Adenovirus-Mediated Gene Therapy Using Human p21WAF-1/Cip-1to Prevent Wound Healing in a Rabbit Model of Glaucoma Filtration Surgery FREE

CONTROL OF the mammalian cell cycle is a highly regulated process that requires the interactions of multiple gene products. Several gene products have been identified that act to prevent progression through the cell cycle. Two of these are the tumor suppressor protein p53¹ and its downstream effector protein p21^WAF-1/Cip-1 (p21).^{2 - 4} The p21 protein negatively regulates the cell cycle by binding to and inhibiting the activity of cyclin-dependent kinases, a family of proteins required for the transition from the interphase(G₁) to the S phase (the phase during which DNA is synthesized) of the cell cycle (Gartel et al⁵ provide a review). Inhibition of cyclin-dependent kinases results in cell cycle arrest at the G₁-S interface, and several studies^{6 - 8} have shown that exogenous expression of p21 can inhibit cell proliferation.

Treatment of cancer has been a primary focus of gene therapy protocols and clinical trials using cell cycle regulatory genes such as p53.^{9 - 10} However, other medical conditions that result from nonmalignant cell proliferation are candidates for gene therapy with cell cycle regulators. One example is the fibroproliferation that occurs as part of the normal wound healing response after trabeculectomy for glaucoma, resulting in failure of the surgery. The use of antiproliferative agents, such as fluorouracil and mitomycin, has greatly increased the success of filtration surgery, but these agents are associated with complications ranging from corneal epithelial defects to late-onset bleb leaks that can lead to endophthalmitis.¹¹ These complications have prompted the search for alternative reagents that provide the desired antiproliferative effect without the associated complications.^{12 - 14} In this study, we examined the potential of gene therapy using human p21 to prevent wound healing in a rabbit model of filtration surgery. The results show that an adenovirus containing the human p21 gene (rAd.p21) can effectively block the proliferation of Tenon fibroblasts and inhibit scarring during a 30-day period in the rabbit eye, without some of the extreme effects found with mitomycin.

New Zealand white rabbits (Pasteurella free) were used for this study. Experimental procedures were conducted according to the guidelines established by the Association for Research in Vision and Ophthalmology statement for the use of animals in research and approved by the Animal Use Committee at the University of Wisconsin, Madison. These animals were used to evaluate the effect of rAd.p21 in glaucoma filtration surgery in 3 independent series of experiments. A total of 20 rabbits were used to evaluate changes in intraocular pressure (IOP) and surgical site histological features, 18 were used to determine changes in bleb characteristics, and 15 were used to measure outflow facility. An additional 12 rabbits were used for separate experiments to determine the localization and persistence of rAd.p21 expression. All experimental analyses using rabbits were conducted in a masked fashion, including surgery, IOP measurements, bleb scoring, outflow facility measurements, and the determination of histological features.

An E1/partial E3–deleted recombinant adenovirus was used for gene transfer in these experiments. Essentially, the p21 coding region, driven by the human cytomegalovirus immediate early promoter, was cloned into the E1 region of the adenovirus (rAd.p21). Control viruses were created in a similar manner; they encode bacterial β-galactosidase, jellyfish green fluorescent protein, or no added transgene (rAd.control) in place of the p21 coding sequence. Deletion of the E1 region of adenovirus renders the virus replication deficient. Details of the construction of the recombinant adenovirus are published elsewhere.¹⁵ Recombinant viruses were grown and propagated in the human embryonic kidney cell line 293 (American Type Culture Collection, Rockville, Md) and purified using standard protocols.¹⁶

Tenon fibroblasts were harvested from explants of the Tenon capsule dissected from freshly enucleated rabbit eyes and grown in Dulbecco Modified Eagle Medium (DMEM), containing 10% fetal bovine serum (FBS); penicillin G sodium, 100 U/mL; streptomycin sulfate, 100 µg/mL; and amphotericin B, 0.25 µg/mL. Only cells before passage 10 were used for in vitro analyses. Before recombinant adenovirus treatment, cells were made quiescent by serum starvation (with 0.5% FBS) for 24 hours. For thymidine incorporation experiments, quiescent cells plated in 96-well plates were incubated overnight with rAd.p21 or rAd.control (in triplicate) with increasing virus particle concentrations(4 × 10⁷-3 × 10⁹ virus particles per milliliter). Cells were then washed and refed with complete media containing tritiated thymidine. After 24 hours, plates were harvested and incorporated tritiated thymidine was counted (TopCount Microplate Scintillation Counter; Packard BioScience Co, Meriden, Conn).

For cell growth assays, quiescent cells were incubated with rAd.p21 or rAd.control (6.7 × 10⁷ virus particles per milliliter) for 24 hours, after which the recombinant adenovirus was removed by washing. The cells were stimulated to grow by the addition of complete media. Five days later, cell growth was evaluated by counting viable cells using trypan blue exclusion.

For cell cycle analyses, primary human ocular fibroblasts from the Tenon capsule obtained during surgery were isolated and cultured as previously described. Cells were arrested by culturing in DMEM plus 0.5% FBS for 3 days. Arrested cells were treated overnight with 1 × 10⁸ virus particles per milliliter of rAd.p21 or rAd.control. Cells were washed in DMEM plus 10% FBS to remove the adenovirus, and released from G₀ (the quiescent phase of the cell cycle) or G₁ by culturing in DMEM plus 10% FBS for 24 hours, at which time they were treated with trypsin and fixed in 70% ethanol at 4°C overnight. Cells were washed with phosphate-buffered saline(PBS), centrifuged at 1500 rpm, and treated with RNase, 5 µg/mL, for 30 minutes at 37°C. Cells were stained with propidium iodide, 50 µg/mL(Molecular Probes, Inc, Eugene, Ore), for 30 minutes, followed by flow cytometric analysis. Computer software (Cell Quest; Becton, Dickinson and Company, Franklin Lakes, NJ) was used to quantify percentages of cells in each phase of the cell cycle.

Surgery was conducted on one eye of each rabbit for IOP and bleb evaluation studies. In some cases, both eyes of one rabbit were operated on in animals used for outflow facility measurements. The bilateral surgery protocol was approved by the Animal Use Committee at the University of Wisconsin, providing that one eye received balanced saline solution (BSS) only as the antiproliferative agent. Rabbits were anesthetized with an intramuscular injection of ketamine hydrochloride, 40 mg/kg body weight, and xylazine hydrochloride, 5 mg/kg body weight. A wire eyelid speculum was placed, and the eye was fixed with a corneal suture (6-0 nylon). A limbal-based flap of the conjunctiva and the Tenon capsule was made in the superior nasal quadrant of either left or right eyes, and the sclera was cauterized. Each reagent was soaked into a 4 × 4 × 1-mm cellulose sponge, which was placed on the sclera under the flap for 5 minutes. After application, the area was irrigated extensively with BSS. The irrigation step was omitted in some cases, with no effect on any of the variables examined in this study (such as the level of transduction of fibroblasts). For the sclerostomy, a limbal groove angled toward the anterior chamber was first made with a No. 57 blade, followed by the production of a fistula first with a No. 75 blade and then with a Kelly punch. About 2 to 3 punches were generally taken in each eye. An iridectomy was then performed through the fistula, and the conjunctiva was closed with a 9-0 polyglactin suture. After surgery, 1 cm each of 1% atropine sulfate ophthalmic ointment and neomycin sulfate, 3.5 mg/g; polymyxin B sulfate, 10 000 U; and dexamethasone, 1 mg/g, ointment was applied to the eye.

Immediately following euthanasia, eyes were enucleated and rinsed in PBS. The surgical site was dissected, placed in a cryovial, and snap frozen in liquid nitrogen. Total RNA was extracted from approximately 100 mg of tissue harvested from the surgical sites of 3 eyes, using a reagent (Tri-Reagent; Molecular Research Center, Inc, Cincinnati, Ohio), per the manufacturer's protocol. Total RNA was treated with DNase I (Roche/Boehringer-Mannheim, Berkeley, Calif) to remove residual DNA. Quantification of human p21 messenger RNA was performed using reverse transcription–polymerase chain reaction with real-time detection (PE Applied Biosystems, Foster City, Calif), as described previously.¹⁷ The primers used were as follows: forward primer, 5′-AACGGTACTCCGCCACC-3′; reverse primer, 5′-TTCTGACATGGCGCCTACT-3′; and probe, 5′-FAM-TCCGCATCGACCGGATCGG-TAMRA-3′. The protein from the remaining tissue homogenate was then extracted for the detection of p21 by enzyme-linked immunosorbent assay per the manufacturer's protocol (Oncogene Research Products, San Diego, Calif).

Expression of the p21 transgene in vivo was detected by immunohistochemistry using a monoclonal antibody against human p21 that exhibited no cross-reactivity to the rabbit homolog (Oncogene Sciences, Inc, Uniondale, NY). Briefly, globes from rabbits treated with each reagent were harvested at different times after surgery, fixed in 4% paraformaldehyde in phosphate buffer (pH, 7.2) for 48 hours at 4°C, and embedded in paraffin. Paraffin sections (5 µm) were cut and stained for p21 expression, as previously described.¹⁸

Intraocular pressure was measured in awake rabbits before and after surgery using an applanation tonometer (Tono-Pen XL; Mentor Corp, Norwell, Mass).¹⁹ All IOPs were obtained at the same time each day by one of us (J.A.K.). Baseline IOPs were determined from average measurements taken one to several days before surgery after the rabbits became familiar with their handler. Intraocular pressure was monitored every 2 to 3 days after surgery for the first 10 days and every 3 to 4 days thereafter for the remaining 20 days.

Blebs were evaluated during slitlamp examinations on a qualitative scale of 1+ to 4+, reflecting increasing bleb height and size as follows: 1+, minimal height, conjunctiva thickening, and no microcysts; 2+, microcysts are present; 3+, elevated bleb covering 3 to 4 clock hours of the eye; and 4+, greatly elevated bleb covering longer than 5 clock hours of the eye. A score of 0 indicates no observable bleb. Controls for these experiments were conducted using sponges soaked in PBS containing 3% sucrose, which is a solution used to maintain viral stability during cryostorage.

For this study, the rabbits were killed humanely 30 days after surgery. A small burn was made with an ophthalmic cautery in the cornea of each eye to mark the surgical area before enucleation. An incision was made 90° away from the surgical site, and the whole globe was fixed in 4% paraformaldehyde in phosphate buffer (pH, 7.2) for 48 hours at 4°C and embedded in paraffin. Serial sections were then cut through the sclerostomy site. Approximately every fifth section was stained using Gomori trichrome.²⁰ Sections were masked and evaluated by 2 pathologists who scored the extent of fibroproliferation and cellular infiltrate using the following scale: 0, no change; 1, minimal change; 2, mild change; 3, moderate change; and 4, severe change. A final score for each variable was made by consensus between the 2 observers.

Total outflow facility was determined using 2-level constant-pressure anterior chamber perfusion with Bárány²¹ solution. Animals were evaluated 14 days after surgery, after the time when sclerostomies performed with no antifibrotic agent typically fail. Generally, each eye was perfused for 60 minutes, during which 9 outflow facility measurements were taken. The calculated outflow facility for each eye was then taken as the average of these measurements.

Treatment of quiescent rabbit Tenon fibroblasts with rAd.p21 resulted in a dose-dependent inhibition of DNA synthesis within 24 hours after infection, as measured by tritiated thymidine incorporation (Figure 1A). No inhibition was observed with control treatments. Fluorescent-activated cell sorter analysis of similarly treated human Tenon fibroblasts showed that inhibition of DNA synthesis by rAd.p21, but not by rAd.control, resulted in a G₁ arrest of the cell cycle (Table 1). The rAd.p21-treated cells were viable, as there was no sub-G₁ population indicative of debris from dead cells (not shown). To demonstrate a prolonged inhibition of cell growth, quiescent cells were infected with 6 × 10⁷ virus particles per milliliter; this dose was chosen based on results from the dose-response thymidine incorporation experiment showing approximately 80% to 90% inhibition relative to control (Figure 1A). Five days after treatment, rAd.p21-treated cells showed a significant 3-fold decrease in cell number compared with controls (P<.001, unpaired t test). Treatment with rAd.control did not significantly depress overall cell growth relative to untreated cells(P>.05, unpaired t test), demonstrating a p21-specific effect (Figure 1B).

Experiments were conducted to determine the pattern and persistence of transgene expression. Immunohistochemical staining for human p21 protein in rabbit eyes 14 days after rAd.p21 treatment and surgery showed numerous fibroblasts expressing the transgene throughout the flap region (Figure 2B). No p21 staining was evident in eyes treated with BSS (Figure 2A), rAd.control, or mitomycin (data not shown). Other experiments, conducted with the β-galactosidase–expressing vector to identify transduced cells, showed that vascular endothelial and conjunctival epithelial cells were transduced in addition to Tenon fibroblasts. No evidence of p21 expression in scleral fibroblasts, ciliary body epithelial cells, retina, iris, or any cells of the cornea was found using this mode of recombinant adenovirus application(data not shown).

The persistence of p21 transgene expression was investigated at the messenger RNA and protein levels. Quantitative reverse transcription–polymerase chain reaction analysis identified transgene-specific transcripts as late as 60 days after treatment, with levels decreasing throughout the experiment(Figure 3A). Enzyme-linked immunosorbent assay analysis of protein fractions from the same eyes showed detectable recombinant protein up to 28 days after infection (day 60 was not evaluated) (Figure 3B), which also decreased with time.

Rabbits were examined for bleb height at various points after surgery. Figure 4 shows the change in the bleb score during a 30-day period for eyes treated with rAd.p21 compared with PBS containing 3% sucrose and mitomycin. Statistical evaluation of the data during the experiment showed significantly larger blebs for mitomycin- and rAd.p21-treated eyes relative to the eyes treated with PBS containing 3% sucrose (P<.001, unpaired t test), while blebs in the mitomycin-treated rabbits were slightly larger than blebs in the rAd.p21-treated rabbits (P = .03).

The surgical sites were examined 30 days after surgery from sections cut through the surgical site and stained with Gomori trichrome. Figure 5 shows a panel of representative sections from eyes of each treatment group. Eyes treated with either BSS or control virus (recombinant adenoviruses containing either green fluorescent protein or β-galactosidase) exhibited nearly complete scarring over the sclerostomy site, including evidence of new collagen deposition in the scleral gap created by surgery (Figure 5A and B, respectively). In contrast, eyes treated with rAd.p21 had moderate bleb cavities and minimal evidence of new collagen deposition in the sclera(Figure 5C). Eyes treated with mitomycin exhibited large acellular bleb cavities with substantial damage to surrounding tissues, including thinning of the conjunctiva and Tenon layer in the region of the surgery (Figure 5D).

Sections (range, 5-15) through each surgical site were masked and scored by 2 pathologists for the 2 criteria of fibroproliferation and cellular infiltrate into the surgical site. The consensus score of each criterion, for each eye, is shown in Table 2. For fibroproliferation, most eyes treated with BSS or control virus received scores of 3, indicative of a completely healed wound, while 5 of 6 rAd.p21-treated eyes and 4 of 5 mitomycin-treated eyes received scores of less than 3. On average, the eyes treated with BSS, control adenovirus, or rAd.p21 displayed minimal cellular infiltrate. Eyes treated with mitomycin, however, showed mild to moderate signs of inflammatory cells, including 2 that exhibited an abscess with hypopyon and hyalitis extending from the bleb cavity into the anterior chamber (Figure 6).

Intraocular pressure measurements taken on awake rabbits with a handheld applanation tonometer exhibited a high degree of variability. In general, the IOP decreased below baseline immediately as a result of the sclerostomy regardless of treatment. Surgical procedures that failed did so within 10 days, as determined by a return to the baseline IOP (data not shown). The data for each treatment are as follows:^els10039ta

The data were taken directly from the handheld applanation tonometer and not calibrated to true millimeters of mercury for the rabbit eye. The rAd.p21- and the mitomycin-treated eyes showed a trend toward maintaining a reduced IOP during the experiment, although these data do not show a significant decrease over the BSS- or control vector–treated eyes.

Figure 7 shows the results of outflow facility measurements in rabbit eyes 14 days after surgery, after the point when BSS-treated eyes were observed to fail based on IOP measurements. The mean ± SD outflow facility of unoperated on rabbit eyes (n = 11) was 0.26 ± 0.04 µL/min per millimeter of mercury. A sclerostomy using BSS or rAd.p21 had no significant (P = .10, Mann-Whitney test) influence on the outflow facility of a normal-tension eye after 14 days, but mitomycin treatment created a significant increase in facility vs unoperated on eyes (P = .02).

Several measurements were taken to judge the effect of p21 gene therapy in controlling wound healing in the rabbit eye. Observations of the bleb characteristics and histological evaluation of the wound indicate that the p21 transgene attenuates fibroproliferation and scarring in the rabbit model at a level comparable to mitomycin-treated eyes. Intraocular pressure measurements were also taken as a measure of performance of the sclerostomies. Although there was a trend toward a decrease in IOP for eyes treated with rAd.p21 and mitomycin, these data were variable and not significantly different from those of eyes treated with BSS or the control virus. Several investigations with the rabbit model have used IOP as an indicator of efficacy for antiproliferative treatments. Jampel and Moon¹² showed that paclitaxel (Taxol) and mitomycin were able to prevent wound healing as judged by the change in IOP relative to the unoperated on eye of each animal. Intraocular pressure was also used to evaluate antibodies against transforming growth factor β₂ by Cordeiro and coworkers.¹⁴ Their study failed to find any effect on IOP in rabbits, although effects on bleb morphologic characteristics were detected. Other studies have questioned the usefulness of IOP as a measure of successful filtration surgery and bleb function in normal-tension rabbit eyes because the ciliary epithelium and residual iris have a tendency to extend forward into the anterior chamber and block the fistula.²² This problem is evident in histological sections taken through the eyes used in this study (Figure 5), which often show the iris either pushing into the fistula or extended toward the cornea.

Our study also included examination of outflow facility in treated compared with unoperated on eyes 14 days after surgery, when control blebs had typically failed (based on IOP measurements). No change in outflow facility was noted for the rAd.p21-treated eyes relative to the BSS-treated eyes or the unoperated on eyes. However, the sclerostomy provided only a small option for outflow in these eyes, which contained an undamaged outflow pathway that was presumably still functioning normally around the entire remaining circumference. It is not surprising that a functioning bleb may not be reflected in these measurements. Mitomycin-treated eyes did show a significant increase in outflow relative to normal eyes. This observation is consistent with studies²³ showing an increased risk of hypotonia in human eyes that undergo trabeculectomies using mitomycin. Mitomycin also produced complications in our study that were not observed with rAd.p21. The most noticeable were the increased incidence of inflammatory cells in mitomycin-treated eyes at 30 days after surgery (2 of 5 eyes) and the thinning of the Tenon capsule and conjunctiva in the surgical area (5 of 5 eyes). Although the numbers of cases reported in this study were too few to assign a definitive correlation with mitomycin use, these complications were not observed in any of the other eyes. These data are also consistent with numerous reports^{24 - 26} that mitomycin treatment can lead to excessively leaky and thin-walled blebs that are prone to late-onset infections.

The sequence of events that lead to wound healing includes the proliferation and migration of cells to the site of damage, the deposition of new extracellular matrix, the contraction of the wound margins, and the subsequent programmed death of infiltrating fibroblasts at the end of the scarring process.²⁷ The strategy of using antiproliferative agents to prevent wound healing is to block the proliferation step, which disrupts the subsequent wound-healing cascade. Mitomycin is an alkylating agent that has numerous toxic effects on cells, although its primary mode of action is to cross-link DNA.^{28 - 29} Cultured Tenon fibroblasts treated with mitomycin undergo apoptotic cell death,³⁰ and electron microscopic evaluations of mitomycin-treated rabbit conjunctiva and Tenon capsule show extensive tissue damage.¹⁴ Previous studies of p21 function show that this protein can also cause apoptotic cell death in some circumstances,^{31 - 32} but in most cases it causes cell cycle arrest.^{6 - 8} Treatment of cultured Tenon fibroblasts with rAd.p21 results in reduced DNA synthesis and cell division, while fluorescent-activated cell sorter analysis shows that these cells accumulate in the G₁ phase of the cell cycle.

The present in vivo experiments do not conclusively demonstrate that p21 inhibits wound healing by preventing cell division and subsequent extracellular matrix deposition, but Gomori trichrome staining clearly shows reduced collagen deposition and fibroproliferation in the surgical area. To expand these results, experiments are under way to characterize the time course of extracellular matrix deposition and to demonstrate inhibition of cellular proliferation in situ.

In summary, our results demonstrate the p21-specific inhibition of fibroproliferation and wound healing in a rabbit model of glaucoma surgery, which resulted in patent fistulas. These results are comparable to those obtained with mitomycin, but the toxic adverse effects, such as thin filtration blebs and persistent inflammation, observed with mitomycin use are eliminated. Consequently, p21 gene therapy may make an attractive alternative to mitomycin in those undergoing trabeculectomy.

Submitted for publication August 28, 2001; final revision received March 13, 2002; accepted March 20, 2002.

This study was supported by a grant from Canji, Inc, San Diego (Dr Nickells), and by grant EY02698 from the National Eye Institute, Bethesda, Md (Dr Kaufman).

We thank Ken Wills for adenovirus vector creation; Shufen Wen for oversight of polymerase chain reaction/reverse transcription–polymerase chain reaction efforts; the members of the Process Sciences Department, Canji, Inc(Josie Beltran, Doug Cornell, Daniel Giroux, Wendy Hancock, Beth Hutchins, Diane McAllister, Susan Miller, Margarita Nodelman, Thomas Schluep, Anastasia Sofianos, Barry Sugarman, and Suganto Sutjipto) for the production and characterization of all viruses used; Steven Schwartz, MD, for providing human Tenon fibroblasts; Jennifer Seemans, DVM, for her assistance with the rabbit surgical procedures; and Daniel Albert, Morton Smith, T. Michael Nork, MD, and Cassandra Schlamp, PhD, for their constructive comments during the study.

Corresponding author and reprints: Robert W. Nickells, PhD, Department of Ophthalmology and Visual Sciences, University of Wisconsin, Madison, 1300 University Ave, Room 6640 MSC, Madison, WI 53706-1532 (e-mail: nickells@facstaff.wisc.edu).