Adenoviral Gene Transfer of Active Human Transforming Growth Factor-β2 Elevates Intraocular Pressure and Reduces Outflow Facility in Rodent Eyes

Abstract

Purpose.: Glaucoma is a leading cause worldwide of blindness and visual impairment. Transforming growth factor-β2 (TGFβ2) has been implicated in the pathogenesis of primary open-angle glaucoma (POAG) based on elevated levels in glaucomatous aqueous humor and its ability to induce extracellular matrix (ECM) remodeling in the trabecular meshwork (TM). The goal of this study was to generate a rodent model of POAG using viral gene transfer of human TGFβ2.

Methods.: Latent (hTGFβ2^WT) or active (C226S, C228S; hTGFβ2^226/228) TGFβ2-encoding cDNA was cloned into the pac.Ad5.CMV.K-N.pA shuttle vector for generation of replication-deficient adenovirus. Empty adenovirus (Ad5.CMV.K-N.pA) was used as a control. Adenoviral expression of active and total TGFβ2 was assayed in vitro by the transduction of Chinese hamster ovary and trabecular meshwork cells. BALB/cJ mice or Wistar rats were injected either intracamerally or intravitreally with the adenovectors and assessed for changes in intraocular pressure (IOP) using the rebound tonometer. At peak IOP, aqueous outflow facility and total TGFβ2 levels in aqueous humor were measured. Mouse eye morphology was assessed by hematoxylin and eosin staining.

Results.: Adenoviral gene transfer of hTGFβ2^226/228, but not hTGFβ2^WT, to the rodent eye elevated IOP in rat (43%, P < 0.001) and mouse (110%, P < 0.001) and reduced aqueous humor outflow facility in the mouse. The TGFβ2-induced ocular hypertension correlated with anterior segment TGFβ2 expression levels (P < 0.0001).

Conclusions.: The adenoviral TGFβ2 rodent model displays the glaucoma risk factors of elevated IOP and decreased aqueous outflow facility and may potentially serve as a model for studying glaucoma.

Glaucoma is a leading cause of visual impairment and blindness, affecting approximately 60 million persons worldwide. ¹ Elevated intraocular pressure (IOP) is a causative risk factor for both the development and the progression of primary open angle glaucoma (POAG). ^2–5 The exact mechanism in the pathogenesis of glaucoma is yet to be fully elucidated. However, pathologic changes in the aqueous humor outflow pathway, particularly the trabecular meshwork (TM), are related to increased outflow resistance and elevated IOP. These changes include abnormal accumulation of extracellular matrix (ECM), ^6–9 abnormal cytoskeletal changes in the TM cells, ^10,11 and in MYOC glaucoma, stress of TM cells associated with mutant myocilin-induced protein misfolding and abnormal catabolism. ^12–17  

In addition, transforming growth factor β (TGFβ) may play an important role in the development of glaucoma. Of the three isoforms of TGFβ (TGFβ1, TGFβ2, TGFβ3), TGFβ2 has been most widely studied in glaucoma. Elevated levels of TGFβ2 in the aqueous humor (AH) of glaucoma patients have been associated with glaucoma pathogenesis. ^18–26 TGFβ2 is expressed by TM cells and can affect many TM cell functions, including inducing ECM remodeling and overexpression in ocular tissues, ^27,28 inhibiting TM cell proliferation, ²⁹ and increasing TM cell phagocytosis. ³⁰ TGFβ2 perfusion in human anterior eye segments increases IOP and ECM deposition in the AH drainage structure, ^28,31,32 supporting the causative role of TGFβ2 in this aspect of glaucoma pathogenesis. Despite these important findings implicating the involvement of TGFβ2 in glaucoma, the effect of this cytokine on IOP in animals has not been demonstrated. 

To evaluate the role of TGFβ2 in glaucoma pathogenesis and to develop in vivo models of glaucoma, we sought to express TGFβ2 at sustained pathophysiological levels in the anterior segment of the rodent eye by viral vector transduction. Viral vectors have been shown to transduce cells of the AH outflow pathway and to result in expression of the delivered transgene. ^33,34 Adenoviral vectors delivered intracamerally or intravitreally readily infect the TM, iris, ciliary body, and corneal endothelium of rodents. ¹⁶  

TGFβ2 is produced in the cell as a latent 390-amino acid precursor that forms dimers through association of its latency-associated peptide (LAP). ³⁵ Dimeric, mature, biologically active C-terminal 112-amino acid TGFβ2 is generated from the cleavage and dissociation of LAP. Removal of LAP can be readily achieved in vitro by treatment of the protein with acid, heat, or chaotropic agents. In vivo removal of LAP occurs primarily through the protease furin, which is regulated by Emelin-1. ³⁶  

Our first attempt at adenoviral overexpression of wild-type, latent TGFβ2 in rat eyes resulted in no IOP change, suggesting little in situ activation. Based on bioactivating mutations introduced into the LAP-binding portion of the orthologous monkey TGFβ1, ³⁷ pig TGFβ1, ³⁸ human TGFβ1, ³⁹ cow TGFβ1, ⁴⁰ rat TGFβ1, ⁴¹ mouse TGFβ1, ⁴² and mouse TGFβ2, ⁴² we introduced similar mutations into human TGFβ2 to generate spontaneously active human TGFβ2 (Fig. 1). In vitro and in vivo characterization of the viral vector encoding this mutant TGFβ2 demonstrated the production of spontaneously active TGFβ2 with biological activity, including marker gene upregulation, ocular hypertension in rats and mice, and reduced AH outflow facility in the mouse. Our findings suggest that virus-expressed human mutant TGFβ2 may be useful for evaluation of the pathophysiological activities of TGFβ2 in the eye and for the development of an animal model that mimics the major risk factors of elevated IOP and decreased aqueous outflow facility of glaucoma in humans. 

Methods

Rodents

All animals were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and all protocols were approved and monitored by the Animal Care and Use Committee of Alcon Research, Ltd. BALB/cJ mice (2–5 months old; Jackson Laboratory, Bar Harbor, ME) and Wistar rats (2–4 months old; Charles River Laboratories, Wilmington, MA) used in this study were housed and handled as previously described. ⁴³ The animals were maintained on a 12-hour light/12-hour dark cycle (lights on 6:00 a.m.). Food and water were available ad libitum. 

Recombinant Plasmid and Adenoviral Vectors

Wild-type (WT) human TGFβ2 cDNA (NM_003238) was obtained from Origene (Rockville, MD) in the pCMV6-XL5 vector. To produce cDNA coding for the active human TGFβ2, coding sequences for the cysteines at positions 226 and 228 were converted to serines (Cell & Molecular Technologies, Inc., Phillipsburg, NJ). The region extending from the HindIII site (nt 1976) to the PpuMI site (nt 2395) was synthesized such that nucleotides 1985 and 1991 were changed from G to C, converting the cysteines to serines at amino acids 226 and 228 of the hTGFβ2 protein. This fragment was cloned into pCMV6-XL5.hTGFβ2. The WT and mutant coding region cDNAs were then subcloned into the pacAd5.CMV.KN.pA shuttle vector (Source: University of Iowa, Gene Transfer Vector Core) using EcoRI/Xba1 restriction. Viruses were purified by CsCl gradient centrifugation and dialyzed against TE (Tris-EDTA) buffer containing 10% sucrose and stored at −80°C (University of Iowa, Gene Transfer Vector Core). Each virus was tested for WT revertants and for titer by PCR and A549 plaque assay, as described. ⁴⁴ Empty adenovirus (Ad.Empty) was used as a negative control. 

Cell Culture

Human GTM3 cells were cultured as previously described. ⁴⁵ Chinese hamster ovary (CHO) cells were maintained in F12K medium (Invitrogen, Carlsbad, CA). The medium was supplemented with 10% FBS (Hyclone, Logan, UT), 100 U/mL penicillin G, 100 μg/mL streptomycin sulfate, 2 mM l-glutamine (Invitrogen), and all cells were incubated in a humidified 5% CO₂ atmosphere at 37°C. Adenoviral transductions were performed essentially as described. ^14,16 CHO cells were transiently transfected with plasmids using reagent as instructed by the manufacturer (Lipofectamine 2000; Invitrogen). 

Adenovirus Injections and IOP Measurements

For in vivo studies, animals were examined at day −1 by direct ophthalmoscopy (hand-held ophthalmoscope, model 11710; Welch-Allyn, Skaneateles Falls, NY) to confirm a normal appearance, free of any signs of ocular disease. At day 0, rodents were anesthetized using either a rat anesthesia solution (acepromazine 3 mg/kg; [Vetus; Burns Veterinary Supply, Westbury, NY]; ketamine 33 mg/kg [Ketaset; Fort Dodge Animal Health, Fort Dodge, IA]; and xylazine 7 mg/kg, intramuscular injection [Vetus; Burns Veterinary Supply]) or a mouse anesthesia solution (acepromazine 1.8 mg/kg, ketamine 73 mg/kg, and xylazine 1.8 mg/kg; intraperitoneal injection). Eyes given intracameral injection were pretreated with 1 to 2 drops 1% cyclopentolate (Mydriacyl; Alcon Laboratories, Fort Worth, TX) to dilate the pupil. A randomly assigned eye of each animal was then topically anesthetized with 1 to 2 drops of 0.5% tropicamide (Alcaine; Alcon Laboratories) and given a single intravitreal or intracameral injection of a suspension of adenovirus of a specified titer in a volume of 5 μL for the rat eye and 2 μL for the mouse eye. Ocular injections were administered using a Hamilton (Reno, NV) glass microsyringe fitted with a custom-made 1-inch, 33-gauge needle with a 10° bevel, as described previously, ¹⁶ while the uninjected contralateral eye served as a control. Each injection was made over the course of approximately 30 seconds The needle was then left in place in the anterior chamber or vitreous for a another minute before being rapidly withdrawn. IOP measurements were taken in conscious rats and mice using the rebound tonometer (TonoLab; Tiolat Oy; Helsinki, Finland), as described. ⁴³ The IOP investigator was masked to the identity of the injected adenovirus and to which eye was injected. 

Tissue Collection and RNA Isolation

At specified time points after adenoviral vector delivery, animals were euthanatized with CO₂ asphyxiation. Eyes were enucleated and bisected into anterior and posterior sections, and each section was placed in a separate microfuge tube containing 1 mL reagent (Trizol; Invitrogen), 1 μL glycogen covalently linked with blue dye (GlycoBlue; Ambion, Foster City, CA), and one 5-mm stainless steel ball (Qiagen, Valencia, CA). Samples were homogenized for 5 minutes at 25 Hz (TissueLyser system; Qiagen) and processed according to the reagent manufacturer's recommended procedure (Invitrogen). RNA was dissolved in 10 μL nuclease-free H₂O, and the concentration was determined by absorbance (A260). 

First-Strand cDNA Preparation

Contaminating adenoviral DNA was removed from the RNA by treating 1 μg total RNA with 1 U DNase according to the manufacturer's recommendation (Ambion). First-strand cDNA was generated from 1 μg total RNA using random hexamers and reverse transcription reagents (TaqMan; PE Biosystems, Foster City, CA) according to the manufacturer's instructions. cDNA was diluted to a working concentration of 0.5 ng/μL in water containing 50 μg/mL glycogen (GlycoBlue; Ambion). Effective removal of contaminating adenoviral DNA was verified in control reactions minus reverse transcriptase. 

Quantitative PCR

Transcript levels were measured by quantitative RT-PCR (QRT-PCR) using a sequence detection system (ABI Prism 7700; Applied Biosystems, Foster City, CA) essentially as described. ⁴⁶ Primers (TaqMan; Applied Biosystems) for hCTGF (Hs00170014_m1; ⁴⁷ ), mCTGF (Mm00515790_g1), hPAI-1 (Hs00167155_m1), hCOL1A1 (Hs00164004_m1), hTGFβ2 (Hs00234244_m1), mCOL4A1 (Mm00802377_m1), mPAI-1 (Mm00435860_m1), and mNOX4 (Mm00479246_m1) were purchased. Primer/probe for the EDA splice variant of human fibronectin (NM_002026, exon 32) was generated (Assay By Design; Applied Biosystems): CCTACTCGAGCCCTGAGGAT (forward), FAM-AATCCATGAGCTATTCCC-MGBNFQ (probe), and TGCAGTGTCTTCTTCACCATCAG (reverse). Amplification of all target gene mRNA was normalized to 18S ribosomal RNA expression using primers designed to the 18S rRNA gene (GenBank accession no. X03205) GTCCCTGCCCTTTGTACACAC (forward), VIC-CGCCCGTCGCTAC-MGBNFQ (probe), and CGATCCGAGGGCCTCACTA (reverse). Quantification of relative RNA concentrations was made using the comparative Ct method, as described in PE Biosystems User Bulletin 2 (http://docs.appliedbiosystems.com/pebiodocs/04303859.pdf). 

Determination of TGFβ2 Protein by ELISA

TGFβ2 concentrations were determined in aqueous humor or cell culture medium using a human TGFβ2 ELISA kit (R&D Systems, Minneapolis, MN), according to the manufacturer's instructions. Total TGFβ2 levels were determined by acid-activation of samples. Optical density was measured with a spectrophotometer (HTS7000plus [Perkin Elmer, Waltham, MA] or SpectraMax Plus384 [Molecular Devices, Sunnyvale, CA]). 

Histology

Eyes were removed and fixed for 24 hours in 4% paraformaldehyde and embedded in paraffin wax for sectioning. Sections (5 μm) were stained with hematoxylin and eosin reagent. 

Rebound Tonometer Calibration

Uninjected and hTGFβ2^226/228-expressing eyes were cannulated intracamerally with a 30-gauge steel needle and connected by flexible tubing to a variable-height physiologic saline manometer. Incorporated in series within the tubing was a previously calibrated (sphygmomanometer) pressure transducer (BLPR-2; World Precision Instruments, Sarasota, FL) for the continuous monitoring of actual (manometric) pressure. In this way pressure in the system could be set (10–40 mm Hg in increments of 5 mm Hg) and recorded. Ten tonometer (TonoLab; Colonial Medical Supply) readings were taken against each eye at each pressure, and then tonometer-indicated pressure was plotted as ordinate against manometrically set pressure as abscissa. A simple linear regression was performed on each curve (Prism 5; GraphPad Software, San Diego, CA). 

Aqueous Humor Hydrodynamics

For aqueous humor dynamics assessment, animals were anesthetized with intraperitoneal injection of an anesthetic cocktail (ketamine [72.7 mg/kg, Fort Dodge Animal Health], xylazine [1.8 mg/kg Vetus; Burns Veterinary Supply], and acepromazine [1.8 mg/kg Vetus; Burns Veterinary Supply]). Maintenance doses of the anesthetic cocktail (one-quarter to one-half the initial dose) were given subsequently as required. Each animal was then placed on a heated pad (37°C) for maintenance of body temperature and left for 30 minutes to allow a plane of surgical anesthesia to develop, judged to be of sufficient depth when the animal failed to react to toe-pinch (absence of flexor withdrawal response) and the blink reflex was absent. One to two drops of 0.5% proparacaine HCl (Alcaine; Alcon Research, Ltd.) were then applied (topical ocular) to each eye for local anesthesia. The eyes were then cannulated intracamerally with a 30-gauge steel needle inserted through the peripheral cornea approximately 1 to 2 mm from the limbus and pushed toward the region of the opposing chamber angle, taking care not to strike the corneal endothelium, anterior surface of the iris, anterior lens capsule, or chamber angle itself. The needle was connected by tubing to an in-line pressure transducer (BLPR; World Precision Instruments) for the continuous determination of pressure within the system. The opposing terminal of the pressure transducer was connected by tubing to a 3-mL syringe loaded into a microdialysis infusion pump (SP101I Syringe Pump; World Precision Instruments). The pump was then switched on and set to a flow rate of 0.1 μL/min. The pump remained running until the pressure (which was continuously recorded) in the system stabilized (typically 10–15 minutes). Flow rate was then increased sequentially to 0.2 and 0.3 μL/min, and stabilized pressure values at each flow rate were recorded. For each eye, aqueous humor outflow facility (μL/min/mm Hg) was calculated as the reciprocal of the slope of the respective pressure-flow rate curves. 

Statistical Analysis

Data are presented as mean ± SEM. The Student's t-test was used to compare between two groups of results. One-way ANOVA followed by Dunnett's test was used to compare among three or more groups. Linear regression was evaluated by Pearson correlation. P < 0.05 was regarded as statistically significant. Statistical significance between groups was determined with statistical analysis software (Prism 4 [GraphPad Software] or SigmaPlot 10 [Systat Software, Inc., Chicago, IL]). 

Results

In Vitro Characterization of TGFβ2 Constructs

Our initial attempt to produce elevated levels of active human TGFβ2 in the rodent eyes used adenovirus (Ad5) to drive the expression of full-length, wild-type hTGFβ2 (Ad.hTGFβ2^WT). Subsequently, we developed an adenovirus expressing full-length hTGFβ2 with two point mutations incorporated into the LAP-binding portion (C226S, C228S) (Ad.hTGFβ2^226/228), as described previously for the TGFβ1 isoforms in monkey, ³⁷ pig, ³⁸ human, ^39,48 bovine, ⁴⁰ rat, ⁴¹ and mouse ⁴² and the TGFβ2 isoform in the mouse ⁴² (Fig. 1). To verify the vector-mediated expression of TGFβ2, we transfected CHO cells (Fig. 2A) with either p.hTGFβ2^WT or p.hTGFβ2^226/228 and transduced GTM3 cells ⁴⁵ (Fig. 2B) with either Ad.hTGFβ2^WT or Ad.hTGFβ2^226/228. After 48 hours, culture medium was analyzed by ELISA for active and total (active plus latent) TGFβ2. Total TGFβ2 was determined by acidification of the samples, which converts latent molecules to active molecules. In both cell types, the level of active TGFβ2 produced by the hTGFβ2^226/228 vector was approximately 5 ng/mL (Fig. 2). Total TGFβ2 levels were twofold higher in p.hTGFβ2^226/228- than in p.hTGFβ2^WT-transfected CHO cells, and the level of active TGFβ2 accounted for 25% of total TGFβ2 in the p.hTGFβ2^226/228 transfected cells compared with 8% in the p.hTGFβ2^WT-transfected cells (Fig. 2A). In the GTM3 cells, total TGFβ2 levels were 10-fold higher in the Ad.TGFβ2^226/228- than in the Ad.TGFβ2^WT-transduced cells, and the level of active TGFβ2 was 31% of total TGFβ2 in the Ad.TGFβ2^226/228-transduced cells compared with 13% in the Ad.hTGFβ2^WT-transduced cells (Fig. 2B). 

Figure 1.

 

Comparison of human TGFβ2 amino acid sequence (NP_003229) with that of monkey TGFβ1 (XP_512687), ³⁷ pig TGFβ1 (NP_999180), ³⁸ human TGFβ1 (P01137), ³⁹ cow TGFβ1 (P18341), ⁴⁰ rat TGFβ1 (NP_067589), ⁴¹ mouse TGFβ1 (NP_035707), ⁴² and mouse TGFβ2 (NP_033393). ⁴² Gray: identical amino acids. Asterisk: cysteine residues targeted for mutation to serine.

Figure 2.

 

Levels of active and total (active + latent) hTGFβ2 in the culture medium of (A) CHO cells transfected with p.Empty, p.hTGFβ2^WT or p.hTGFβ2^226/228 (n = 2) or (B) GTM3 cells transduced with Ad.Empty, Ad.hTGFβ2^WT or Ad.hTGFβ2^226/228 (n = 3). Samples were collected 48 hours after transfection or transduction and assayed for hTGFβ2 by ELISA. Results are expressed as mean ± SEM (n = 3). **P < 0.01 or ***P < 0.001 compared with the respective empty vector group by one-way ANOVA and then Dunnett's test.

The biological activity of Ad.hTGFβ2^WT and Ad.hTGFβ2^226/228 was determined by QRT-PCR analyses of downstream effectors of TGFβ2 in TM cells, ²⁸ such as connective tissue growth factor (CTGF), plasminogen activator inhibitor-1 (PAI-1), the EDA splice variant of fibronectin (FN-EDA), and collagen 1A1 (COL1A1) (Fig. 3). CTGF and PAI-1 were significantly (P < 0.01) induced by Ad.hTGFβ2^226/228 but not by Ad.hTGFβ2^WT or control Ad.Empty viruses (Fig. 3). COL1A1 was less responsive but still significantly induced by both Ad.hTGFβ2^WT (P < 0.05) and Ad.hTGFβ2^226/228 (P < 0.01) (Fig. 3). Interestingly, FN-EDA ⁴⁹ expression was unaltered by the treatment of viral vectors in these cells. 

Figure 3.

 

Biological activity of Ad.TGFβ2 in vitro. GTM3 cells were transduced with Ad.Empty, Ad.hTGFβ2^WT, or Ad.hTGFβ2^226/228 for 48 hours and assayed for marker gene expression by QRT-PCR. Results are expressed as mean ± SEM (n = 3). *P < 0.05 or **P < 0.01 compared with the respective Ad.Empty group by one-way ANOVA and then Dunnett's test.

In Vivo Characterization of Ad.hTGFβ2^WT and Ad.hTGFβ2^226/228 in Rodent Eyes

Initially, Wistar rats were injected intracamerally with 5 × 10⁷ pfu Ad.hTGFβ2^WT, which encodes latent TGFβ2, with subsequent IOP measurements for 28 days. No change in IOP was noted over this period (data not shown). Next, we tested the vector encoding the bioactive form of TGFβ2, Ad.hTGFβ2^226/228. Intracameral injection of Ad.hTGFβ2^226/228 in the rats significantly increased IOP 5 to 12 days after injection (Fig. 4A). In the Ad.hTGFβ2^226/228-injected eyes, IOP at day 7 equaled 25.4 ± 2.4 mm Hg (mean ± SEM, n = 7), which was statistically significantly different (P < 0.001) from that of the noninjected eyes (17.8 ± 0.2 mm Hg; n = 14). Ad.Empty injection did not affect IOP (18.5 ± 0.6 mm Hg; n = 7; P > 0.05). Ad.hTGFβ2^226/228-induced ocular hypertension lasted for up to 8 days, after which the IOP declined toward the baseline value. 

Figure 4.

 

Effects of Ad.hTGFβ2^226/228 on IOP in rodent eyes. (A) Wistar rats were injected intracamerally with 5 × 10⁷ pfu either Ad.hTGFβ2^226/228 or Ad.Empty. (B) BALB/cJ mice were injected intravitreally with 6 × 10⁷ pfu either Ad.hTGFβ2^226/228 or Ad.Empty. Day of injection was designated as day 0. In both studies, the contralateral eye of each rodent was uninjected and served as a paired control. Results are expressed as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 compared with No Injection group by one-way ANOVA and then Dunnett's test.

The effect of Ad.hTGFβ2^226/228 was also tested in the mouse. Previously, we found that expression of adenovirus in the anterior segment of the mouse eye was more efficient when the vector was injected intravitreally. ¹⁶ Hence, Ad.hTGFβ2^226/228 (6 × 10⁷ pfu) was administered intravitreally in BALB/cJ mice. Significantly elevated IOP occurred starting at day 4 after injection and reached a plateau from days 6 to 15 (Fig. 4B). IOP at day 6 was 26.1 ± 4.0 mm Hg (n = 6) in the Ad.hTGFβ2^226/228-injected eyes, significantly different (P < 0.001) from the noninjected group (12.4 ± 0.9 mm Hg; n = 12). After day 15, Ad.hTGFβ2^226/228-induced ocular hypertension slowly declined, but at day 29 the IOP in these eyes was still statistically higher (P < 0.05) than in the noninjected eyes. Similar to results seen in the rat, intraocular injection of Ad.Empty did not affect IOP (day 6 IOP, 12.0 ± 0.3 mm Hg; n = 6). 

In a separate group of animals, evaluation of tissues from anterior segments of mouse eyes at day 8 after injection revealed a significant correlation (r = 0.88; P < 0.0001) between IOP (measured at day 7) and virally expressed TGFβ2 transcript levels (Fig. 5A). In the uninjected eyes (n = 6), there was no detectable transcript level of TGFβ2. Only the Ad.hTGFβ2^226/228-injected eyes (n = 6) had measurable levels (Fig. 5A). In this study, aqueous humor was collected from mouse eyes on day 14 after injection, pooled and analyzed by ELISA for total TGFβ2 levels. Correspondingly, aqueous humor TGFβ2 levels were significantly (P < 0.01) elevated in Ad.hTGFβ2^226/228-injected eyes (1.58 ± 0.06 ng/mL) versus uninjected eyes (0.46 ± 0.14 ng/mL) (Fig. 5B). In addition, when anterior segment tissue from mouse eyes was examined for marker gene expression after Ad.hTGFβ2^226/228 injection, endogenous mouse CTGF, PAI-1, and NADPH oxidase 4 (NOX4) levels were significantly (P < 0.05) elevated in Ad.hTGFβ2^226/228-treated versus uninjected eyes, whereas COL4A1 levels were not significantly altered (Fig. 6). 

Figure 5.

 

TGFβ2 transcript or protein levels in mouse anterior tissue or aqueous humor after Ad.hTGFβ2^226/228 injection. (A) Correlation of day 7 postinjection IOP with day 8 postinjection hTGFβ2 mRNA levels in anterior tissue (r = 0.88, P < 0.0001; Pearson correlation). All nonzero hTGFβ2 data points were derived from Ad.hTGFβ2^226/228-injected eyes, and the remaining data points were derived from the paired uninjected eyes. (B) Measurement of total (latent + active) TGFβ2 levels in mouse aqueous humor on day 14 after injection. Samples were pooled from 13 eyes and assayed in duplicate, with results expressed as mean ± SD. **P < 0.01 by two-tailed, unpaired Student's t-test.

Figure 6.

 

Marker gene expression in mouse anterior tissue after Ad.hTGFβ2^226/228 injection. QRT-PCR analysis of murine CTGF, PAI-1, NOX4, or COL4A1 gene expression in anterior half-tissue of mouse eyes injected 7 days earlier with 6 × 10⁷ pfu Ad.hTGFβ2^226/228. Contralateral uninjected eye served as a paired control. Data reflect mean ± SEM from triplicate QRT-PCR samples from five mice. *P < 0.05 by paired Student's t-test.

Mouse Eye Histology

Morphologic analysis of the adenovirus-injected mouse eyes showed that compared with control eyes (Figs. 7A, 7B), adenoviral (Ad.Empty) injection alone (Figs. 7C, 7D) caused increased overall corneal thickness, modest anterior chamber inflammation, and a propensity toward juxtaposition of the corneal endothelium to the iris during tissue sectioning, yet showed no signs of decreased lenticular or corneal transparency or elevated IOP. Similar morphologic changes have been noted by others using either AdTGFβ1 or Ad.Empty-injected mouse eyes. ⁵⁰ On the other hand, Ad.hTGFβ2^226/228-injected eyes (Figs. 7E, 7F) appeared similar in morphology to Ad.Empty-injected eyes, yet had elevated IOP that was not seen with Ad.Empty eyes or previously with either Ad.GFP- or Ad.MYOC^WT-injected eyes. ^16,51 It is also noteworthy that Ad.hTGFβ2^WT-injected eyes had similar histologic changes yet did not develop elevated IOP, further supporting the requirement for expression of active TGFβ2 from the adenovirus for elevated IOP. 

Figure 7.

 

Histologic analysis of mouse eyes 12 days after adenovirus injection. (A, B) Control uninjected eyes with an IOP of 13.1 mm Hg. (C, D) Ad.Empty-injected eyes with an IOP of 11.8 mm Hg. (E, F) Ad.hTGFβ2^226/228-injected eyes with an IOP of 34.7 mm Hg. (A, C, E) 4× image of central corneal. Scale bar, 200 μm. (B, F) 20× image of cornea, iris, ciliary body. Scale bar, 50 μm.

Rebound Tonometer Calibration

To assess whether the impact of tonometry against corneas of both uninjected eyes and eyes expressing hTGFβ2^226/228 was yielding accurate and reproducible readings of IOP, we compared IOP readings from cannulation with those from the tonometer (TonoLab; Tiolat Oy) for an uninjected eye with an IOP of 12.4 ± 0.3 mm Hg and paired hTGFβ2^226/228-expressing eye with an IOP of 25.9 ± 0.4 mm Hg (Fig. 8). The tonometer (TonoLab; Tiolat Oy)-generated IOP readings correlated well with the actual IOP in either the uninjected (r ² = 0.98) or the hTGFβ2^226/228-expressing (r ² = 0.99) eyes. The rebound tonometer (TonoLab; Tiolat Oy) yielded a reading ∼9% less than the actual manometric pressure with the uninjected eye over the range 10 to 40 mm Hg (Fig. 8A). When tested against the hTGFβ2-expressing eye over the same pressure range, there was a slightly greater degree of accuracy inherent within the IOP readings, but readings were still ∼16% greater than the actual manometric pressure (Fig. 8B). Thus, although overexpression of hTGFβ2 caused thickening of the cornea, the tonometer readings provided an acceptable degree of accuracy and reproducibility. 

Figure 8.

 

Effect of intravitreal injection of Ad.hTGFβ2^226/228 on tonometer-measured IOP in mice 13 days after adenovirus injection. Correlation between the tonometer-measured IOP and the actual (manometric pressure) IOP in the (A) uninjected eye and (B) paired Ad.hTGFβ2^226/228-injected eye was conducted using an intracameral method. Ten measurements were obtained from each preset actual IOP.

Aqueous Humor Hydrodynamics

To evaluate the effect of hTGFβ2 on aqueous humor outflow rate, one eye each of six mice was injected intravitreally with Ad.hTGFβ2^226/228, and 5 to 7 days later aqueous outflow facility was determined in both eyes. IOP of conscious animals immediately before outflow facility determination was 24.3 ± 2.2 mm Hg in the Ad.hTGFβ2^226/228-injected eyes and 11.4 ± 0.5 mm Hg (P < 0.001) in the contralateral uninjected eyes. When the eyes of the anesthetized mice were infused with balanced salt solution (BSS Plus; Alcon) at various flow rates, a resultant change in IOP was observed (Fig. 9A). The IOP of control mouse eyes correlated linearly with the infusion rate (r ² = 0.994). The mean regression slope of these eyes was 76.6 ± 9.6 mm Hg/μL/min. Similarly, the flow rate-IOP relationship of the Ad.hTGFβ2^226/228-injected eyes was also linear (r ² = 0.994) with a mean regression slope of 134.1 ± 28.3 mm Hg/μL/min, which is statistically significantly different from the control uninjected eyes (P < 0.05; paired Student's t-test). The AH outflow facility calculated from the inverse of the regression slope of the vector-treated eyes was 0.0098 ± 0.0023 μL/min/mm Hg, which is significantly (P < 0.01) lower than that of the control uninjected eyes (0.0146 ± 0.0024 μL/min/mm Hg) (Fig. 9B). These data indicate that Ad.hTGFβ2^226/228 treatment increased mouse IOP by, at least in part, reducing AH outflow facility. These in vivo data are consistent with our previous report of TGFβ2 effects on perfusion cultured anterior segments. ²⁸ Perfusion was performed at constant flow, so the TGFβ2 increase in IOP was exclusively due to a decrease in the aqueous outflow facility. 

Figure 9.

 

Effect of intravitreal injection of Ad.hTGFβ2^226/228 on aqueous outflow facility in the mouse. One eye of each mouse was injected with the viral vector, whereas the contralateral uninjected eye served as control. Aqueous outflow facility was evaluated on both eyes at days 5 to 7 after injection. (A) Correlation between infusion flow rate and IOP. (B) Calculated aqueous outflow facilities. Results are expressed as mean ± SEM (n = 6). *P < 0.05 and **P < 0.01 between vector-treated and control eyes by paired Student's t-test.

Discussion

In this study, we have demonstrated that transfection of CHO and transduction of GTM-3 cells with vectors expressing either hTGFβ2^WT or hTGFβ2^226/228, but not control vectors, increased the expression of total and active human TGFβ2 levels. The effect of Ad.hTGFβ2^226/228, especially on active TGFβ2 expression, was significantly more prominent than that of Ad.hTGFβ2^WT. Previous reports have shown that cells treated with active TGFβ can undergo autoinduction of TGFβ gene synthesis ⁵² and have increased expression of TGFβ-converting enzyme furin. ⁵³ Chagraoui et al. ⁵⁴ showed that mice transduced with adenovirus-expressed monkey TGFβ1^223/225 had high levels of TGFβ1 because of autocrine production by the exogenous bioactive TGFβ1. These authors used TGFβ1 knockout embryonic fibroblasts to show that the effect was autostimulatory. The overexpression of active TGFβ2 in our cells correlated with increased downstream effector molecules known to be involved in the TGFβ2-signaling pathways, such as CTGF, PAI-1, and COL1A1, indicating that the cytokine was biologically functional. The amount of activated human TGFβ2^226/228 detected in our in vitro assays was within the range found for similar orthologous TGFβ bioactive mutations. ^{37,38,54–56} The mutations introduced into hTGFβ2 only make the molecule more susceptible to furin-protease cleavage; therefore, production of the active form is not expected to be quantitative. 

Intraocular injection of Ad.hTGFβ2^226/228, but not Ad.hTGFβ2^WT or Ad.Empty, induced ocular hypertension in the rat and mouse. The lack of effect of Ad.hTGFβ2^WT suggests that rodent eyes have sufficient regulatory mechanisms in place to prevent the activation of exogenous TGFβ. In this sense, Ad.hTGFβ2^WT served as an effective negative control. In the rat, Ad.hTGFβ2^226/228 increased IOP by 43% at day 7 after treatment, whereas intravitreal injection of the vector elevated mouse IOP by 110% at day 6. This TGFβ2-induced pressure elevation was concomitant with an increase in TGFβ2 levels in the AH and enhanced the expression of signaling molecules downstream of TGFβ2 activation, such as CTGF, PAI-1, and NOX4 in the anterior segment of the eye. Any differences in TGFβ2-response genes between our current and previous ²⁸ studies may be attributed to differences in the experimental setup and treatment timing. Previous work by our laboratory using intravitreal injection of Ad.GFP (with a titer similar to that of the present study) detected gene expression in the trabecular meshwork and corneal endothelium and sporadically in iris, ciliary body, and lens epithelium. ¹⁶ Based on the similarities of the adenoviral vectors, we presumed that the localization of TGFβ2 expression in our present study was similar. This assumption was corroborated by the increased expression of TGFβ2 in the anterior segments of the mouse and elevated levels of TGFβ2 in the aqueous humor. Furthermore, within each mouse, the extent of IOP increase correlated with the ocular level of TGFβ2 mRNA. Thus, the ocular hypertensive effect and corresponding changes support the role of TGFβ2 in the glaucoma phenotype. 

Elevation of IOP can be generated, in theory, by an increase in AH production or a reduction of AH outflow facility, by an increase in episcleral venous pressure, by a reduction in uveoscleral outflow, or by some combination of these factors. In patients with POAG, experimental evidence indicates that pathologically impaired AH outflow facility is the mechanism for elevated IOP. Similarly, the IOP increase induced by Ad.hTGFβ2^226/228 treatment was, at least in part, a result of the impaired AH outflow facility. To evaluate the effect of TGFβ2 on aqueous humor hydrodynamics in the mouse eye, we developed a constant flow infusion method modified from that reported by Aihara et al. ⁵⁷ The validity of this method was confirmed by its detection of the outflow-enhancing effect of latanoprost, a prostaglandin agonist, and the lack of outflow effect of betaxolol, a β-adrenergic antagonist (Millar JC, et al. IOVS 2008; 49:ARVO E-Abstract 354; Millar JC, Pang IH, Clark AF, manuscript in preparation). 

These results are, as expected, based on the known mechanism of action of these glaucoma therapeutic agents. ⁵⁸ Using this constant flow technique, we showed that mouse eyes injected with Ad.hTGFβ2^226/228 had a suppressed AH outflow facility at each of the infusion flow rates tested. The biochemical and morphologic changes responsible for this observation are not understood but were likely a result of altered biochemical and physical properties, such as elasticity, compressibility, and response to shear stress, of structures in the outflow pathway in response to the overexpression of active TGFβ2. Because TGFβ2 is known to modify ECM production and metabolism in the TM, we speculate that TGFβ2-induced ECM changes caused such abnormality. However, the involvement of other possibilities, such as TGFβ2-altered expression of various other cytokines and signaling molecules, which further affect cytoskeleton and other cell and tissue functions, cannot be excluded. A recent report using intracameral delivery of adenovirus-expressing active TGFβ1 in male Wistar rat eyes showed morphologic changes that the authors described as peripheral anterior synechiae. ⁵⁹ These authors have also previously reported anterior subcapsular cataract in C57Bl/6 mice. ⁵⁰ Any differences between their studies and ours may be attributed to the choice of delivery route, cytokine (TGFB1 vs. TGFB2) and species/strain. Because our model involves transient elevation of TGFβ2 in adult mice eyes, we did not see, and did not expect to see, some of the severe changes in ocular tissues noted with transgenic mice engineered to chronically overexpress active TGFβ1 in a lens-specific manner. ^39,60,61 However, we did see an effect of adenovirus injection alone on anterior segment morphology, but these effects were independent of IOP elevation. Further, data obtained during the tonometer (TonoLab; Colonial Medical, Franconia, NH) calibration studies with both uninjected and hTGFβ2-expressing eyes confirmed that tonometry was accurate and reproducible regardless of corneal morphology. An ex vivo outflow model with human eyes (isolated and independent of the immune system and blood circulation) perfused continuously with human recombinant TGFβ2 showed decreased outflow facility and increased extracellular matrix material, ^28,31 which is consistent with our in vivo results using adenoviral overexpressed TGFβ2. Altogether, our data support the hypothesis that adenovirus-mediated expression of TGFβ2^226/228 in the anterior segment of the mouse eye is the basis for elevated IOP. 

In this study, the adenovirus-mediated gene expression in the rodent eye was transient, which is consistent with previously published reports using adenovirus in the rodent eye. ^16,62,63 The Ad.hTGFβ2^226/228-elevated IOP started to return to baseline IOP at approximately day 8 in the rat and day 18 in the mouse. This short-lived expression most likely was a result of immune responses generated by the host animal against viral proteins. ^63–65 Recently, with the use of green fluorescent protein as a reporter transgene, we showed that immunomodulation achieved by treating the animals with an anti–CD40L antibody, which specifically suppresses the CD4⁺ T-cell–assisted, but not B-cell–mediated, humoral responses normally mounted against adenovirus can significantly lengthen the duration of adenoviral transgene expression in the mouse eye. ¹⁶ Ongoing studies in our laboratory indicate that ocular hypertension induced by Ad.hTGFβ2^226/228 can be prolonged by anti–CD40L antibody treatment to at least 150 days. 

The use of adenoviral vectors to induce gene expression in tissues of the anterior segment, such as the TM, is a significant achievement in glaucoma research. Adenovirus is one of the most popular techniques to introduce genes into the TM of different animals. ^{34,62,63,66–68} With this method, genes such as myocilin ¹⁶ and secreted frizzled-related protein ⁵¹ have been shown to contribute to the regulation of IOP and the potential pathogenesis of glaucoma. Besides adenovirus, other methods for delivering genes to the anterior segment of the eye include liposomes, ⁶⁹ adenoassociated virus (AAV), ⁷⁰ herpes simplex viral vectors, ⁷¹ and lentivirus. ^72,73 Like adenovirus, each method has distinct advantages and disadvantages, but AAV and lentivirus appear to offer the hope of sustained expression of the delivered gene product. In the future, it would be useful to explore the use of AAV or lentivirus to deliver hTGFβ2^226/228 for sustained expression of TGFβ2 and the potential development of glaucomatous optic neuropathy and retinopathy in an animal model. 

In conclusion, we have developed a rodent model that displays the glaucoma risk factors of elevated IOP and decreased aqueous outflow caused by the overexpression of active, but not latent, TGFβ2. Our model may be useful for the further study of POAG mechanism, development, and progression. 

Footnotes

 Supported by Alcon Research, Ltd.

Footnotes

 Disclosure: A.R. Shepard, Alcon Research, Ltd. (E); J.C. Millar, Alcon Research, Ltd. (E); I.-H. Pang, Alcon Research, Ltd. (E); N. Jacobson, Alcon Research, Ltd. (E); W.-H. Wang, Alcon Research, Ltd. (E); A.F. Clark, None

The authors thank Deborah Lane and Robin Chambers for the culture of CHO and GTM-3 cells.