LIPOADENOFECTION-MEDIATED GENE DELIVERY TO THE CORNEAL ENDOTHELIUM

Prospects for Modulating Graft Rejection¹

Abstract

Background. 

Gene transfer to the corneal endothelium has potential for the prevention or reversal of corneal allograft rejection. Previous work has examined adenoviral vectors for gene transfer to endothelium. These have a number of theoretical and practical disadvantages, both for experimental and clinical applications. We have therefore used lipoadenofection, in which plasmid DNA is delivered using a combination of liposomes and adenovirus, to transfer marker genes to the cornea.

Methods. 

Corneas were obtained from New Zealand White rabbits and cultured ex vivo using standard conditions. The corneas were transfected using either lipofection or lipoadenofection with plasmids encoding marker genes. The efficiency of gene transfer and the location and kinetics of gene expression were determined. We also investigated the delivery of a gene construct containing an inducible promoter that is activated by tumor necrosis factor(TNF), to determine whether expression of the relevant genes could be controlled by exogenous factors such as cytokines.

Results. 

This study shows that gene expression is limited to the endothelium and that expression is transient. Furthermore, we have shown that expression of a gene controlled by an inducible promoter only occurs when TNF is present.

Conclusions. 

These data indicate that lipofection is an efficient method to transfer therapeutic genes to the corneal epithelium, and that it can be used to transfer constructs that utilize an inducible promoter controlled by TNF. As TNF is present in the aqueous humor during allograft rejection, and this is in contact with the corneal endothelium, this has the potential to restrict expression of a therapeutic gene to rejection episodes in the cornea.

In the context of transplantation, gene therapy has the potential to modulate the immune response in order to prolong graft survival. We have investigated the feasibility of genetically modifying corneal allografts before transplantation ^(1-3). The cornea is an ideal tissue for the development of such strategies. Not only is the cornea readily accessible, it can also be maintained in culture for up to 1 month before grafting, allowing ex vivo manipulation. Moreover, the transparency of the cornea allows direct visualization of any consequences of gene transfer. In addition, there is a clinical need to prevent or reverse corneal graft rejection, especially for recipients in high risk groups^{(4, 5)}. In previous work, we have used adenoviral vectors to transfer genes to the corneal endothelium^(1-3). These have proved highly efficient, and high levels of gene expression have been observed for up to 14 days. However, the adenoviral vectors used in current studies are unlikely to reach clinical application on account of the inflammatory response induced^(6-9). Another constraint in experimental evaluation is the necessity to construct recombinant vectors, which is a time-consuming process. In addition, the use of conditional promoters, either tissue-specific or inducible, would be desirable in some applications. However, these promoters can become constitutively active if placed in adenoviral vectors ⁽¹⁰⁾.

We have for these reasons sought to develop nonviral approaches to gene transfer. In preliminary experiments, we have shown that liposome-mediated gene delivery to corneal endothelium is very inefficient (unpublished observations). We have therefore examined the approach first described by Yoshimura and colleagues ⁽¹¹⁾ of using adenovirus-assisted gene transfer or lipoadenofection. These authors showed that adenovirus enhances the delivery of genes by liposomes to a variety of cell lines ⁽¹¹⁾. Other groups have extended this work, showing that lipoadenofection is an efficient method of gene transfer to breast tumor cells ⁽¹²⁾, to adipocytes and hepatoma cells ⁽¹³⁾ and to vascular tissue⁽¹⁴⁾. This has the advantage over adenoviral gene transfer that the gene of interest does not need to be cloned into a viral vector, and that inactivation of the viral particles would reduce the immunogenic nature of the vector ⁽¹⁵⁾.

We have now demonstrated that the use of lipoadenofection is an efficient method of delivering marker genes to an intact organ, the cornea, and so could be applicable in therapeutic gene delivery in corneal transplantation. We have also shown that transfer of a construct that contains a conditional promoter can be accomplished, and that gene expression is then restricted by the promoter, only occurring after addition of the appropriate cytokine to the corneal culture.

MATERIALS AND METHODS

Reagents. RPMI 1640, minimal essential medium(MEM^*), Dulbecco's modified Eagle's medium, fetal calf serum (FCS), Opti-MEM I reduced serum medium, the cationic liposome LipofectAMINE, L-glutamine, penicillin, and streptomycin were purchased from Gibco (Paisley, UK). Cesium chloride and X-gal were from NBL Gene Sciences(Cramlington, UK). Ortho-phenylenediamine tablets were obtained from Dako(High Wycombe, UK). All other chemicals were purchased from Sigma Chemical Co.(Poole, UK), unless otherwise stated.

Cell lines. The in vitro model for gene transduction was provided by the hybrid cell line EAhy 926. EAhy 926 were originally established by fusion of the human lung carcinoma cell line A549 and umbilical cord endothelial cells ⁽¹⁶⁾. The cells were passaged and maintained in growth medium consisting of RPMI 1640 supplemented with 5% FCS, 2 mM L-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin.

Recombinant adenovirus. The recombinant adenovirus (Ad0) has been described previously ⁽⁹⁾ and was a kind gift of Dr. M.J.A. Wood (Human Anatomy, University of Oxford). It is an E1-deleted replication-deficient adenovirus, with no inserted gene in the deleted region of its genome, and so does not direct the production of any nonviral proteins. The virus was propagated and recovered from the transcomplementing 293 cell line, and purified using two rounds of CsCl density centrifugation, as described ^{(1, 3)}. The titer of Ad0 was determined using a plaque-forming assay with 293 cells; in this study, two preparations were used with titers of 1.2×10⁹ plaque-forming units (pfu)/ml and 8×10⁹ pfu/ml. The cells were maintained in growth medium consisting of Dulbecco's modified Eagle's medium, 10% FCS, 2 mM L-glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin.

Plasmids. The following plasmids were used: pCMV/β-gal(Clontech, Palo Alto, CA) and pCDNA3/CAT (Invitrogen, Leek, The Netherlands), which contain the lacZ gene and the chloramphenicol acetyl transferase (CAT) gene, respectively, under the transcriptional control of the human cytomegalovirus (CMV) promoter, and an empty plasmid (with no insert) pCDM8 (Invitrogen). A construct (pELAM-1pro./CAT) containing the CAT gene under control of the E-selectin promoter was made as described below.

Isolation and cloning of the E-selectin promoter. A primer pair(5′ primer: 5′-AGATTGAAGCTTATGCATGTAGACTATGGATGACAAACC-3′ and 3′ primer: 5′-GCAAAGTCTAGATCTCTCAGGTGGGGTATCACTG-3′) recognizing-383 base pairs upstream and +80 base pairs downstream of the transcription site ⁽¹⁷⁾ was designed to include HindIII andXba I sites (italics) at the 5′ and 3′ ends, respectively. The E-selectin promoter was isolated from human genomic DNA prepared from A549 cells using the polymerase chain reaction under the following conditions; 94°C for 1 min followed by addition of Taq polymerase, then 25 cycles of 94°C for 30 sec, 60°C for 30 sec, and 72°C for 30 sec. Amplified DNA was cloned into the pCRII vector(Invitrogen) and then subcloned into a CAT reporter vector pCAT Basic(Promega, Madison, WI) at the HindIII and XbaI restriction sites.

Corneal samples. Corneas were excised from the eyes of New Zealand White rabbits that had been killed by intravenous pentobarbitone injection. All corneas were transfected within 48 hr of removal. The corneas were cultured in MEM supplemented with 10% FCS, 2 mM L-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin (complete culture medium) in 5% CO₂ at 37°C. For long-term incubation, the medium was replaced every 48 hr.

Transfection and culture conditions for corneas. The optimal conditions given here for the transfection of corneas were determined in preliminary experiments (data not shown). In some experiments, rabbit corneas were cut into four segments. For each transfection, the liposome LipofectAMINE, which consists of dioleoylglycero-3-phosphoethanolamine and 2,3-dioleyloxy-N-[2(spermine carboxamido)ethyl]-N,N-dimethyl-1-propanaminium trifluoroacetate, was diluted to a concentration of 4 μg in 100 μl of Opti-MEM I reduced serum medium and the plasmid DNA was prepared at a concentration of 0.5 μg in 100 μl of Opti-MEM I reduced serum medium. The two solutions were then mixed together. For liposome-DNA-Ad0 complexes, the virus was then added at a concentration of 8.0×10⁶ pfu per quarter of cornea. The mixture was incubated at 37°C for 30 min. The corneas were washed once with serum-free medium and the complex, together with 800 μl of Opti-MEM I reduced serum medium, was overlaid. The corneas were then incubated at 37°C in 5% CO₂ for 5 hr before addition of 1 ml of complete culture medium. For the time course, they were cut into eight segments. The medium was replaced with complete culture medium at 24 hr. The corneas were analyzed after 48 hr or as required, using the colorimetric assay or CAT assay on cell lysates or staining for β-gal activity.

Assay for β-gal activity. β-Gal activity was measured in cell lysates by using a substrate for the enzyme o-nitrophenyl-β-D-galactopyranoside. Briefly, corneal specimens were washed five times with phosphate-buffered saline (PBS) and lysed with 90 μl of Tris·HCl (250 mM, pH 8.0) by three freeze-thaw cycles to release cytoplasmic content. Tissue debris was removed by centrifugation at 14,000 rpm in a microcentrifuge, for 10 min. Fifty microliters of 2×buffer (200 mM sodium phosphate, 2 mM magnesium chloride, 100 mM β-mercaptoethanol, 1.33 mg/ml o-nitrophenyl-β-D-galactopyranoside) were added to 50 μl aliquots of each sample. The mixture was incubated at 37°C until the yellow color developed, and the reaction was stopped by adding 150 μl of 1 M sodium carbonate. Absorbance was measured at 414 nm in a microtiter plate reader (Titretek Multiscan MCC/340, Labsystems, Joensuu, Finland). Enzyme levels were determined by a standard curve constructed using various titrations of β-gal. The total protein content of the corneal lysates was estimated using a modified Lowry method ⁽¹⁸⁾, with absorbance read at 560 nm using the microtiter plate reader. β-Gal activity was presented by normalizing to total protein content of the corneal sample.

Histochemical analysis of β-gal. In order to determine the transfection efficiency and tissue localization, the cornea specimens were incubated with X-gal, a substrate for β-gal. Transfected corneas were washed twice in PBS and incubated at 4°C for 10 min in fixative composed of 2% formaldehyde and 0.2% glutaraldehyde in PBS. After washing twice with PBS, the cells were incubated with 1 ml of X-gal solution(5 mM potassium ferrocyanide, 5 mM potassium ferricyanide, 2 mM magnesium chloride, 200 μg/ml X-gal) at 37°C for 5-10 hr. Expression oflacZ results in a nuclear-dominant indigo staining pattern when cells are exposed to the chromogenic substrate X-gal. For histological examination, the specimens were embedded in paraffin wax, 5-μm sections were cut, mounted on glass slides, and counterstained with 0.1% nuclear fast red.

CAT enzyme-linked immunosorbent assay (ELISA). CAT levels were measured by a CAT ELISA following the instructions of the kit manufacturer(Boehringer-Mannheim Biochemica, Mannheim, Germany). Briefly, the wells were rehydrated with 250 μl of sample buffer and incubated for 5 min at room temperature. The buffer was then removed, and 200 μl of the cell lysate, normalized to protein content, was added and incubated for 90 min at 37°C. After washing with 250 μl of washing buffer four times, 200 μl of sheep anti-CAT, labeled with digoxigenin, was added and incubated for 1 hr. The samples were then washed four times, and 100 μl of peroxidase-conjugated anti-digoxigenin antibody was added and incubated 1 hr at 37°C. The peroxidase substrate ABTS (200 μl) was then added after washing. The absorbance was determined using an ELISA reader at 405 nm. CAT activity was presented by normalizing to total protein content of the corneal sample.

Induction of E-selectin promoter with tumor necrosis factor(TNF). Rabbit corneas were transfected as described above with pELAM-1pro./CAT plasmid under the transcriptional control of the E-selectin promoter. They were then cut into quarters and incubated for 24 hr before being stimulated with 10 ng/ml recombinant human TNF-α (Peprotech, London, UK) after 24 hr; lysates were prepared 24 hr later. The plasmid pCDNA3/CAT under the control of CMV promoter was used as a positive control and the empty plasmid pCDM8 as the negative control. The lysates were then analyzed for CAT activity as described above.

RESULTS

Addition of virus to LipofectAMINE significantly enhances marker gene expression. In preliminary experiments, the ability of liposomes, with and without the addition of adenovirus, to transfer DNA to vascular endothelial cells was determined using the EAhy 926 cell line.

The cells were transfected with 0.3 μg of DNA (pCMV/β-gal DNA) and 4 μg of LipofectAMINE, with or without 8.0×10⁶ pfu of Ad0 per well. This concentration of virus was found in preliminary studies to show minimal toxicity to the cells (data not shown). Higher concentrations of virus increased efficiency but at the cost of significant toxicity of the cells(data not shown). It was found that 7 times more β-gal was produced using the liposome-virus-DNA combination (Fig. 1, left). This indicates that, after lipoadenofection, more protein was expressed than by transfection with liposomes alone. Staining of the cells indicated that the efficiency of transfection was significantly increased, as shown inFigure 1, middle; between 40 and 50% of the cells expressed β-gal after lipoadenofection, compared with 7% after lipofection alone. No significant gene delivery was seen after administration of DNA and virus alone.

In order to confirm the capability of the liposome-virus-DNA combination to deliver different gene constructs, a plasmid encoding for chloramphenicol acetyltransferase type I enzyme (CAT) was delivered to the cells. The levels of protein produced were measured by CAT ELISA. As shown inFigure 1, right, the levels of CAT protein are 8 times higher using the liposome-virus-DNA system when compared with the liposome-DNA complex. This demonstrates that the lipoadenofection can efficiently deliver different constructs to the cells.

Lipoadenofection is an efficient method to deliver a reporter gene to the cornea. Once the conditions had been optimized using the in vitro cell line, the strategy was applied to rabbit corneas. New Zealand White rabbit corneas were divided into four, and each quarter was transfected with 0.5 μg of plasmid pCMV/β-gal DNA and mixed with 6 μg of LipofectAMINE and 8.0×10⁶ pfu of the adenovirus. (It should be noted that, at these concentrations, highly efficient gene transfer would be seen using recombinant adenoviral vectors^(1-3).) Forty-eight hours after transfection, specimens were incubated with X-gal, to identify cells expressinglacZ. Lipoadenofection was considerably more efficient in reporter gene transfer than lipofection alone (Fig. 2, top). Using liposome-DNA complex, negligible numbers of endothelial cells stained forβ-gal (typically 0-2 cells per high power field), whereas considerably more cells (≈80 cells per field) were transfected with the liposome-virus-DNA system (Fig. 2, middle). This increase in efficiency after lipoadenofection was associated with an ≈14-fold increase in the total levels of β-gal protein detected by colorimetric assay (Fig. 2, bottom).

Marker gene expression is restricted to the corneal endothelium. On transverse histological sections, lacZ expression was found to be restricted to the corneal endothelium (Fig. 3). There was no detectable β-gal in the epithelium or stroma of transverse corneal sections. Figure 3 shows that the efficiency of transfection was approximately 15% of endothelial cells using lipoadenofection.

Time course of β-gal expression.Figure 4 shows a time course obtained for β-gal expression in transfected corneas. The plasmid pCMV/β-gal was delivered to whole corneas by lipoadenofection using 2 μg of DNA and mixed with 24μg of LipofectAMINE and 32.0×10⁶ pfu of the adenovirus. After a 5-hr incubation period, the corneas were cut into eight segments, which were then incubated for selected periods before determination of β-gal activity. Maximal levels were found between days 3 and 10, after which the levels of the enzyme decreased to undetectable levels by day 28.

Induction of the E-selectin promoter by exogenous TNF-α. In order to determine whether lipoadenofection allowed the use of conditional promoters in the cornea, the pELAM-1pro./CAT plasmid was used. This encodes for CAT under the control of the E-selectin promoter, activity of which is up-regulated by TNF-α. After lipoadenofection with pELAM-1pro./CAT, corneas were cut into quarters, and 24 hr later some quarters were incubated in TNF. The levels of CAT were measured 24 hr later. As can be seen inFigure 5, there is a 9-10-fold up-regulation of CAT expression after TNF stimulation. Thus, lipoadenofection allows for conditional promoters such as the E-selectin promoter to be induced in the cornea. As a positive control, we used pCNA3/CAT, in which the CAT gene is under control of the constitutive CMV promoter. In EAhy 926 cells, expression of CAT from this construct is not affected by the addition of TNF-α(data not shown).

DISCUSSION

Despite the relative immune privilege of the corneal bed⁽¹⁹⁾, there is still a significant graft rejection rate. Those recipients at high risk of rejection include young patients⁽⁵⁾, patients with a previous rejected transplant or a vascularized recipient corneal bed ^{(4, 5)}. Gene-based approaches may become useful for the prevention of allograft rejection. Although adenovirus is an efficient vector for the delivery of genes to the cornea, there is concern about the inflammatory reaction that such a vector may cause. In this article, we report on an alternative strategy, termed lipoadenofection, in which the gene was efficiently delivered to the cornea by a liposome adenovirus complex. This approach has the potential for reducing the immunogenic nature of the vector. Furthermore, in the context of experimental model systems of transplantation, it allows the expression of a number of different constructs, without the need to clone them into the viral vector.

We have used rabbit corneas as our model system because they closely resemble human corneas in their size and in the comparatively low replicative capacity of the endothelial cells. We have shown that lipoadenofection is efficient at delivering genes to the cornea, giving around 14 times more gene expression as liposomes alone. The expression of the gene was restricted to the corneal endothelium, as seen after adenoviral transfection of the cornea^(1-3). This is highly appropriate, as the endothelium is the major target for immunological attack in graft rejection⁽²⁰⁾.

The expression seen after lipadenofection was transient. The reasons for this are not known, although, intriguingly, in an earlier study in which adenoviral vectors were used to transfer genes to human corneal endothelium, we showed that viral DNA persisted for a long period, even though gene expression was transient ⁽³⁾. In this case, therefore, loss of viral DNA could not explain the short term expression, indicating that either the promoters responsible for transcription had been inactivated or that the recombinant DNA had been modified or sequestered in some way as to prevent gene transcription. It is reasonable to suppose that similar mechanisms would operate after lipoadenofection.

The exact role of the adenovirus in the transfection of liposome-DNA complexes is not clear. Previous work on adenolipofection suggests that the virus may act as a by-stander, with the liposome-DNA complex and the adenovirus being internalized into the same endosomal compartment. Endosomal disruption by the virus would then allow the liposome-DNA complex to escape into the cytosol ⁽¹¹⁾. However, the liposome is a vital ingredient, as combinations of the virus and DNA were incapable of transfecting cells (¹³ and this study). Alternatively, the virus may associate with the liposome-DNA system. This is supported by data from Raja-Walia and collaborators ⁽¹⁴⁾, who demonstrated that the three components, liposome, DNA and adenovirus form a complex as determined by gel electrophoresis. There are several potential sites where the adenovirus complexed to the liposomes could assist the gene delivery. First, adenovirus may help attachment to the cell surface. Second, interaction of the penton proteins with α_v integrins may promote internalization of the complex. Third, virus-mediated endosomal disruption at low pH may allow the complex to escape into the cytoplasm. Finally, the virus may help translocation of the DNA to the nucleus. At present, it is not known which of these pathways contributes to the increased gene expression.

In many settings, regulated expression of the therapeutic gene, using conditional promoters, would be desirable. For example, tissue-specific promoters would allow restricted expression of therapeutic proteins, and inducible promoters the exogenous regulation of expression. However, many viral vectors (including adenoviral vectors) have a tendency to override such conditional promoters, leading to constitutive expression of the gene⁽¹⁰⁾. In this study, we explored the possibility of delivering constructs containing the conditional promoters using lipoadenofection. We chose a construct containing the E-selectin promoter, which is activated by TNF and other cytokines ^{(17, 21)}. High levels of TNF are present in aqueous humor during corneal allograft rejection ^{(22, 23)}, and so the use of a TNF responsive promoter might allow expression of any potential therapeutic gene to be restricted to rejection episodes. In this study, we have shown that E-selectin promoter remains conditional after lipoadenofection to the cornea, with strong induction of gene expression after incubation with TNF-α. This therefore affords a useful approach to the control of expression of a therapeutic gene in the context of corneal transplantation.

The lipoadenofection does utilize adenovirus, and so cannot prevent all the inflammatory responses of adenovirus-mediated gene delivery. However, inactivation of the viral DNA by short wave UV light or long wave UV light in combination with psoralen would arrest the production of viral proteins in transfected cells, thus greatly reducing the immunogenic and inflammatory nature of the vector ⁽¹⁵⁾.

In summary, the major important advantages of the lipoadenofection strategy demonstrated in these studies are (a) the feasibility of moderately efficient transfer of exogenous DNA without the necessity of cloning into recombinant viral vectors and (b) the feasibility of conditional promoter control of transcription after lipoadenofection, a facility found to be poorly conserved in orthodox adenovirus-mediated gene transfer to other cell types⁽¹⁰⁾. Lipoadenofection therefore has potential in the development of gene based approaches to a number of disorders of corneal endothelium, including modulation of allograft rejection.

Acknowledgments. The authors thank Professor Mary Ritter for comments on the manuscript; Dr. M.J.A. Wood for supply of Ad0; and Professor Thiemermann and Julie Piper, Fiona Myint, and Joanne Bowes of the William Harvey Research Institute for providing corneal specimens.

Figure 1: Lipoadenofection of EAhy.926 cells. The endothelial hybridoma cell line was transfected with the pCMV/β-gal plasmid using either liposomes alone (LD), liposomes and adenovirus (LDV), or virus alone (DV). Untransfected cells were used as a control (-). After 48 hr, the amount ofβ-gal expressed was measured using a colorimetric assay (left) (results expressed as milliunits of β-gal/10 μg of total protein) or the efficiency of transfection was determined by counting blue cells after incubation with X-gal (middle). Similar data was obtained for the pCDNA3/CAT vector (right), with the total level of CAT expressed determined by ELISA. All results represent the mean ± standard deviation of triplicate cultures.

Figure 2: Lipoadenofection of corneas. Rabbit corneas were divided into four sections and transfected with the pCMV/β-gal plasmid using liposomes alone (LD) or liposomes and adenovirus (LDV). Untransfected corneas were used as a control (-). After 48 hr, the efficiency of gene transfer was assessed by X-gal staining (top and middle), and the total amount of β-gal expressed was determined by colorimetric assay (bottom). The photograph of the cornea shows an “en face” view of the endothelial surface, with transfected cells staining blue (original magnification, ×20; Olympus CK2 inverted microscope); the curvature of the cornea causes the image to be out of focus. The efficiency (middle) is indicated as the number of blue cells per random microscope field. The results represent the mean ± standard deviation of triplicate cultures.

Figure 3: Localization of gene expression in corneal endothelium. New Zealand White rabbit corneas were lipoadenofected with pCMV/β-gal plasmid DNA. Forty-eight hours later, the corneas were incubated with X-gal and embedded in paraffin wax. Transverse sections were cut and counterstained with 0.1% nuclear fast red. β-Gal expression (arrows) is restricted to the endothelial monolayer that forms the posterior surface of the cornea and lies on the Descemet's membrane (DM), with no staining seen in the epithelium(anterior surface, not shown) or the stroma.

Figure 4: Time course of β-gal expression after lipoadenofection. Rabbit corneas were transfected with pCMV/β-gal plasmid DNA using lipoadenofection and then cut into eight segments. At selected times, theβ-gal content of the corneas was determined using the colorimetric assay. Results shown are the aggregate of eight different corneas and represent the mean ± standard error of the mean. Maximal expression is seen between days 3 and 10, after which β-gal levels decline, falling to undetectable levels between days 21 and 28.

Figure 5: Stimulation of E-selectin conditional promoter by TNF after lipoadenofection. Rabbit corneas were lipoadenofected with pCDNA3/CAT (CMV promoter), pELAM-1 pro./CAT, or pCDM8 (negative control). Twenty-four hours later, they were incubated in 10 ng/ml TNF-α (+TNF) or left in normal medium (-TNF). Twenty-four hours later, CAT levels were determined by ELISA. A strong induction of CAT expression was seen in corneas transfected with pELAM-1 pro./CAT after TNF stimulation, with low levels seen in the absence of TNF. The results represent the mean ± standard deviation of triplicate corneal samples.