Abstract
purpose. To quantify and localize plasmid transfection of filtration surgery
tissues using two delivery techniques.
methods. Full-thickness filtering procedures were performed on eyes of New
Zealand albino rabbits. In 10 eyes, naked plasmid DNA in saline was
either injected beneath Tenon’s capsule at the filtration site or
absorbed into a collagen shield that was then placed external to the
sclerostomy and under the Tenon’s capsule. Forty-eight hours after
surgery, levels of the reporter gene, chloramphenicol acetyltransferase
(CAT) were measured in samples of ocular tissues. In two
additional eyes, the β-galactosidase (β-Gal) reporter
gene expression was localized histologically.
results. Injection of plasmid DNA in saline vehicle into the filtration bleb
produced readily detectable CAT activity in bleb tissue (conjunctiva,
Tenon’s capsule, and sclera) whereas CAT activity was nearly
undetectable in samples of the cornea, iris–ciliary body, and tissues
located opposite the bleb site. Delivery of the plasmid DNA into the
bleb through a collagen shield increased CAT activity 30-fold over
injection of plasmid in saline (2711 ± 567 mU/mg versus 92 ± 38 mU/mg). β-Gal activity was imaged only in the region of the
bleb, and microscopic examination showed β-Gal activity localized to
Tenon’s capsule fibroblasts, with minimal β-Gal activity observed in
inflammatory cells or scleral fibroblasts.
conclusions. Transfection of filtration tissues is enhanced by absorption of naked
DNA into a collagen shield. Furthermore, transfection is localized to
the fibroblasts and inflammatory cells of the filtration bleb site.
Gene therapy using naked plasmid DNA and a simple collagen shield
delivery vehicle may be useful for regulating wound healing after
glaucoma surgery.
Attempts to use gene transfer to correct diseases in humans began
in 1991,
1 and recently, the first successful clinical
outcome was reported in two young patients who had severe combined
immunodeficiency.
2 To date, there are no reports of
successful treatments of ocular diseases in patients using gene
transfer. However, there are a limited number of reports investigating
gene transfer with human cells in culture and animal ocular tissues.
Adenovirus vectors were used to transduce genes into intraocular
tissues by injecting the vector into the anterior chamber and vitreous
cavity of mice and rabbits.
3 4 Corneal graft donor buttons
were infected in organ culture by an adenovirus vector expressing theβ
-
Gal reporter gene.
5 A cationic lipid
preparation was used to assist transfection of a plasmid vector
expressing β-gal applied topically and intraocularly in
rats,
6 and a gene gun was used to transfect a plasmid
expressing green fluorescent protein into corneal epithelial cells of
rabbits.
7 A human retinal pigmented epithelial
(RPE) cell line transduced by an adenoviral vector containing the human
heme oxygenase gene
8 and replication-deficient adenoviral
vector to transduce the anterior segments from postmortem human
eyes.
9 These experiments demonstrated the general
feasibility of using viral vectors and plasmids to deliver genes to
ocular tissues.
Complications that have been reported for adenoviral vectors include
severe inflammatory reactions and development of high neutralizing
antibody titers that limit repeated applications.
10 11 12 Adenoassociated virus (AAV) vectors do not produce significant
inflammation but instead generate a neutralizing antibody
response.
13 Retrovirus vectors in general only are
effective on dividing cells and have the potential to integrate at
oncogenic sites in the genome.
The objective of many gene therapies, such as those to correct
metabolic disorders including cystic fibrosis, hemophilia, or sickle
cell disease, is the long-term expression of the transgene by stable
integration into the genome. However, reduction of scarring during
wound healing presumably requires only short-term expression of a
transgene. Transfection of cells with naked plasmid vectors generally
results in short-term expression of the transgene and has the
additional advantage of not inducing inflammation, which can increase
scarring. The success rate of glaucoma filtration surgery has been
improved by the use of antimetabolites that reduce scarring at the
filtration site and subsequent failure of the filter.
14 However, antimetabolite therapy is a nonspecific manipulation of the
wound-healing process that is reported to increase the risk of bleb
leaks and hypotony with a possible increased risk of
infection.
15 16 17 18 The complex process of wound healing
involves the gene-regulated expression of multiple growth factors,
angiogenic factors, enzymes, and inhibitory substances. Therefore, the
use of plasmid transfection to inhibit or increase selected gene
expression could provide an opportunity to regulate the process in a
targeted fashion.
Before such new therapeutic approaches can be used, it is necessary to
develop practical methods of gene delivery that minimally alter current
surgical techniques. In this study, we used a rabbit model of
filtration surgery to compare the efficiency of two delivery techniques
for plasmid vectors, to measure transfection in different locations of
the eye, and to identify the specific tissues and cells that were
transfected.
The pMP6-CAT plasmid expressing the chloramphenicol
acetyltransferase (
CAT) gene was kindly provided by Mohan
Philip (RPR-Gen Cell, Santa Clara, CA) and Lyle Moldawer (Department of
Surgery, University of Florida, Gainesville).
19 The
plasmid contains a mammalian expression cassette consisting of the
cytomegalovirus (CMV) immediate early promoter and enhancer followed by
a hybrid intron consisting of an adenovirus major late intervening
sequence, a mouse immunoglobulin intervening sequence, and a simian
virus (SV)40 polyadenylation sequence. The 5′ heterologous intron and
SV40 polyadenylation signal sequence are designed to increase the
stability of the transgene mRNA and to direct the mRNA out of the
nucleus into the cytoplasm where it can be effectively translated. The
expression cassette containing the
CAT reporter gene is
flanked by AAV left and right terminal repeats. The AAV terminal
repeats are intended to enhance transgene expression in primary and
slowly dividing cells and to serve as origins of replication. The
pCMVβ plasmid expressing the β-
Gal reporter gene was
kindly provided by Barry Byrne (Department of Pediatrics, University of
Florida, Gainesville). The β-
Gal insert was excised from
the pCMVβ plasmid (Clontech, Palo Alto, CA) and cloned into the
SnaBI and
SphI sites of pTR-UF5 under the
transcriptional control of the CMV promoter with a human growth hormone
polyadenylation sequence. Plasmid DNA was isolated from bacterial
cultures using alkaline lysis followed by anion-exchange column
chromatography (Qiagen, Valencia, CA). Endotoxin levels were measured
to be below 0.05 endotoxin units (EU)/μg DNA, by using a kit
(E-Toxate; Sigma, St. Louis, MO).
This study conformed to the ARVO Statement for the Use of Animals
in Ophthalmic and Vision Research and was approved by the Animal Care
and Use Committee of the University of Florida College of Medicine. Six
New Zealand albino rabbits were anesthetized with intramuscular
injection of ketamine (50 mg/kg) and xylazine (10 mg/kg), and topical
anesthesia was achieved with proparacaine eye drops. Five animals were
used for the CAT experiment, and one rabbit was used for theβ
-Gal experiment. Viscoelastic (Amvisc; Iolab, Claremont,
CA) was injected into the anterior chamber through a paracentesis
before entering through the filtration site. No attempt was made to
remove the viscoelastic at the end of the procedure. This was intended
to maintain the chamber, repress bleeding, and prevent flow of the
plasmid into the anterior chamber and away from the desired site of
transfection in the sclerostomy and bleb tissues. Filtration surgeries
were performed using a limbus-based approach, and a full-thickness
sclerostomy was made using a Kelly Decemet punch (Stortz
Instrument Co., St. Louis, MO) inserted through a limbal stab incision.
In the right eye of five rabbits, the pMP6-CAT plasmid was applied by
injection after closure of the conjunctiva and Tenon’s capsule with an
8.0 Vicryl suture. A 30-gauge needle was used to inject 100 μg of
plasmid dissolved in 100 μl of phosphate-buffered saline (PBS)
beneath the conjunctiva and Tenon’s capsule. Care was taken to
approach the bleb site from as far away as possible and to inject with
the needle tip centered just posterior to the sclerostomy site.
Antibiotic ointment was placed on the eye at the end of surgery.
In the left eye of the same five rabbits, the pMP6-CAT plasmid was
applied using a collagen shield placed under the conjunctiva and
Tenon’s layers before closure of the wound. A collagen shield (Bio-Cor
II 24-Hour Collagen Corneal Shield; Bausch & Lomb, Tampa, FL) measuring
16 mm in diameter was cut into four equal pie-shaped pieces. One of the
pieces was soaked for 10 minutes in 20 μl of PBS containing 100 μg
of plasmid. The shield piece was then placed under the conjunctiva and
Tenon’s capsule on top of and posterior to the sclerostomy site. The
wound was closed with an 8.0 Vicryl suture, and antibiotic ointment was
placed in the eye. No corticosteroid was administered.
Bilateral, nonblinding surgery was performed on the rabbits to permit
fewer numbers of animals to be used. Postoperative care of the animals
included daily eye examinations, observation of eating and drinking
behavior, and monitoring for signs of distress and pain. No animals
displayed signs of infection, distress, or dehydration during the
course of the experiment.
Eyes were harvested 48 hours after the surgeries were performed.
For each of the 10 eyes used in the CAT assay, samples were analyzed
from four locations. The filtration site specimen consisted of a
rectangular block of conjunctiva, Tenon’s capsule, and sclera
measuring 8 mm concentric with the limbus and 6 mm radially posterior
to the limbus. An identical specimen 180° opposite the sclerostomy
site was also harvested. The third specimen consisted of the entire
clear cornea, and the fourth specimen was the entire iris and ciliary
body complex. Tissue samples were stored at −80°C until measured for
CAT activity using the radioactive enzymatic protocol described by the
manufacturer’s technical bulletin (No. 84; Promega, Madison, WI).
Briefly, tissue samples were weighed, homogenized in 1 ml of lysis
buffer (250 mM Tris [pH 8.0], 5 mM EDTA, 0.02% sodium azide, 1 mM
paraoxon, and 10 μg/ml aprotinin), and centrifuged at
17,000g for 10 minutes at 4°C. Aliquots of the supernatant
solutions (100 μl) were added to 1.5-ml centrifuge tubes containing
25 μl of substrate solution creating final concentrations of 24 μM 14C-chloramphenicol (50 mCi/mmol; duPont NEN
Research Products, Beverly, MA) and 214 μM n-butyryl
coenzyme A. After incubation overnight at 37°C, 1.25 ml xylene was
added to the tube, and the reaction was vortexed for 1 minute. The
xylene phase containing the butyrylated 14C-chloramphenicol reaction product was
transferred to a new 1.5-ml centrifuge tube and extracted three times
with 200 μl of 250 mM Tris (pH 8.0). Levels of radioactivity were
measured in 500 μl of the xylene phase using liquid scintillation
counting, and the amount of CAT activity present in the sample was
calculated using the best-fit equation generated from a standard curve
of CAT activity, which ranged from 0.01 mU to 100 mU. Samples were
assayed in triplicate and the average CAT units were expressed as
milliunits per milligram tissue. Levels of CAT activity in the tissue
samples tissues were compared using analysis of variance (ANOVA) and
Tukey’s post hoc test with P = 0.05 considered
statistically significant.
Plasmid transfection at tissue and cellular levels was localized
by histochemical detection of β-gal activity using a β-gal staining
kit (Boehringer Mannheim; Indianapolis, IN). Briefly, 100 μg of the
plasmid expressing β-Gal was applied to the eyes of a
rabbit using the previously described injection method in one eye and
the collagen shield method in the other eye. Two days after surgery,
the rabbit was killed, and the eyes were fixed in situ for 5 minutes
with ice cold 2% formaldehyde and 0.02% glutaraldehyde in PBS. The
eyes were then removed and placed in the fixative solution for 2 hours
at 4°C. The fixed eyes were permeabilized using 0.02% nonidet P-40
for 15 minutes at room temperature, then rinsed and incubated in the
X-Gal substrate solution (Boehringer Mannheim) overnight at 37°C.
Tissues were photographed to document gross anatomic localization ofβ
-Gal expression and then were embedded in paraffin.
Sections were counterstained with nuclear fast red and photographed to
document cellular localization of β-Gal expression.
Figure 1 shows the average levels of CAT activity measured in different ocular
tissues 2 days after injection of plasmid into the filtration bleb or
absorption of the plasmid into the collagen shield. Two major results
were apparent. First, the average level of CAT activity in the
filtration bleb was approximately 30 times greater when the plasmid was
delivered by a collagen shield compared with injection of the plasmid
(mean ± SE of CAT activities, 2711 ± 567 mU/mg versus
92 ± 38 mU/mg,
P = 0.0001). Second, the average
levels of CAT activity in the other ocular tissue samples (cornea,
iris–cilliary body, and conjunctiva–Tenon’s–sclera opposite the
bleb site) was nearly undetectable for both delivery methods (mean ± SE of CAT activity, 1.6 ± 0.1 mU/mg,
P =
0.0001 compared with bleb tissue transfected by the shield method).
Thus, both plasmid delivery systems produced transfection that was
highly localized to the filtration site, and the use of a collagen
shield to deliver the plasmid greatly enhanced the level of
transfection of the filtration bleb without increasing transfection of
other ocular tissues.
As seen in
Figure 2A , β-
Gal staining was localized in the bleb filtration
tissues after treatment with the β-
Gal plasmid delivered
by the collagen shield method. The cornea, iris and surrounding
conjunctiva did not generate β-
Gal staining. The observed
pattern of localized staining of β-
Gal reporter gene in
the filtration bleb is consistent with the quantitative measurements of
CAT activity found in homogenates of the ocular tissues transfected
with the
CAT reporter plasmid.
Figure 2B shows the pattern
of β-
Gal staining at the cellular level. This field shows
an area of prominent staining of β-
Gal in fibroblast-like
cells of the Tenon’s capsule surrounding the bleb, although other
areas of Tenon’s capsule within the bleb showed fewer cells staining
for β-
Gal. β-
Gal staining was not found in
cells of the sclera or conjunctival epithelium. Inflammatory cells were
observed in the bleb tissues but minimal β-
Gal staining
was present in these cells (
Fig. 2B , inset). Identification of
fibroblasts and inflammatory cells was based on morphologic features
that are characteristic of these two cell types. Specifically, cells
were identified as fibroblasts based on the presence of multiple,
irregular, cytoplasmic extensions and an oval-shaped nucleus.
Inflammatory cells, specifically polymorphonuclear leukocytes, were
identified based on the presence of a segmented nucleus and multiple
cytoplasmic inclusion bodies. Scleral fibroblasts and conjunctival
epithelium were identified by their location in the tissue sections.
Similar patterns of β-
Gal staining were observed in the
companion eye transfected by injecting the β-
Gal plasmid
into the bleb.
Conjunctival wound healing after glaucoma filtration surgery is a
major determinant of the long-term clinical success of the procedure.
Failure of glaucoma filtration surgery is most often due to scarring in
the conjunctiva level at the bleb and sclerostomy sites. Information
from previous investigations in animal models and in patients have
helped to formulate a general model of conjunctival
scarring.
20 Similar to skin wound healing, conjunctival
scarring progresses through a sequence of phases. Initially, after
tissue damage and blood vessel rupture, the clotting cascade is
activated, which leads to degranulation of platelets and release of
multiple growth factors including platelet-derived growth factor
(PDGF), transforming growth factor (TGF)-β, epidermal growth factor
(EGF), insulin-like growth factors (IGFs), and fibroblast growth
factors (FGFs). The growth factors, together with arachidonic acid
metabolites, promote chemotaxis of inflammatory cells to the
injury site neutrophil are the first inflammatory cells of the immune
system to enter the wound area, accumulating within 6 hours and
disappearing by the third day after wounding.
21 Macrophages are the next inflammatory cells to migrate into the wound,
becoming the predominant inflammatory cell type at approximately 12
hours after injury, reaching peak numbers on approximately the third
day and decreasing by the fifth day. Lymphocytes comprise approximately
70% of the inflammatory cells in the wound area by the 10th day after
injury. Newly synthesized collagen fibers and elastin fibers are
detected in the subconjunctival wound area as early as 3 days after
surgery in animal models, reach a maximum approximately 7 days after
surgery, and then slowly decrease in subsequent weeks as the scar
tissue remodels.
Intuitively, a major goal of glaucoma filtration surgery is to identify
the molecule(s) that play key roles in regulating conjunctival scarring
and to develop agents that selectively and controllably inhibit excess
scarring and bleb failure. Chemotherapeutic drugs such as
5-fluorouracil and mitomycin C have powerful, although nonspecific
effects, on cells by inducing cell death or apoptosis. Both drugs are
used clinically to reduce conjunctival scarring in certain high-risk
patients, but these drugs have potential side effects. A unique aspect
of glaucoma filtration surgery healing is the bathing of wound tissues
by aqueous humor. The presence of multiple growth factors in normal
aqueous humor and tears, especially latent TGF-β2, suggests that
components in aqueous humor can influence conjunctival
scarring.
22 23 Recently, repeated injections of a
recombinant monoclonal neutralizing antibody to human TGF-β2
significantly inhibited conjunctival scarring in a rabbit model of
filtration surgery.
24 Transfection of conjunctival cells
with plasmids expressing antisense oligonucleotides,
ribozymes, or other genes that block scarring or promote
antiscarring activities may provide an additional approach to
preventing excessive scarring and failure of glaucoma filtration
surgery.
Plasmids generally are not considered optimal vectors for gene therapy,
in part because of their presumed inefficiency in cellular
transfection. These experiments demonstrated that substantial levels of
reporter gene expression could be achieved in ocular filtration surgery
tissues by simple injection of naked plasmid DNA into the bleb, and
furthermore, absorption of the plasmid into a collagen shield increased
the level of transfection by many times. Perhaps even more important
for clinical applications, the transfection was almost exclusively
localized to the bleb site without significant uptake in other ocular
tissues. This suggests that potential unwanted side effects that could
result from transfection of other tissues may be avoidable. In
contrast, postoperative injections of 5-fluorouracil can produce
well-known side effects in tissues away from the surgical site, most
notably toxicity to the ocular surface.
25
The localized gene transfection was achieved using a rabbit filtration
surgery model that mimics many of the key parameters of human
filtration surgery. However, certain anatomic aspects of the rabbit eye
require alterations in technique. The rabbit sclera is very thin and
soft, which prevents the construction of a reliable partial-thickness
scleral flap and necessitates the use of a full-thickness sclerostomy.
In addition, the anterior chamber is very shallow in the rabbit, which,
when combined with the absence of a scleral flap, results in a
prolonged period of postoperative flat chamber. This model, therefore,
necessitates the use of viscoelastic in the anterior chamber to prevent
a flat chamber, which would probably promote retrograde flow of plasmid
from the bleb into the anterior chamber and away from the desired
location of transfection. In human surgery where the sclerostomy is
guarded by a scleral flap, the use of a viscoelastic may not be
necessary. However, many surgeons routinely leave viscoelastic in the
anterior chamber in human surgery,
26 27 28 which together
with a guarded sclerostomy, should minimize retrograde flow of material
into the anterior chamber.
The reason that increased transfection was achieved with the collagen
shield is not known. It could be because a slow release of the plasmid
from the dissolving collagen shield resulted in longer exposure time of
the target tissues to the vector. Plasmid injected into the bleb
probably diffused away from the site much more rapidly. If this
explanation is correct, it suggests that other slow-release vehicles
may also enhance transfection.
It is also known that plasmid vectors generally produce relatively
short-term gene expression. The level of gene transfection at various
time points after intraoperative plasmid application must be delineated
in further experiments. However, gene expression that decreases over a
period of days or weeks may be desirable in this setting where the
primary goal is to regulate wound healing in the short term until a
stable fistula for aqueous egress is established. Furthermore,
short-term expression could be amenable to clinical scenarios in which
postoperative injections are titrated according to the monitored
clinical response. Retroviral and AAV vectors have longer term
expression due to integration into the genome, which limits the ability
to adjust gene expression. To overcome this potential problem, efforts
are under way to develop promoters that can be incorporated into
vectors and can be regulated by small molecules, such as tetracycline.
Another advantage of naked plasmid vectors is the minimal inflammatory
cell and neutralizing antibody humoral response produced compared to
adenovirus vectors or lipid–plasmid complexes.
The cellular localization of gene expression predominantly in
fibroblasts is an important finding. This seems to indicate that cells
that were present in the tissues at the time of exposure to the plasmid
were transfected, but that inflammatory cells that appeared later in
the healing process were not transfected. Therefore, if gene expression
in the inflammatory cells is desired in an eventual therapeutic model,
this may require repeated injections of plasmids into the filtration
bleb.
Short-term gene transfection may provide advantages over current
anti-scarring treatments for glaucoma surgery such as chemotherapeutic
drugs (5-fluorouracil and mitomycin C) because of the theoretical
ability to selectively target genes that play key roles in regulating
conjunctival scarring. In addition, because gene transfection should be
effective for days or weeks, it is possible that only a single
application to tissues is needed. In contrast, treatment with
neutralizing antibodies or other proteins typically require repeated
applications to be effective because the therapeutic proteins diffuse
away from the site of injection or are inactivated. Transfected cells
should continue to synthesize the transgene protein (or an antisense
oligo or a ribozyme) for an extended period. Before plasmid
transfection can be evaluated for effectiveness with therapeutic
transgenes, basic parameters for transfection of conjunctival tissues
must be established. The data reported here have helped to delineate
many of the issues of surgical technique, vector choice, transfection
quantitation, and transfection localization that are unique to glaucoma
filtration surgery. The results justify further investigations of gene
transfection as a therapeutic adjuvant to manipulate wound healing in
glaucoma filtration surgery.