Search International and National Patent Collections
Some content of this application is unavailable at the moment.
If this situation persists, please contact us atFeedback&Contact
1. (WO1997015330) HYALURONIC ACID AS DNA CARRIER FOR GENE THERAPY AND VEGF ANTISENSE DNA TO TREAT ABNORMAL RETINAL VASCULARIZATION
Note: Text based on automatic Optical Character Recognition processes. Please use the PDF version for legal matters

HYALURONIC ACID AS DNA CARRIERFORGENE THERAPY AND VEGF ANΗSENSE DNATO TREAT ABNORMAL RETINALVASCULARIZATION

This invention relates to use of hyaluronic acid to target active agents which ablate the function of
targeted genes in the control or treatment of disease. In 5 one embodiment, this invention relates to a method and
composition for treating ocular diseases, in particular
retinal disease involving neovascularisation of the choroid and/or retina. It makes use of the phagocytic
characteristic of specific cells in the eye to provide an 0 effective manner of delivering an active agent to the
target, for either short term or long term treatment of
neovascularisation. The methods and compositions of the invention are useful for delivering DNA, RNA, anti-sense nucleotides, peptides or other therapeutic agents to
5 phagocytic cells or surrounding cells .

BACKGROUND OF THE INVENTION
A) Hyaluronic Acid as an Adjuvant or Targeting Agent
Hyaluronic acid (HA) is a large, complex
oligosaccharide consisting of up to 50 000 pairs of the 0 basic disaccharide glucuronic acid-β(l-3) N-acetylglucos- amine β(l-4) . It is found in vivo as a major component of the extracellular matrix. Its tertiary structure is a
random coil of about 50 nm in diameter.
HA has the ability to bind a large amount of 5 water, which in vivo makes it a viscous hydrated gel with viscoelastic properties. It is found in this form in the mammalian eye, both in the vitreous and in the
extracellular matrix.
HA has been used in the treatment of certain 0 diseases and conditions of the human body both systemically and topically, because of its ability to target an active agent to sites where the disease or condition is localised (International Patent Publications No. WO 91/04058 and No. WO 93/16733) . It has been shown that HA forms depots, for 5 example at the injured carotid artery (relative to uninjured contralateral arteries) and in colorectal tumours growing in experimental animals, and is retained in the skin of such animals. In all these cases, the sites of the deposits are areas of high HA receptor expression,
indicating that HA targets specifically to tissues that are expressing high levels of these receptors, particularly to tissues undergoing unusual proliferation and migration, including tissues responding to injury, inflammation, development, and tumorigenesis.
The characteristic of HA which is important to its action as a potential adjuvant iβ its ability
simultaneously to bind to other molecules and to bind to cell membranes. Cell surface receptors specific for HA have been identified, including the histocompatibility antigen CD-I4, the receptor for hyaluronic acid-mediated motility (RHAMM) , intercellular adhesion factor (ICAM), and some homologous proteins in the CD44 family. The binding of virus to the cell membrane facilitated by HA would allow the usual endocytotic mechanisms of viral uptake to be more effective.

B) Diseases of the Bye
A variety of ocular diseases such as macular degeneration and diabetic retinopathy are characterised by neovascularisation of the choroid and/or retina. This process is the major cause of blindness in patients
suffering from these conditions.

Prior Art Treatments
In age-related macular degeneration (ARMD), the formation and haemorrhaging of a subretinal neovascular membrane (SRNVM) results in rapid and substantial loss of central vision. Various treatments are available, but all are unreliable. Laser photocoagulation is the most acceptable type of treatment, but it still suffers from the disadvantages that damage by the laser rays causes dense, permanent scotoma (Schachet, 1994; Ibanez et al, 1995 and Hudson et al, 1995) resulting in temporary loss of vision, and inability to prevent progression of the condition in the long term because of recurrence of the neovascular membrane.
Thus this treatment provides an advantage only in terms of preventing profound visual loss.
Similarly, surgical removal of the SRNVM or of subretinal blood, or re-positioning of the fovea by
rotating the retina have largely been unsuccessful, due to post-surgical complications and to minimum or temporary improvement in vision. These invasive forms of treatment and the corresponding complications therefore far outweigh the advantages gained, and are limited in usefulness.
Administration of interferon α2a, which has some anti-angiogenic activity (Fung, 1991; Guyer et al, 1992 and Engler et al, 1994) and transplantation of retinal pigment epithelial (RPE) cells (Algvere et al, 1994) have also proved to be of limited usefulness, and initial promising results obtained with small groups of patients have not been confirmed in larger trials.
In addition to laser photocoagulation which, as described above, suffers from various disadvantages, the other main method of treating diabetic retinopathy is the control of blood glucose and blood pressure. The efficacy of such forms of treatment is limited by the motivation and compliance of the patient involved.
About 30% of the population above age 75 suffers from macular degeneration, and about 3 in 1000 individuals suffer from diabetic retinopathy. As each of these numbers will increase due to the aging of the population, and the increase in incidence of diabetes, there is a need for a more effective manner of treating these and other ocular diseases mediated by neovascularisation.

Mechanism of Neovascularization
Vascular endothelial cell growth factor (VEGF) is a dimeric, disulphide-bridged glycoprotein which is well- known to be synthesised and secreted by a variety of normal as well as tumour cells . Recent observations indicate that VEGF is frequently detected in the neovascular retinal membranes of patients with diabetes (Malecaze et al, 1994), and in the ocular fluid from patients with either diabetic retinopathy or with central retinal vein occlusion (Aiello et al, 1994) . More recently, it was found that VEGF expression was induced in conditions such as central vein occlusion, retinal detachment and intraocular tumours . In a rabbit model, levels of VEGF mRNA were elevated in the hypoxic region of the retina following induction of retinal vein occlusion. (Pe'er et al, 1995) . Stimulation of VEGF expression by hypoxia has also been observed in other animal models (Pierce et al, 1995; Miller et al, 1994), and in vitro in all types of cell cultures (Simorre-Pinatel et al, 1994; Hata et al, 1995 and Thiema et al, 1995).

C) Anti-Sense DN and Gene Therapy in Treatment of
Diseases
The suppression of expression of genes encoding proteins which mediate undesirable activity has been achieved in a variety of situations by the introduction or in situ production of 'anti-sense' DNA sequences in the target cells. These anti-sense sequences are DNA sequences which, when transcribed, result in synthesis of RNA whose sequence is antiparallel to the sequence encoding the protein. Such anti-sense sequences have been tested in a number of viral diseases. Alternatively, anti-sense oligodeoxynucleotideβ can be introduced into target cells; such short sequences are not themselves transcribed, but inhibit transcription and/or subsequent translation of the corresponding sense DNA sequence in the target cell.
Until recently it was widely thought that the minimum sequence length necessary in order to effect antisense inhibition of gene expression was 12 to
14 nucleotides (Wagner, 1994) . However, it has now been shown that the specificity of binding to the target sequence can be sufficiently enhanced by use of modified oiigonucleotides comprising C-5 propyne pyrimidines and phosphorothioate internucleotide linkages that sequences as short as 7 or 8 nucleotides are effective in providing gene-selective, mismatched sensitive, ribonuclease
H-dependent inhibition, in which flanking sequences of the target RNA are important in determining specificity (Wagner et al, 1996) .
However, successful use of anti-sense nucleotides to counter expression of a gene in vivo is limited by factors such as the need for specific suppression of mutant gene expression (Milan, 1993; Mclnnes and Bascom, 1992), or the need for high concentrations of the anti-sense
nucleotides (Akhtar and Ivinson, 1993).
To date, this form of therapy has largely
involved use of anti-sense sequences packaged in liposomes, or direct application of antisense cDNA or oiigonucleotides to the site of disease. Thus attempts to increase uptake of anti-sense sequences into the target cell by
encapsulating these sequences in liposomes have been largely unsuccessful. It is also difficult to target liposomes efficiently, and uptake iβ even lower than with viruses .
The targeting may also be achieved by virus-mediated DNA transfer, using viruses such as the Sendai virus. Sendai virus is an RNA virus which has been shown to deliver DNA and proteins into cells with more than 95% efficiency (Kaneda et al, 1987). In this gene transfer system, DNA nuclear protein complex in liposomes iβ
directly introduced into the cytoplasm of the cell by the fusion activity of Sendai virus . The DNA is delivered rapidly into the nucleus with nuclear protein. Sendai virus-mediated gene transfer occurs by fusion of the virus with the cell membrane, and bypasses the endocytic pathway. Recently, highly efficient delivery of anti-sense or plasmid DNA into target cells by Sendai virus has been observed. Both the anti-sense and plasmid DNAs retained their activity not only in culture but also in vivo (Kaneda et al, 1987) . However, the use of this virus is limited by the fact that there are no suitable constructs available at present to use as vectors. In addition, the transferred DNA can only be expressed for a limited period of time since the gene transfer is mediated by fusion.
Retroviruses have been widely used for somatic tisβue gene therapy (Boris-Lawrie and Temin, 1993) . They can target and infect a wide variety of host cells with high efficiency, and the transgene DNA integrates into the host genome. Theoretically, the integration of the DNA will provide permanent production of the transgene which could result in permanent rescue of the cells. However, retroviruses cannot infect non-dividing cells (Salmons and Gτinzburg, 1993) . Furthermore, the retrovirus particles are unstable in vivo, which makes it difficult to achieve high virus titre with inoculation. In addition, there are significant concerns regarding the oncogenicity of the integrated viruses. The inability of retroviruses to infect non-dividing cells means that they cannot be
selected aβ candidateβ for gene transfer in the eye, as the most important target cells such as photoreceptors and RPE cells are non-dividing cells.
The usefulness of herpes simplex virus vectors has been limited by their poor efficiency of infection (Culver et al, 1992) . Two types of vectors have been developed, namely the replication defective recombinants and the plasmid-derived amplicons. The latter requires a helper virus. Although the toxic genes can be removed from the herpes simplex virus with difficulty, the constructs remain cytotoxic (Johnson et al, 1992) . In addition, the long term expresβion of the sequences inserted has been unsuccessful to date, and there are problems with the regulation and stability of the constructs . The
application of modified herpes simplex viruses to the eye in gene therapy poses major concerns because of their pathogenicity. Herpes zoster virus infection causes serious infections in the eye, frequently resulting in blindness requiring corneal transplantation.
Adenoviruses have been widely uβed for gene transfer in both non-dividing and proliferating cells.
They can accommodate DNA up to 7.5 kb, and provide
efficient transfection and high viral titre. The main advantage of using these rather than retroviruses is the ability to infect a wide range of non-dividing target cells (Kozarsky and Wilson, 1993). Replication-defective
adenoviruses are considered to be relatively safe, in that these viruses are common pathogens in humane, usually causing relatively benign conditions such as colds . The vectors carry tumour genes with a deletion mutation, lowering the possibility of becoming oncogenic (Siegfried, 1993) . In the first experimental gene therapy trial approved by the ϋS National Institutes of Health
Recombinant DNA Advisory Committee, recombinant
adenoviruβeβ were uβed to treat individuals suffering from cystic fibrosis.
However, the main disadvantage of adenoviruβes is their transient gene expression. This is a result of the lack of integration of the transgene into the cellular genome. Furthermore, few attempts at gene delivery to non-dividing cells have been successful. The first successful gene transfer into the brain, which consists of non-dividing cells, was reported in 1993 using adenoviruses (Le Gal La Salle et al, 1993).
These results indicate that gene therapy is a theoretically viable approach in the treatment of diseases, but that the technical difficulties of efficient targeting and uptake need to be overcome by using viruses which adhere to and are taken up by the target cells . This process is inefficient, and the use of viruses may entail an undesirable level of risk of iatrogenic disease.
Positive results have, however, been published that teach that regulation of biological processes by gene therapy is feasible.

There iβ therefore a need for improved methods of targeting gene therapy for the treatment of disease, and for suitable compositions comprising hyaluronic acid for use in such treatment.

D) Gene Therapy and Ocular Disease
In Australian Patent Application No. 75168/94 (Hybridon Inc), it was shown that in vitro expression of murine VEGF could be inhibited in COS-1 or NB41 cells by incubation with 19- to 21-mer anti-sense oiigonucleotides based on murine VEGF. A 21-mer antisense nucleotide targeted against the translational stop site was shown to be the effective sequence. There is no disclosure or suggestion of specific targeting of sequenceβ to any tissue in the eye, or of treatment of any ocular conditions other than diabetic retinopathy.
In U.S. Patent No. 5,324,654, a method of
stimulating proliferation of non-malignant cells iβ
disclosed. The method comprises the in vitro treatment of cells with an anti-sense nucleotide corresponding to the retinoblastoma (Rb) gene to inhibit expression of the Rb gene product, resulting in suppresβion of the expression of proteins which inhibit cell growth. In this way,
proliferation of cells iβ encouraged. The proliferated cells can then be re-implanted if desired, and the cells may be genetically engineered to replace a βpecific gene prior to re-implantation. However, there iβ no reference to use of this anti-sense sequence to treat conditions of the eye. The invention of US-5324654 iβ directed to establishing cell lines capable of long-term proliferation and to treatment of conditions such as muscular dystrophy and diabetes, caused by failure to express a gene.
The targeting of a specific gene to a specific cell has not been attempted, and no one ocular type has been singled out. Specific targeting using adenovirus alone is expected to be difficult, as the virus has the ability to transfect a large variety of cell types.

For treatment of ocular diseases, in which other βiteβ in the body are largely or entirely unaffected, it iβ highly desirable to deliver the therapeutic agent
selectively to the target tissue in the eye. For anti-sense DNA, it is essential that the DNA be actually taken into these target cells.
The advances in gene therapy referred to above have led to further βtudieβ of the delivery and expreββion of transgenes into target cells, such as the
β-galactosidase transgene into the retina (Bennett et al, 1994, Li et al, 1994 and Mashmour et al, 1994) using recombinant adenovirus as a delivery system. The retinal pigment epithelium (RPE) is a non-renewable single cell layer in the eye, situated between the neural retina and the choroid. The cells of the RPE are phagocytic
neuroepithelial cells which form the outer most layer of the retina. The phagocytic properties of these cells have long been known, and have been reviewed (Bok and Young, 1979) . High levels of transgene expression within 3 days in the RPE layer and within two weeks in the photoreceptor cells of the neural retina in young animale were observed. The expression of the reporter gene was followed up to 9 weeks. In older animals, neither subretinal nor
intravitreal injections induced the expreββion of the β-galactosidase transgene in the photoreceptor cells (Li et al, 1994).
Australian Patent Application No. 61444/94 shows that replication-defective recombinant adenovirus is taken up by various tissues in the eye following injection into the anterior chamber, the vitreous humour, or the
retrobulbar space, and that the reporter gene β-galactosidaee is expressed. However, this document does not show that such forms of viruses successfully
incorporate the active agent into the target cell or area. Nor is there any disclosure or suggestion that VEGF can be used to heal any ocular condition.

One specific obstacle to success of using antisense nucleotides as a form of therapy for the eye is the inability of the nucleotide to enter the target cells, and the limited stability of the oiigonucleotides which have been modified, eg. phoβphorothioate oiigonucleotides
(Helene 1991) . These factors greatly restrict the success of gene therapy in vivo, particularly in the long term. In the treatment of retinal diseases, the ability to delay progression of the conditions by about 12 months would greatly increase the value and effectiveness of long term therapy.
Cytotoxicity has been observed in association with use of adenoviruses as a transport vector for retinal gene therapy. This cytotoxicity has been shown to be dose-dependent (Mashmour, 1994) and poses another difficulty in using such a vector. In order to decrease the dose of a given vector but retain its transfer efficiency, an
adjuvant may be used. Adjuvants such as lipofectin have been shown to increase the uptake of "naked" DNA by cells.
Even though HA has been widely uβed in eye surgery aβ a replacement for vitreous humour lost during the surgical procedure, we are not aware of any suggestion in the art that HA promotes uptake of any pharmaceutical agent into any cells or tiβsues in the eye. Similarly, although HA has been suggested to promote penetration of pharmaceutical agents such as antibiotics or anti-cancer agents, as set out in Australian Patent Application No. 52274/93 by Norpharmco, this specification does not suggeβt that HA promotes uptake of any agent, let alone DNA or viruses, by individual cells of any type. In particular, this specification does not teach the use of HA via intraocular injection.
We have now found that the phagocytic nature of the RPE cells will increase the uptake of molecules such as oiigonucleotides and viruses, following injection into the vitreous space in vivo. These RPE cells show increased uptake of virus compared to other cell types. Our findings enable the induction of both long-term and short-term inhibition of VEGF expression in retinal or choroid
epithelial cells, and hence inhibition of
neovascularisation of the retina or the development of SRNVM.

SUMMARY OF THE INVENTION
According to one aspect, the invention provides a compoβition comprising a nucleic acid and a hyaluronic acid or a derivative thereof, together with a pharmaceutically-acceptable carrier.
The nucleic acid may be a DNA or RNA, and/or may be a nucleotide sequence which is in the anti-sense
orientation to a target sequence. The target sequence is a nucleic acid sequence which is implicated in the causation or exacerbation of a pathological condition. This target nucleic acid sequence may be a genomic DNA, a cDNA, a messenger RNA or an oligonucleotide. Where the target nucleic acid sequence iβ a genomic DNA, it may be present in a coding region, or in a regulatory region, such as a promoter sequence.
Alternatively, the nucleic acid may be present in a vector comprising a nucleic acid sequence to be
transferred into a target cell. Again the nucleic acid sequence may be genomic DNA, cDNA, messenger RNA, or an oligonucleotide. However, in this case the nucleic acid may either be a sense βequence to be provided to a target cell in order to exert a function, or may be an anti-sense sequence to be provided to inhibit the functioning of a nucleic acid present in the target cell .
The vector comprising the DNA to be transferred may be a virus, βuch aβ an adenovirus, an adeno-asβociated virus, a herpes viruβ or a retrovirus. The use of all of theβe claββeβ of viruβ aβ vectorβ for gene therapy has been extensively canvassed in the art. Alternatively the vector may be a liposome.

The invention also provideβ a method of treatment of a pathological condition in a subject in need of such treatment, comprising the step of administering an
effective dose of a composition according to the invention to βaid subject.
It will be clearly understood that the dose and route of administration will depend upon the condition to be treated, and the attending physician or veterinarian will readily be able to determine suitable doses and routes. It is contemplated that the compositions of the invention may be administered parenterally, for example by intravenous or subcutaneous injection, topically, for example adsorbed on gels or sponges, or directly into the tissue to be treated, for example by intra-ocular or intra-tumoral injection.
The subject to be treated may be a human, or may be an animal, particularly domestic or companion mammalβ βuch as cattle, horse, sheep, goats, cats and doge.
In the compositions of the invention the nucleic acid or vector may simply be mixed with the hyaluronic acid, or may optionally be physically or chemically coupled to hyaluronic acid. Methods for attaching DNA to
hyaluronic acid have been disclosed in "Synthesis of
Sulfonated Hyaluronan Derivatives containing Nucleic Acid Bases, Chemistry Letters, 1994 2027-2030 and "Transport Performance of Nucleosides Through Nucleic Acid Baseβ Conjugated to Hyaluronan"; Chirachanchai, S., Wada, T., Inaki, Y. and Takemoto, K, Chemistry Letters. 1995 2_ 121- 122.
In a preferred embodiment this aspect of the invention provides compositions and methods for treatment of a retinal disease mediated by abnormal vascularization, in which the nucleic acid iβ an anti-sense nucleic acid sequence corresponding to at least a part of the sequence encoding vascular endothelial growth factor (VEGF) , and is administered together with a hyaluronic acid as described below.

Many forme of HA are suitable for use for the purposes of the invention. In particular, both low and high molecular weight forms of HA may be uβed. The only requirement iβ that the HA be of a degree of purity and sterility to be suitable for pharmaceutical use; preferably the HA is also pyrogen-free. High molecular weight
preparation of HA may require dilution prior to use. In particular, commercially-available HA products suitable for use in the invention are those supplied by Hyal
Pharmaceutical Corporation, Misβiββauga, which is a 2% solution of HA having a mean average molecular weight of about 225,000; sodium haluronate produced by Life Core™ Biomedical, Inc.; Pro Vise (Alcon Laboratories); and
"HEALON" (Pharmacia AB, Uppsala) . It will be clearly understood that for the purposeβ of thiβ βpecification, the term derivativeβ of HA encompasses homologues, analogues, complexes, esters and fragments and sub-unite of HA.
Derivatives of HA which may be uβed in the invention include pharmaceutically-acceptable salts thereof, or fragments or subunits of HA. The person skilled in the art will readily be able to determine whether a given preparation of HA, or a particular
derivative, complex etc. of HA, iβ suitable for use in the invention.
According to a second aspect, the invention relates to a composition for treatment of a retinal disease mediated by abnormal vascularisation, comprising an antisense nucleic acid sequence corresponding to at least a part of the sequence encoding vascular endothelial growth factor (VEGF) , and optionally further comprising one or more adjuvants such as hyaluronic acid or a dendrimer compound for increasing cellular uptake, together with a pharmaceutically acceptable carrier. The use of dendrimer compounds to transport genetic material into target cells is disclosed in International Patent Application
No. WO 95/24221 by Dendritech Inc et al.

The VEGF iβ most preferably human retinal pigment epithelial (RPE) or choroidal endothelial VEGF.
In separate embodiments, thiβ aspect of the invention is directed to treatment for such retinal disease in the short term (up to about two months), the long-term (up to about one year), and indefinitely (for the life of the patient) . In the first embodiment, for short-term treatment the invention provides one or more anti-sense oiigonucleotides having 100% complementarity to a
corresponding region of the VEGF gene. The oligonucleotide should have 16 to 50 nucleotides, preferably 16 to 22, and more preferably 16 to 19 nucleotides . Modified
oiigonucleotides of the kind described by Wagner et al (1996) may be used, and enable the lower limit of sequence length to be reduced to 7 nucleotideβ.
For long-term inhibition, the invention provides a recombinant virus comprising VEGF DNA in the anti-βenβe direction. Thiβ VEGF DNA iβ a long sequence, which for the purposeβ of thiβ βpecification iβ to be understood to represent a VEGF sequence of greater than 20 nucleotides in length, preferably greater than 50 nucleotideβ, ranging up to the full length βequence of VEGF. In thiβ embodiment, the recombinant viruβ iβ accumulated in RPE cells, and produces anti-sense VEGF in situ, thereby inhibiting VEGF expression in the RPE cell.
For indefinite inhibition, the invention provides a viruβ compriβing VEGF DNA in the anti-βenβe direction in which the viruβ iβ one capable of integrating the anti- βenβe βequence into the genome of the target cell.
Preferably the virus is an adeno-associated or similar virus. As in the embodiment directed to long-term
treatment, this VEGF DNA iβ of at least 20 nucleotideβ, preferably greater than 50 nucleotides. The adeno- asβociated or similar virus facilitates integration of anti-βenβe VEGF DNA into the RPE cell genome, thuβ enabling expression of anti-sense VEGF for as long as the cell remains functional.

Eye diβeaβeβ which may be treated using the compositions and methods of the invention include, but are not limited to, age-related macular degeneration (ARMD) and diabetic retinopathy. Other ocular conditions and tissues in which neovascularisation occurs, for example branch or central retinal vein occlusion, retinopathy of prematurity (also known as retrolental fibroplasia) , rubeosis iridie or corneal neovascularisation, may also be treated by the invention.
In another aspect, the invention provides a method of prevention or amelioration of a retinal diseaβe mediated by abnormal neovascularisation, comprising the step of administering an effective amount of an anti-βenβe nucleic acid sequence directed against VEGF into the eye, thereby to inhibit neovascularisation.
The anti-sense βequence may be carried in a replication-defective recombinant viruβ, aβ a vector or vehicle. The vector preferably comprises replication-defective adenovirus carrying promoters such as the
respiratory syncytial virus (RSV), cytomegalovirus (CMV), adenovirus major late protein (MLP), VAl pol III or β-actin promoters. The vector may aleo comprise a polyadenylation signal sequence such aβ the SV40 signal sequence. In a particularly preferred embodiment, the vector is pAd.RSV, pAd.MLP, or pAd.VAl. In a more particularly preferred embodiment the vector is Ad.RSV.aVEGF or Ad.VAl.aVEGF.
In a preferred embodiment, human VEGF is
subcloned into the vector, in order to create the
restriction βiteβ necessary for insertion, to form an adenovirus plasmid carrying VEGF or partial sequenceβ thereof in an anti-βenβe direction, which can then be linearized by restriction enzyme digestion. The linearized plasmid can then be co-transfected with a linearized replication defective adenovirus, in a suitable permissive host cell such aβ the kidney 293 cell line.
The compositions of the invention may be
delivered into the eye by intra-vitreal or sub-retinal injection, preferably in an appropriate vehicle or carrier. Such methods of administration and vehicleβ or carriers for such injection are known in the art. Alternatively, ex vivo delivery of the compositions of the invention may be achieved by removal of RPE cells from the patient to be treated, culturing the cells and subjecting them to
infection in vitro with a replication-defective adenovirus or an adeno-associated virus as defined above. RPE cells carrying the virus are then injected into the βub-retinal layer of the eye of the patient .
While the invention iβ specifically described with reference to conditions of the eye, the person skilled in the art will be aware that there are many other
pathological conditions in which VEGF is of importance. Such a person will understand that the antisense
oiigonucleotides and the recombinant viruses of the invention are applicable to treatment of such other conditions. Similarly the skilled person will understand that while the invention iβ specifically illustrated with reference to VEGF the methods described herein are
applicable to use with other proteins.

BRIEF DESCRIPTION OF THE FIGURES
Figure la shows the results of GeneScan analysis of persistence of anti-sense oiigonucleotides in vivo in the retina following a single intra-vitreal injection.
Figure lb shows a confocal microscopic image of the retina of a RCS-rdy+ rat at different times following injection of CATSCF.
Figure 2 is a graphical representation of the number of phagosomes in the RPE layers of Long-Evans rats. Doses were as follows: Low 6.6 μg, medium 66 μg and high 132 μg of CATSC anti-sense oligonucleotide. Each column shows the mean and standard deviation of the number of phagoβomes in five randomly selected areas in the rat retinae.

Figure 3 is a graphical representation of the number of phagosomeβ in the RPE layers of RCS-rdy-f- rats . Experimental animals were injected with 66 μg of sense oiigonucleotides (Sl) and 66 μg of antisense
oligonucleotide (CATSC) .
Figure 4 shows the effect of increasing the titre of adenoviral vector on the number of cells expreβsing the adenoviral transgene. In all cases, the incubation period was 16 hours. RPE7 denotes Human retinal pigment
epithelial cells from a 7 year old donor; F2000C denotes F2000 fibroblastic cells. The C suffix on the F2000 key indicates that the counts for the F2000 cell expression have been corrected for direct comparison with the RPE7 cells.
Figure 5 shows the effect of increasing the time of incubation with the adenoviral vector on the number of cellβ expressing the adenoviral transgene. In all cases, the concentration of the adenoviral vector was 2 x 10£ p.f.u./ml. The C suffix on the F2000 key indicates that the counts for the F2000 cell expression have been
corrected for direct comparison with the RPE7 cells.
Figure 6 iβ a graphical representation of the effect of Hyaluronic Acid (HA) on the number of RPE7 cells expressing an adenoviral transgene for a fixed viral titre. The three bars indicate the effect of 0.001% HA, 0.005% HA and no HA (control) . The error bar indicates one standard deviation.
Figure 7 iβ a graphical representation of the effect of Hyaluronic Acid (HA) on the number of F2000 cells expressing an adenoviral transgene for a fixed viral titre. The three bars indicate the effect of 0.001%HA, 0.005%HA and no HA (control) . The error bar indicates one standard deviation.
Figure 8 shows the immunofluorescent staining of HA receptors in RPE7 and F2000 fibroblasts 8a. CD44 staining on RPE7; 8b. ICAM staining on RPE7; 8c. RHAMM staining on RPE7; 8d. CD44 βtaining on F2000 fibroblasts;

8e. ICAM staining on F2000 fibroblasts; 8f. RHAMM staining on F2000 fibroblasts.
Figure 9 showβ micrographβ of choriocapillary endothelial cellβ isolated from porcine eye, illustrating their characteristic appearance (top panel), presence of Factor Vlll-related antigen (middle panel), and ability to take up acetylated low-density lipoprotein into the cytoplasm (bottom panel) .
Figure 10 showβ the effects of a variety of hyaluronic acid preparations on tube formation by
choriocapillary endothelial cellβ.
Figure 11 βhowβ the alkaline phosphatase staining of CD44 antigen in retinal pigment epithelium cells . In each case the epithelium is at the bottom of the picture with choroid above.
A. Unbleached pigment epithelium layer
B. Pigment epithelium layer bleached to remove melanin granules.
C. Bleached pigment epithelium stained with alkaline phosphatase-labelled anti-CD44 antibody.
Figure 12 showβ the results of DNA PCR and RT-PCR analysis of transfection of a retinal pigment epithelial cell line with VEGF165.
Figure 13 showβ the effect of VEGF165 produced by transfected RPE cells on tube formation by choriocapillary endothelial cells.

DETAILED DESCRIPTION OF THE INVENTION
The invention will now be described by way of reference only to the following non-limiting examples. In some of these examples, the feasibility of the methods utilised in the invention is demonstrated using anti-sense oiigonucleotides complementary to cathepsin S (CATSC) .

Example 1 Accumulation of Antisense Oiigonucleotides
in the RPE Cell Layer
Human retinal pigment epithelial cells were cultured and on the third passage were used for in vitro experiments. Confluent cultures were incubated with bovine rod outer segments (ROS) to mimic the in vivo situation. A fluorescein-labelled anti-sense oligonucleotide
complementary to human cathepsin S (CATSCF) was added to the medium of these cells and after 7 days of incubation, the cellβ were harvested. The presence of fluorescein-labelled oiigonucleotides within the RPE cells was detected by fluorocytometry (FACS) . A GeneScan DNA analyser was used to assess the presence and stability of the
oiigonucleotides in the cells. The fluorescence of
cultured RPE cells was increased by about 100-fold, demonstrating the presence of the anti-sense
oiigonucleotides within the RPE cellβ. Theβe results are summarised in Table 1.

Table 1
Fluorocytometer measurements of human RPE cells
incubated with or without complementary CATSCF



It waβ not known if the fluorescence was emitted by the full length CATSC or by degraded oiigonucleotides. Using GeneScan, it was demonstrated that the fluorescence was largely due to a 19-mer oligonucleotide, which appeared at a position similar to that of CATSCF. Using a similar procedure, it was obβerved that CATSC oiigonucleotides were still intact after 7 days of incubation.

Example 2 Cellular Distribution of Oiigonucleotides
in Retinal Cellβ and Stability of
Oiigonucleotides Following Injection Into
Eyes
One nmole of CATSCF was injected into the
vitreous humour of 6-week old non-pigmented RCS-rdy+ rats, and the movement of the oiigonucleotides were followed by confocal fluoromicroscopy. Fluorescein (1 nmole) was also injected as a control. Animals were euthanised 2 hours, 3 days and 7, 14 and 28 and 56 days after injection.
Following euthanasia, the injected eyes were enucleated, frozen, sectioned and immediately used for confocal
microscopy without fixation.
Two hours after intravitreal injection of CATSCF the penetration of the oiigonucleotides were observed in the ganglion cell layer at 2 hours and also in the
photoreceptor and pigment epithelial layers at 3 days.
However, 7 days following injection, only the RPE layer had significant amounts of CATSCF. At 14, 28 and 56 days, a fluorescent βignal waβ maintained in the RPE layer, and no signal was observed in any other cell types. Theβe results show that a large proportion of CATSCF was taken up by the phagocytic RPE cells.
Following intravitreal injection as described above, eyes were diββected, the retina waβ removed, and the DNA extracted. The purified DNA waβ subjected to GeneScan analysis. The presence of undegraded fluorescein-labelled oligonucleotide was demonstrated in the rat retinas after 7, 14, 28 and 56 days of injection, as shown in Figure la. The intensity of the signal had significantly diminished by 56 days.
Confocal microscopic analysis was performed following a βingle injection of 10 nmol CATSCF into non- pigmented RCS-rdy+ rate . Retinae were examined at
intervale after injection, and the results are shown in

Figure lb, in which g represents the ganglion cell layer, i the inner nuclear layer, o the outer nuclear layer, and r the retinal pigment epithelial layer. The panels show retinas 2 hours (B), 3 days (A), 7 days (C), 28 days (D) and 56 days (E) after injection of 10 nmol CATSCF, and 3 daye (F) after injection of FITC aβ a control.
Theβe reβults demonstrate that following
intravitreal injection, oiigonucleotides accumulate in the RPE cells. The oiigonucleotides are present in the RPE layer up to 56 days, and remain in a biologically active form during this period of time.

Example 3 Biological Activity of Anti-Sense
Oiigonucleotides
Female sixty day-old pigmented rats of the Long-Evans strain were obtained from Charles River Breeding Laboratories, Wilmington, MA.
Sixty day old non-pigmented RCS-rdy+ rats were obtained from our colony. The animals were acclimatised to a 12 hr light/ 12 hr dark lighting cycle, with an average illuminance of 5 lux for at least 10 daye prior to
experimentation.
Animals were anaeβthetiβed by intraperitoneal injection of sodium pentobarbital (50 mg/kg body weight) . Intravitreal injections through the pare plana were made using a 32 gauge needle. The left eyes served as controls, and the right eyee were injected with 3μl of 150 mM eodium chloride (βaline), or with 3 μl of saline containing 6.6, 66 or 132 μg of CATSC respectively, an anti-sense
oligonucleotide described earlier (Rakoczy et al, 1994) or 66 μg of sense oligonucleotide Sl, 100% complementary to CATSC. Injected animals were allowed to recover from anaesthesia, and at one week post-injection were sacrificed by an overdose of eodium pentobarbital and uβed for
morphological examination. All animals were killed within half an hour at the same time of the day, approximately 4 hours after light onβet. Two to three animals were used for each dose.

Following enucleation, whole eyee were immersed in 2.5% glutaraldehyde and 1% paraformaldehyde in 0.125M sodium cacodylate buffer, pH 7.35. The cornea and lens were diββected free and the eyecup trimmed for orientation purposes. The tissue waβ fixed overnight at 4°C and then post-fixed for 1 hour in 1% osmium tetroxide at room temperature. After ethanol dehydration, the tissue was embedded in epoxy resin. Retinal sections were prepared for transmission electron microscopy as described
previously (Kennedy et al, 1994) .
Histological data were obtained by light
microscopy. Semi-thin 1 μm sections were cut using a LKB 2088 Ultratome (LKB-Produkter, Sweden) with a diamond knife and stained with toluidine blue. The number of phagosomee that accumulated in the RPE cellβ of each βpecimen injected with βaline, low ( 6 .6 μg) , medium (66 μg) or high 132 μg) dose of CATSC and 66 μg of Sl sense oligonucleotide was determined. From each eye, five sets of counts were made at 40 fold magnification and the standard deviation was calculated. Each set consisted of the total number of phagosomee in 250 μm length of RPE from 6 different
randomly selected areas. The number of phagosomes that accumulated in the RPE of the control eyes, low medium and high doβes of CATSC were analysed and graphically
represented. Comparisons were made using the analysis of variance following the general linear models procedure of the SASR (version 6) statistical package (SAS Institute Inc. , USA) .
The reβults show that we successfully tested an anti-βenβe oligonucleotide (CATSC) in two strains of rats. The number of phagosome-like inclusion bodies present in control Long-Evans and RCS rdy + rats was not significantly different, 35.84-11.6 and 47.29+14.8 (mean ± SD) ,
respectively. The intravitreal injection was non- traumatic. Light microscopic examination of the retinas of the saline injected eyes revealed no damage to the outer layers of the retina, and there was no increase in the number of phagosome-like inclusion bodies in the RPE layer when compared to the control non-injected animals. Long-Evans rats were used to identify the minimum amount of CATSC required to induce biological changes in the RPE layer. In the control eyee and in thoβe injected with low dose (6.6 μg) of CATSC, the number of phagosome-like inclusions within the RPE cells were 35.84-11.6 and 35.04-7.4 respectively. In animals injected with higher doses (66μg and 132 μg) , the number of phagosome-like inclusions were 96.24-13.6 and 141.04-34.7, respectively, and the difference was statistically significant when compared to the control and low dose samples (Figure 2) .
RCS-rdy4- rats injected with 66 μg of CATSC also demonstrated a statistically significant increase in the number of phagosome-like inclusion bodies, ie 204.204-39.3 when compared to the 47.204*14.8 in controls. In contrast, the injection of 66 μg of sense oligonucleotide (Sl) did not increase the number of phagosomee (Figure 3) preβent in the RPE Layer, (34.44*12.54).
The inclusions found in RPEs of CATSC-injected

Long-Evans and RCS-rdy-i- animals were spherical in shape, and clearly distinguishable from the very dark, small elliptical melanin granules present in Long Evans rats. In the presence of 66 μg of CATSC, the tips of the outer segments showed βignβ of disorganisation and there were some vacuoles preβent in the outer nuclear layer. However theβe changeβ were not obβerved in Sl sense
oligonucleotide-injected animals .
Electron microscopic examination of the RPE layer of a CAT SC-injected eye revealed no significant changes in the morphology of RPE cells. Melanin granules appeared smaller and less concentrated due to regional differences. Individual mitochondrial profiles were smaller in the treated group than in the controls, although the number was greater in the treated than in the untreated animals.
Electron microscopic examination confirmed that the structures of the undigested material was similar to that of phagosomee. The numerous phagosomee seen in the RPE layer of rate treated with CATSC were paranuclear, and contained mainly compacted phospholipid membranes,
resembling undigested photoreceptor outer segment (POS) and confirming their photoreceptor origin. There were no other morphological changes observed in the POS layer, except for the disorganised appearance of the apices in treated animals .

Example 4 Gene Transfer to the RPE Cell Layer
The nature and dynamics of gene transfer using an adenoviral vector were examined. The effects of adjuvants on the uptake of the adenovirus waβ also studied.
Human RPE cultures (HRPE7) were obtained from a 7-year old Caucasian donor and prepared as described in Rakoczy et al (1992) . Human F2000 fibroblast cells were cultured, harvested and pooled in Minimal Eagles Medium (MEM, Multicel ™ Trace Bioβciences, Australia), with 10% FBS (Multiser™, Trace Bioscienceβ) and containing 125μl gentamicin (Delta West, Bentley, Australia) per 100 ml medium. One ml aliquots of the pooled cell suspension were placed into each well of a 24 well plate, to ensure equal seeding of wells. Experiments were carried out with cells at confluence, and at least two parallel sets of each experimental points were obtained.

Expression of Adenoviral Transgene
Replication-deficient Adenovirus 5 carrying a RSV promoter and β-Galactosidase gene (Ad.RSV.βgal)
(Stratford-Perricaudet, 1992) was cultured and purified as described by Graham and Prevac, 1991. Ad.RSV.βgal was added to each well as a 1 ml aliquot, in MEM, at a
concentration of 4xl06 p.f.u./ml. for the time-based trials, giving a final concentration of 2xl06 p.f.u./ml. For the titre-based trials, concentrations of 8xl03, 4xl04, 8xl04, 2.4xl05, 4x10s p.f.u./ml were added to the wells in a 1 ml aliquot, making the total volume 2 ml in each well (the final viral concentration is half of that added) . All of the trials examining the effect of increasing viral titres involved incubation of the culture with the viral suspension for a fixed period of 16 hours.
Experiments were terminated by removing the medium from each well and fixing the cells with 0.5 ml of 0.5% glutaraldehyde. The glutaraldehyde was removed after 5 minuteβ and the cellβ waehed once with Phoephate Buffered Saline (PBS). Following this, 0.5 ml of X-gal stain [For 1 ml of solution (concentration in final solution) : 25μl X-Gal (0.5mg/ml, BioRad, Hercules, California), 44μl HEPES buffer (44mM), lOOμl K4Fe(CN)s (3mM) lOOμl K3Fe(CN)6(3mM, lOOμl NaCl(15mM), lOOμl MgCl2(1.3mM) , sterile distilled water to make 1 ml (531μl)] waβ added to each well and incubated overnight (about 16 hours) at room temperature.

Cell Counting
An Olympus T041 phase contrast microecope
(Olympus Optical Co Ltd, Tokyo, Japan) at a magnification of 200x was uβed. Counting waβ carried out by a βingle observer. A second observer then blind counted 25% of the samples as a countercheck. A counting graticule in the microscope was used to define the region for counting when averaging was uβed.
All cells staining positively with the X-Gal stain were counted. At low expression of transgene
(< approximately 2000 cells/well), the entire plate waβ counted. When the cell count was higher, averaging was used. Cells were counted in five standardized regions and their average was used to calculate the total count for each well.
In the trials comparing the rate of expression in HRPE7 and F2000 fibroblastβ, the figure for the number of F2000 cellβ expressing the gene was corrected. This correction reflects the different total cell number of each cell type in a confluent culture in a 24 well plate. The count for HRPE7 is 3x10s per well and for F2000, it is 2x10s per well. The graphical figures (Figures 4 and 5) also contain corrected counts to allow direct comparison. Where there is no comparison between cell types, no
alteration of the raw count is carried out.
In the titre-baβed trials, the profiles of expresβion were markedly different in terms of rate of increaβe and absolute expresβion. For HRPE7 cells, the expresβion rate appeared to have an exponential form, while in F2000 fibroblasts the profile was more linear. There was a widening gap in expresβion throughout the trial comparing titre. At higher viral titre, HRPE7 expression was an order of magnitude greater than F2000 cells. For the conditions and titres tried in this experiment there was an overall and constant increase in the number of cells expressing with increasing vector titre (Figure 4) .
In the study of the effect of incubation time on the profile of transgene expreββion, the concentration of AdV.RSV.βgal waβ kept constant at 2xl06 p.f.u./ml. The profiles of expreββion of transgene in the two cell types were markedly different, both in terms of rate of increase and magnitude of number of cells expressing the gene.
There was also a notable delay between the sharp increase in number of HRPE7 and F2000 fibroblasts expressing the gene. For HRPE7 cells, the upturn in expression rate occurred at 4 hours while in F2000 fibroblasts, it occurred at 24 hours. There is a "window" period between 4 and 24 hours where the HRPE7 expresβion iβ an order of magnitude greater than that of F2000 cellβ (Figure 5) .

Example 5 Effect of HA ae an Adjuvant on the Uptake
and expresβion of the β-gal Gene ueing a
Viral Vector
HRPE7 and F2000 cells were aliquoted into 24 well plates . The cells were incubated as described in
Example 4, and allowed to reach 95% of confluence.
Solutions of 0.001% to 0.005% buffered eodium hyaluronate (HA) (1% Hyaluronic acid from rooster comb; HEALON, Pharmacia AB, Uppβala, Sweden) were prepared with MEM. A dose of 10 μl of viral solution at a concentration of
4 x 106 p.f .u. was added to 10 ml of each of the diluted HA solutions and 10 ml of MEM for the control, and incubated for 30 minuteβ at 25°C with intermittent gentle shaking. To separate wells of the 24 well plate, 1 ml of each of the test and control solutions was added. There were four parallel samples for each test concentration and for the control, which were counted and averaged.
The viral/HA solutions were incubated with the cell cultures for 16 hours. Each experiment was terminated according to the procedure given in Example 4.

Table 2
Experiment 1: Expreββion in HRPE 7 Cellβ

1 2 3 4 Mean

RPE 7/HA 17114 20776 18730 17998 19168 (0.001%)
RPE7/HA 17688 22186 20258 22236 20592 (0.005%)
RPE 10782 15480 16326 15266 14705

7/Cont

The mean number of HRPE7 cellβ expressing the transgene in each well for adenovirus alone waβ 14 705 (SD±2228) . For adenovirus with 0.001% HA the mean number of expressing cells was 19 168 per well (SD 1561) and for 0.005% HA the mean was 20592 (SD 2143) (Figure 6). This shows an increase in number of cells expressing the transgene of 30.4% for 0.001% HA and of 40.0% with 0.005% HA.
As assessed by Student's t test, the probability of the significance of the increase in number of HRPE7 cells expressing the gene, when 0.005% HA is used, compared with the control, is 0.0097, which shows a level of significance of p<0.01. The significance reflects the large difference between the means (20592 (test) v 14705 (control)) and the separation of the means by more than two standard deviations.
The t test probability of the significance of the increase in number of RPE7 cells expreββing the gene, when 0.001% HA iβ uβed compared with the control, is 0.02931, which showβ a level of significance of p<0.05. The reduced significance reflects the smaller difference between the means (19168 (test) v 14705 (control)).

Table 3
Experiment 2: Expresβion in F2000 Cellβ.

1 2 3 4 Mean

F 2000/HA 4358 4620 4195 NA 4391 (0.001%)
F 2000/HA 4506 3914 4759 4332 4378 (0.005%)
F2000 3844 3652 3875 3748 3780 Cont

The protocols for examining the effect of HA on the expresβion of a tranegene in F2000 fibroblaβtβ were the same aβ that for HRPE7. The numbers of cells expressing transgenes were significantly less than for HRPE7, which is consistent with the results demonstrated in Example 4. The mean number of cells expressing in each well for adenovirus alone was 3780 (SD±IOO) . For adenovirus with 0.001% HA, the mean number of expressing cells was 4391 per well
(SD±214) and for 0.005% HA the mean waβ 4378 (SD355) (Fig. 7.). Thiβ shows an increase of 15.8% for 0.001% HA and of 15.5% with 0.005% HA in the number of cells expressing the adenoviral transgene.
The two-tailed Student's t test was used to assess the significance of the difference between the means for each set of experimental data. For each experiment, the means, the Standard error of the differences of the means and the p value for the t test are given. In both experiments, HA gave very significantly increased uptake (p < 0.05) .
The t test probability of the significance of the increase in number of cells expreββing tranegene for the F2000 fibroblaβtβ with 0.005% HA, compared with the
control, iβ 0.0044, which shows a level of significance of p<0.01. The high significance here reflects the large difference between the means (4391(test) v 3790 (control) ) and the small variation within the two samples. The standard deviation is 214 (test) and 111 (control) .
The t test probability of the significance of the increase in number of cellβ expreβsing tranegene for the F2000 fibroblaβtβ with 0.001% HA, compared with the control, iβ 0.0195, which shows a level of significance of p<0.05. There iβ a greater variation in the raw figures, and the standard deviation is higher than for the 0.005% sample (355 v 214), which accounts for the higher p value.
Preliminary trials of chondroitin sulphate and lipofectamine as adjuvants were also carried out in order to assess the likely efficacy. These agents had no significant effect on gene expresβion in HRPE7 cells.
The following doses of adjuvants were alβo uβed:-

Table 4
HA Concentration

Amount of viral 0.05% 0.01% 0.005% 0.001% Control Control solution
5 μl 176a 318 319 316 279 282

10 μl 305* 906 802 645 623 609

25 μl _• 714b 1682 1822 1478 1184

50 μl _» 2772 2692 3328 2250 1822

The figures represent the effect of HA
concentration on the uptake and expression of β-gal transgene. Increasing virus concentration resulted in an increase in the number of β-gal expressing cells. The numbers represent the number of RPE cells staining positive for β-gal following 16 hours incubation of virus in the presence of HA in a 24 well plate (cc 2x10" pfu/ml) .

* The viscosity of these solutions precluded adequate
dispersion of the HA and made them very difficult to manipulate.
b It was not clear why this figure fell so far outside of the normal distribution of the other results.

Example 6 Effect of HA Molecular Weight on the Uptake
and Expression of the βgal Gene ϋeing a
Viral Vector
Adenovirus with a β-galactoeidase marker gene and a RSV promoter (Adv.RSV. βgal) was cultured in cells of the K293 embryonic human kidney cell line. Supernatant was collected, and the concentration of virus was determined by serial dilution with 4 replicates of each dilution. The concentration of the virus waβ calculated to be
5 x 10s pfu/ml. The viruβ was suspended in MEM medium with 10% fetal bovine serum (FBS) and 125μl/100ml gentamicin.
Human Retinal Pigment Epithelial Cells (HRPE) were from a 20 year old donor and cultured in medium as described above. They were aliquoted into 24 well plates from the same stock and allowed to reach confluence.
Fourth passage cells were used.
The following HA preparations were tested:
1. Hyal (MW approx. 300 000)
2. Provisc (MW approx. 1 900 000)
3. Healon GV (MW approx. 5 000 000)
Each of the preparations was diluted to a solution of 0.002% in MEM without FBS.
The virus solution as above was mixed in a 1:1 ratio with the adjuvant solution giving a final viral concentration of 2.5 x 108pfu and an HA concentration of 0.001%. The two solutions were incubated in this mixture for 30 minutes at room temperature with gentle shaking. The control solution consisted of a mixture of the virus with MEM without FBS with no HA present.
To each of the 24 well cellβ 1 ml of the viral/HA mixture was added. Incubation was for 24 hours in a C02 incubator (5% C02) at 37°C. The experiment waβ terminated by removing the viral/HA mixture and adding 0.5ml of 0.5% glutaraldehyde for 5 minuteβ to each well. The well was washed once with PBS and reacted with X gal stain.
An Olympus T041 phaβe contrast microscope
(Olympus Optical Co. Ltd. Tokyo, Japan) at a magnification of 10OX was used throughout. Counting was carried out by a single observer and checked against a second blind observer who counted a quarter of the samples . A counting graticule in the microscope waβ uβed to define the region for counting. All cellβ βtaining positively blue with the X-gal stain were counted as positive. Cells were counted in five βtandardized regions and their average was used to calculate the total count for each well. The results and statistical analysis are presented in Tables 5 to 9.

Table 5

(count is of sample only) Control ay i Protri.BC Healon GV

Number of cells expresβion 2043 2486 2424 2756 β-gai
Statistics
Anova: Single Factor Between all groups

Table 6
SUMMARY

Groups Count Sum Average Variance

Control 3 6129 2043 15769

Hyal 3 7458 2486 4225

Provisc 3 7271 2423 .667 36677 . 33

Healon GV 3 8268 2756 36928 Table 7
ANOVA

Source of Variation SS df MS F
Between Group 111561 . 0 3 259189 11.07653

Within Groups 187198.7 8 23399..83

Total 964765.7 11

Anova: Single Factor Between Adjuvants

Table 8
SUMMARY

Groups Count Sum Average Variance

Hyal 3 7458 2486 4225 Provisc 3 7271 2423.667 36677.33 Healon GV 3 8268 2756 36928

Table 9
ANOVA

Source SS df MS F P-value Fcrit of
Variation
Between 187230.9 2 93615.44 3.61 0.094 5.14 Groups
Within 155660.7 6 25943.44
Groups

Total 342891.6 8

There was an increase in transgene expression in all of the HA-containing samples relative to the control (P < 0.003). The percentage increase was 21.7%, 18.6% and 34.8% for Hyal, Provisc and Healon GV HA preparation respectively. There is no significant difference between the effect of different molecular weights of hyaluronic acid (p = 0.09) .

Theβe reβults demonstrate that hyaluronic acid increases viral vector uptake, demonstrating an adjuvant effect. In addition it was shown that the adjuvant effect is independent of the molecular weight of hyaluronic acid between MW 300 000-5, 000 000.

Example 7 Demonstration of HA Receptors on the cell
membrane of HRPE7 and F2000.
Polyclonal RHAMM (Receptor for Hyaluronan
Mediated Motility) antibodies were kindly provided by
Dr E Turley, Manitoba Institute of Cell Biology, Canada. The antibody was used at a dilution of 1:75. Monoclonal Intercellular Adhesion Molecule 1 (ICAM-1) antibodies
(Boehringer-Mannheim) were uβed at a concentration of
4μg/ml and monoclonal homing receptor CD44 antibody (CD44) waβ uβed at a concentration of 4μg/ml (Boehringer Mannheim Biochemica, Germany) . Monoclonal anti-human IgG antibody and rat non-immune serum were kindly provided by
Dr M Baineβ, Lions Eye Institute, Perth, Australia. They were uβed at a concentration of 4μg/ml and a dilution of 1:75 respectively. Anti-Mouse IgG (Fab specific) -FITC conjugate secondary antibody was used at a 1:64 dilution and anti-Rabbit IgG (whole molecule) -FITC conjugate
secondary antibody waβ uβed at a 1:100 dilution (Sigma Immunochemicals, St Louie, Missouri) .
HRPE7 and F2000 fibroblast cells were cultured in

Lab Tek β-well slide chambers (Nunc Inc. Naperville,
Illinois) . Cell cultures were fixed with methanol at -20°C for 10 minuteβ before immunofluoreβcent staining. All primary antibody solutions were incubated for 1 hour. The primary antibodies used for each of the two cell types were monoclonal anti ICAM-1, anti-CD44 as test and monoclonal anti-Human IgG aβ control, and polyclonal anti-RHAMM with a non-immune rabbit serum as control. Following the removal of the primary antibody, each well was washed three times with PBS and the secondary antibody was applied for 1 hour. The secondary antibody to the monoclonal antibodies was antimouβe IgG and the polyclonal waβ anti-rabbit IgG. The βecondary antibodieβ were applied to tiββue without primary antibody aβ a further control. Finally, on removal of the βecondary antibody, each well waβ waehed a further three timeβ before the well chambers were removed and the slides mounted with Immuno Fluore Mounting Medium (ICN Biomedical's Inc, Aurora, Ohio) .
Immunohistochemical staining for CD44 using a monoclonal antibody demonstrated positive staining for both HRPE7 cells and F2000 fibroblasts, as shown in Figures 8a and 8b respectively. The staining had a distribution consistent with the cell surface, as the staining pattern was the same aβ the cellular outline of cultured tiββue.
A monoclonal human anti-IgG waβ uβed aβ control, and was negative for both HRPE7 and F2000 fibroblastβ. A βecond control, using βecondary fluoreβcent antibody with no primary antibody waβ alβo negative for both cell types.
Immunohistochemical βtaining using a monoclonal antibody for ICAM-1 demonstrated positive staining for both HRPE7, and F2000 fibroblastβ, aβ shown in Figures 8c and 8d respectively. The βtaining had a similar distribution to that of CD44, but the signal was slightly weaker. The same controls as for CD44 were uβed for ICAM-1 βtaining, and were alβo negative.
Staining for RHAMM receptorβ uβing a rabbit polyclonal antibody waβ positive for both HRPE7 and F2000 fibroblaβtβ, aβ shown in Figures 8e and 8f respectively. The distribution of βtaining, however, waβ markedly
different in the two cell types. In HRPE cells the
staining pattern was predominantly nuclear, with a very faint cytoplasmic outline (Figure 8e) . The distribution of staining in F2000 fibroblaβtβ waβ similar to that of CD44 and ICAM-1, with no significant nuclear signal observable over the cytoplasmic or cell outline pattern.
The control serum waβ a rabbit non-immune serum, which waβ negative for HRPE7 but gave a very weak signal in F2000 fibroblastβ. In both cases, the secondary fluoreβcent antibody alone did not lead to a positive signal from either cell type.

Example 8 The Effects of Hyaluronic Acid Preparations
of Different Molecular Weight on Tube
Formation
Reagents
Hank's balanced salt βolution (Hank'e BSS) without calcium or magnesium, medium Hams F12, minimum essential medium with Earlee salts (EMEM) , foetal calf serum (FCS), penicillin-streptomycin, amphetericin B, and trypsin-EDTA were obtained from Australian Biosearch
(Perth, Western Australia) . Collagenase A, endothelial cell growth supplement (ECGF) , mouse anti-human monoclonal antibody against factor VIII-related antigen, and anti-mouse Ig-fluorescein were acquired from Boehringer Mannheim Australia Pty. Ltd. (Perth, Western Australia) . Gelatin, heparin, ascorbic acid were purchased from Sigma Chemical Company (Sydney, Australia), acetylated low-density
lipoprotein (Dil-ac-LDL, l,l'-dioctadecyl-3,3,3 ' ,3 '-tetramethyl-indocarbo-cyanine perchlorate) from Biomedical Technologies, Inc. (Stoughton, Massachusetts), Matrigel from Collaborative Research (Bedford, Massachusetts), recombinant human vascular endothelial cell growth factor (VEGF) from Pepro Tech EC Ltd. (Rocky Hill, New Jersey), ProVisc (MW 1.9 x IO6) from Alcon Laboratories, Healon

(MW 2.5 x IO6) and Healon GV (MW 5.0 x IO6) from Pharmacia.

Isolation and Culture of Porcine Choriocapillary
Endothelial Cells
Porcine eyes were obtained from a local abattoir 2-4 hours after death of the animals. The choriocapillary endothelial cells (CECs) were isolated as previously described (Morse et al, 1990, Sakomoto et al, 1995) .
Briefly, Hank's balanced ealt βolution (Hank'e BSS) without calcium or magnesium, but with 0.1 % collagenase A was used to release endothelial cells at 37°C for 1 hour. After washing twice in Hank's BSS, the cells were plated in 1 % gelatin-coated 75-cm2 cell culture flasks in 5% C02, 95% air at 37°C. The growth medium consisted of Hams F12 plus 10% fetal calf serum (FCS) , 100 U penicillin-100 μg
streptomycin/ml, 2.5 μg/ml amphotericin B, 37.5 μg/ml endothelial cell growth supplement (ECGS), heparin
100 μg/ml, and ascorbic acid 25 μg/ml. After 24 or
48 hours of plating the capillary βegments, the coloniee of endothelial cellβ showing a cobblestone appearance
flattened and spread. On the third or fourth day, the non-endothelial colonies were recognised and were circled with a permanent marker pen on the top of 75-cm2 flasks. A glass pipette which had been drawn through a flame to produce a bead tip was uβed to remove and crush any non-endothelial colonies within the circles (Folkman et al., 1979). This technique was carried out under a phase contrast microscope (xlO phase objective) in a laminar flow hood. The medium was changed twice to remove floating cellβ. Thiβ procedure waβ repeated three to five times to enrich the primary cells for endothelial cellβ before they became confluent. The cellβ were identified aβ vaβcular endothelial cellβ by typical cobblestone morphology, presence of factor Vlll-related antigen (Sakomoto et al, 1995), and positive staining (uptake) with Dil-ac-LDL
(Folkman et al, 1979).

The Effects of Hyaluronic Acid on Tube Formation
The tube formation assay was performed aβ
previously described (Haralabopoulos, et al, 1994.
Briefly, Matrigel (16.1 mg protein/ml) waβ prepared from the Engelbreth-Holm Swarm tumour waβ uβed to coat 24 well cluster plates (250 μl/well) as recommended by the product sheet. After polymerisation of the Matrigel at 37°C for 30 minuteβ in C02 incubator, 0.5 ml medium containing 10 or 20 μg/ml of hyaluronic acid preparations of different molecular weightβ (ProViβc, MW 1.9 x IO6; Hyal MW 2.5 x IO6 and HealonGV MW 5.0 x IO6 reepectively) in MEM with 10% FCS was added to the Matrigel coated wells . 10% FCS in
0.5 ml MEM waβ uβed for comparison of a relative unit of the tube area. The CECs (passage 3-7) were lifted from flasks by 0.25% trypsin-0.02% EDTA, suspended in 5% MEM, and added to the coated wells (50,000 cells/well in 0.5 ml medium) . To evaluate the areas of tube-like structured on the gel, photographe were taken with a phase-contraβt microecope after six hours. Five to seven fields (xlO objective) were chosen randomly in each well for
quantitative study.

Choriocapillary Endothelial Cells
Primary cultures of capillary endothelial cells have a characteristic appearance that distinguishes them from other cell types. In addition they were characterized by staining for factor Vlll-related antigen, and assaying for the ability to phagocytize Dil-ac-LDL. More than 95% of the CECs showed a positive reaction to factor VIII-related antigen. Almost every cell showed uptake of
Dil-ac-LDL into the cytoplasm, as shown in Figure 9. This indicates that at least 95% of the cells were
choriocapillary endothelial cells (CEC cellβ) .

Quantification of Tube Formation and Statistical Analysis
The tube areas from duplicate wells were measured uβing a Computer Imaging Analyzer System (Professional Image Proceeding for Windows, Matrox Inspector) . The slide photographs were βcanned into a computer and the background adjueted to obtain the beet contraβt between the tubee and Matrigel. Tube formation waβ then quantified by meaβuring the total tube area of each photograph. The results were expressed as the mean and the standard error of the percentage of tube area in the presence of 7.5% FCS alone (the final concentration) and were analyzed by Student's t-test for at least two experiments.

Tube Formation
After 1 hour of being βeeded on the top of
Matrigel, the CECβ became attached. Within 2-3 hours the CECs rapidly migrated into a reticular network of aligned cellβ. After 3 hours the CECs started to flatten and form capillary-like structures on the βurface of Matrigel'. By 6 hours capillary-like structures became apparent, showing an anastomosing network like vessel tubes. Tube formation in control and experimental samples containing different preparations of hyaluronic acid at a biologically active concentration waβ assessed after six hours, and the results are summarized in Figure 10 and Tables 10 to 13.

Table 10

Pro VisK Pro Visk Healon Healon

(5 μg/ml) (10 μg/ml) (5 μg/ml) (10 μg/ml)
60.3 139.91 119.37 118.33
49.64 122.96 134.01 111.22
90.03 47.38 39.96 47.48
41.78 36.1 37.12 68.9
59.04 99.4 142.32 126.36
129.2 106.05 72.63 117.69

Table 11

Healon GV Healon GV Control
(5 μg/ml) (10 μg/ml)
115.22 86.16 155.09
111.57 62.57 118.71
114.42 78.88 85.08
105.06 145.59 43.34
22.64 136.12 94.91
104.08 102.94 Table 12

Anova: Single Factor

SUMMARY

Groups Count Sum Average Variance

Column 1 6 429.99 71.67 1062.86

Column 2 6 551.8 91.97 1724.36

Column 3 6 545.41 90.90 2226.80

Column 4 6 589.98 98.33 1035.70

Column 5 6 572.99 95.50 1295.74

Column 6 5 509.32 101.86 1351.08

Column 7 6 600.07 100.01 1370.50

Table 13

ANOVA

Source SS df MS F P-value Fcrit of
Variation
Between 3655.25 6 609.21 0.42 0.86 2.38 Groups
Within 48984.10 34 1440.71
Groups

Total 52639.35 40

There was no statistically significant difference in CEC tube formation between the control and hyaluronic acid-containing samples, demonstrating that hyaluronic acid of molecular weight between MW 300,000-5,000,000 does not induce neovascularisation in the absence of another agent.

Example 9 Demonstration of the Preβence of CD44 HA
Receptor in the Human Retina
Preparation of Human Retina
A human Eye Bank donor eye was diββected
following the removal of the cornea. After diβcarding the anterior segment the vitreous waβ carefully removed, leaving behind some parts of the neural retina and the complete layer of pigment epithelium attached to the choroid. The eye cap was filled with 2.5% glutaraldehyde for fixation. Sections of the fixed tissue were subjected to paraffin embedding. Paraffin blocks were cut and sections were transferred on to glass histochemical slides, dewaxed in xylane and ethanol, and washed in distilled water and Tris-buffered βaline pH 7.2 (TBS).

Alkaline Phosphatase Staining of Sections
Removal of melanin granules was achieved by incubating the eye sections in 50 μl 0.25% potassium permanganate for 45 minuteβ followed by 50 μl 1.0% oxalic acid for 5 minuteβ. Bleaching waβ carried out following incubation with serum 50 μl/section of 10% normal horse serum/TBS (Commonwealth Serum Laboratories, Perth,
Australia) for 30 minuteβ. Sections were then washed twice in TBS, 5 minutes per wash and incubated in 50 μl mouβe anti-CD44 monoclonal antibody (Boehringer Mannheim
Biochemica, Mannheim, Germany) or 50 μl of mouβe anti-81-11 monoclonal antibody (non-immune control) for 60 minuteβ. After incubation sections were again washed twice in TBS, 5 minutes per waβh, incubated in 50 μl of 1/250 horse anti-mouse IgG (H4-L) conjugated to alkaline phosphataee
conjugated to alkaline phosphataee (βecondary antibody) for polyclonal antibodieβ for 60 minuteβ and waehed twice in TBS, 5 minutes per wash. Sections were incubated in 50 μl FAST RED (Sigma Aldrich, St louiβ, USA) for 20 minutes, washed twice in TBS, 5 minutes per wash and counterstained in Meyer's Haemotoxy1in for 10 minutes, followed by
5 minutes in tap water. Sections were allowed to dry, then mounted using glycerol jelly.
Bleaching of melanin was carried out successfully without causing damage to the tissue sections, as shown in Figure 10. Staining of glutaraldehyde-fixed human eye sections with the mouse control monoclonal antibody and FAST RED resulted in clear βtaining of the RPE layer in unbleached tiββue (Figure 10A) and following bleaching after incubation with 10% normal horse serum (Figure 10B) . A strong pink βignal demonβtrating the βpecific preeence of CD44 HA receptorβ in the retinal pigment epithelium but not in the choroid waβ obβerved in tiββue stained with anti-CD44 monoclonal antibody (Figure IOC) . As in choroid there was no βignal detected in the neural retina.
Theβe results demonstrate that the retinal pigment epithelial cellβ preferentially express HA
receptors, thus facilitating an enhanced uptake of HA complexes.

Example 10 Up and Down Regulation of Cathepsin D
Expreββion in NIH 3T3 Cells
A 1620 bp Hindlll fragment of human cathepsin D was subcloned into pHBApr-1-neo vector in both βenβe and anti-βenβe directionβ. Poβitive cloneβ were selected, and the orientation of the fragments was confirmed by EcoRI restriction enzyme analysis. For the transfections of NIH 3T3 cells the clones carrying cathepsin D in the antisense and sense directions were on caesium chloride density gradients .
NIH 3T3 cells were seeded on to 6-well tissues culture plates at a concentration of 2 X 10s in 2 ml DMEM supplemented with 10% fetal bovine serum (FBS) . The cells were incubated overnight at 37°C until they became 70% confluent. Having reached confluency, the cells were waehed twice with serum and antibiotic-free medium.
Lipofection reagent (10 μl) (GIBCO-BRL) diluted in 100 μl of OPTI-MEM (GIBCO-BRL) were gently mixed and incubated at room temperature for 15 minutes. Following incubation, an additional 800 μl of OPTI-MEM waβ added to the mixture.
Thiβ diluted mixture was gently overlaid onto the washed NIH 3T3 cells. The cells were incubated for 16-20 hrs before the transfection media was removed and replaced with DMEM supplemented with 10% FBS. After a further 48 hrs incubation the cells were trypsinised and subcultured at 1:5 in media containing 10% FBS and Geneticin 418 (GIBCOBRL) at 1 ng/ml concentration. Succesβfully transfected cellβ eelected with Geneticin 418 were maintained in media supplemented with FBS and Geneticin 418 as described above. Confluent transformed cultures were frozen for storage and subcultured for further analysis. The presence of
cathepsin D in the tranβformed NIH 3T3 cells was detected with polyclonal antibody against cathepsin D, using a conventional cytochemical technique and an alkaline
phosphataee-labelled βecond antibody.
The preeence of cathepsin D fragment of the vector was demonstrated with Hindlll digestion. Positive clones showed the presence of a 1620 kb fragment. The orientation waβ established by ECO RI restriction enzyme digestion, which gave two fragments at 5.7 and 5.9 kb in the case of the anti-βenβe orientation and 4.3 and 7.3 kb in the case of the βenβe orientation. All NIH 3T3 cellβ surviving Geneticin 418 selection carried cathepsin D cloneβ, which are antibiotic resistant. The transformed control NIH 3T3 cellβ did not survive the selection
procedure. The immunocytochemistry results suggeβt that NIH 3T3 cellβ carrying cathepsin D in the sense direction up-regulated cathepsin D production, while those carrying cathepsin D in the anti-βenβe direction down regulated cathepsin D production.

Example 11 Production of a VEGFlpc-Expreseing RPE Cell
Line
Cell Culture
The human RPE cell line 407A (Davis et al, 1995), was maintained at 37°C in a humidified environment containing 5% C02. The culture medium consisted of Minimal Essential Medium (MEM, Trace Bioβciences, Sydney, NSW, Australia) supplemented with 10% FCS (Trace Biosciences, Sydney, NSW, Australia) and 100 lU/ l Penicillin/100 μg/ml Streptomycin (P/S) (ICN Pharmaceuticals Inc, Costa Mesa, CA, USA). Cells were passaged 1 in 5 with 0.25% trypsin (Trace Biosciences, Sydney, NSW, Australia) /0.05% EDTA (BDH Chemicals Australia Pty Ltd, Kilsyth, VIC, Auβtralia) approximately every 5 daye.

Cloning of VEGFlf5 into the Expression Vector
Mouβe VEGF16S in Bluescript KS was obtained from Dr Georg Breier, Max Planck Institut, Germany (Breier et al, 1992) . VEGF1{S was inserted into the Bam HI site of pHβAPr-1-neo (Figure 1) (Gunning et al, 1987) . This cloning was performed via pGem 7Zf (4-) (Promega, Madison,

WI, USA) for the addition of a Bam HI site to the 3' end of VEGFlβ5. A sense VEGF-pHβAPr-1-neo clone was identified by Eco RI digeetion (Promega, Madison, WI, USA) .
VEGF-pHβAPr-1-neo DNA was prepared using the Qiagen Plasmid Midi Kit (Qiagen GmbH, Hilden, Germany) . The extraction was carried out as described in the manufacturer' s
protocol, and the reeulting pellet waβ reβuβpended in
500 μl TE buffer (10 mM Trie HCL, pH 8.0, 1 mM EDTA).

Transfection of RPE Cell Line
VEGF-pHβAPr-1-neo DNA waβ transfected into 407A cellβ using Lipofectin (Gibco BRL, Gaithersburg, MD, USA) as described in the manufacturer's instructions. Briefly, 2 mg of VEGF-pHβAPr-1-neo DNA in 100 μl OPTI-MEM (Gibco BRL, Gaithersburg, MD, USA) was mixed with 5 μl Lipofectin reagent in 100 μl OPTI-MEM. The mixture was allowed to stand at room temperature for 15 minutes, then made up to 1 ml with OPTI-MEM, and overlaid on to 60% confluent 407A cells. The cells were incubated at 37°C overnight in a humidified environment and 5% C02, then 4 ml MEM with
10% FCS and P/S was added. The cells were re-incubated for 24 hours before 1 mg/ml Geneticin (Gibco BRL, Gaithersburg, MD, USA) waβ included in the cell culture medium. After one week a βerieβ of discrete colonies was selected, and grown in 1 mg/ml Geneticin until established. The
concentration of Geneticin waβ then decreaβed to 300 μg/ml cell culture medium.
A control cell line consisting of 407A cells transfected with pHβAPr-1-neo only (407A-pHβAPr-l-neo) was also produced using Lipofectin. Both cell lines were maintained in MEM containing 10% FCS, P/S and 300 μg/ml Geneticin.

Selection of Primers for DNA and RT PCR
Primers were selected to allow βpecific
amplification of transfected mouse VEGFlβ5, without
background amplification of human VEGF1C5 from the human 407A cell line. The sequences of mouse VEGF165 and human VEGF1(5 as listed on the GenBank database were compared using the IBI Pustell Analyβis Software (IBI Ltd,
Cambridge, England) . 19mer regionβ which were less than 70% homologous with human VEGF1C5 were selected from mouse VEGF165. Primer sequenceβ were: "VEGFM01", 115-134 bp on mouβe VEGFlβs, 5 ' -AGG AGA GCA GAA GTC CCA T; "VEGFM02", 300- 318 bp on mouβe VEGFlβ5 5 ' -CGT CAG AGA GCA ACA TCA C.
Analysis of primer sequences by the Basic Local Alignment Search Tool (BLAST, National Centre for Biotechnology
Information, Bethesda, MD, USA) demonstrated homology to mouse VEGF forms only.

DNA PCR
Cells were harvested using 0.25% trypsin/0.05% EDTA. Samples of 2 x IO6 cells were collected and washed with PBS, then incubated for 3 hours, 37°C, in the presence of 100 ng/ml Proteinase K (Boehringer Mannheim, Mannheim, Germany) and 0.5% w/v Sodium Dodecyl Sulphate (SDS) (BDH Chemicals Australia Pty Ltd, Kilsyth, VIC, Australia) . DΝA waβ isolated by phenol/chloroform extraction and sodium acetate/ethanol precipitation. DNA pellets were
resuspended in 100 μl TE buffer.
All PCR reagents, including Ultra Pure Water, were obtained from Biotech International Ltd. (Bentley, WA, Australia) . The PCR reaction mixture consisted of 5 μl 5X Polymerisation Buffer, 25 mM MgCl2, 1U Tt Plus DNA Polymerase, 50 ng VEGFMOl, 50 ng VEGFM02 and Ultra Pure Water to 25 μl. 1 μl of each DNA sample was used for PCR. For each series of PCR reactions carried out, a positive control containing 20 ng VEGF-pHβAPr-1-neo DNA, and a negative control containing Ultra Pure Water in the place of DNA, were included. PCR reactions were carried out using a Perkin Elmer GeneAmp PCR System 2400 Thermocycler (Perkin-Elmer Corporation, Norwalk, CT, USA) . Cycles used were 1 cycle of 92°C for 5 minutes, 55°C for 1 minute, 74°C for 1 minute; 35 cycles of 92°C for 1 minute, 55°C for 1 minute, 74°C for 1 minute; 1 cycle of 92°C for 1 minute, 55°C for 1 minute, 74°C for 10 minuteβ. The PCR productβ were eiectrophoresed on a 2% agarose gel, and visualiβed by ethidium bromide staining.

Reverse Transcription PCR (RT PCR)
RNA waβ extracted using RNAzolB (Biotecx
Laboratories Inc., Houston, Texaβ, USA). The procedure waβ carried out aβ deβcribed in the manufacturer' s protocol, with RNA being extracted directly from confluent 25 cm3 flasks of cells (4 x IO6 cells per flask) . The resulting pellets were resuspended in 50 μl Diethyl Pyrocarbonate (DEPC) (BDH Ltd, Poole, Dorset, England) treated water.
RT PCR was performed using the GeneAmp
Thermostable rTt Reverse Transcriptase RNA PCR Kit
(Perkin-Elmer Corporation, Norwalk, CT, USA) . Reverse transcription and PCR reactions were carried out as described in the manufacturer's instructions. 200 ng RNA was used for each reaction. Water used for all reactions was Ultra Pure Water. The RT PCR positive control
contained 20 ng of VEGF-pHβAPr-1-neo DNA. The negative control received Ultra Pure Water in the place of RNA.
Controlβ for DNA contamination were produced by the
addition of rTt DNA Polymerase after completion of the Reverse Transcription step. RT PCR products were
precipitated using sodium acetate/ethanol. Samples were washed in 70% ethanol and resuspended in TE buffer to 1/5 the PCR reaction volume. PCR products were eiectrophoresed on a 2% agarose gel and visualised by ethidium bromide staining.

Production of a VEGF 1SS -Express ing Retinal Pigment
Epithelial Cell Line
VEGFlβ5 was successfully cloned into the Bam HI site of pHβAPr-1-neo. The identity of the clone was confirmed using a Bam HI digest which yielded two fragments of 10.0 kb, corresponding to pHβAPr-1-neo, and 656 bp, corresponding to mouse VEGF165. Eco RI digestion of the VEGF-pHβApr-1-neo clone produced two fragments of 5.7 kb and 5.0 kb, confirming that VEGF was in the sense
orientation.
VEGF-pHβApr-1-neo was transfected into the 407A cell line using Lipofectin. The presence of mouse VEGF1SS DNA in the transfected 407A cell line was confirmed using DNA PCR. DNA was extracted from VEGF-pHβApr-1-neo
transfected 407A colonies, along with DNA from the control 407A-pHβApr-l-neo cell line. PCR of the VEGF-ρHβApr-1-neo transfected 407A DNA reβulted in the production of a 200 bp DNA fragment in every colony teβted. This fragment was the size predicted from the position of the primers on the mouse VEGF165 gene, and agreed with the fragment size produced from the VEGF-positive control. One established colony of transfected cellβ waβ chosen for the remainder of the experiments (407A- EGF) . No signal waβ detected on PCR of 407A-pHβApr-l-neo. The results are illustrated in Figure IIA. DNA from untransfected 407A cells alβo produced PCR βignal, confirming that the primere being uβed were βpecific to mouse VEGF165.

RT PCR waβ uβed to verify the production of mouβe VEGFlβs mRNA by 407A-VEGF. On RT PCR of 407A-VEGF total RNA, a fragment of 200 bp was produced, corresponding to the fragment size predicted from the position of the mouse VEGF16S primere. No βignal was received from 407A-pHβApr-l-neo total RNA. Both RNA samples were shown to be free of contaminating DNA by omission of the cDNA synthesis step during RT PCR. The reβults are shown in Figure IIB. RT PCR using untransfected 407A RNA did not produce any βignal.

Tube Formation Assay
The assay was performed as described in
Example 8. CEC adhered to the Matrigel support within 1 hour of seeding. After 2 to 3 hours of culture, the CEC had migrated rapidly to form a reticular network of aligned cells, and subsequently began to form capillary-like structures on the surface of Matrigel . By 24 hours the CEC had the appearance of an anastomosing network, which is typical of vascular tubules. The quantitative analysis of tube formation, obtained from computer images, iβ
summarised in Figure 13.
The most extensive capillary network was seen in CEC cultured in the presence of 100 ng/ml human recombinant VEGF (Figure 13B) . The amount of capillary tube formation induced by the 407A-VEGF conditioned medium was similar to that from the human recombinant VEGF. In contrast the level of tube formation from conditioned medium of the control 407A-pHβAPr-l-neo cell line was significantly less, and was comparable to the control cultures containing Ham'e F12 medium with 5% FCS and P/S only.
There waβ a 100% increaβe in the amount of tube formation induced by 407A-VEGF conditioned media when compared to 407A-pHβAPr-l-neo. This difference was found to be significant (P = 0.009, Student's t-test). The difference between the control culture and the culture containing lOOng/ml human recombinant VEGF was also found to be βignificant (P = 0.002, Student's t-test).

Example 12 Cloning and Characterisation of Human RPE
Vascular Endothelial Growth Factor (RPE- VEGF)
Human RPE cells, available in our laboratory, are grown in tissue culture. To upregulate VEGF expresβion, cell cultures are treated in hypoxic conditions . The upregulation of VEGF expression is monitored with
immunohistochernistry. The mRNA is extracted from IO7 RPE cells, and a cDNA library carrying all genes expressed in the RPE/choroid is established using methods known in the art.
VEGF iβ a highly conserved molecule which is highly homologous between different βpecies. A murine VEGF cDNA clone, available in our laboratory, is uβed to βcreen the human RPE cDNA library in order to identify the full length human RPE-VEGF clone. Positive clones are analysed by restriction enzyme analysis and finally by DNA
sequencing. Full length RPE-VEGF clones are analysed to elucidate their genomic structure (initiation sequences, start and stop codons, putative exons etc.).
Clones carrying the full length RPE-VEGF are analysed for the expresβion of VEGF protein with in vitro translation. The identified clones are used to derive the anti-sense molecule for insertion into the vehicle system, and for the selection of the anti-sense oiigonucleotides.

Example 13 Pharmaceutical Agent for the Short-Term
Inhibition of VEGF Expresβion
Anti-βenβe DNA technology enables the sequence- specific inhibition of target molecules without affecting the non-targeted functions of the cell. As described above, we have demonstrated both in vitro and in vivo that anti-βenβe DNA can be uβed effectively to inhibit the antisense oligonucleotide into the vitreous .

A panel of 16 to 19-mer oiigonucleotides, 100% complementary to parts of the VEGF gene, is selected from the 5 ' and 3 ' ends of the DNA βequence . Sense and
scrambled sequenceβ are alβo uβed as control.
Phosphorothioate-protected oiigonucleotides are syntheeized on a DNA βyntheβizer and subjected to purification.

Example 14 Anti-Sense Agent for the Long-Term
Inhibition of VEGF Production
Preparation of VEGF-pAd.RSV for Homologous Recombination
VEGF165 in Bluescript IIKS (Stratagene) waβ uβed to produce Kpnl sites. Kpn I restriction enzyme βiteβ were obtained at both the 5 ' and 3 ' ends of VEGF165 by
subcloning. VEGF165 waβ removed from Bluescript II KS using an Xba I (5' cut) /Kpn I (31 cut) restriction enzyme digest, and cloned into pGem 7Zf(4*) (Promega). A Kpn site was then added to the 3 ' end by cloning VEGF165 into pGe 3Zf(+) (Promega), using a Hind III (3' cut) //Xba I (5' cut) digest .
VEGF was removed from pGem 3Zf(4-) with a Kpn I restriction enzyme digest and cloned into the unique Kpn I site on pAd.RSV. Thiβ plaβmid contains two segments of the adenovirus genome separated by cloning sites for the insertion of foreign DNA. The resulting clones were screened for the presence of sense and antisense clones, which were used in homologous recombination (VEGF-pAd.RSV) . VEGF165 was shown to be preβent and intact within pAd.RSV by restriction enzyme cleavage and sequencing.
VEGF-pAd.RSV DNA was prepared using the Qiagen Plasmid Midi Kit, as per the manufacturer's instructions. The DNA was linearised by Xmn I restriction enzyme
digestion, purified by sodium acetate/ethanol precipitation and resuspended in TE buffer. The DNA was then stored at -20°C until required.

Generation of Ad.RSV-VBGF or Ad.RSV-aVBGF by Homologous Recombin tion
The adenovirus type 5 deletion mutant, dl324, was used to generate the recombinant adenovirus carrying VEGF. dl324 is unable to replicate due to deletion of the El region and, in addition, carries a partial deletion in the E3 region. In order to generate viral particles this mutant was propagated in 293 cells, which supply the missing El region in trans. The linearised plasmid DNA pAdRSV-VEGF or pAd.RSV-aVEGF was co-transfected into 293 cells with dl324 viral DNA which has had its extreme left-hand sequenceβ removed by a Clal digeetion. Thiβ reduces the infectivity of dl324 and allows for easier
identification of recombinants. After transfection using the calcium phosphate precipitation method, screening of the resultant plaques yielded recombinant AdRSV-VEGF virus carrying VEGF in sense or antisense orientation.

Example 15 Construction of a Vehicle for the Permanent
Expresβion of Target Molecules
The vehicle described in Example 14 is suitable for long-term treatment in that it provides temporary
(maximum one year) expresβion of the anti-βenβe VEGF DNA molecule, and consequent protection against
neovascularisation. To achieve indefinite treatment, we use a vector system which enables the integration of VEGF in the anti-βenβe direction into the human genome preβent in RPE cells using an adeno-associated virus (AAV) vector, which means that the protection against neovascularisation can be provided for the rest of the life of the patient, as long as the RPE cells remain functional.
Adeno-associated viruses are non-pathogenic, are able to infect non-dividing cells, and have a high
frequency of integration. We use AAV-2, which is a replication defective parvovirus which can readily infect other cellβ βuch aβ RPE cellβ, and integrate into the genome of the hoβt cells. Recent characterisation has - 51 -revealed that AAV-2 specifically targets the long arm of human chromosome 19.
AAV constructβ use varying promoter sequences in combination with a reporter gene. The expresβion of the reporter gene mRNA is detected with PCR amplification or in situ PCR, and the integration of the reporter gene is identified by chromosomal analysis of RPE cells .
Using the appropriate restriction sites, the reporter gene is replaced by anti-sense VEGF DNA. The new construct iβ co-transfected with the complementing plasmid (pAAV/ad) into kidney 293 cells previously infected with adenovirus type 5 to make the rAAVaVEGF construct. The conetruct produced iβ uβed to infect RPE cellβ, and the expression of anti-sense VEGF is detected by PCR
amplification.

Example 16 Model Systems for Testing Inhibition
In Vitro
Human VEGF is cloned into COS cellβ to produce a culture system (VEGF-COS) in which the effective inhibition of VEGF expresβion can be teβted. The inhibition of VEGF expresβion iβ tested by Northern and Western blot analyses and quantified by immunoassay.
The toxicity of increasing concentrations of oiigonucleotides on COS cells is assessed with the trypan blue assay. The proliferation of COS cells is monitored with or without increasing concentrations of
oiigonucleotides. The inhibition of the expresβion of VEGF in controlβ and in cultures maintained in the preeence of anti-βenβe oiigonucleotides is monitored by Northern and Western blot analyses, immunocytochemistry and by a quantitative immunoassay.
RPE cells are cultured in hypoxic conditions and the up-regulation of VEGF expresβion iβ monitored in the preeence of increaβing concentrations of oiigonucleotides for an extended period of time. Toxicity, proliferation assay and the monitoring of VEGF expression are performed aβ described above.
CEC cells are cultured in normal and hypoxic conditions with or without increasing concentration of oiigonucleotides . In addition to the toxicity,
proliferation assay and VEGF detection, the effect of antisense oligonucleotide-mediated inhibition of VEGF
expresβion on tube formation iβ analyβed. RPE/CEC dual cultures produced in normal and hypoxic conditions will be subjected to similar tests. The same model systems are used to assess the long-term and permanent agents of the invention.

Example 17 In Vivo Model for Sub-Retinal Neovascular
Membrane (SRNV )
In addition to the above examples an animal model for study of the particular inhibition of the development of SRNV was developed. The model uses laser treatment of rats to induce symptoms similar to those observed in humane

Pigmented rate (Dark Agouti, DA) weighing between 175 and 250 g were anaeβthetized with an intramuscular injection of xylazine hydrochloride (2 mg/kg of body weight), acepromazine maleate (0.5 mg/kg), and ketamine hydrochloride (100 mg/kg of body weight) and given topical 0.5% proparacaine hydrochloride. The pupils were dilated with 2.5% phenylephrine hydrochloride.
Krypton laser radiation (647 nm) waβ delivered through a Zeiss slip lamp (Coherent Model 920
Photocoagulator, Palo Alto, Calif) with a handheld
coverslip (22 c 30 mm) serving as a contact lens. Laser parameters used were aβ follows: a spot size of 100 μm, a power of 150 mW, and an exposure duration of 0.1 s. An attempt was made to break Bruch's membrane, as clinically evidenced by central bubble formation with or without intraretinal or choroidal hemorrhage. We found that a treatment power of 150 mW most consiβtently produced this effect.

Approximately 40% of animals treated the above described way developed growth of blood vessels into the retina from the choroid. This growth iβ accompanied by the upregulation of VEGF expreββion, providing an excellent system to test our oiigonucleotides and constructs .

Example 18 Inhibition of RPE-VEGF Expreseion with
Anti-Sense Oiigonucleotides, Ad.RSV.aVEGF
and rAAVaVEGF Jn Vivo in Rats
Neovascularisation can be induced using pocket implants in the choroid or the subretinal layer. One of the disadvantages of these models is that the process of neovascularisation might not follow the same biochemical steps which naturally occur in humans suffering from ARMD. To overcome these difficulties we use an animal model in which choroidal neovascularisation is induced by VEGF overexpression in the RPE cells. Using recombinant adenoviruses carrying VEGF, for example Ad.RSV.VEGF, for the in vivo trials all animal models described above are utilised to provide us with a wide range of information. Tests are conducted to demonstrate the expresβion of a VEGF expression over a period of one year. Using Northern and Western blot analysis, VEGF down-regulation is monitored and immunohistochernistry is used to demonstrate the down-regulation of VEGF expresβion in a cell-specific manner. Using the above described animal models, choroidal
neovascularisation is monitored by histology and
angiography. These models are applicable to all the embodiments of the invention.

It will be -appreciated that the present invention is particularly useful in the study, treatment or
prevention of age-related macular degeneration, by virtue of the successful adenoviral gene transfer to the RPE.
Without wishing to be bound by any proposed mechanism for the observed advantages, the higher degree of gene
expreββion in the HRPE7 cells, compared with the F2000 cellβ, may indicate the ability of RPE cellβ to phagocytoee large molecules and hence increase the uptake of
adenovirus. The level of expression of the transgene may also be increased by increasing the time of expoβure or the viral titre.
The comparison studies between the HRPE7 cells and the F2000 fibroblast show that there are marked
differences in the pattern of expresβion between the different cell types under the same conditions . These differences could be exploited for targeting of different cells, for example RPE. The upβtroke in the
time/expression curve for RPE cells (Figure 5) was at 4 hours, while for F2000 cells it waβ 24 hours. There is, therefore, a window during which RPE cellβ are taking up Ad.RSV.βgal. and expreββing the tranegene at a
significantly higher level than F2000 fibroblastβ.
Transfection for periods of less than 24 hours would allow use of thiβ window aβ a targeting tool (eg. viruβ solutions could be aspirated from subretinal blebs or the vitreouβ after 24 hours) . The titre/expresβion curves (Figure 5) also show that there was a difference between the cellβ, with RPE cellβ beginning to express highly at a lower concentrations. Once again, low concentration could be used to preferentially target RPE cells. A combination of lower titres for less than 24 hours would combine the two effects and provide targeted delivery.
Aβ shown in some of the embodiments, the present invention may alβo be uβed in conjunction with adjuvants to keep viral toxicity to a minimum by reducing the titre required to effect gene transfer and expreββion.
We have shown a consistent and significant adjuvant effect for adenoviral gene transfer using HA.
This waβ the case in both phagocytic and non-phagocytic cell lines . The advantage of HA is its presence as a normal component of human vitreous and extracellular matrix, and its long history of therapeutic acceptance as a viscoelastic aid to surgery.

The important feature of HA in terms of its acting aβ a potential adjuvant is its ability to bind cell membranes and other molecules simultaneously. We propose that the HA molecule can bind adenovirus and the cell membrane at the same time, and therefore increase the contact time or concentration of virus in the vicinity of the cell membrane using this mechanism. We have identified cell surface receptorβ βpecific to HA identified on both F2000 and RPE7, aβ each cell teβted positive for the presence of CD44, RHAMM and ICAM-1 receptors.
Interestingly, the RHAMM receptors on RPE showed a nuclear distribution, and thiβ could account for the slightly higher adjuvant effect of HA in RPE than in F2000. Our preliminary studies of in vivo immunofluorescent βtaining for CD44 show no βignal in the neuroretina, βuggeβting that HA aββociation of the adenovirus may also be a potential targeting mechanism for RPE in vivo.

It will be apparent to the person skilled in the art that while the invention has been described in the Examples, various modifications and alterations to the embodiments described herein may be made without departing from the scope of the inventive concept disclosed in this specification.

References cited herein are listed on the following pages, and are incorporated herein by this reference.

REFERENCES

Aiello et al
New England Journal of Medicine, 1994 331 1980-1987

Akhtar, S. and Ivineon, A.J.
Nature Genetice, 1993 4.215-217

Algvere, P.V.
Graefeβ Arch. Clin. Exp. Ophthalmol . , 1994 232 707-716

Bennett et al
IOVS, 1994 3_5 2535-2542

Bok, D. and Young
The Retinal Pigment Epithelium
Ed Zina K and Marmor M (Harvard University Press, 1979)

148-174

Boris-Lawrie, K.A.
Genetics and Development, 1993 3_ 102-109

Breier, G., Albrecht, U., Sterrer, S. and Risau, W. Development, 1992 114 521-532.

Culver, K.W. et al
Science, 1992 256. 1550-1552

Davis, A.A., Bernstein, P.S., Bok, D., Turner, J.,

Nachtigal, M. and Hunt, R.C.
Inveβt. Ophthalmol. Vie. Sci., 1995 3_6_ 955-964.

Engler, CB. et al
ACTA Ophthalmologica, 1994 11.27-31

Folkman, J., Haudenβchild, CC. and Zetter, B.R.
Proc. Natl. Acad. Sci., USA, 1979 76 5217-5221 Fung, W . E .
Ophthalmol., 1991.110. 1383-1384

Graham, F.C et al
Gene Transfer and Expresβion Protocole, The Humane Preβs

Inc., Clifton, NJ, USA 1991:101-128

Gunning, P., Leavitt, J., Muscat, G., Ng, S. and Kedes, L. Proc. Natl. Acad. Sci. USA, 1987 84.4831-4835.

Guyer, D.R. et al
Am. J. Ophthalmol., 1992 112 349-350

Haralabopoulos, G.C, Grant, D.S., Kleinman, H.K.,
Lelkes, P.I., Papaioannou, S.P. and Maragoudakie, M.E. Lab. Inveβt., 1994 Tl 575-582.

Hata et al
Virchows Arch., 1995 426 479-486

Helene, C
Eur. J. Cancer, 1991 22 1466-1471

Hudson, H., Frambach, D. and Lopez, P.
Brit. J. Ophthalmol., 1995 79.417-423

Ibanez, H.E. et al
Arch. Ophthalmol., 1995 113 62-69

Johnson, P.A.
J. Virology, 1992 .66 2952-2965

Kaneda, Y. et al
Exp. Cell Res., 1987 173 56-69

Kennedy, CJ. et al
Exp. Cell Res., 1994 210 209-214 Kozareky, K.F. and Wilβon, J.M.
Genetice and Development, 1993 3 499-503

Le Gal La Salle, G. et al
Science, 1993 259. 988-990

Li et al
Inveβt. Ophthalm. Vie. Sci., 1994 35.2543-2549

Mclnnes, R.R. and Bascom
Nat. Gen., 1992 1 155-157

Malecaze et al
Arch. Ophthamol., 1994 112 1476-1483

Mashmour et al
1994 1 122-126

Milam, A.H.
Neurobiology, 1993 3.797-804

Miller et al
Am. J. Physiol., 1994 4_5 574-584

Morse, L.S. and Sidikaro, Y.
Curr. Eye Reβ., 1990 9. 631-642

Pe'er et al
Lab. Inveβt., 1995 72. 638-645

Pierce et al
Proc. Natl. Acad. Sci., 1959 .92 905-909

Simorre-Pinatel et al
Inveet. Ophthal. Vis. Sci., 1994 35 3393-3400 Rakoczy, P .E .
Invest. Ophthalm. Vis. Sci., 1994 3_5_ 4100

Rakoczy, P.E. et al
Biol. Cell, 199226.49-54

Sakomoto, T., Sakomoto, H., Hinton, D.R., Spee, C, Ishibashi, T. and Ryan, S.J.
Curr. Eye Res., 1995 14_ 621-627

Salmons, B. and Gύnzburg, W.H.
Science, 1992 256. 1550-1552

Schachat A.P.
Arch. Ophthalmol., 1994 112 459-461

Siegfried, W.
Exp. Clin. Endocrinol., 1993 101 7-11

Stratford-Perricaudet LD et al
J.Clin Inveβt. 1992 .90. 626-630

Thiema et al
Diabetes, 1995 44. 98-103

Wagner, R.W.
Nature, 1994 372 333-335

Wagner, R.W. et al
Nature Biotechnology, 1996 .14 840-843