Experimental and Numerical Studies of Adenovirus Delivery to Outflow Tissues of Perfused Human Anterior Segments

Abstract

purpose. To investigate the efficacy of two different methods of adenoviral transfer of genes to trabecular meshwork (TM) and Schlemm’s canal (SC) cells in cultured human anterior segments, using both experimental and numerical analyses.

methods. Replication-deficient adenoviruses having coding sequence for β-galactosidase (β-gal) under the control of the cytomegalovirus promoter were used. Efficiency of gene transfer over time was verified by infecting cultured human TM cells and assaying for β-gal activity. Next, ostensibly normal paired human eyes were prepared by standard techniques and perfused for 2 to 5 days to measure baseline facilities. Eyes were then infected by one of two methods: standard transcorneal puncture, or injection into a 1 mm diameter silastic segment of supply tubing immediately upstream of the perfusion dish. In both cases, the nominal total dose was 2 × 10⁸ viral particles. Five days after viral injection, eyes were harvested and fixed, and wedges from each of four quadrants were examined histologically. Sections were assayed for β-gal activity and/or stained with toluidine blue. In a parallel study, flow and viral transport within perfused anterior segments were numerically simulated for conditions that approximated those used experimentally.

results. Eyes receiving viral particles by transcorneal injection showed variable levels of β-gal activity and highly variable TM cellular morphology, ranging from excellent preservation to cellular lysis. Eyes receiving an equivalent viral dose via the supply tubing showed higher transfer efficiency, as judged by almost complete TM cell loss (indicative of viral toxicity) and intense extracellular β-gal activity from the residual cytoplasm. At lower doses (1/3 to 1/1000 of that used in transcorneal injection) β-gal activity was still present, while TM cell morphology was good at the lower viral doses. Computer modeling showed that the region beneath the cornea was nearly stagnant, and consequently virus introduced into this region by transcorneal injection was delivered very slowly to the TM. This caused the effective delivered viral dose to be low and sensitively dependent on the volume and shape of the transcorneally injected virus bolus.

conclusions. Injection of adenovirus into supply tubing led to consistent delivery of reporter gene and approximately 300-fold greater efficiency of gene transfer compared to the transcorneal injection method, and is therefore the preferred method for introducing viral particles into perfused anterior segments. These findings were consistent with computer modeling of flow and mass transport in perfused anterior segments. Although these quantitative results are specific to adenovirus, this general trend should hold for a wide range of perfused compounds.

Despite its central role in ocular hypertension, the biology and physiology of the human outflow tract remain poorly understood in both normal and glaucomatous eyes. Two powerful tools for studying conventional outflow in human eyes are viral gene transfer and anterior segment organ culture. Several different types of viruses have been used for gene transfer in ocular tissues, ¹ with adenoviruses being preferred for gene transfer to outflow pathway cells and tissues. For example, Loewen et al. used feline immunodeficiency virus (FIV) to obtain prolonged expression with a single injection, ² while engineered adenoviruses have been successfully used to deliver genes to outflow cells in culture, ^{3 4} in living animals, ^{5 6 7} and in cultured human anterior segments. ^{3 8}  

Human anterior segment organ culture has become an extremely useful experimental tool to study the conventional aqueous outflow pathway. It allows medium-term (2 to 3 week) experiments to be conducted using human tissue, which is ideal for many types of experiments, such as virally-mediated gene transfer to outflow cells, including trabecular meshwork (TM) and Schlemm’s canal (SC) cells. 

In living eyes or anterior segments in organ culture, delivery of adenovirus has been accomplished by transcorneal injection of a viral suspension into the anterior chamber, where a combination of diffusion and bulk flow (convection) transports viral particles to outflow tissues. While this technique successfully delivers viral particles to outflow tissues in organ culture, complications of this technique are: inconsistent injection of viral suspension into the anterior chamber due to corneal leakage/reflux; introduction of small air bubbles into the anterior chamber; nonspecific delivery of genes to corneal endothelium; and segmental variation in delivery of virus to outflow cells. 

An alternate method of viral delivery to outflow tissues in normal perfused human eyes was studied in an anterior segment organ culture model. The goals were twofold: to determine whether a protocol could be developed for more reliably infecting TM cells, and to analyze the physical factors controlling delivery of viral particles to the TM, or indeed, the delivery of any type of agent to the TM in anterior segment organ culture. 

Methods

Adenovirus

The adenovirus backbone was a replication-deficient “first-generation” adenovirus with deletions of the E1 and E3 genes. ⁹ This “empty” adenovirus contains the cytomegalovirus promoter and bovine growth hormone polyadenylation site separated by a polylinker that was used to insert the β-galactosidase (β-gal) reporter gene. The β-gal adenovirus construct was a generous gift from Karsten Peppel, Duke University. 

Individual adenovirus DNA titers were determined by three different methods: 1) plaque titration on human embryonic kidney 293 cells; 2) immunofluorescence microscopy of adenovirus protein expression (anti-penton group antigen, clone 143; Biodesign, Kennebunk, ME) in 293 cells infected with serial dilutions of adenovirus; and 3) absorbance at 260 nm. 

The infective half-life of the β-gal adenovirus construct at 37°C in serum-free Dulbecco’s Modified Eagle Medium (DMEM; Invitrogen, Carlsbad, CA) was determined empirically. Human TM cells (see below) were seeded onto 12-well culture plates and allowed to attain confluence. After at least 1 week at confluence, cells were rinsed once in prewarmed serum-free DMEM and exposed to β-gal adenovirus at a multiplicity of infection of 1.0 in serum-free medium for 2 hours. During the incubation with adenovirus, plates were maintained at 37°C in humidified air containing 5% CO₂ and rocked every 15 minutes. Before contact with TM cells, adenovirus was incubated in serum-free DMEM at 37°C for 0, 2, 4, or 8 hours in a polypropylene tube. Five days after infection, cells were rinsed in prewarmed phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde in PBS for 5 minutes. Cells were rinsed with phosphate-buffered saline and assayed for β-gal activity by incubation with developer (2 mM MgCl₂, 4 mM potassium ferricyanide, 4 mM potassium ferrocyanide, and 0.2 mg/mL X-gal; Sigma, St. Louis, MO) for 2 hours at 37°C. Next, cell nuclei were labeled using 4′,6-dimidino-2-phenylindole (DAPI, 1 μg/mL; Sigma) in PBS for 5 minutes. Digital photographs of three fields from each well containing infected cells were taken using both ultraviolet and visible light microscopy, and total cell number and blue cells per field were counted. Efficacy of adenoviral delivery of β-gal was evaluated as the ratio of blue cells to total cells. 

TM cells were isolated using a blunt dissection technique in conjunction with extracellular matrix digestion and cultured as previously described. ¹⁰ The cell strain used in this study was isolated from a nonglaucomatous donor eye (TM85). 

Perfusion Studies

Paired, ostensibly normal human eyes were obtained postmortem from the Eye Bank of Canada (Ontario Division, Toronto) and NDRI (Philadelphia). Eyes were free of any known ocular disease, and were stored in moistened chambers at 4°C until use. The perfusion protocol was similar to that described by Johnson and co-workers, ^{11 12 13} with modifications described in detail elsewhere. ¹⁴ The perfusate was DMEM to which antibiotics (0.17 mg/mL gentamicin, 0.25 μg/mL amphotericin-B, 100 units/mL penicillin, and 100 μg/mL streptomycin; all from Sigma), 1% fetal bovine serum (FBS; Sigma), and 250 μg/mL bovine serum albumin (Sigma) were added. After dissection and mounting in culture dishes, the anterior segments were perfused at a constant flow rate of 2.5 μL/min and intraocular pressure was continuously measured. After 3 to 5 days of perfusion, when a stable baseline facility was reached, virus was injected as described below into experimental eyes, while most contralateral control eyes received a similar injection with perfusion media only. Eyes were then further perfused for 5 days while measuring facility. 

Viral Injection Protocol

Aliquots of adenovirus-containing solution were thawed immediately before use. Twenty μL of solution (nominal 10¹⁰ PFU/mL) were mixed with 80 μL of DMEM to create 100 μL of a “full strength” solution containing nominally 2 × 10⁸ PFU. For some eyes this was diluted 1:3, 1:10, 1:30, 1:100, 1:300, or 1:1000 using DMEM. Delivery of virus using a standard anterior chamber exchange (from a perfusion syringe) was considered, but irreversible binding of virus to the long inlet tubing and inline filter was of concern. Therefore, in all cases, 100 μL of final solution was delivered to perfused anterior segments, using one of the following two protocols. 

1. In some eyes, the delivery was by careful transcorneal injection into the anterior chamber using a 30-gauge needle and a tuberculin syringe. Despite trying several different injection protocols, there was usually some transcorneal leakage of injected fluid (perhaps 10 to 20 μL), which, although small in absolute terms, could be a significant fraction of the injected volume. 

2. In most eyes, a short (∼2 cm) piece of tubing from a blood collection set (Vacutainer; Becton Dickinson and Company, Franklin Lakes, NJ) was interposed between the inlet cannula of the perfusion dish and the polyethylene supply tubing before sterilization of the dish and attached tubing. The eye was mounted on the dish and perfused in the normal manner. Just before the viral injection, a tuberculin syringe was used to draw 100 μL of virus-containing solution into the tubing of a second Vacutainer blood collection set with attached 25G needle. The supply pump was shut off and 200 μL of media were withdrawn from the eye into the media supply syringe at the pump. This depressurized the eye and reduced the possibility of leaks in subsequent steps. The 2-cm segment of Vacutainer tubing was then pierced with the 25G needle immediately adjacent to the inlet cannula, and the needle barrel was inserted inside the cannula. The virus-containing solution was then injected, after which the needle was withdrawn and the 2-cm segment of Vacutainer tubing was slid further onto the inlet cannula of the perfusion dish. This ensured that no fluid leaked out through the needle puncture site. Fluid (100 μL) was then infused into the eye from the supply syringe to repressurize the eye, and the supply pump was restarted. The entire procedure required several minutes. 

Morphology and β-Galactosidase Assay

At the conclusion of the perfusion, anterior segments were removed from the culture dishes, and washed with PBS. Several wedges (1- to 2-mm wide) were rapidly cut from each quadrant containing outflow tissues. One wedge from each quadrant was incubated for 10 minutes in fixation buffer from the Gal-S β-galactosidase reporter gene staining kit (Sigma), and the X-gal substrate was developed according to the kit instructions supplied by the manufacturer. Testing showed essentially complete colorimetric development by 7 hours of incubation at 4°C, so all samples were incubated for 7 to 12 hours. Selected wedges were then embedded in paraffin and sagitally oriented 5-μm sections of the chamber angle region were cut and counterstained with hematoxylin and eosin stain. 

Other wedges were immersed overnight in universal fixative (2.5% glutaraldehyde and 2% paraformaldehyde in 0.1 M Sörensen’s buffer, pH 7.3). Sagitally oriented semithin sections (0.8 μm) of the chamber angle region were cut and stained with toluidine blue. 

Computational Modeling

To better understand processes controlling viral transport to the TM, computer modeling techniques were used to simulate fluid flow patterns and adenovirus delivery in cultured anterior segments. In this technique, the equations governing fluid flow and species transport are numerically solved in a defined geometry, using a well-validated engineering technique known as the finite element method. ^{15 16 17 18 19} The anterior segment was modeled as a hemisphere of radius 9.7 mm containing a 300-μm wide drainage region representing the TM (Fig. 1) . The TM was located such that the TM “strip” had a mean diameter of approximately 11.6 mm. Fluid of viscosity 1.0 cPoise and density 1.0 g/cm³ was perfused at 2.5 μL/min from the injection port of the culture dish. 

For fluid flow modeling, the steady, axisymmetric Navier–Stokes equations were solved using a commercial code (ANSYS, version 6.0; ANSYS Inc., Canonsburg, PA). No slip conditions were imposed on all surfaces except the TM and inflow port, and fluid rheology was assumed to be Newtonian. A parabolic velocity profile was applied at the entrance to the 0.2 mm long inlet port (radius = 0.43 mm), and a constant pressure boundary condition was applied at the “trabecular meshwork”. In a preliminary study, we used meshes of different density in the inlet and TM regions to verify global mass conservation. The final mesh contained 97,281 linear triangular elements (49,087 nodes), with 11 elements spanning the inlet and 20 elements spanning the TM, and globally conserved mass to within 2.4%. 

Time-resolved viral concentration fields were computed by solving the unsteady, axisymmetric convection-diffusion equation assuming a constant viral particle diffusion coefficient of 3.67 × 10⁻⁸ cm²/s. ²⁰ Four runs were carried out, with the following initial conditions for three cases.

1.  
   A 100-μL bolus injection of virus (2 × 10⁸ PFU) from the lower port (2.5 μL/min for 40 minutes), followed by perfusion with media only. This simulated viral transport in the eye after injection of virus through the inlet cannula of the eye dish.
2.  
   A 50-μL (½ of 2 × 10⁸ PFU) transcorneal bolus, assumed to be originally distributed as a hemisphere of viral-containing fluid adjacent to the cornea (Fig. 1) . Fluid entering the anterior segment from the inflow port was assumed to be virus-free. This was designed to simulate the case of a transcorneal viral injection with some subsequent leakage of virus out through the needle track in the cornea.
3.  
   A 100-μL (2 × 10⁸ PFU) transcorneal bolus, with geometry and inflow conditions similar to that described in case 2. This simulated a transcorneal injection with no viral leakage.
4.  
   A 100 μL bolus of virus (2 × 10⁸ PFU) was uniformly distributed throughout all fluid within the anterior segment (total volume, 1911 μL), followed by perfusion with media only. This simulated viral transport in the eye after ideal, uniform mixing of virus.

Each species transport simulation was run for a total time of 25 hours using a finite element code developed by one of the authors (SW). This code is based on linear shape functions and the streamline upwind Petrov-Galerkin (SUPG) formulation ²¹ ; it has been validated against several standard test problems and used in a variety of mass transfer studies. ^{22 23 24} The mesh was obtained by subdividing each element of the fluid flow mesh into 16 equal subelements, to give a total of 1,556,496 elements (780,031 nodes), and the time step was 0.5 seconds. 

The effective viral dose reaching the trabecular meshwork was determined by:  

\[V_{eff}(t){=}\frac{{{\int}_{0}^{t}}a({\tau})J({\tau})d{\tau}}{c_{0}a_{0}}\]

where c ₀ is the viral concentration in the injected bolus, a ₀ is the infection efficiency of the virus at time 0, a(τ) is the virus activity in the fluid reaching the TM at time τ, and J(τ) is the mass flow rate of virus entering the TM at time τ (mass virus/unit time). J(τ) is computed as:  

\[J({\tau}){=}{{\int}_{A}}c(\mathrm{\mathbf{x}},{\tau}){\nu}(\mathrm{\mathbf{x}})d\mathrm{\mathbf{x}}\]

where c( x,τ) is the concentration of the virus entering the TM at location x and time τ, v( x) is the fluid entry speed into the TM, and the integral extends over the entire surface of the TM exposed to the anterior chamber. V _eff (t) can be interpreted as the effective volume of viral particles reaching the trabecular meshwork at a given time t; it is influenced by how much virus reaches the TM, as well as the time when the virus reaches the TM, since this influences the virus activity, a. If all the virus was delivered to the TM at time 0 (when the virus has maximum biological activity), V _eff would be constant and equal to the injected volume of viral containing fluid; in the other extreme, if none of the virus delivered to the TM had biological activity, V _eff would be zero for all time. 

Results

Adenovirus Half-life Measurements

At a multiplicity of infection of one (one viral particle per cell), 51 ± 9% of TM cells were infected by freshly thawed β-gal adenovirus particles after 2 hours of exposure (mean ± SEM; n = 9 wells in total from two separate experiments). This efficiency decreased with increasing duration of adenovirus particle incubation in serum-free DMEM at 37^°C before contact with TM cells (Fig. 2) . Fitting the experimental data to an exponential decay curve gave a viral half-life of 5.3 hours. 

Perfusion and Histologic Findings

The donor age was 76 ± 4 years (all values are mean ± SEM), the postmortem time to enucleation was 4.9 ± 3.3 hours, and the postmortem time to beginning of perfusion was 31.3 ± 2.1 hours (Table 1) . The facility before injection of virus/control solution was 0.24 ± 0.03 μL/min per mm Hg. Injection of vehicle, either transcorneally or via the inlet tubing, caused no change in facility (net change = 1 ± 7% from just before injection to termination of perfusion, n = 4 eyes). Transcorneal injection of virus reduced facility by 16 ± 12% (to termination of perfusion), which was not statistically significant (P = 0.24; n = 6 eyes). Injection of virus via the inlet tubing caused a slight facility decrease that was not statistically significant for any single dosage; however, when all eyes receiving β-gal virus were pooled, the facility decrease reached statistical significance (17 ± 6% decrease to termination of perfusion; P = 0.014, n = 14 eyes). There was no obvious relationship between magnitude of facility decrease and dose of virus. 

Macrophotographs of wedges from anterior segments that received 2 × 10⁸ PFU of virus by transcorneal injection showed variable β-galactosidase activity, ranging from intense blue color development to a light blue, patchy development (Fig. 3 ; compare Fig. 4 ). This was consistent with paraffin sections through angle tissues. In eyes that received a full dose (2 × 10⁸ PFU) of virus via the inlet tubing, the color development was remarkably consistent and intense, with blue coloration in the sclera and corneal endothelium. Lower incremental doses of virus delivered via the inlet tubing showed a dose-dependent decrease in blue coloration, with significant coloration still present at 1:300 of a full dose (Fig. 3 ; compare Fig. 4 ) but less consistent coloring at 1:1000 of a full dose. Delivery of low doses of adenovirus via the inlet tubing resulted in predictable β-gal activity in the trabecular meshwork with little collateral activity in surrounding tissues such as cornea. 

Semi-thin sections showed variable appearance in eyes that received 2 × 10⁸ PFU by transcorneal injection. In some sections, there was extensive loss and lysis of TM cells, while other sections showed only modest cell loss, primarily in the uveal meshwork. In contrast, almost all cells had lysed in the TM of eyes that received 2 × 10⁸ PFU via the inlet tubing. In fact, the blue color in the TM of these eyes was coming from cytoplasmic remnants of lysed cells. As the viral dose delivered via tubing injection was decreased incrementally, overall TM cellular appearance improved at each step. At a dose of 1:300 (i.e., 2 × 10⁸/300 PFU) the cells looked very similar to those seen in control eyes, with the expected loss of uveal cells and focal loss of corneoscleral cells. 

Computational Modeling

Computed flow patterns showed a preferential route from the injection port to the trabecular meshwork (Fig. 1) . It is important to note that the “corner” where the sclera meets the perfusion dish was a relatively stagnant region, as was the central zone immediately underneath the central cornea. 

Viral transport patterns depended in a surprisingly sensitive manner on the initial distribution (injection protocol) for the virus. Virus adjacent to the central cornea (simulating the case of a transcorneal injection) showed very little motion for the first 3 to 5 hours (Fig. 5) due to the stagnant zone near the central cornea. Further, if the volume of virus adjacent to the cornea was decreased by a factor of two, it took much longer for virus to start to reach the TM. In contrast, when virus was delivered via the supply tubing, it was fairly rapidly convected toward the TM, with the bulk of the virus arriving in 3 to 4 hours and being carried directly into the TM. If the virus was uniformly mixed into the anterior segments then the central zone of virus was rapidly delivered to the TM, but stagnant fluid zones were “left behind” for extended periods of time. 

These differences in transport patterns resulted in notable differences in viral dose delivered to the TM (Fig. 6) . In particular, the effective dose delivered to the TM was about twofold less when 100 μL of fluid was injected transcorneally than when the same volume of fluid was delivered via the inlet tubing. Furthermore, there was a very sensitive dependence of effective delivered viral dose on the volume of viral-containing fluid adjacent to the cornea. This indicates that the effective delivered dose would be very sensitive to small amounts of leakage through the cornea, at least for injected volumes of 100 μL or less. 

Discussion

The efficiency of two different adenoviral delivery techniques was evaluated in the perfused human anterior segment culture model. Results clearly demonstrated that reliable, consistent, and efficient adenoviral delivery to the TM was best obtained when the virus was injected through the inlet port, using a protocol such as described here to avoid leakage. This is in contrast to transcorneal injection of virus, which resulted in significantly less efficient and generally more inconsistent delivery of adenovirus to the trabecular meshwork. In fact, injection of adenovirus through the cornea resulted in the appearance of only approximately 1/300th of the β-galactosidase activity in the TM when compared to inlet port injection. More consistent delivery of genes to TM cells increases the power of future experiments to deduce the effects of specific genes and their protein products on TM function and outflow facility, and hence further enhances the utility of the anterior segment perfusion model for studying human outflow function. 

The experimental results were qualitatively consistent with computer modeling studies that indicated a significant dependence of viral delivery patterns on delivery method. In particular, the presence of a large stagnant region near the “top” of the eye (immediately under the center of the cornea) made this an inefficient region for delivery of material into the TM. The combination of a stagnant zone below the cornea, with the relatively short biological half-life of the adenovirus, resulted in significant loss of viral activity by the time the viral particles were carried into the TM. 

One limitation of the present study is that the computer modeling studies overpredicted the efficiency of viral delivery to the TM, compared to the experimental measurements. For example, the computer results suggested that transcorneal injection was 2 to 8 times less efficient than delivery of adenovirus through the inlet tubing, while experiments suggested that this ratio was as large as 300. This discrepancy was probably due to a combination of factors:

•  
  The half-life of the virus in the anterior segment may be shorter than what was measured experimentally. For example, over time, viral particles may bind to themselves, non-TM tissue in anterior segments, and/or serum proteins in the perfusion media (absent for in vitro infection experiments) thus shortening virus half-life. A shorter half-life penalizes delivery schemes where there are significant stagnant regions. For example, if the half-life is reduced to 1 hour, simulation showed that the 50 μL transcorneal injection protocol became approximately 43 times less efficient than the 100 μL tubing injection protocol. However, this effect alone seems unable to explain all the experimentally observed efficiency differences.
•  
  Despite taking great care, transcorneal injection sometimes resulted in small bubbles forming in the eye. This created an air–fluid interface which is a preferred site of accumulation for many biological materials that have both hydrophobic and hydrophilic domains. Such molecules tend not to be delivered to the TM.
•  
  The computer modeling assumed that the virus existed as an unmixed bolus in the inlet tubing when injected via the lower port. In reality, manipulation of the eye and syringes during the injection protocol produced some mixing in the anterior segment. A near-optimal strategy for delivery of active virus to the TM was to mix virus with fluid in the anterior segment while avoiding virus entering into stagnant regions near the sclera and cornea (Figs. 1 and 5 ; right column). Presumably this was approximated by a central port injection associated with some mixing in the vicinity of the port outlet. Because of these reasons, the experimental delivery of virus via the lower port was probably more efficient than was modeled.

The quantitative results were specific to adenoviral delivery in human anterior segments perfused at 2.5 μL/min. However, the qualitative features of the study can be extended to other situations (e.g., delivery of different agents, with different diffusivities). In general, if the biological half-life of the delivered agent is longer than approximately 20 hours, there seems to be little advantage of one technique over the next, since most of the delivered agent will make it to the TM eventually. On the other hand, if the half-life is 5 to 10 hours, then there is a clear advantage of the lower port injection protocol. 

It is particularly interesting to consider the case of an agent with a very short half-life, say only a few hours. There is a time delay of at least 2 to 3 hours before any material is delivered to the TM by any method. Therefore, the preferred way to deliver active agent to the TM in such a case is to somehow place the agent immediately adjacent to the TM. This may be difficult, and a second alternative is to well-mix the contents of the anterior segment before restarting perfusion; this lowers the effective concentration, but ensures that at least some of the agent will reach the TM while still biologically active (Fig. 6) . Future anterior segment perfusions should consider such half-life factors in experimental design. 

How do these findings relate to the in vivo transcorneal injection of virus? In the living eye there is natural mixing resulting from thermally-induced convection in the anterior chamber, saccades, and head motions, all which are absent in organ-cultured anterior segments. This would lead to fewer “dead zones” in the eye, and better delivery to the TM than would occur with transcorneal injection in organ cultured eyes. 

 Supported by the American Health Assistance Foundation (WDS and CRE), Research to Prevent Blindness Foundation (WDS), National Eye Institute (Grant EY12797; WDS), Canadian Institutes of Health Research (Grant MA-10051; CRE) and the Doerenkamp–Zbinden Foundation (CRE).

 Submitted for publication October 13, 2003; revised December 30, 2003; accepted January 13, 2004.

 Disclosure: C.R. Ethier, None; S. Wada, None; D. Chan, None; W.D. Stamer, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Computed streamlines in a perfused anterior segment. Color coding on streamlines is indicative of local fluid speed (see calibration bar in units of mm/s). Fluid is delivered from a centrally located inflow port at 2.5 μL min and leaves the eye through the porous trabecular meshwork region.

Figure 2.

 

Graph showing decay of viral activity in cultured trabecular meshwork cell assay. The vertical axis is the percentage of infected cells in a plate, normalized by the percentage infected cells at time 0 for the same plate. The horizontal axis is the duration of viral incubation before exposure to TM cells. The fitted line assumes an exponential decay, and gives a half-life time constant of 5.3 hours. Error bars are SEM.

Table 1.

 

Summary of Characteristics of Eyes and Experimental Conditions

Pair	Donor Age/Sex	Postmortem Time to			Virus Administration Protocol	Viral Dose†
				Enucleation (hrs)			Perfusion (hrs)
521/522	77/M	2.5	31	Trans-corneal	521 vehicle; 522 1:1
525/526	69/M	8	33.5	Trans-corneal	525 1:1; 526 vehicle
537/538	31/M	4.5	43	Trans-corneal	538 1:1; 537 1:1 (S)
551/552	81/M	1.5	20	Trans-corneal	551 1:1 (S); 552 1:1
557/558	96/F	10.5	40	Trans-corneal	557 1:1 (S); 558 1:1
567/568	73/F	12.5	31	Trans-corneal	567 1:1; 568 1:1 (S)
577/578	77/F	6	32	Tubing inject	577 1:1 (S); 578 1:1
599/600	83/M	5.5	32	Tubing inject	599 1:3 (S); 600 1:3
601/602	78/F	1	25	Tubing inject	601 1:10; 602 1:3
603/604	95/M	5	33	Tubing inject	603 1:10; 604 1:3
605/606	78/F	1.5	21	Tubing inject	605 1:10; 606 1:30
607/608	82/M	6	37.5	Tubing inject	607 1:30; 608 1:100
609/610	56/F	4	33	Tubing inject	609 1:30; 610 1:100
611/612	70/M	3	39	Tubing inject	611 vehicle; 612 1:300
613/614	78/M	1.5	28	Tubing inject	613 1:300; 614 1:300
615/616	86/F	2.5	41	Tubing inject	615 vehicle; 616 1:300
619/620	86/F	7	11.5	Tubing inject	619 vehicle; 620 1:1000

†  PFU = Plaque forming units. 1:x means an x-fold dilution from the full-strength dose of 2 × 10⁸ PFU. (S) means that the adenovirus contained an insert coding for AQP-1 rather than β-gal.

Figure 3.

 

Comparison of β-galactosidase activity and histology in human anterior segments perfused with adenovirus. Top row: macrophotographs of wedges; middle row: paraffin sections; and bottom row: plastic semi-thin sections. The first column: eyes that received 2 × 10⁸ PFU by transcorneal injection; middle column: an eye that received 2 × 10⁸ PFU by infusion into the supply tubing; and the last column: an eye that received 1:300 of 2 × 10⁸ PFU by infusion into the supply tubing. SC, Schlemm’s canal. Numbers in first column refer to eye number; all images in columns two and three taken from Eyes 578 and 616, respectively. Note the variability in blue staining in Eye 567: the wedge shows blue staining while the paraffin section from another quadrant shows none.

Figure 4.

 

β-galactosidase activity and histology in human anterior segments perfused with control solution (vehicle only). Left panel: a macrophotograph of wedges; right panel, middle row: a plastic semi-thin section. SC, Schlemm’s canal. Both photographs taken from Eye 615.

Figure 5.

 

Contour plots of time-resolved adenovirus concentration profiles in perfused anterior segments, for different delivery methods. Color indicates virus concentration, normalized to lie in the range 0 to 1 (see scale bar). Note that in the rightmost column of plots, all concentrations have been multiplied by a factor of 19.11 (anterior segment volume/100 μL) so that all images can be plotted to the same color scale.

Figure 6.

 

Computed volume of virus-containing fluid delivered to the trabecular meshwork in perfused anterior segments versus time. Solid lines represent the actual volume of fluid; dashed lines represent the effective volume, accounting for the decay of viral activity with time according to equation (1) . Results for four different delivery protocols are shown.

The authors thank Tasnuva Koachar for excellent technical assistance and the Eye Bank of Canada (Ontario Division) for supplying eyes.