristics in the cat, a readily available larger animal with an outflow tract that is anatomically and physiologically similar to that of the human. 

Methods

FIV Vectors

Vesicular stomatitis virus glycoprotein G-pseudotyped FIV vectors were generated with pMD-G, ¹⁸ packaging plasmid pFP93, ¹⁸ and three different FIV transfer vectors: pGINWF, ^{17 18} which encodes the red-shifted, human codon–optimized version of green fluorescent protein (i.e., eGFP) from Aequorea victoria ¹⁹ ; pRGWF, which encodes codon-optimized GFP from Renilla reniformis ^{20 21} (Vitality hrGFP Mammalian Expression Vector, cat. no. 240062; Stratagene, La Jolla, CA); and pCT26, which encodes β-galactosidase (lacZ). ^{17 18} Vectors were produced as described previously. ^{17 18 22} Titers of GINWF and RGWF vector preparations were determined by flow cytometry and that of CT26 by β-gal staining of transduced cells. ¹⁸ All preparations were tested for reverse transcriptase (RT) activity. 

Cultured TM Cell Transduction

Primary human TM cells (gift of Terete Borras, University of North Carolina, Chapel Hill, NC) were transduced with an escalating multiplicity of infection (MOI) of GINWF to test for potential GFP toxicity in vitro before in vivo experimentation. Doses were chosen to have an MOI equivalent to that in a feline eye injected with a bolus of 10⁶, 10⁷, 10⁸, and 10⁹ transduction units (TU). ^{23 24} The medium was replaced weekly, and TM cell layers were photographed (Coolpix 990 and Eclipse TE300; Nikon, Melville, NY) at the experimental end point of 1 month. 

Animals and Group Assignments

Animals in this study were handled in accordance to the Institutional Animal Care and Use Committee and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Experiments were conducted in specific pathogen-free domestic cats (Harlan, Indianapolis, IN). Ten days before vector application, cats were anesthetized with 10 mg/kg intramuscular tiletamine HCl/zolazepam HCl (Telazol; Fort Dodge Laboratories Inc., Fort Dodge, IA) for ocular examination with slit lamp (Haag-Streit, Mason, OH) and a handheld pneumatonometer (Model 30 Classic; Medtronic, Fridley, MN) to determine IOP. Fluorescence of transduced TM was observed with a standard gonioscope (Posner; Ocular Instruments, Bellevue, WA) and a microscope (Eclipse E400; Nikon) equipped with a GFP-optimized filter (EF-4 B-2E/C FITC, cat. no. 96107; Nikon). Two days before vector application, cats received a bilateral conjunctival injection of 200 μL triamcinolone acetonide (Kenalog-40; Bristol-Myers Squibb, Princeton, NJ) to blunt nonspecific inflammatory reactions during initial intraocular procedures, as were seen after saline injection in pilot studies. Anterior chambers of feline eyes were injected with a bolus of 50 μL PBS (Dulbecco’s PBS; Cellgro; Mediatech, Herndon, VA) containing 10⁸, 10⁷, or 10⁶ TU of vectors RGWF, GINWF, or CT26. 

When 7 pilot animals given 10⁸ TU of GFP vector showed intense expression of GFP in the entire TM, 12 more animals were assigned to dose groups (group 1, 10⁸ TU; group 2, 10⁸ TU; group 3, 10⁷ TU; and group 4, 10⁶ TU) in which each group of 3 animals was injected transcorneally with different vectors (see text and Table 1 for vector identities) in each eye. 

Noninvasive Monitoring of Transgene Expression In Vivo

Anesthetized cats were examined with slit lamp, pneumatonometer, and biomicroscopy on the day after arrival; 3 days before, on the day of, and 3 days after vector application; and weekly for 2 months, biweekly for 4 months, and monthly thereafter. IOP was always measured at the same time of day; three readings per eye were taken and averaged. The entire TM circumference was inspected. Photographs of fluorescent TM were taken at each examination and grades determined in a masked manner by two observers rating independently, using the scale photographs shown in Figure 1B . A digital microscopy camera (DXM 1200; Nikon, Melville, NY) was used, with 2-second exposures and image capture software (Automatic Camera Tamer [ACT-1]); Nikon). 

Quantification and Statistics

In vivo TM fluorescence, as seen during gonioscopy was graded on a scale of 0 to 4, with representative photographs shown in Figure 1B . Grade 0 was defined as no detectable fluorescence; grade 1 as single fluorescent spots in the TM; grade 2 as numerous, nonconfluent fluorescent spots with some confluent areas; grade 3 as extensive, mostly confluent, midlevel transduction; and grade 4 as extensive, high-level and completely confluent fluorescence. Because of the high linear correlation, expression grades were handled as continuous data during the statistical comparison of marker protein accumulation. To determine whether the rate of fluorescent marker protein accumulation (V _F) influenced the duration of expression, V _F was measured as the change from initial expression grade (grade_I) to peak expression grade (grade_P) over time from the first examination (t ₁) to the time point of peak expression (t _P): V _F = Δ grade/Δ time = (grade_P − grade_I)/(t _P − t ₁). Time was measured in days. A large increase in expression grade over a short period would result in a larger V _F than a small increase in the same period. IOP readings before vector injection were compared to peak pressures of the same eye as well as to the experimental end point, with the paired Student’s t-test. 

Histologic Assessment and Immunolabeling

Animals were killed with an injection of 175 mg/kg pentobarbital sodium intravenously (Sleepaway; Fort Dodge Laboratories, Sligo, Ireland) and eyes were enucleated. To correlate gonioscopic expression grades with transduction efficiency, three quadrants in one eye per expression grade were analyzed. One random quadrant was removed from the anterior segment for X-Gal staining, and the remaining three were incubated with PBS and 1000-fold–diluted Hoechst 33342 (Molecular Probes, Eugene, OR) to counterstain nuclei of living cells. The transduction efficiency was determined by counting GFP-positive, living TM cells and Hoechst 33342–positive nuclei in the same visual field at 100× magnification with a histology microscope (Eclipse E400; Nikon). For a more sensitive assessment of the extent of transduction with GINWF, other eyes were labeled with an anti-GFP antibody, as described. ^{16 17} Eyes transduced with CT26 were X-Gal stained overnight, photographed with an operating microscope (SMZ800 and DXM 1200, with ACT-1 software; Nikon) and embedded in paraffin. Sections were counterstained with fast nuclear red. 

Anterior segment sections from cats in which iritis developed after GFP overexpression were analyzed for presence of feline macrophages ^{25 26 27} with a primary murine anti-myeloid/histiocyte antibody (1:50 dilution, MAC 387, M0747; Dako, Carpinteria, CA) and a secondary Alexa 488–labeled goat anti-mouse antibody (1:100 dilution, A-11001; Molecular Probes). To detect feline T-cells, ^{25 26 28} sections were incubated with peroxidase-conjugated anti-CD3 (anti-CD3/HRP, U0026; Dako), according to protocol. A horseradish peroxidase-labeled immunoglobulin mix (EPOS Negative Control solution, U0951; Dako) was used as a control. 

Sera of cats with proven iritis infiltrate were analyzed for antibodies against GFP and VSV-G. A 10% SDS-PAGE gel with a continuous well was loaded with GFP protein produced by transient transfection of 293T cells and transferred. Membrane strips were incubated with serum before transduction and 4 weeks after animals iritis had developed. A rabbit anti-GFP antibody (1:1000 dilution; NB 600-303; Novus Biologicals, Littleton, CO) and a monoclonal mouse anti-VSV-glycoprotein antibody (1:1000 dilution, protein clone P5D4, C-7706; Sigma-Aldrich, St. Louis, MO) served as a positive control. After washing, wells with feline serum as the primary incubation were incubated with a secondary peroxidase-conjugated goat anti-cat antibody (1:1000 dilution; cat. no. 55293; IGN Pharmaceuticals, Aurora, OH). A peroxidase-conjugated goat anti-rabbit antibody (1:1000 dilution; Calbiochem, San Diego, CA) served as the secondary antibody for GFP detection and a peroxidase-conjugated goat anti-mouse antibody (1:1000 dilution; Calbiochem) as the secondary antibody to the anti-VSV-G antibody. The same protocol was used to detect anti-VSV-G antibodies, except that 293T cells were transfected with VSV-G expression plasmid pMD-G. 

Results

Primary Human TM Cell Transduction

Because enabling real-time noninvasive monitoring was a central goal, GFP-expressing vectors were the principal focus of the study. For preliminarily assessment of feasibility and potential toxicity of GFP expression, cultured primary human TM cells were transduced with a 4-log range of FIV vector input (Fig. 1A) . Transduction at the MOI, which we estimated to be the approximate equivalent of 10 times the maximum in vivo multiplicity used in the animal experiments that follow, produced marked GFP overexpression in all TM cells (Fig. 1A) without morphologic change, detachment, vacuolization, change in doubling time, or other evidence of vector or transgene toxicity (data not shown). The plateau of GFP fluorescence was reached within 24 to 48 hours. As shown by the 1-month post-transduction results in Figure 1A , this transgene expression was stable. When Renilla GFP (from the sea pansy ^{20 21} ) was substituted for Aequorea ^{19 20} GFP, cells transduced with equivalent MOI appeared paler and reached peak fluorescence later, at 48 to 72 hours (data not shown). Unless otherwise specified below, GFP designates the Aequorea protein. 

Transgene Expression In Vivo

Nineteen domestic cats were used in this study. Seven were used as pilot animals to establish techniques and assess feasibility of gene transfer with eGFP vector, using what was later established to be a highly effective dose (10⁸ TU per eye). When these animals showed extensive expression of GFP in the TM, 12 more cats were assigned to groups organized by dose level and marker gene in a design that used comparisons of different marker transgenes in right and left eyes of the same animal (see the Methods section and Table 1 for animal group assignments). Vectors were injected transcorneally into the anterior chambers of lightly anesthetized animals by single 50-μL bolus injections through a 27-gauge needle. Animals injected with 10⁷and 10⁸ TU of GFP-transducing vector showed high-grade (grade 4; see Fig. 1B for grading scale), persistent GFP expression in the TM that was readily visualized and photographed through a conventional gonioscope (Figs. 1B 2) . Notably, expression was confined to the TM. Numerical results are summarized by group in Table 1 . GFP expression in the TM steadily increased after transduction, reaching a maximum within days to weeks. It then was observed in most animals to plateau at the same or a lower grade, where it persisted for at least 10 months, at which time the animals were killed (Fig. 3A ; Aequorea GFP panels). There was a high linear correlation (R ² = 0.9) of gonioscopic expression grade and histologically determined transduction efficiency, with antibody labeling for GFP, confirming that expression was limited to the TM (Fig. 4) . Overlay with 4′,6′-diamino-2-phenylindole (DAPI; nuclear) staining demonstrated normal cellularity in both highly transduced and untransduced TM in the same animal (Fig. 4) . Of note, the expression grade data in this study were scored in a masked manner by two independent observers who used the scale photographs shown in Figure 1B , did not participate in the experiments, and were blinded to vector assignments. The robustness of this rating scale is evident in the outcome that both of the masked observers agreed independently on each scored expression grade, except for one point in which a grade of 2 was assigned by one observer and 3 by the other. In this one instance, a grade of 2 was recorded. 

Temporal and spatial patterns of cats in group 1, which were injected with 10⁸ TU GFP vector in one eye and 10⁸ TU Renilla GFP vector in the other eye were consistent with the human TM cell culture outcomes described earlier. Companion eyes were initially similar (Figs. 3A 5A , top panel). However, eyes expressing Renilla GFP showed smaller median expression (maximum grade 2 compared with 4 in the GFP-injected eye of the same animal, P = 0.04, Table 1 ). Renilla GFP expression was also shorter lived than GFP expression (35 ± 22 days versus >10 months). High-level GFP expression remained stable in two of the three group 1 animals (grade 2 in one, grade 3 in the other) to the end point of the study (i.e., death at 10 months). The third animal also had grade 4 GFP expression initially. Strikingly, however, the intense GFP fluorescence was observed to disappear abruptly on day 38, concurrent with brief iritis (discussed later). The rate of marker protein accumulation V _F, expressed as Δ grade/time was 0.2 ± 0.1 grades/d for GFP and 0.2 ± 0.1 grades/d (P = 0.4) for Renilla GFP (Table 1) . 

These results indicated that high-level and sustained focal transgene expression in the TM was feasible after a single transcorneal injection, that it could be monitored gonioscopically, and that GFP (from Aequorea) was preferable to Renilla GFP. Because of the high level of GFP expression achieved with 10⁸ TU, we considered that the loss of expression in one of these maximally transduced animals might be due to the previously reported phenomenon of GFP-specific overexpression toxicity. ^{19 20 29 30 31 32 33 34 35} We therefore proceeded to examine further the parameters affecting GFP expression in dose-response studies, this time using β-galactosidase vector ¹⁷ as a control in the companion eye rather then Renilla GFP vector (Table 1) . In group 2 (10⁸ TU for each vector), the extent of GFP and β-galactosidase expression were comparable (Fig. 3B) , and all animals showed sustained grade 4 GFP expression. The plateau of GFP expression was attained five times faster (Fig. 5B , Table 1 ) than in the animals of group 1 (P = 0.03). In group 2, the rate of marker protein accumulation V _F was 1.1 ± 0.5 grades/d. Consistent with toxicity from rapid GFP overexpression, the intense fluorescence in all group 2 animals was observed to terminate abruptly after a mean of 12 ± 5 days, concurrent with brief iritis. In contrast, and consistent with the results of β-galactosidase vector transduction in human eyes, ^{16 17} high-level β-galactosidase expression was found in the same animals at death after GFP expression had disappeared in the other eye (Fig. 3B) . Thus, the cytotoxicity observed was marker-protein-specific and the result of rapidly developing GFP overexpression. Immunolabeling with leukocyte-lineage–specific antibodies in histologic sections of eyes of cats exhibiting this short-lived GFP overexpression showed numerous T cells, but not neutrophils or macrophages, in the TM, as well as loss of TM cells (data not shown). Sections of Renilla GFP-expressing eyes also demonstrated occasional T-cells. No infiltrates were present in companion eyes expressing β-galactosidase. Sera of cats with short-term expression did not react in immunoblots with GFP, which was readily detected with commercial anti-GFP sera (data not shown). 

These results were encouraging, because they indicated that the method was effective enough to enable marked protein overexpression in the TM. We therefore administered reduced vector doses (animal groups 3 and 4, receiving 10⁷ and 10⁶ TU, respectively, by single transcorneal injections). A dose of 10⁷ TU also produced a median expression grade of 4, which was not significantly different from group 2 (P = 0.37). However, at this lower dose, the grade 4 GFP expression was reached noticeably slower than in group 2 (V _F was 0.2 ± 0.03 grades/d, Table 1 ). A clear difference between groups 2 and 3 was that GFP expression persisted until death at the 10-month postinjection end point of the study, and remained grade 4 throughout. Thus, 10⁷ TU emerged as the optimal dose of the GFP vector, whereas β-galactosidase was tolerated well at 10⁸ TU. Group 4 animals (10⁶ TU GFP versus β-galactosidase vector, or 2 log less vector input than animals in groups 1 and 2) corroborated these results and clarified dose requirements. They had a maximum median expression grade of 3, with a V _F of only 0.1 ± 0.1. The expression persisted at grade 2 until death at 10 months. The results indicate that an optimal GFP vector dose substantially below the peak of the achievable dynamic range can be selected (10⁷ TU in the case of this GFP vector) to achieve long-term gene expression in the TM. 

Intraocular Pressure

The initial IOP of all transduced eyes was not different from IOP at the end of the study (P = 0.4). Days to weeks after application, cats had a brief increase in IOP (33% ± 34%, P < 0.01) followed by return to normal pressure, which may be consistent with corticosteroid-induced ocular hypertension. ^{36 37} In support of this inference, IOP change in nonpremedicated pilot animals was minimal with GFP vector (n = 3, IOP change = 3% ± 5%, P = 0.4) or β-galactosidase vector (n = 3, IOP change = 1% ± 1%, P = 0.4). 

Discussion

Gene therapy for glaucoma and proper investigation of glaucoma-associated genetic mutations requires a method that is simple to administer and provides long-term, stable, high-grade, and properly targeted transgene expression within the anterior chamber. More generally, several long-standing technical obstacles that concern scale and accessibility are well recognized to have thwarted the clinical realization of many forms of gene therapy. Among the most daunting has been achieving an adequate level of gene transfer and gene expression in most cells of a physiologically relevant target tissue. When target cells cannot divide or when there is no inherent selection pressure driving preferential expansion of a population of gene-altered cells, as in the case of X-linked severe combined immunodeficiency (SCID), ⁹ and when expression must persist for the lifetime of the host, this obstacle has been especially problematic. 

We applied lentiviral vectors to this distinctive problem of glaucoma gene therapy. We were able to target the relevant tissue, the TM, by using the natural fluid dynamics of the anterior chamber outflow pathway after transcorneal injection. By focusing on the tissue involved in this disease, with its advantages of scale and accessibility, we enabled long-term, serial in vivo imaging of transgene expression in permanently genetically modified living animals. Stable transgene expression that persists for at least 10 months was enabled. The effectiveness of transduction was such that at the highest doses, rapidly accumulating GFP was seen to result in toxicity (i.e., these animals were effectively overtransduced), while administration of equivalent β-galactosidase vector also produced high-level expression without toxicity. Toxicity of GFP is well-described, particularly at maximum expression levels, and appears to be cell type dependent. ^{19 20 29 30 31 32 33 34 35} Because eGFP is not amplifiable (one molecule equals one fluorophore, whereas enzymatic markers can convert many molecules of substrate to detectable dye), at least 10⁵ molecules per cell are needed for the cell to yield twice the background fluorescence. ¹⁹ This estimate is a lower boundary, as it assumes perfect eGFP maturation; incomplete maturation would raise the threshold higher. ¹⁹ Undoubtedly, many more than 10⁵ molecules per cell are needed to image through the cornea and to produce the high-grade fluorescence observed in this study. That lower doses circumvented this problem while resulting in GFP expression that was still readily imaged in the living animal makes clear that a high dynamic range of gene delivery and expression has been achieved. Consistent with results of others, ²¹ the use of Renilla GFP did not offer advantages in the TM. 

The preferential transduction of the TM over other ocular structures in these animals is consistent with our previous results in perfused human anterior chambers. ^{16 17} This targeting may be an in vivo correlate of the enhancement produced by perpendicular convective flow of retroviral vector supernatants through target cell monolayers grown on porous substrates. ³⁸ In addition, the tropism of these VSV-G pseudotyped lentiviral vectors may favor certain cell types, as seen for example in the subretinal space, where the retinal pigment epithelium is preferentially transduced. ^{39 40} Thus, aqueous humor flow dynamics and cell-specific permissiveness of transduction by these vectors may both be factors. As in human donor eyes, ^{16 17} high-level transduction was performed without detriment to aqueous humor outflow or the histologic fine structure, indicating that these replication-defective lentiviral vectors have favorable toxicity profiles for further investigation of glaucoma gene therapy. 

There is a critical need for a realistic animal model for glaucoma. The results establish that prolonged, targeted, and multigrade expression of transgenes in the disease-relevant tissue is now possible. This approach appears promising for testing candidate disease-associated gene alleles as well as therapeutic genes in the TM. 

 Supported by a Fight For Sight research grant (NL), National Institute of Allergy and Infectious Diseases Grant AI47536 (EMP), and National Eye Institute Grants EY14411 (EMP) and EY07065 (DHJ).

 Submitted for publication April 1, 2004; revised May 17, 2004; accepted May 30, 2004.

 Disclosure: N. Loewen, None; M.P. Fautsch, None; W.-L. Teo, None; C.K. Bahler, None; D.H. Johnson, None; E.M. Poeschla, 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.

 

Table 1.

 

Characteristics of Transgene Expression Imaged Serially In Vivo

Group	Cats (n)	Vector	Median Expression Grades			Duration	V F = \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\frac{{\Delta}Grade}{Day}\) \end{document}	IOPstart (mm Hg)	IOPend (mm Hg)	IOPstart vs. IOPend
											Max.	Final
1	3	R = 108 TU GFP	4	2*	>10 months*	0.2 ± 0.10	20 ± 5.0	15 ± 10	P = 0.2
		L = 108 TU rGFP	2	0	35 ± 22 days	0.2 ± 0.10	20 ± 5.0	25 ± 11	P = 0.03
2	3	R = 108 TU GFP	4	0	12 ± 5 days	1.1 ± 0.5	18 ± 4.0	12 ± 3	P = 0.6
		L = 108 TU β-gal	NA	NA	>12 ± 5 days	NA	22 ± 6.0	15 ± 3	P = 0.08
3	3	R = 107 TU GFP	4	4	>10 months	0.2 ± 0.03	22 ± 4.0	23 ± 5	P = 0.9
		L = 107 TU β-gal	NA	NA	>10 months	NA	22 ± 3.0	25 ± 4	P = 0.07
4	3	R = 106 TU GFP	3	2	>10 months	0.1 ± 0.1	19 ± 0.3	20 ± 6	P = 0.9
		L = 106 TU β-gal	NA	NA	>10 months	NA	19 ± 3.0	18 ± 3	P = 0.06

*  One animal in this group terminated expression at 38 days.

Figure 1.

 

Vector dose escalation experiments. (A) Dose-escalation experiment with GINWF in primary cultured human TM cells. Shown are contact-inhibited TM cells at 1 month after transduction. The vector encodes (codon-humanized) eGFP under the transcriptional control of the human CMV promoter. No toxicity was observed at any dose. (B) GFP fluorescence in the TM of living cats was serially photographed and graded by direct gonioscopy (a cornea-shaped lens with a mirror is used to enable visualization of the anterior chamber angle). There was a high linear correlation (R ² = 0.9) between expression grade and histologic transduction efficiency (grade 1, 2% ± 0.3%; grade 2, 21% ± 1%; grade 3, 39% ± 3%; and grade 4, 94% ± 5%).

Figure 2.

 

View of TM in a living animal in standard (top and bottom) and corresponding UV light gonioscopic views (middle). All four quadrants from an eye with grade 4 expression after transduction with 10⁸ TU are shown. Transduction was virtually complete and was confined to the TM. SN, superonasal; IN, inferonasal; IT, inferotemporal; ST, superotemporal.

Figure 3.

 

Representative examples of reporter gene expression in animals in groups 1 and 2. (A) GFP (right eye) and Renilla GFP (left eye) expression in group 1. (B) Paired comparison of GFP (right eye) and β-galactosidase expression (left eye) in group 2. Shown are photographs of TM as seen by UV light gonioscopy (the 13-day β-galactosidase eye was fixed and stained with X-gal after death). Detectable Renilla GFP expression was always shorter lived than that of GFP, whereas β-galactosidase expression persisted longer than GFP only at the highest dose, when overexpression toxicity of GFP was observed.

Figure 4.

 

Antibody labeling for GFP confirms gonioscopic extent of transduction. GFP expression was limited to the TM and collector channels. Comparison DAPI staining of transduced (left) and control (right) TM demonstrated preserved cellularity in GFP-expressing TM. AC, anterior chamber; P, plexus.

Figure 5.

 

The rate of initial GFP accumulation determined whether successful long-term GFP expression would be established. (A) Expression slowly reached peak levels in two cats from group 1 transduced with 10⁸ TU GFP vector (right eye) and Renilla GFP vector (left eye), resulting in stable fluorescence of GFP, but not Renilla GFP. (B) Rapid GFP overexpression in animals of group 3 that were also transduced with 10⁸ TU GFP vector triggered iritis, which eliminated transduced cells. Dotted red line: onset of expression. Insets: actual gonioscopic view of TM, as indicated.

The authors thank Mary Peretz for technical assistance.