WO2009081154A1

WO2009081154A1 - Targeted delivery of macromolecules

Info

Publication number: WO2009081154A1
Application number: PCT/GB2008/004247
Authority: WO
Inventors: Leif Lindholm; Maria Magnusson
Original assignee: Got-A-Gene Ab; Rowe, Daniel, Thomas, David
Priority date: 2007-12-21
Filing date: 2008-12-22
Publication date: 2009-07-02
Also published as: GB0725065D0

Abstract

The present invention provides a method for the transportation of macromolecules or aggregates thereof across a targeted layer of epithelial cells comprising (a) selectively displaying a transcytosis receptor (TrCytR) on the surface of the targeted epithelial cells; and (b) applying the macromolecules or aggregates thereof to the targeted epithelial cells, wherein said macromolecules or aggregates thereof comprise at least one region which induces transcytosis of said macromolecules or aggregates thereof upon binding to said TrCytR. Cell lines, vectors and compositions for performing the method of the invention are also provided.

Description

Targeted Delivery of Macromolecules

The present invention relates to the targeted delivery of macromolecules and aggregates thereof to epithelial tissues and/or tissues associated with epithelial tissues. The endothelium Js a special type of epithelium that lines, inter alia, the luminal side of blood vessels and, in particular, the invention relates to the delivery of gene therapy vectors to tissues associated with the endothelium of the vasculature of tumours, such as the tumour tissue.

Epithelial tissues are those tissues of an organism that form cellular barriers between two environments. To varying extents, epithelial tissues act as barriers because the cell junctions between cells can restrict uncontrolled paracellular transport (i.e. transport between the epithelial cells) and promote the more controllable transcellular transport (i.e. transport through the epithelial cells). In tight epithelial tissue the cell junctions are tight junctions and numerous. In these tissues paracellular transport is highly restricted, the blood-brain barrier is an extreme example of the level of control that can be achieved. Delivery of macromolecules and aggregates thereof to tissue associated with epithelial tissues is often difficult on account of the restrictive permeability of the epithelial tissue. Even in relatively permeable epithelial tissues, transport across these tissues could be enhanced by exploiting the transcellular pathways.

Passage across epithelial tissue can occur via transcellular transport. One of the mechanisms involved in this type of transport is transcytosis (Tuma, P. L. et al., 2003, Physiol Rev 83, 871 ). Transcytosis is the transport of macromolecular cargo from one side of a cell to the other within a membrane-bound carriers). Examples of macromolecules transported in this manner are albumin, orosomuciod, IgG, LDL, transferrin-iron and IgA. In tight epithelial tissues this process is a key mechanism for transport across the epithelium as paracellular transport is restricted in these tissues.

Access of macromolecular therapeutic agents to tumour cells is often restricted by the presence of epithelial (e.g. endothelial) cells. This is especially the case if the therapeutic agent is administered systemically and enters the tumour by the vasculature of the tumour. In this instance the therapeutic agent must cross the endothleium of the tumour vasculature to reach the tumour cells. It is commonplace that a tumour will have a high interstitial fluid pressure compared to healthy tissue and this results in convective flow of fluid and solutes from the tumour to the circulation rather than the usual flow from the circulation into the tissue observed in healthy tissue. This discourages entry of a therapeutic agent into the tumour by passive paracellular mechanisms.

The ideal cancer therapeutic agent would be a very potent, highly specific and relatively safe substance that can be administered systemically. There are many cancer therapeutic agents and gene therapy as a cancer treatment has attracted much interest. Within the field of gene therapy there have been a number of gene delivery systems developed, e.g. liposome mediated, polymer mediated, dendrimer mediated and viral vectors. So far, adenoviral vectors have proven to be safe but are still far away from reaching the above criteria. In particular the successful systemic application of adenoviral vectors is still not a reality. The innate and the acquired immune system, as well as physical (i.e. intratumoural pressure, extracellular matrix, stroma) and biological (i.e. scarcity of viral receptors) factors generate high barriers for such routes of administration. As a result few, if any, successful clinical applications have emerged.

Adenovirus (Ad) vectors are promising candidates as delivery vehicles in gene therapy studies. They are well characterized, are genetically stable, can be grown to high titres, lack integration into the host cell genome and their genetic make-up is modifiable by routine molecular biology techniques. Several thousand patients have received genetically modified adenoviruses in experimental gene therapy studies. However, their utility in the clinic is limited since the virus infects many different cells in the body and cells relevant for targeting (such as cancer cells and dendritic cells in case of vaccine vectors) are often resistant to infection. The vectors are also powerfully immunogenic

Enhanced targeting of Ad vectors is needed both to decrease the transduction of undesired cells in vivo and to increase the transduction of target cells. Hence, the ideal vector should be detargeted from its natural receptors and retargeted to a new, selected receptor. In cancer therapy this is especially important since many tumour cells are poorly transduced by adenoviral vectors due to low levels of the coxsackie and adenovirus receptor (CAR). There is also the additional problem of neutralizing antibodies, formed as a consequence of earlier infections. Vectors therefore need to be protected against pre-formed antibodies.

However, adenovirus biodistribution is not only determined by vector specificity and cellular receptor expression but also by anatomical barriers such as the vascular endothelium and extracellular stroma that has to be bypassed to reach the desired cell types. So far, too few investigations and even fewer solutions have been described for these barriers. The first barrier is the many abnormalities of the tumour microcirculation. Most detrimental is probably the high interstitial fluid pressure within tumours, resulting in a convective flow of fluid from the tumour into the circulation. This convective flow makes it very hard for viruses and macromolecules to enter the tumour bed from the circulation by passive paracellular routes. Normalisation of the vessels by anti-angiogenic agents has to some extent resulted in decreased intra-tumoural pressure and improved distribution of anti cancer drugs (Tong, R. T., et al., 2004, Cancer Res 64(11 ): 3731-6) but alternative approaches to this problem would be beneficial. A problem facing the art is to provide further methods for the targeted delivery of macromolecules or aggregates thereof to cells associated with an epithelial tissue.

A particular problem facing the art is to provide gene therapy vectors and methods for administration that (i) show enhanced selectivity for target cells by having increased transduction of target cells but decreased transduction of non-target cells in vivo and (ii) overcome the anatomical barriers to the target cells presented by epithelial tissues.

The present invention seeks to address these problems.

In a first aspect the invention provides a method for the transportation of macromolecules or aggregates thereof across a targeted layer of epithelial cells comprising:

(a) selectively displaying a transcytosis receptor (TrCytR) on the surface of the targeted epithelial cells; and

(b) applying the macromolecules or aggregates thereof to the targeted epithelial cells, wherein said macromolecules or aggregates thereof comprise at least one region which induces transcytosis of said macromolecules or aggregates thereof upon binding to said TrCytR.

By "layer of epithelial cells" it is meant an arrangement of epithelial cells that acts as a barrier between two compartments. This barrier could be more than one cell thick in places or in its entirety (known as stratified epithelium). Preferably the layer of epithelial cells is predominantly one cell thick (known as a simple epithelium). In other words, the layer of epithelial cells is one cell thick across 70%, preferably 85%, most preferably 95% of its surface area. Epithelial cell layers can also be defined in terms of the shape of the cells, i.e. squamous (irregular flattened shape) cuboidal (substantially cube- shaped, i.e. height, width and depth are substantially the same), columnar (substantially column-shaped, i.e. the height of the cells is greater than the width and depth), and transitional (the shape of the cells depends on whether the layer is contracted or distended). Simple and stratified layers can be formed from any of these cell types. Typically an epithelial layer will be homogenous for a particular cell shape but heterogenous layers are contemplated. Preferably the epithelial layer of the invention is a simple epithelial layer and preferably squamous, cuboid or columnar, most preferably squamous.

Non-limiting examples of epithelial cell layers that can be targeted in the present invention are the epithelial cells of the skin and those that line the lumen of the lungs, the gastrointestinal tract, the reproductive and urinary tracts, the exocrine and endocrine glands, the walls of the pericardium, pleurae and peritoneum and, in particular, the endothelium (the inner lining of blood vessels, the heart, and lymphatic vessels) Preferably the epithelial surface is the endothelium of the circulatory system. Most preferably the epithelial surface is the endothelium of the vasculature of tumours (sometimes referred to as the neoendothelium).

In a prefered embodiment the invention provides a method for the transportation of macromolecules or aggregates thereof across a targeted layer of endothelial cells comprising :

(a) selectively displaying a transcytosis receptor (TrCytR) on the surface of the targeted endothelial cells; and (b) applying the macromolecules or aggregates thereof to the targeted endothelial cells, wherein said macromolecules or aggregates thereof comprise at least one region which induces transcytosis of said macromolecules or aggregates thereof upon binding to said TrCytR.

The macromolecules to be transported include, but are not limited to, biological polymers such as proteins, nucleic acids, polysaccharides and lipids and macromolecules that are combinations thereof such as glycoproteins or lipoproteins. The nucleic acid can be DNA, RNA or polynucleotide analogues such as morpholino modified nucleotides, 3'-5' phosphoroamidates or peptide nucleic acids. The nucleic acid will typically be a vector, for instance encoding a gene of interest or an antisense construct, or the nucleic acid might be the antisense construct itself or a probe or primer. Non-limiting examples of proteins/glycoproteins transported in accordance with the invention are albumin, insulin, IgG, IgA, transferrin, melanotransferrin (MTf, also known as P97).

Aggregates of macromolecules comprise two or more macromolecules. These aggregates can range from simple aggregates with no observable order, to highly ordered structures comprising many individual macromolecules. The macromolecules in these ordered structures might be the same (i.e. the aggregate is homogeneous) but more often there will be a number of different species of macromolecules in the aggregates (and each species of macromolecule might be represented by multiple copies), i.e. the aggregate is heterogeneous. Preferably, there will be at least two non- identical macromolecules in the aggregate, more preferably there will be between 3 and 5 different species of macromolecule, most preferably more than 7 different species of macromolecule will be in the aggregate (e.g. about 10).

Typical examples of homogenous aggregates would be a homogenous lipid bilayer or liposome. Heterogeneous aggregates include, but are not limited to viruses, viral vectors, virus-like particles (VLPs)₁ polycationic complexes containing DNA (also known as polyplexes), complex liposomes, liposomes containing nucleic acid vectors, complexes of cationic and neutral lipids containing DNA (also known as lipoplexes), inert particles carrying macromolecular markers or labels and LDL or HDL. Most preferably the macromolecular aggregate is a viral vector, most preferably an adenoviral or a baculovirus vector, e.g. adenovirus Ad 5. A TrCytR is any receptor which, when present at the surface of a cell, can bind a macromolecule or aggregates of macromolecules and induce the transport of that macromolecule or aggregate to another surface of that cell. TrCytRs are typically transmembrane proteins.

Selective display of the TrCytR can be achieved in a number of ways. Selective display of the TrCytR involves the selective introduction of a TrCytR that is not expressed in the cells of the targeted epithelial cell layer or a significant increase in the display in the targeted epithelial cell layer of a TrCytR that is already expressed therein. The display is selective in that it is designed to occur in the target epithelial cell layer in preference to other tissues, there may be some display in surrounding or other tissues but the display is greater in the target epithelium. Preferably there is negligible or substantially less (as compared to the target epithelium) display in non- target tissues.

This is conveniently achieved by selectively introducing the TrCytR to the surface of the target epithelial cell layer or by selectively increasing the display of a TrCytR on the surface of the target epithelial cell layer by means of a vector targeted to the epithelial cell of interest. For instance, the targeted vector could be any of the viral vectors mentioned above, or a fusiogenic VLP. Conveniently the vector is a fusiogenic VLP, a baculovirus vector or an adenovirus vector. By targeting the vector to the epithelial cells of interest, the display of the TrCytR is selective to those cells.

Targeting of the vector that introduces the TrCytR to the surface of the target epithelial cell layer or selectively increases the display of a TrCytR on the surface of the target epithelial cell layer to the epithelial cell layer of interest can be achieved in any conventional way. Conveniently this will be by displaying a targeting molecule on the surface of the vector that selectively binds a partner on the target epithelial cell layer. By "selectively bind" it is meant that the targeting molecule binds a binding partner on the target epithelial cell layer in preference to molecules on other tissues, there may be some binding to molecules in surrounding or other tissues but the amount of binding is greater in the target epithelial cell layer. Preferably there is negligible or substantially less binding in non-target tissues (as compared to the binding to the binding partner in the target epithelial cell layer). The skilled man would be able to measure the amount of binding of a particular vector to target and non-target tissues or cells using routine techniques available to him such as ELISA, FACS or plasmon resonance. By comparing the relative amounts of binding to target and non-target tissues or cells the skilled man will be able to determine if a putative targeting molecule selectively binds the target tissue.

Preferably the targeting molecule binds a binding partner on the target epithelial cell layer and binds with equal or greater affinity to less than 10, preferably less than 5, more preferably less than 3 and most preferably no other different partners in their physiological contexts and that are accessible to the vector at the time the vector is administered. Typically the binding affinity of the target molecule to a binding partner will be at least in the micromolar (mM) range, preferably at least in the nanomolar (nM) range and more preferably in the picomolar range (pM).

By "physiological context" it is meant that the putative other different partners are bound by the targeting molecule when they are in the form that they adopt when they are in their normal physiological environments or in in vitro models thereof. By "accessible" it is meant that the vector can encounter the other different partners during or following administration to the targeted epithelial cell layer. This is therefore dependent on factors such as the mode of administration and the recipient host, e.g. an intravenous administration of a vector will expose the vector to a significant amount of the host's body whereas administration of the same vector to an ex vivo or in vitro tissue would result in the vector only being exposed to the isolated tissues and the culture media. In another example if a targeting molecule binds a single protein in a human and a single protein in a mouse, intravenous administration of a vector displaying that targeting molecule to a human will not show any cross-reactivity because the mouse protein is not present in the human.

Most preferably the sole binding partner of this targeting molecule is the binding partner on the target epithelial cell layer. The skilled man would be able to identify and select appropriate targeting molecules for his needs without undue burden.

Conveniently, these targeting molecules will be molecules that bind molecular markers on the surface of the target epithelial cell layer that are substantially exclusive to the target epithelial cell layer. By substantially exclusive it is meant that the molecular markers are only displayed and accessible to the targeting molecule on the target epithelial cell layer and less than 5, preferably less than 3, more preferably less than 2 other different tissues in the host. Most preferably the molecular marker is unique to the epithelial cell layer. Alternatively, or addition, the molecular marker is one that is displayed the surface of the target epithelial cell layer at a level that exceeds the display in other tissues. Preferably there is at least two, preferably at least 5, more preferably at least 10 times the level of display in the target epithelial cell layer as compared to non-target tissue. Most preferably there is negligible or substantially less display of the molecular marker in non-target tissues (as compared to the display in the target epithelial cell layer). Levels of display of molecular markers are routinely testable using techniques such as western blotting, immunohistochemistry, ELISA, FACS, plasmon resonance, reverse transcriptase analysis.

Preferably, the target epithelium is endothelium, particularly tumour endothelium, and representative examples of suitable targeting molecule binding partners (molecular markers) are prostate specific membrane antigen (PSMA; Milowsky, M. I., et al., 2007, J Clin Oncol 25, 540), endoglin (Nettelbeck, et al., 2001 , MoI Ther 3(6): 882-91.; Nicklin, S. A., et al., 2004., J Gene Med 6(3): 300-8), the RGD4C and NGR peptides (Arap, W., R. , et al., 1998, Science 279(5349): 377-80) endosialin (also known as tumour endothelial marker 1 or TEMI) and delta like ligand 4 (DII4) (Thurston, G., et al., 2007, Nat Rev Cancer 7, 327). Still further examples include TEM5, TEM7, TEM8, CD34; LYVE1 (lymphatic vessel endothelial hyaluronan receptor 1), VE-cadherin (CD144), von Willebrand Factor (vWf); platelet/endothelial cell adhesion molecule (CD31 antigen or PECAM1 ); intercellular adhesion molecule-1 (ICAM-1 ), CD146, endocan (ESM-1 ), Endoglyx-1 , vascular endothelial growth factor receptor 1 and 2 (VEGFR-1 and VEGFR-2).

The targeting molecules can conveniently be the natural ligands of the target epithelial cell molecule markers or molecules that selectively bind thereto. For instance the vector might be engineered to display antibodies to the receptor. Camelid antibodies (antibodies from the members of the taxonomic family Camelidae, e.g. animals such as camels, dromedaries, llamas, alpacas, vicunas, and guanaco) also known as nanobodies are antibodies of note in this embodiment. An alternative would be to use Affibody™ molecules. Affibody™ molecules use the disulfide bond independent 58 aa three-α-helix bundle structure Zwt as scaffold for new binding specificities (Nord, K., et al., 1997, Nat Biotechnol 15, 772). Zwt is a stable and robust framework that importantly for Ad retargeting, is soluble in the cell cytoplasm. By randomization of the Zwt domain novel variants with new binding specificities can be selected from phage display libraries. In accordance with the invention, Affibody™ molecules can be produced that will specifically bind molecular markers unique to the target epithelial cells, e.g. the molecular markers described above.

A different approach is the two-component system. In this system bispecific conjugates are used that bind both to the vector and to the molecular marker. They can be polymers conjugated to targeting ligands, chemically crosslinked targeting ligands, recombinant bispecific fusion proteins or bispecific peptides. Antibody binding vectors represent a simple system to coat vectors with antibodies recognizing desired molecular markers (Henning, P., et al., 2005, Gene Ther 12, 211). Another system is based on incorporation of biotin acceptor peptides into vector proteins. The vector is then metabolically biotinylated during production and can be retargeted by coating with biotinylated versions of the targeting molecules described above via tetrameric avidin (Parrott, M. B₁ et al., 2003, MoI Ther 8(4): 688-700). A still further system is based on the snake venom bungarotoxin (Btx). Peptides which bind to this protein with high affinity have been identified and these binding pairs can be used in the same way as the avidin/biotin system described above (McCann, CM., et al., 2005, Biotechniques, 38, 945).

A variation to these approaches involves encapsulation of vectors in a polymer coating which comprises the targeting molecules or molecules capable of interaction with the bispecific conjugates discussed above. This is discussed below. Such coats have the added benefit of shielding the vector from the host's immune system.

In one embodiment the TrCytR is directly introduced to the epithelial cells by a fusiogenic VLP. Such vectors are described in detail in WO2006/059141 and are specifically contemplated for use in the present invention. VLPs are self-assembling, non-replicating particles lacking the viral genome that are formed by one or several viral structural proteins (Ramqvist, T., et al., 2007, Expert Opin Biol Ther, 7, 997). VLPs are typically formed when a gene encoding a viral structural protein is overexpressed in a host cell in isolation from other viral genes. In the cytosol, the structural proteins assemble into the VLP in a process analogous to the process in which a bona fide virus particle assembles. Of course, without the other viral genes being present a true virus particle cannot be formed. Formation of VLPs results in their release from the host cell. This may be by cell lysis. In the case of VLPs derived from enveloped viruses this process is by budding from the host cell and this results in the VLP being enveloped by a lipid bilayer derived from the plasma membrane of the host. For example, Sf9 insect cells infected with the baculovirus Autographa californica nucleopolyhedrovirus (AcMNPV) expressing retroviral Gag polyproteins, e.g. HIV-1 (AcMNPV-GagHIV) produce large quantities of membrane-enveloped VLP's which are constituted of retroviral Gag polyproteins and are devoid of viral genome (Boulanger, P., et al., 1996, Curr. Top. Immunol., 214, 237). By incorporating fusiogenic proteins (e.g. the baculovirus gp64 envelope fusion protein) into these coated VLP the bilayer of the VLP can fuse with the bilayer of recipient cells.

Fusiogenic proteins are typically viral proteins that can induce the fusion of the plasma membrane derived envelope of the VLP to the membrane of the recipient cell. It is this mechanism that results in entry of the proteinaceous component of the VLP to the cytosol. The envelope glycoproteins of RNA viruses and retroviruses are well known to bind cell receptors and induce this fusion. Accordingly these proteins are responsible for the infectivity of these viruses. Examples of fusiogenic proteins include, but are not limited to, influenza haemagglutinin (HA)₁ the respiratory syncytial virus fusion protein (RSVFP), the E proteins of tick borne encephalitis virus (TBEV) and dengue fever virus, the E1 protein of Semliki Forest virus (SFV), the G proteins of rabies virus and vesicular stomatitis virus (VSV) and baculovirus gp64.

The bilayer coat of fusiogenic VLPs can incorporate homologous viral glycoproteins from the same virus, e.g. Gp160/Gp120 from HIV-1 in HIV Gag-based VLPs, but also heterologous viral glycoproteins from unrelated viruses, e.g. VSV-G in HIV Gag-based VLPs₁ a phenomenon which has been called 'pseudotyping' (WO2006/059141 ). These proteins can contribute fully or in part to the fusiogenic nature of the VLP and/or to the targeting of the VLP by being modified to carry the targeting molecules discussed above. The use of HIV gp160 comprising a targeting molecule in a fusiogenic VLP is contemplated specifically. In addition, Gag polyprotein of HIV-1 binds to an auxiliary viral protein called Vpr, via the C-terminal p6 domain of Gag (Wu, X., et al., 1995, J Virol 69, 3389). Thus, Vpr protein is encapsidated into HIV-1 virions in stoechiometric ratio with Gag polyprotein. As an application of this property, fusion proteins Vpr-X or X-Vpr (e.g. X = GFP or luciferase) can be coencapsidated with Gag polyprotein with nearly the same efficiency as nonfused Vpr (Wu, X., et al., 1995, supra). A further mechanism to incorporate proteins in to the VLP bilayer is to coexpress membrane bound proteins at the plasma membrane of the host cells. These proteins will be incorporated into the VLP as the nascent VLP bud off from the host cell membrane. In this regard, WO2006/059141 describes how the transmembrane proteins (e.g. the TrCytR) can be modified in order to optimise incorporation into VLPs.

Through one or more of these mechanisms the bilayer of the fusiogenic VLP can be loaded with a protein of interest (e.g. the TrCytR described above) and targeting molecules. Preferably the VLP of the invention will also display an antibody (e.g. scFv, Fab fragment or camelid antibody), alternatively an Affibody™ molecule, against PSMA or DII4 in order to target the VLP to tumour endothelium. Preferably this display is achieved by attaching these targeting molecules via the biotin/avidin bridge system described above.

Alternatively or in addition to the use of a fusiogenic VLP a viral vector can be use to introduce the TrCytR. Viral vectors would typically introduce an expression construct encoding the TrCytR although alternative approaches are also contemplated. For instance a factor that upregulates the expression of a native TrCytR could be introduced by the vector or, in the case of enveloped viral vectors, the lipid bilayer of the virus might comprise the TrCytR and the TrCytR can be incorporated into the host cell membrane when the vector enters the host cell in much the same way as a fusiogenic VLP. The targeting strategies described above apply to varying degrees to viral vectors depending on the physical properties of the viral vector used. The skilled man would immediately recognise which strategy could apply to which viral vector and be able to accommodate any necessary modifications. In any event, the same targeting molecules can be used and the above discussion in relation to the targeting molecules applies mutatis mutandis. Conveniently, the proteins of the viral vector are modified to carry one or more of the targeting molecules described above.

Preferably the viral vector is an adenovirus or baculovirus, preferably a vector based on adenovirus A5.

The following discussion of the host-virus interaction makes reference to the Ad5 vectors, although the concepts are shared by other adenoviral vectors.

The first step of adenovirus infection is mediated by binding of the globular carboxy-terminal "knob" domain of the adenovirus type 5 fibre protein and the adenovirus cellular receptor, identified as the coxsackievirus group B and Ad receptor, called CAR. The trimeric fibre protein protrudes from each of the 12 vertices of the viral icosahedron where it is attached noncovalently to the penton base. The second step is internalization by receptor-mediated endocytosis mediated by the interaction of Arg-Gly-Asp (RGD) sequences in the penton base with cellular integrins αVβ3 and αVβδ. After internalization the virus is localized in clathrin-coated pits and then in cell endosomes, from which the virions escape and enter the cytosol where the virus is disassembled and the DNA is transported to the nucleus. Due to the adenovirus entry pathway described above, a way of targeting adenoviral vectors is to modify the fibre protein to display binding motifs specific for receptors on the target cell. To this end peptide ligands have been inserted into various virus capsid proteins e.g. the Hl-surface loop of the fibre knob, to the C-terminus of the fibre knob, the plX protein and the loops of the hexon protein (Curiel, D. T. in Vector targeting for therapeutic gene delivery Wiley-Liss Inc., Hoboken, 2002, 171-200; Barnett, G., et al., 2002, MoI Ther 6, 377; Vellinga, J., 2004, J. Virol, 78, 3470). Ad vectors with dual specificity can also be used by inserting two different peptides in the same fibre (Hl- loop and C-terminally) or even with two different fibres in the same capsid (Wu, H.,et al. 2002, Hum Gene Ther 13(13): 1647-53; Henning, P. et al., 2006, J Gen Virol 87(R 11): 3151-60)

Tumour cell transduction can also be greatly improved through modification of the knob domain by removing CAR-binding residues and inserting sequences that target alternative receptors (Henning, P. et al., 2005 supra; Henning, P. et al., 2002, Hum Gene Ther 13, 1427). Preferably linkers are introduced between the targeting molecules and the flanking knob sequences as described in (Magnusson, M. K., et al., 2007, Cancer Gene Ther 14, 468). The resulting virus is re-targeted and exhibits almost wild type growth characteristics.

A further strategy involves the removal of the entire knob of the fibre and replacement with a heterologous trimerization domain (Henning, P. et al., 2002, supra), e.g. the neck region domain from human lung surfactant protein D, carrying the targeting molecule.

An important restriction on targeting ligands for Ad5 is that complex ligands dependent on disulfide bonds for folding and ligand reactivity cannot be rescued into viable virus when fused to viral proteins. Therefore a preferred class of targeting molecules are those that are not dependent on disulfide bonds for folding and reactivity. Such modified viral vectors are described in WO02/08263. These targeting molecules may be based on a three α-helical bundle structure, the so-called Affibody™ molecules, that can be efficiently used for genetic re-targeting of Ad5 (Henning, P. et al., 2002, supra). These molecules have been described above. The targeting molecules can also be based on hyperstable single chain antibody fragments (hyperstable scFv) which are capable of correct folding in the cytosol and have been used to retarget adenoviral vectors (Martineau, P., et al., 1998, J MoI Biol 280, 117) and camelid antibodies.

The targeting molecules can also be provided in or on an artificial polymer coat surrounding the viral vector. This process is known from the art, e.g. (Green, N. K., et al., 2004, Gene Ther 11, 1256; Ogawara, K., et al. 2004, Hum Gene Ther 15, 433). Preferably the targeting molecule is an antibody (e.g. a camelid antibody) or an Affibody™. Preferably the polymer is a polymer such as polyethylene glycol (PEG) or poly[N-(2-hydroxypropyl) methacrylamide] (PHPMA), i.e. polymers that are uncharged, hydrophilic, linear, nonimmunogenic and have a low order of toxicity. PEG is approved by the Food and Drug Administration (FDA) for use in drugs (parenterals, topicals, suppositories, nasal sprays), foods, and cosmetics. The targeting molecule can be incorporated in the polymer coat by linking the target molecule to the polymer and using the polymer conjugated target protein alone or with unconjugated polymer to coat the vector. The PEGylation of proteins, that is, the covalent attachment of polyethylene glycol (PEG) chains of defined length mainly to the lysine amino group, is long established and currently applied to numerous active proteins approved for therapeutic use.

The demonstrated advantages of PEGylated proteins versus their unmodified forms include increased plasma half-life, reduced antigenicity and increased resistance to proteolysis. Investigators have reported the use of PEGylated Ads that retain their ability to transduce cells and tissues, show reduced cytotoxic T cell production, extend the time of gene expression, are protected from antibody neutralization, and allow expression after administration to animals previously immunized with unmodified virus. Accordingly this approach to providing targeting molecules to vectors has other advantages. As discussed below, a polymer coat can also protect the vector from the host's immune system. This improves the half life of the vector in the circulation and inhibits the generation of an immune response to the vector.

Alternatively or in addition to the above transductional targeting, a vector carrying an expression construct can be effectively targeted to the epithelial tissue of interest by placing the expression product of that vector under the control of a promoter specific to the epithelial tissue. In the case of replication competent viral vectors further control can be exerted by making replication specific to the epithelial cells of interest. These mechanisms are both known as transcriptional targeting.

Conditionally replicative adenovirus (CRAds) are genetically modified to preferentially replicate in specific cells by either (i) replacing viral promoters with tissue specific promoters or (ii) deletion of viral genes important for replication that are compensated for by the target cells only. The skilled man would be able to identify epithelial cell specific promoters.

The vector responsible for the targeted display of the TrCytR can be targeted by more than one targeting molecule and by both transductional and transcriptional targeting methods.

In a prefered embodiment the invention provides a method for the transportation of macromolecules or aggregates thereof across a targeted layer of epithelial cells comprising : (a1 ) applying to the targeted epithelial cells a fusiogenic VLP which is targeted thereto and which comprises a transcytosis receptor (TrCytR); and/or

(a2) applying to the targeted epithelial cells an adenoviral vector which is targeted thereto and which encodes or comprises a transcytosis receptor (TrCytR); and

Non limiting examples of the TrCytRs that can be displayed according to the invention are the polymeric immunoglobulin receptor (plgR), the melanotransferrin (MTf₁ also known as P97) receptor (MTfR/P97R), gp60, the IgG receptor (FcRn), the transferrin receptor, LDL receptor, insulin receptor, megalin. These are all naturally occurring TrCytRs. Further examples are listed in Tuma et al. {supra). Preferably plgR and P97R are used. In an in vitro model of the blood-brain barrier (monolayers of bovine brain microvascular endothelial cells) the P97R expressed naturally in those cells has been shown to transcytose adenovirus when adenovirus particles are targeted thereto with a bi-specific adaptor protein containing adenovirus knob and P97R binding regions of CAR and P97, respectively.

Functional derivatives or fragments of these TrCytRs can also be used. Such derivatives are those that display at least 70%, preferably at least 85%, more preferably at least 95% and most preferably at least 99% of the transcytotic function of the native sequence (i.e. the ability to induce transcytosis of a macromolecule appropriately bound thereto). The skilled man will be aware of how to determine if a particular derivative or fragment retains transcytotic function. These functional derivatives of the TrCytRs will typically have at least 40%, preferably 50 or 60% or more, particularly 70 or 80% or more sequence homology with the native sequence (e.g. at least 90% or at least 95%). For the purposes of the present invention, and in accordance with common understanding in the art, "sequence homology" is not used to refer only to sequence identity but also to the use of amino acids that are interchangeable on the basis of similar physical characteristics such as charge and polarity. Substitution of an amino acid within an amino acid sequence with an amino acid from the same physical group is considered a conservative substitution and would not be expected to alter the activity of the peptide. Thus a derivative which just replaced leucine with isoleucine throughout would be considered to have 100% "sequence homology" with the starting sequence. Convenient groups are, glycine and alanine; serine; threonine, asparagine, glutamine and cysteine; lysine arginine and histidine; glutamic acid and aspartic acid; valine, leucine, isoleucine, methionine, phenylalanine, tryptophan and tyrosine. Preferred subgroups within this last group include leucine, valine and isoleucine; phenylalanine, tryptophan and tyrosine; methionine and leucine. Sequence homology may be calculated as for 'sequence identity¹ discussed below but allowing for conservative substitutions as discussed above. Preferably, the functional derivatives of the TrCytRs exhibit at least 50%, preferably at least 60% or 70%, e.g. at least 80% sequence identity to a naturally occurring TrCytR (as determined by, e.g. using the SWISS-PROT protein sequence databank using FASTA pep-cmp with a variable pamfactor, and gap creation penalty set at 12.0 and gap extension penalty set at 4.0, and a window of 2 amino acids). The skilled man would understand that these receptors might have regions that are very functionally tolerant of sequence diversion and others that are very intolerant (e.g. the ligand binding regions and the transmembrane regions). These change intolerant areas are typically those areas essential to maintaining the transcytotic function of the TrCytR. The above percentages are a total percentage based on the entire native TrCytR and not regions thereof. Accordingly, overall a derivative might have a low sequence identity or homology to the native TrCytR but still have most or all of the amino acid sequence of the change intolerant sequences and therefore displays similar transcytotic function to the native TrCytR.

One cellular structure that is involved in transcytosis are caveolae. Caveolae are flask-shaped pits in the plasma membrane of certain cells (e.g. endothelial cells, smooth muscle cells, and adipocytes) that are lined with caveolin (Tuma, et al., supra). Three members of the caveolin gene family (caveolin-1 , -2 and -3) and multiple isoforms of caveolin-1 have so far been identified (Cohen, A. W., et al., 2004, Physiol Rev, 84, 1341 ) Other proteins such as gp60 and aminopeptidase P are also associated with caveolae. Artificial TrCytRs can be also be generated and used in addition to or in place of naturally occurring TrCytRs such as those discussed above. An artificial TrCytR is a macromolecule that can function as a TrCytR but does not act as a TrCytR in nature. Artificial TrCytRs can be generated by modifying molecules that can be bound by the moiety to be transported across the epithelial cell layer (i.e. the macromolecule or the macromolecule aggregate) in such a way that the modified molecule can be located at the plasma membrane and is capable of interaction with one or more caveolae proteins in such a way that the binding of the moiety to be transported to the modified molecule induces transcytosis. For instance, the artificial TrCytR can be an Affibody™ molecule or a hyperstable scFv or a camelid antibody (or a functional fragment or derivative thereof) that is designed to bind the moiety (e.g. virus) to be transported that is displayed on a transmembrane scaffold that has a region capable of interaction with one of the caveolins described above (e.g. caveolin-1 , -2 and -3), gp60 or aminopeptidase. The artificial TrCytR could alternatively comprise a natural binding partner of the moiety to be transported. "Natural binding partner" encompasses the entirety of an entity that binds the moiety to be transported in nature as well as the relevant regions of that entity that participate in the interaction.

Preferably the artificial TrCytR comprises the Affibody™ molecule Zztaq.

Artificial TrCytRs can also be generated by creating fusion proteins comprising caveolae proteins and proteins that can be bound by the moiety to be transported across the epithelial cell layer (i.e. the macromolecule or the macromolecule aggregate) such that the moiety to be transported across the epithelial cell layer can interact with the appropriate part of the fusion protein and induce transcytosis. For instance, the fusion protein can comprise one of the caveolins described above, gp60 or aminopeptidase P (or a functional fragment or derivative thereof) and an Affibody™ molecule or a hyperstable scFv or a camelid antibody (or a functional fragment or derivative thereof) that is designed to bind the moiety (e.g. virus) to be transported. The fusion protein could alternatively comprise a natural binding partner (as defined above) of the moiety to be transported. Preferably the fusion protein consists of a fusion of a caveolin described above (e.g. caveolin-1 , -2 and -3) and the Affibody™ molecule Zztaq.

In a separate aspect the invention provides a fusion protein comprising the amino acid sequence of a caveolae protein or a functional derivative or fragment thereof and another protein sequence wherein said another protein sequence is capable of binding to the macromolecules or aggregates thereof that are to be transported across a targeted layer of epithelial cells. The above discussion of macromolecules and aggregates thereof applies mutatis mutandis, to this aspect of the invention. By functional derivative it is meant an amino acid sequence that displays at least 70%, preferably at least 85%, more preferably at least 95% and most preferably at least 99% of the transcytotic function of the native sequence. The skilled man will be aware of how to determine if a particular fragment or derivative retains function. These functional derivatives of the caveolae proteins will typically have at least 40%, preferably 50 or 60% or more, particularly 70 or 80% or more sequence homology with the native sequence (e.g. at least 90% or at least 95%). Preferably, the functional derivatives of the caveolae proteins exhibit at least 50%, preferably at least 60% or 70%, e.g. at least 80% sequence identity to the native sequence. "Sequence identity" and "sequence homology" are defined above.

Following the targeted display of a TrCytR on the epithelial cell surface the moieties to be transported across the epithelial cell layer are then applied. Each comprises at least one region which can bind said TrCytR and upon which transcytosis of said moiety will be induced. This region can be present as a part of a larger macromolecule, or itself be the macromolecule, or be one of, or a part of, the macromolecules in the aggregate. The region can be presented as a part of the macromolecule from which it is naturally derived although the region can be present in a macromolecule that is not the macromolecule from which it is normally derived. For instance, in the case of an adenovirus vector, the TrCytR binding region can be presented in the fibre knob. In the case of a VLP the TrCytR binding region can be presented as a part of the viral structural proteins that make up the VLP and/or as a part of a protein, or as a protein, embedded in the plasma membrane derived lipid bilayer of such a coated VLP. In a similar way, in the case of polymer coated viral vectors or liposome vectors the coat/bilayer can be studded with macromolecules carrying or consisting of the binding region. As there is less constraint on the size of the binding region in these embodiments, these embodiments can cope with the inclusion of the entire, or a substantial part of, the macromolecule from which the binding region is derived. Conveniently the entire macromolecule from which the binding region is derived is used. Preferably only those regions involved in the binding will be used and regions not involved will be substantially or entirely absent.

Binding partners for the above mentioned naturally occurring TrCytRs are IgA for plgR, melanotransferrin/P97 for MTfR/P97R, albumin for gp60, IgG for FcRn, transferrin for the transferrin receptor, LDL for the LDL receptor, insulin for the insulin receptor, vitamin B1 for megalin. Preferably IgA and melanotransferrin are used. Preferably the binding regions in the moieties to be transported will substantially or entirely correspond to the regions of these TrCytR binding partners that interact with their TrCytR and cause transcytosis. Conveniently, the binding partner or binding region thereof is an antibody which selectively binds the TrCytR and the binding of which induces transcytosis (Tang, Y., et al., supra). An Affibody™ molecule could be used similarly.

Functional derivatives or fragments of the binding partners and their binding regions can also be used. Such derivatives are those that display at least 70%, preferably at least 85%, more preferably at least 95% and most preferably at least 99% of the transcytotic function of the native sequence (i.e. the ability to bind to the corresponding TrCytR and induce transcytosis). The skilled man will be aware of how to determine if a particular fragment or derivative retains transcytotic function. These functional derivatives of the TrCytR binding partners and their binding regions will typically have at least 40%, preferably 50 or 60% or more, particularly 70 or 80% or more sequence homology with the native sequence (e.g. at least 90% or at least 95%). Preferably, the functional derivatives of the TrCytR binding partners and their binding regions exhibit at least 50%, preferably at least 60% or 70%, e.g. at least 80% sequence identity to the native sequence. "Sequence identity" and "sequence homology" are defined above.

As discussed above, in principle Affibody™ molecules can be generated which bind to any protein. Therefore, in another embodiment the TrCytR binding region is an Affibody™ molecule directed to the naturally occurring TrCytRs described above.

For artificial TrCytRs comprising the Affibody™ molecule Zztaq, the TrCytR binding region is Ztaq (Eklund, M., et al., 2002 Pro. Struc. Func. Gen., 48, 454).

The macromolecules or aggregates thereof transported across the target layer of epithelial cells as described above can be therapeutic products. By "therapeutic product" it is meant a product that is capable of eliciting a therapeutic effect or having a therapeutic benefit (e.g. the product has a diagnostic or an imaging application) in that tissue. "Therapeutic" is used broadly herein to include both therapy (in the sense of curative or palliative therapy of a pre-existing or diagnosed condition) and prophylaxis and diagnosis. Therefore, the method of the invention can be considered to be a method of medical treatment and/or diagnosis. For example, the therapeutic product could be an anticancer agent (e.g. a cytotoxic protein or a cytotoxic bioactive RNA molecule) and the method of the invention can be used accordingly in the treatment of cancer. In the field of diagnostics the therapeutic (diagnostic) product could be, inter alia, fluorescent, luminescent or radioactive (e.g. GFP, luciferase, or propidium iodide, FITC, Technetium- 99m or lodine-123 labelled macromolecules) or a nucleic acid encoding such fluorescent or luminescent proteins

In one embodiment the method of the invention is used to transfer therapeutic genes to tissue associated with an epithelial layer. By "therapeutic genes" it is meant a nucleic acid that is capable of being transcribed in the tissue and which is capable of eliciting a therapeutic effect or having a therapeutic benefit (e.g. the gene has a diagnostic or an imaging application) in that tissue.

Therapeutic genes can be heterologous to the nucleic acids transcribed in the tissue. For instance, this means that the therapeutic gene can be from a different species or can be a nucleic acid that is from the same species but which is not transcribed under normal conditions in the tissue into which it is introduced. It can also be a nucleic acid that is a non-mutated version of a nucleic acid that is transcribed in the tissue (e.g. a "rescue" gene). Alternatively the therapeutic gene is a nucleic acid transcribed naturally in the tissue but an increase in the level of that expression would be beneficial. Preferably the therapeutic gene encodes a protein or a bioactive RNA molecule (e.g. an antisense molecule, a ribozyme, siRNA and miRNA). In this aspect of the invention the macromolecule aggregates being transported across the epithelial layer (the therapeutic product) will typically be gene therapy vectors (e.g. the vectors described above) carrying the therapeutic genes described above.

As shown in Figures 1 and 2 the type of vector used to selectively display the TrCytR can be the same as or different to the vector used to transfer the therapeutic gene. For instance, in Figure 1 a VLP is used to selectively display the TrCytR and an adenoviral vector is used to transfer the therapeutic gene whereas, in Figure 2, an adenoviral vector is used to both selectively display the TrCytR and an adenoviral vector is used to transfer the therapeutic gene. As discussed above, targeting strategies can be shared by the two steps and the vectors used therein.

In a particularly preferred embodiment the invention provides a method for the targeted delivery of an adenoviral vector to tissue associated with tumour endothelium comprising:

(a) applying to the tumour endothelium a first adenoviral vector wherein said first adenoviral vector is uncoated or coated and is targeted to the tumour endothelium and encodes or comprises a transcytosis receptor (TrCytR); and

(b) applying a second adenoviral vector to the tumour endothelium, wherein said second adenoviral vector is uncoated or coated and comprises at least one region which upon binding to said TrCytR induces transcytosis of said second adenoviral vector to the associated tissue.

The preferred embodiments of the adenoviral vectors and TrCytRs of the invention described above apply to this embodiment also.

A further layer of control can be introduced to this system by targeting the vector carrying the therapeutic gene to the tissue of interest associated with the epithelial layer. For instance, the targeting strategies described above can be used. Preferably, the vector carrying the therapeutic gene is an adenoviral vector that has been modified to express a targeting molecule on its surface that is selective for a binding partner expressed in the tissue of interest. For example, the target tissues could be breast or ovarian tumour tissue. HER2/neu is a member of the EGF-R receptor family that is over- expressed on certain breast- and ovarian cancers and the Affibody™ molecule ZH binds to HER2/neu (Wikman, M., et al., 2004, Protein Eng Des Se1 17(5), 455-462). This molecule has been successfully genetically engineered into the Hl-loop of a CAR ablated fibre knob and been shown to successfully target adenoviral vectors to these tumours. In a preferred embodiment therefore, the macromolecule aggregate is a gene therapy vector that is in addition targeted to HER2/neu by the display of Affibody™ ZH on its surface. In another example the target tissue could be prostate tumour tissue. PSMA is expressed in prostate tumour tissue and so this receptor could therefore be targeted in this instance.

The method of the invention can use any combination of the above discussed techniques to achieve selective display of the TrCytR in combination with any of the above discussed macromolecules or aggregates thereof. As shown in Figure 1 the selective display of the TrCytR is achieved with a fusiogenic VLP carrying the TrCytR in its lipid bilayer targeted to the target epithelial cell layer. In a second step the macromolecule aggregate to be transported is an adenoviral vector with a polymer coat comprising an antibody to the TrCytR and a target tumour cell binding protein. As shown in Figure 2, adenoviral vectors can be used in both steps. In the second step the virus is functionally the same as the vector used in Figure 1 , although an Affibody™ molecule is used to bind the TrCytR. In the first step however selective display of the TrCytR is achieved by infecting the targeted epithelial cells with an adenoviral vector which is targeted to the targeted epithelial cells and which encodes the TrCytR.

Typical examples of therapeutic genes would be coding sequences for growth factor receptors, oncogenes, tumour suppressor genes, antisense oncogenes, suicide genes, genes for immune modulatory substances, genes for tumour antigens, genes for anti-angiogenic factors, cytokines, genes for vascular endothelial growth inhibitors, genes for fusiogenic membrane glycoproteins, cytotoxic genes or genes encoding enzymes which convert a pro-drug to cytotoxic substance. In particular, the therapeutic gene can be a coding sequence for genes that are mutated in monogenetic diseases (e.g. cystic fibrosis). Most preferably the vector comprising the therapeutic gene will be targeting ^' tumour cells and will be designed to kill those cells, thus resulting in the destruction of the tumour, a reduction in the size of the tumour or at least a deceleration or interruption in its growth.

Viral vectors can kill tumour cells directly by viral oncolysis and/or by expression of a cytotoxic therapeutic gene. Replication incompetent vectors typically encode therapeutic genes for tumour cell killing rather than engaging in oncolysis. There are several types of therapeutic genes available in cancer gene therapy (cancer therapeutic genes) such as prodrug activating enzymes, immune system modulators, apoptosis inducers and toxins. A mixture of cancer therapeutic genes for molecules with different activities is particularly preferred.

Prodrug-activation gene therapy (also known as suicide gene therapy) delivers genes to the cancer cells, enabling them to convert non-toxic prodrugs into active chemotherapeutic agents. Drug activation occurs primarily in the cancer cells, thereby maximizing damage to the cancer cells, while keeping the systemic toxicity low. Two of the most frequently studied prodrug strategies is based on transduction with herpes simplex virus type 1 thymidine kinase followed by ganciclovir administration and cytosine deaminase followed by the prodrug 5-fluorocytosine (Yaghoubi, S. S., et al., 2006, Nat Protoc 1 , 3069; van Dillen, I. J., et al., 2002, Curr Gene Ther 2, 307).

Certain transduced tumour cells also appear to be capable of inducing the death of neighbouring untransfected cells. This cell kill is called the "bystander effect". Melanoma differentiation associated gene-7/interleukin- 24 (mda-7/IL-24) is a cancer selective apoptosis inducing and immunomodulatory cytokine with potent bystander antitumour activity with potential for eradicating not only primary but also metastatic disease (Sarkar, D., et al. 2007, Expert Opin Biol Ther 7(5): 577-86). Therefore, non limiting examples of suitable cancer therapeutic genes for use in the invention are cytolethal distending toxin B (cdtB), yeast cytosine deaminase (yCD), uridine phophoribosyl transferase (UPRT, Drosophila melanogaster deoxynucleoside kinase gene (Dm-dNK/B5), cytosine deaminase, thymidine kinase (e.g. HSV1-tk), CD40 ligand, granulocyte- macrophage colony-stimulating factor (GM-CSF), the interleukins IL-2, IL-2 and IL-18, and mda-7/IL-24.

To control the expression of toxic products the therapeutic genes can be put under the control of inducible promoters such as the tetracycline-responsive system.

In another aspect there is provided a method for gene therapy, comprising (a) selectively displaying a TrCytR on the surface of a layer of epithelial cells associated with the tissue to undergo gene therapy; and

(b) applying a gene therapy vector comprising a therapeutic gene to the epithelial cells associated with the tissue wherein said gene therapy vector comprises at least one region which induces transcytosis of said gene therapy vector upon binding to said TrCytR.

Preferably the gene therapy vector is a viral vector, more preferably an adenoviral vector, e.g. Ad 5 and the therapeutic gene is one or more of cytosine deaminase, mda-7/IL-24, cdtB, yCD, yCD/UPRT, Dm-dNK/B5, HSV1-tk, CD40 ligand, GM-CSF, IL-2 IL-12, IL-18 and CFTR. The selective display of the TrCytR can be achieved in any of the above discussed ways.

In another aspect there is provided a method for the treatment of cancer, comprising (a) selectively displaying a TrCytR on the surface of a layer of epithelial cells associated with a tumour; and (b) applying cancer therapeutic macromolecules or aggregates thereof to the epithelial cells associated with the tumour wherein said macromolecules or aggregates thereof comprise at least one region which induces transcytosis of said macromolecules or aggregates thereof upon binding to said TrCytR.

The discussion of macromolecules and aggregates thereof provided above applies mutatis mutandis to this aspect of the invention. Preferably the cancer therapeutic macromolecule aggregate is a vector as discussed above (e.g. a gene therapy vector). Most preferably the gene therapy vector is a viral vector, more preferably an adenoviral vector, e.g. Ad 5 and the vector carries one or more therapeutic genes selected from cytosine deaminase, mda-7/IL-24, cdtB, yCD, yCD/UPRT, Dm-dNK/B5, HSV1-tk, CD40 ligand, GM-CSF, IL-2 IL-12 and IL-18. The selective display of the TrCytR can be achieved in any of the above discussed ways.

Further selectivity can be imparted by the transcriptional targeting of the vectors carrying the therapeutic gene to the target tissue as discussed above. If the target tissue is breast or ovarian tumour tissue non-limiting examples of promoters that could be used to target the vector to such tissues are the v-erb-b2 erythroblastic leukemia viral oncogene homolog 2 (ERBB2) promoter (Hung, M. C₁ et al., 1995, Gene 159, 65), CXCR4 (also known as fusin; Stoff-Khalili, M. A., et al., 2005, Breast Cancer Res 7, R1141 ), human small breast epithelial mucin promoter (Hube, F., et al., 2004, DNA Cell Biol 23, 8420) and cyclooxygenase 2 (COX2;.Bauerschmitz, G. J et al., 2006, MoI Ther, 14(2):164).

When using viral vectors (e.g. in either of the steps of the method of the invention) it may be preferable to detarget the vectors from their cellular receptors to increase the survival time of the vector in the circulation. This can, as retargeting, be achieved by providing the mature vector with a coating or by genetic engineering of the vector proteins. Preferably the vectors used in the various stages of the method of the invention are adenovirus vectors which have been detargeted by substituting the amino acids in the knob involved in CAR binding and/or the integrin binding RGD peptide of the penton base is ablated by amino acid substitutions. (Smith, T. A., et al., 2003, Hum Gene Ther 14, 777; Smith, T. A., et al., 2003, Hum Gene Ther 14, 1595). Detargeting by providing the mature vector with a coating can be achieved using the polymer coatings described above.

Coating of the viral vector as described above also serves the purpose of reducing interaction between the viral vector and circulating antibodies or other circulating proteins. A similar effect can be achieved by changing the viral proteins to alter the naturally immunogenic epitopes such that their jmmunogenecity is reduced. In the case of adenoviral vectors these epitopes are located mainly on the fibre knob and on the hexons. Preferably the adenoviral vectors of use in the invention will comprise a shielding protein in the virus capsid to prevent exposure of the immunogenic epitopes to the immune system of the host. Conveniently the protein shield would be anchored by the minor capsid protein plX that sits between the hexons in the capsid (Hedley, S. J., et al., 2006, Cancer Immunol lmmunother 55(11 ), 1412). An alternative approach is to incorporate a serum protein binding domain (Johansson, M. U., et al., 2002, J Biol Chem 277(10), 8114) into plX. When the virus is released after replication or upon delivery to the subject the domain would bind the serum protein of the host and that would then shield antibody binding epitopes on the hexons. preferably the serum protein is one or more of albumin, alpha-1 -globulin, alpha-2-globulin or β- globulin. Most preferably the serum protein is albumin and the binding domain is ABD from the streptococcal protein G. Such sequences can be inserted in the C-terminus of plX. In another preferred embodiment the vector is a polymer coated vector and the polymer coat comprises the serum protein binding domains described above thereby providing a proteinaceous coating in addition to the polymer coating. In another aspect the invention provides a vector that displays, encodes or upregulates the expression of a TrCytR and is capable of selectively introducing said TrCytR to the surface of a target epithelial cell layer or selectively increasing the display of said TrCytR on the surface of a target epithelial cell layer for use in the above described methods.

In a separate aspect the invention provides a virus-like particle (VLP) having a plasma membrane-derived lipid bilayer envelope, said VLP further comprising: a) a viral structural protein, or fragment or derivative thereof, capable of forming an enveloped VLP; b) a fusiogenic protein; and c) a TrCytR.

A "viral structural protein" is a protein that contributes to the overall structure of the capsid protein or the protein core of a virus. The viral structural protein can be obtained from any virus which can form enveloped VLPs. These are typically proteins from viruses that are naturally enveloped. Such viruses include, but are not limited to, the Retroviridae (e.g. HIV, Moloney Murine Leukaemia Virus, Feline Leukaemia Virus, Rous Sarcoma Virus), the Coronaviridae, the Herpesviridae, the Hepadnaviridae, and the Orthomyxoviridae (e.g. Influenza Virus). However, naturally non-enveloped viruses may form enveloped VLPs and these are also encompassed by the invention. Naturally non-enveloped viruses include the Picornaviridae, the Reoviridae, the Adenoviridae, the Papillomaviridae and the Parvoviridae.

Preferred structural proteins are the Retroviridae Gag proteins. Particularly preferred as the structural protein is the protein corresponding to the HIV-1 gag gene. The gag gene of the lentivirus HIV-1 codes for the polyprotein Pr55Gag which is a precursor of the structural proteins p17 matrix (MA), p24 capsid (CA), p7 nucleocapsid (NC) and p6. Gag is cleaved into the individual proteins in mature, infectious virions of HIV-1 , however, in Gag VLPs Gag remains as a single protein since the required viral protease is absent. The mechanisms underlying and proteins involved in Gag VLP formation are extensively discussed in the art. As in the case of HIV1 Gag, encompassed by the term "structural protein" are pro-structural proteins wherein the structural protein is produced upon post translation cleavage of a pro-protein or structural polyproteins wherein multiple structural proteins are derived from a single polypeptide. These proteins may not need to be cleaved to be able to form a VLP

Fragments and derivatives of the structural proteins that retain the ability to form VLPs are encompassed by the invention. The skilled man will be aware of how to determine if a particular fragment or derivative retains the ability to form VLPs. Carriere et al., 1995 J. Virol. 69:2366-2377 and WiIk et al., 2001 J. Virol. 75:759-77130 and references cited provide direction as to the identification of regions and fragments of Gag that retain the ability to form VLPs. Such techniques can be readily applied to other viral structural proteins. These derivatives of naturally occurring sequences will typically have at least 40%, preferably 50 or 60% or more, particularly 70 or 80% or more sequence homology (e.g. at least 90% or at least 95%) with the native sequence. Preferably, the derivatives of naturally occurring virus structural proteins or active fragments thereof exhibit at least 50%, preferably at least 60% or 70%, e.g. at least 80% (e.g. at least 90% or at least 95%) sequence identity to a naturally occurring structural protein or portion thereof. "Sequence identity" and "sequence homology" are defined above. Naturally occurring structural proteins, or fragments or derivatives thereof, may be provided as a fusion protein with one or more domains of structural proteins belonging to different species, subgroups families or subfamilies of viruses (e.g. Lentivirus and spumavirus; see Carriere et al., supra), or with non-viral protein sequences.

Preferably the VLP further comprises a targeting molecule, preferably the targeting molecule is selective for a target epithelial cell. In particular the targeting molecule will be selective for the endothelium of a tumour. In another aspect the invention provides a method for the production of a VLP as defined above, said method comprising the coexpression of a viral structural protein, or fragment or derivative thereof, capable of forming an enveloped VLP, together with a fusiogenic protein, together with a TrCytR in an in vitro cultured cell and isolating the VLP from the culture media. The host cell can be any cell, preferably eukaryotic, and more preferably mammalian. Most preferably the source of the cell will be the same or compatible with the cell to which the VLP are designed to fuse with. Preferably the cell is in a stable cell culture.

In some cases, nucleic acids encoding (i) a viral structural protein, or a fragment or derivative thereof, capable of forming an enveloped VLP, (ii) a fusiogenic protein, and/or (iii) a TrCytR can be stably integrated individually, pairwise or together into the genome of a cell leading to a stable cell line capable of continuous growth in vitro. Such a cell line will preferably express constitutively the VLP or constituents thereof.

In another aspect the invention provides an in vitro host cell line comprising a) a nucleic acid encoding a viral structural protein, or fragment or derivative thereof, capable of forming an enveloped VLP; b) a nucleic acid encoding a fusiogenic protein; and c) a nucleic acid encoding a TrCytR.

Any and all preferred embodiments for the VLPs used in the method of the invention as described above apply to these aspects also.

In a separate aspect the invention provides a viral vector encoding or comprising a TrCytR.

Preferably the viral vector further comprises a targeting molecule, preferably the targeting molecule is selective for a target epithelial cell. In particular, the targeting molecule will be selective for the endothelium of a tumour. Any and all preferred embodiments for the VLPs used in the method of the invention as described above apply to this aspect also.

In another aspect the invention provides a method for the production of a viral vector as defined above, said method comprising expressing the components of the viral vectors in an in vitro cultured cell and isolating the viral vector from the culture media. The host cell can be any cell, preferably eukaryotic, and more preferably mammalian. Most preferably the source of the cell will be the same or compatible with the cell to which the viral vectors are designed to infect. Preferably the cell is in a stable cell culture.

In a further aspect the invention provides the VLPs or the viral vectors described above for use in therapy, preferably gene therapy, in particular cancer therapy. In other words, the VLPs or the viral vectors described above can be used to eliminate tumours, reduce the size of tumours or at least decelerate or interrupt their growth.

The mode of administration of the vectors or the moieties to be transported in the methods described above will vary depending on the physiological situation and/or the disease being treated since, for instance, different diseases will require administration of the vectors or the moieties to be transported at different sites in the body. Conveniently the vectors or the moieties to be transported will be administered systemically, preferably via the circulatory system, preferably intravenously. Typically the vectors or the moieties to be transported will be administered in a pharmaceutically acceptable composition.

By "pharmaceutically acceptable" is meant that the ingredients must be compatible with other ingredients of the composition as well as physiologically acceptable to the recipient. The pharmaceutical compositions may be formulated according to any of the conventional methods known in the art and widely described in the literature. Thus, the active ingredient may be incorporated, optionally together with other active substances, with one or more conventional carriers, diluents and/or excipients, to produce conventional galenic preparations such as tablets, pills, powders, lozenges, sachets, cachets, elixirs, suspensions, emulsions, solutions, syrups, aerosols (as a solid or in a liquid medium), ointments, soft and hard gelatin capsules, suppositories, sterile injectable solutions sterile packaged powders, and the like. Preferably the composition is adapted for administration by injection or aerosol.

Examples of suitable carriers, excipients, and diluents are lactose, dextrose, sucrose, sorbitol, mannitol, starches, gum acacia, calcium phosphate, alginates, tragacanth, gelatin, calcium silicate, microcrystalline cellulose, polyvinylpyrrolidone, cellulose, water syrup, water, water/ethanol, water/ glycol, water/polyethylene, glycol, propylene glycol, methyl cellulose, methylhydroxybenzoates, propyl hydroxybenzoates, talc, magnesium stearate, mineral oil or fatty substances such as hard fat or suitable mixtures thereof. The compositions may additionally include lubricating agents, wetting agents, emulsifying agents, suspending agents, preserving agents, sweetening agents, flavouring agents, and the like. The compositions of the invention may be formulated so as to provide quick, sustained or delayed release of the active ingredient after administration to the patient by employing procedures well known in the art.

The present invention therefore also provides a pharmaceutical composition comprising a VLP or a viral vector as defined above together with at least one pharmaceutically acceptable carrier, diluent or excipient. The VLP or the viral vector in such compositions may comprise from 0.05% to 99% by weight of the formulation, more preferably 0.05% to 50%, 0.1% to 20% or 0.1 % to 10%. In a further aspect the present invention provides a product containing (a) a vector that displays, encodes or upregulates the expression of a TrCytR and is capable of selectively introducing said TrCytR to the surface of a target epithelial cell layer or selectively increasing the display of said TrCytR on the surface of a target epithelial cell layer (b) a therapeutic (including diagnostic) macromolecule or aggregate thereof wherein said macromolecule or aggregate thereof comprises at least one region which is capable of inducing transcytosis of said macromolecules or aggregates thereof upon binding to said TrCytR when said TrCytR is present on the surface of the target epithelial cell layer as a combined preparation for simultaneous, separate or sequential use, preferably in therapy. In one embodiment the therapy is the treatment of cancer. In another embodiment the therapy is gene therapy. Preferably the therapy is the treatment of cancer by gene therapy. Preferably the methods of the invention described above are used.

Components (a) and (b) can be any of the various aspects of the invention and the preferred embodiments thereof that have been described in detail above. For example, component (a) can be a fusiogenic VLP that directly introduces the TrCytR to the target cells. Also for example, component (a) can be a baculovirus vector or an adenovirus vector that directly introduces the TrCytR to the target cells, encodes the TrCytR and/or upregulates the expression of the TrCytR. Preferably these vectors are targeted to the target epithelial cell layer as described above. For example, component (b) is an adenoviral vector carrying a therapeutic gene as described above.

Where two or more of components (a) and/or components (b) are administered, they may be given simultaneously to the patient or times of administration may be staggered throughout the day or treatment cycle.

The, vectors or the moieties to be transported (or compositions) may also be administered to cells ex vivo prior to implantation or re-implantation. In vitro applications such as delivery of macromolecules to target tissues during tissue and organ culture studies are also contemplated.

The invention will be further described with reference to the following non- limiting Examples in which:

Figure 1 shows an embodiment of the invention in which an adenovirus vector is delivered to Her2/neu(+) cells from the circulation. A: Cell attachment of VLP displaying a TrCytR, a fusiogenic protein (Gp64) and a tumour endothelium targeting molecule (DII4). B: Fusion of VLP with endothelial cell membrane. C: Transfer and display of TrCytR on the surface of the endothelial cell. D: Binding of an adenoviral vector to the TrCytR transferred by the VLP in step C via an antibody displayed on the vector (Ab/TrCytR). D: Endocytosis of vector induced by interaction of antibody and TrCytR. E: Exit of the vector at the basal pole. F: Exposure of vector to Her2/neu(+) cells underlying the endothelial cells and delivery of therapeutic gene. Targeting to Her2/neu(+) cells by ZH Affibody™ displayed on vector.

Figure 2 shows another embodiment of the invention in which an adenovirus vector is delivered to Her2/neu(+) cells from the circulation. A: Cell attachment of adenoviral vector encoding a TrCytR (Caveolin/Zztaq fusion protein) and displaying a tumour endothelium targeting molecule. B: Expression and membrane incorporation of Caveolin/Zztaq. C: Binding of adenoviral vector to the Caveolin/Zztaq via Ztaq Affibody™ displayed on the vector. D: Endocytosis of vector induced by interaction of Ztaq and

Caveolin/Zztaq. E: Exit of the vector at the basal pole. F: Exposure of vector to Her2/neu(+) cells underlying the endothelial cells and delivery of therapeutic gene. Targeting to Her2/neu(+) cells by ZH Affibody™ displayed on vector. Examples

Example 1

1.1. Coating of the Ad/HI-ϋnk:ZH2 virus with PEG

Heterobifunctional PEGs are used for the ligand-coupling approaches. Purified virus is dialysed into a coupling buffer and used for PEGylation by standard methods where PEG is slowly added to the virus under agitation and reacted for 3h at room temperature. Size exclusion chromatography or CsCI gradient centrifugation is used to purify the PEG-Ad after the coupling reaction and SDS-PAGE is used to determine PEGylation percentage. The size of PEG and its molar excess in the coupling reaction is evaluated in order to efficiently shield the virus from neutralizing antibodies. Standard infectivity assays on HER2/neu expressing cells are used to study detargeting and dot blot is used to determine loss of binding to neutralizing antibodies. The virus should be completely shielded.

1.2. Introduction of the ZH Affibodv™ into the PEG coat (ZH-PEG)

In order to retarget the Ad/HI-Link:ZH2/PEG virus a ligand conjugation step is performed between the PEGylation and purification steps described in 1.1. After incubation with PEG a PD10 column is used to transfer the virus into an appropriate buffer for conjugation to purified ZH Affibody™ molecule after which purification will take place. The percentage of PEG, detargeting and shielding is analyzed as in 1.1 and ZH incorporation quantified by SDS- PAGE.

1.3. In vitro studies of the transductional ability of ZH-PEG coated virus This is done using HER2/neu expressing cells and read-out of transduction frequency by FACS. The infection studies are made with and without neutralizing antibodies to determine the shielding properties. It is determined if the RGD binding motif is needed for infection. The Ad/HI-Link:ZH2/ZH- PEG virus is evaluated for in vivo tumour targeting and blood circulation time in mice with human xenografts subcutaneously (sc). An animal model with the ovarian cancer cell line SKOV3 has already been established with sc tumours in the flank. The virus (10¹¹ physical particles) is injected i.v. and tumour targeting measured by quantitative PCR (Q-PCR) or live PET scan. The calculation of blood circulation times, is performed by injecting virus and collecting blood samples at different time points. The blood is then subjected to Q-PCR and titration.

1.4. Introduction of other proteins into the PEG coat Introduction of an antibody against TrCytR (aTrCytR) into the PEG coat to create aTrCytR-PEG is as described above for ZH-PEG. Introduction of the peptide -CNGRC- and an Affibody™ molecule with specificity for PSMA into the PEG coat to create CNGRC-PEG and PSMA-PEG is as described above for ZH-PEG.

1.5. Creation of Ad/HI-Link:ZH2 virus simultaneously coupled with ZH-PEG and aTrCvtR-PEG.

PEG coated virus are produced as above with different ratios of ZH-PEG and aTrCytR-PEG in order to obtain the best result. Transductional ability of these virus is assessed as in 1.3. Testing of the ability of ZH-PEG and aTrCytR-PEG coated virus to be transcytosed over neoendothelium is performed using the models described in Example 4.

1.6. Targeting of PEG coated virus to neoendothelium or neoendothelium markers

PEG coated virus with PSMA antibody or neoendothelium-seeking peptide CNGRC in the PEG coat is produced as described above. Performance is tested in vitro and in vivo, (i) In vitro 293 cells expressing membrane bound PSMA as targets are used and infection of the PSMA PEG virus is detected by FACS analysis, (ii) In vivo mice carrying the murine B16 melanoma or the murine colorectal carcinoma CT26 are used. Infection of the CNGRC virus is detected by quantitative PCR of DNA samples purified from tumours or by live PET scan.

The viruses described above contain the green fluorescent protein gene (GFP) for convenient detection of infection by FACS analysis or HSV1- sr39tk for PET scan.

Example 2

Production of Tumour Targeted Replication Competent Adenovirus Particles (TRAP) with E1 A under the tumour specific HER2/neu (erbB2) promoter.

2.1. Construction of erbB2.ZZ. CRAd

CRAds expressing the transgene Firefly Luciferase (Luc), with the HER2/neu capsid modification are constructed using standard molecular cloning and virology techniques. Virus construction is done using the "AdEasy system" that employs recombination between Ad5 genome containing plasmid and a shuttle plasmid containing erbB2.E1A with flanking sequences homologous to the relevant viral genome region.

To generate erbB2.ZZ.CRAd the following strategy is followed:

To insulate the promoter against interference from Ad5 vector elements, bovine growth hormone gene poly A (~280bp) sequence is cloned at 5' end of the erbB2 promoter (-500/+40) + Mud enhancer (-598/-485) (kind donation from Prof Ian Mcneish, UK). The final construct is then be cloned in into pShuttle (pSc, Stratagene) containing E1A elements from Adenovirus in 5' position to E1A (pSc.BGH.MUC1/erbB2.E1A).

The resulting construct, pSc.BGH.MUC1/erbB2.E1A is recombined with the plasmid containing Adenoviral genome with fibre modified to incorporate HER2/neu Affibody™ (ZH) (Ad/HI-Link:ZH2) in recombination competent bacteria to generate perbB2.ZZ.CRAd.

The resulting recombined plasmid, perbB2.ZZ.CRAd, on transfection into the 293 cells expressing HER2/neu receptor (293/HER2) rescues the virus, erbB2.ZZ.CRAd.

Luciferase gene (from pGL3 plasmid, Promega) is incorporated into the E3 region using the same methods to construct erbB2.ZZ.CRAd.Luc. In brief, shuttle plasmid containing "CMV.Luc" cassette flanked by sequences homologous to Ad sequences flanking E3 region is constructed and then recombined with perbB2.E1A.ZZ.CRAd. This generates perbB2.ZZ.CRAd.Luc, which can then be rescued in 293/HER2 cells to generate erbB2.ZZ CRAd.Luc.

As a control non tumour-specific wild type Ad5 with HER2/neu modification

(WT.ZZ.AdLuc) is produced in a similar fashion.

2.1.1. In vitro analysis of gene expression and tumour-specificity of erbB2.ZZ.CRAd

This is done in breast and (ovarian cancer) OC cell lines with variable HER2/neu expression. Luciferase expression in the infected cell lysates is measured by standard enzyme conversion assays (Luminometer). Replication potential of CRAds in cell lines is assessed by titration of lysates from virus-infected OC cells on HEK293A cells (propagation cell line for Ad5s) and HEK293HER2 cells (expressing HER2/neu) using standard limiting dilution plaque assays. By comparing infectivity of tumour-specific, erbB2.ZZ.CRAd.Luc & non-targeted WT.ZZ.AdLuc in OC and non-OC cell lines specificity is measured. Cell lines: Human OC cell lines representing different histological subtypes, growth rates and chemosensitivity to conventional cytostatic agents are used: Epithelial adenocarcinoma lines; A2780, OVCAR-3 and SKOV3, Fibroblastic; clear cell carcinoma, ES2. Human breast cancer cell lines representing different phenotypes: epithelial adenocarcinoma; MCF7, MCF- 7M, ductal adenocarcinoma; MDA-MB-134 and ductal carcinoma; T47D-7.

2.1.2. Evaluation of erbB2.ZZ.CRAds for efficacy and specificity in targeting human OC and breast cancer in vivo.

The efficacy and specificity of the viruses are also evaluated in xenogenic models of human breast and ovarian cancer. An example of orthotopic human ovarian cancer model is given below for reference. The strategy for testing will be similar in all models.

As OC usually remains confined to the abdominal cavity throughout its course, an orthotopic intra peritoneal (ip) tumour model of epithelial serous OC (>50% incidence) is used.

Orthotopic mouse model of peritoneally disseminated ovarian cancer (SK0V3/ip): 6-8 week athymic nu/nu female mice are implanted intraperitoneally (ip) with 10⁷ SKOV3 cells and tumours develop in the ip cavity by 12-14 days. The mice will also be injected with 10⁷ cells sc at the same time as the ip implantation depending on the aims of the experiment.

Evaluation of specificity (Tumour on/Liver off phenotype) of erbB2.ZZ.CRAd in SKOV3/ip mice: Initially, the specificity and optimal dose of viruses for in vivo study is established. SKOV3/ip mice (n=4) are given tumour specific erbB2.ZZ.CRAd.Luc or non specific WT. Ad Luc ip at 10⁸, 10⁹ and 10¹⁰ viral particles on day 12. Mice are imaged using the IVIS Lumina machine for Luciferase bioluminescence before necropsy. Tumours and other organs (lungs, liver, kidney, spleen and lymph nodes) are harvested 2, 7, and 10 days pi. (36 mice) for the following analyses: a.1 : Frozen tissue lysates for Luc expression, a.2: Frozen tissue lysates for viral replication by real time PCR. a.3: Formalin fixed tumours/organs; histological analysis for treatment related toxicity, a.4: Treatment-related toxicity in mouse sera for markers of liver and kidney function. Efficacy of erbB2.CRAd.Luc in SKOV3/ip mice: Mice (n=16) are injected ip with a pre-determined dose of erbB2CRAd.Luc or WTAd. Luc (day 12). The following analyses are carried out: b1 : Evaluation of effects of oncolytic virus on tumour growth: Power calculations show that for a 40% difference at a significance of p<0.05, given a 100% take rate, 10 mice/group are required. Regression of the tumour in response to treatment indicates efficacy. At necropsy (day 30), (a) tumours are measured, weighed and stored for histological analysis, (b) immunohistochemical analysis for proliferation (Ki-67+) and apoptosis

(Tunel+) will be done as described (X. Y. Wang et al. Gene Therapy, 11 , 1559).

2.2. Bioselection of the CRAd. and characterization of the phenotvpe and genotype.

Mutations are generated by chemical mutagenesis or alternatively using the error-prone variants of the adenovirus DNA polymerase. Such mutants have been characterized and shown to facilitate the bioselection of CRAds with improved replication capacity in tumour cells. Adenoviruses have not been evolved for efficient replication in tumour cells and variants with enhanced replication capacity in tumour cells have been isolated after chemical mutagenesis and bioselection in cultured tumour cells. However, chemical mutagenesis will only yield mutants of which the new phenotype is the result of single mutations, rather than more complex compound mutants. To facilitate the isolation of more complex mutants a system in which the adenovirus polymerase is expressed in the cell line is used for bioselection.

2.2.1. Generation of mutant pol and wt pol expressing HER2/neu positive tumour cell lines to be used in bioselection By using mutant polymerases with alterations in the regions involved in the proof-reading activity, polymerase mutants have been generated that have an error-prone phenotype. Lentivirus vectors are generated for the expression of these mutants in tumour cells. These pol mutant tumour cell lines are infected with a CRAd from which the pol gene had been deleted, and as a result replication of the virus is strictly dependent on the mutant polymerase and thereby progeny viruses which carry multiple point mutations are produced.

2.2.2. Bioselection of mutant CRAd

Serial propagation of the pol-deleted CRAd in the mutant-pol expressing tumour cell line is performed. This yields compound mutant viruses that replicate more efficiently in these cells.

2.2.3. Characterization of the bioselected viruses in tumour cell lines (burst size & replication kinetics)

The compound mutant viruses are then stably propagated on tumour cell lines that express the wild-type Ad polymerase. In a suitable virus the wt r polymerase gene is re-introduced to generate a clinically applicable CRAd.

2.2.4. Characterization of the bioselected viruses by 'solexa sequencing' Subsequently viral DNA from individual mutants is isolated and sequenced by 'Solexa sequencing'. This acquires sequence information of the entire genome in a single run.

2.3. Bioevaluation in laboratory animals

In vivo studies are based on methods described in 2.1.2 above but with the bioselected CRAd.

Example 3

3.1. Incorporation of the albumin-binding domain (ABD) from streptococcal protein G into the C-terminus of the minor capsid protein IX.

A sequence coding for the albumin binding domain has been genetically linked to the codons for the adenovirus minor capsid protein IX, via a region coding for an alpha-helical spacer (Vellinga, J., 2004, supra). This construct is evaluated by expression in helper cells using a lentivirus vector system for expression of plX variants. The plX-ABD gene is introduced into the Ad/HI- Link:ZH2 virus. Albumin binding is determined by assays involving binding of iodinated human serum albumin. The ability to bind HSA and to neutralize serum antibodies is further optimized by varying the length of the spacer used to connect the C-terminus of plX and the ABD.

3.2. Comparison of vectors self-coated with human serum albumin using optimized spacer length versus vectors coated with PEG

This is assayed in vitro using defined anti-adenovirus sera, as well as in vivo in pre-immunized mice compared to non-immunized recipients examining virus plasma-circulation time and biodistribution using bioluminescence imaging. The optimal constructs are evaluated and inserted in the backbone of adenovirus vectors. The effects on virus propagation kinetics and virus yields are then determined. A production system in which the use of human and bovine serum albumins is omitted is used.

3.3. Comparison of virus shielded in different ways In addition to virus with ABD on the plX molecule, virus is constructed carrying ABD-PEG in varying amounts (30-5000 molecules per particle). ABD-PEG coated particles are compared to PEG-coated virus and to virus with ABD-plX for ability to evade preformed antibodies as described in 3.2.

Example 4a

4a.1. Production of VLP carrying transcvtosis receptor (TrCvtR) Recombinant baculovirus AcMNPV expressing Pr55GagHIV has been used and described in detail in previous studies. The AcMNPV-Pr55GagHIV clone produces vast amounts of VLP devoid of any retroviral genome (Boulanger, P, et al., 1996, supra). Coinfection of insect cells with AcMNPV-PHPr- TrCytR which carries the gene for TrCytR under the control of the polyhedrin promoter (PHPr) and AcMNPV-Pr55GagHIV results in the budding and extracellular release of VLPs carrying TrCytR inserted in their membrane envelopes, as well as the fusiogenic baculoviral envelope glycoprotein Gp64. In contact with naive, recipient cells, the VLP carrying Gp64 and TrCytR (VLP-TrCytR) fuses with the cell plasma membrane and delivers the TrCytR receptor to this compartment.

4a.2. Large-scale production and purification of VLP-TrCvtR

Sf9 cells in mass culture are infected by recombinant AcMNPV at MO1 10 for 48 h. Extracellular particles budding and released into the culture medium are purified by a two-step procedure comprising (i) step gradient centrifugation through a 20 % sucrose cushion, and (ii) isopycnic ultracentrifugation through sucrose-D₂0 density gradient. VLP (mean density, d = 1.15-1.20) can be separated under these conditions. As much as 10¹² VLP, are usually recovered per 10⁶ cells.

4a. 3. Engineering neoendothelium-targeted VLP carrying transcvtosis receptor

To target VLP-TrCytR specifically to the endothelial cells of tumour microvasculature, VLPs-TrCytR are biotinylated in vitro, and coupled to biotinylated DII4 Mab (aDII4) via an avidin bridge. This gives rise to VLP- TrCytR-AV-aDII4. The nonspecific fusion performed by the viral Gp64 will therefore be replaced by a specific retargeting to endothelial cell surface molecules. As an alternative to aDII4 antibodies to PSMA (commercially available, e.g., the anti-PSMA antibody J591 ) are used to give rise to VLP- TrCytR-AV-aPSMA. Alternatively, an Affibody™ molecule against PSMA is used.

4a.4. Testing the efficiency of TrCvtR transfer to mammalian cells in vitro in a neoendothelium model and validation. Monolayers of human endothelial cells (HUVEC) are grown on a permeable support. The apical pole of this polarized monolayer is exposed to VLP- TrCytR-AV-aDII4 and assayed for different parameters: (i) Efficiency of VLP-cell membrane fusion at the apical pole is calculated using VLP-TrCytR-AV-aDII4 which have encapsidated the fusion protein luciferase-Vpr, and testing for luciferase activity in the HUVEC cells, (ii) The efficiency of TrCytR transfer and membrane localisation in human endothelial target cells is assayed by FACS analysis, using specific anti- TrCytR antibody. (iii) In order to assess the functionality of TrCytR molecules transferred to polarized endothelium, HUVEC cells transduced first with VLP-TrCytR-AV- aDII4 are exposed at their apical pole to Ad/GFP vector coated with anti- TrCytR antibody via PEGylated bridges produced as shown in Example 1. Efficient transcytosis of the Ad/GFP vector is occurring if GFP is detectable in the basolateral medium. The efficiency of transcytosis is measured by titrating Ad/GFP in the lower chamber by conventional techniques. See also Figure 1 and Tang, Y., et al. (supra).

Transcytosis of the HER2/Neu-targeted Ad5 vector Ad/GFP/HI-Link:ZH2 produced in Example 1 is tested in recipient tumour cells added basolaterally in the lower chamber of the culture device (i.e. beneath the HUVEC layer) as represented in Figure 1. The intensity of GFP signal in tumour cells will reflect both the efficiency of transcytosis and the efficiency of tumour cell targeting.

4a.5. Testing the efficiency of TrCvtR transfer in vivo in tumour-bearing mouse model

To be able to target neoendothelium in mice the -CNGRC- peptide mentioned in Example 1 is linked to a VLP as in 4.3. Mice bearing tumours are injected with neoendothelium-targeted VLP-TrCytR-AV-CNGRC containing encapsidated luciferase-Vpr. The efficiency and specificity of targeting are evaluated by the level of luciferase signal analysed in situ using a CDD camera.

4a.6. Testing the ability of targeted VLP-TrCvtR to mediate transcvtosis of virus in vivo

Using the same model as in 4.5 the ability of VLP-TrCytR-A V-CNG RC to mediate uptake into tumour tissue of Ad/HI-Link:ZH2 coated with aTrCytR- PEG (Example 1 ) is assessed with routine assays (e.g. GFP signal) and compared to the uptake of Ad/HI-l_ink:ZH2 coated with CNGRC-PEG (Example 1 ).

Example 4b

4b.1. Production of nonreplicative AcMNPV-derived baculoviral vector. carrying gene for transcvtosis receptor (TrCvtR)

The gene for the TrCvtR is cloned under the control of the mammalian CMV- IE promoter, giving rise to AcMNPV-CMVPr-TrCytR. Sf9 cells are infected with AcMNPV-CMVPr-TrCytR and produce baculovirions carrying their normal envelope Gp64. After fusion with and entry into mammalian recipient cells, TrCytR protein is expressed and delivered to the plasma membrane by intracellular trafficking pathways.

4b.2. Large-scale production and purification of AcMNPV-CMVPr-TrCvtR. Sf9 cells in mass culture will be infected by recombinant AcMNPV at MO1 10 for 48 h. Extracellular particles budding and released into the culture medium will be purified by a two-step procedure comprising (i) step gradient centrifugation through a 20 % sucrose cushion, and (ii) isopycnic ultracentrifugation through sucrose-D₂θ density gradient. Membrane- enveloped AcMNPV virions (d = 1.08-1.10) can be separated under these conditions. As much as 10¹⁰ baculoviral particles are usually recovered per 10⁶ CeIIs. 4b.3. Engineering neoendothelium-tarqeted baculovirus carrying transcytosis receptor.

To target AcMNPV-CMVPr-TrCytR specifically to the endothelial cells of tumour microvasculature, AcMNPV-CMVPr-TrCytR are biotinylated in vitro, and coupled to biotinylated aDII4 via an avidin bridge. This will give rise to AcMNPV-CMVPr-TrCytR-AV-Dltø. The alternative antibodies described in 4a.3 are also used to give rise to AcMNPV-CMVPr-TrCytR-AV-aPSMA.

4b.4. Testing the efficiency of TrCvtR transfer to mammalian cells by AcMNPV-CMVPr-TrCytR in vitro in a neoendothelium model and validation. The protocols described in 4a.4 are used.

Example 5

Incorporation of HSV1-tk into the adenoviral vectors and complemention with the Dm-dNK/B5 gene

The mutant HSV1-sr39tk gene is cloned into Ad/HI-Link:ZH2 to allow for PET scan analysis of virus distribution (HSV1-tk is a PET reporter gene). A cassette with a CMV promoter and the HSV1-sr39tk gene is inserted into the E3 region of the Ad/HI-Link:ZH2 virus. The level of HSV1 -sr39tk expression and its ability to kill cells together with gancyclovir is determined in vitro in HER2/neu expressing cells. The Ad/HI-Link:ZH/HSV1-sr39tk viral backbone is then used together with the different modifications derived from Examples 1 , 2 and 3 to produce adenoviral vectors with those modifications and encoding for the HSV1-tk gene. In vivo evaluation of these vectors is done as described in Example 6.

The Dm-dNK/B5 gene is ligated to the E3 region of the Ad/HI-Link^":ZH2 virus as mentioned above. The cell killing abilities and bystander effects together with studies of viral replication are performed in 293/HER2 and SKOV3 cells in vitro with addition of Gemzar as prodrug substrate. The in vivo studies are performed by injecting virus i.p. in nude mice with i.p. xenografts, derived from SKOV3 cells, followed by Gemzar® treatment.

Example 6

6.1. Virus distribution studies

To study virus survival time in the circulation and uptake in different normal organs as well as ability to target tumour tissues virus distribution studies are required. Studies are performed in normal rats and mice passively injected with human neutralizing serum. Virus distributions are monitored by PET or HSV1-tk encoding virus) luminescence (luminescent labelled virus) as well as Q-PCR and transgene expression in different tissues after necropsy. The circulation time is measured by taking blood samples from the injected animals at different time points followed by Q-PCR and titration on HER2/neu expressing cells.

6.2. Efficacy studies

The ability of the different viruses (including TRAPs described in previous Examples) to kill xenografted human tumours are investigated. Six-eight week old athymic nu/nu female mice are implanted ip with 10⁷ SKOV3 cells for local or orthotopic growth of tumours, respectively. SKOV3 tumours develop in the ip cavity by 12-14 days post implantation. These mice are also injected with 10⁷ cells sc at the same time as the ip implantation depending on the aims of the experiment. Sc tumours develop after 4-8 weeks (depending on whether the injections are with or without prior mixing of the cells with Matrigel, BD) and the tumour take is 90%. When using sc tumours the injection of virus is iv to study targeting. In both models the mice are monitored for tumour progression/regression using PET or luminescence (depending on vector label) and tumours are weighed, stained for Ad replication and studied for transgene expression after necropsy.

Claims

1. A method for the transportation of macromolecules or aggregates thereof across a targeted layer of epithelial cells comprising: (a) selectively displaying a transcytosis receptor (TrCytR) on the surface of the targeted epithelial cells; and

2. The method of claim 1 wherein the epithelial cells are simple epithelial cells.

3. The method of claim 1 or claim 2 wherein the epithelial cell layer is selected from the epithelial cells of the skin, the lumen of the lungs, gastrointestinal tract, reproductive and urinary tracts, the exocrine and endocrine glands, the walls of the pericardium, pleurae and peritoneum and the endothelium, preferably the endothelium of the circulatory system.

4. The method of claim 3 wherein the endothelium is the endothelium of the vasculature of tumours.

5. The method of any preceding claim wherein the selective display of the TrCytR involves the selective introduction of a TrCytR that is not expressed in the epithelial cells of the targeted epithelial cell layer by means of a vector targeted to the epithelial cells of the targeted epithelial cell layer.

6. The method of claim 5 wherein the vector is a fusiogenic VLP, a baculovirus vector or an adenovirus vector.

7. The method of claim 5 or claim 6 wherein the vector displays a targeting molecule on its surface that selectively binds a partner molecule on the target epithelial cell layer.

8. The method of claims 7 wherein the targeting molecule binding partner is selected from PSMA, endoglin, RGD4C peptide, NGR peptide, endosialin, delta like ligand 4, TEM5, TEM7, TEM8, CD34; LYVE1 , VE- cadherin, von Willebrand Factor, PECAM1 , ICAM-1 , CD146, endocan, Endoglyx-1 , VEGFR-1 and VEGFR-2.

9. The method of claim 7 or claim 8 wherein the targeting molecule is a natural ligand on the binding partner, an antibody, preferably a camelid antibody, directed to the binding partner or an affibody directed to the binding partner.

10. The method of any one of claims 5 to 9 wherein the vector is a viral vector comprising an artificial polymer coat.

11. The method of any one of claims 5 to 10 wherein the targeting of the vector involves transcriptional targeting.

12. The method of any preceding claim wherein the macromolecule or aggregate thereof is selected from biological polymers, preferably proteins, nucleic acids, polysaccharides and lipids and combinations thereof; viruses, viral vectors, virus-like particles, polyplexes, liposomes, lipoplexes, inert particles carrying macromolecular markers or labels, LDL, HDL, preferably an adenovirus or a baculovirus vector.

13. The method of any preceding claim wherein the macromolecule or aggregate thereof is a therapeutic or diagnostic product.

14. The method of claim 13 wherein the therapeutic product comprises an anticancer agent, preferably a cytotoxic protein or a cytotoxic RNA molecule.

15. The method of claim 13 wherein the therapeutic product comprises a therapeutic gene, preferably the therapeutic product is a vector carrying a therapeutic gene.

16. The method of claim 15 wherein the therapeutic gene encodes a protein, preferably a growth factor receptor, an oncogene, a tumour suppressor, an antisense oncogene, a suicide protein, an immune modulator, a tumour antigen, an anti-angiogenic factor, a cytokine, a vascular endothelial growth inhibitor, a fusiogenic membrane glycoprotein, a cytotoxic gene or an enzyme which is capable of converting a pro-drug to cytotoxic substance; or a bioactive RNA molecule, preferably an antisense molecule, a ribozyme, an siRNA or an miRNA.

17. The method of claim 16 wherein the therapeutic gene is selected from cytosine deaminase, mda-7/IL-24, cdtB, yCD, yCD/UPRT, Dm-dNK/B5, HSV1-tk, CD40 ligand, GM-CSF, IL-2 IL-12, and IL-18.

18. The method of any one of claims 15 to 17 wherein the vector is a gene therapy vector, preferably a viral vector or a VLP

19. A virus-like particle (VLP) having a plasma membrane-derived lipid bilayer envelope, said VLP further comprising: a) a viral structural protein, or fragment or derivative thereof, capable of forming an enveloped VLP; b) a fusiogenic protein; and c) a TrCytR.

20. A method for the production of a VLP as defined in claim 19, said method comprising the coexpression of a viral structural protein, or fragment or derivative thereof, capable of forming an enveloped VLP₁ together with a fusiogenic protein, together with a TrCytR in an in vitro cultured cell and isolating the VLP from the culture media.

21. An in vitro host cell line comprising a) a nucleic acid encoding a viral structural protein, or fragment or derivative thereof, capable of forming an enveloped VLP; b) a nucleic acid encoding a fusiogenic protein; and c) a nucleic acid encoding a TrCytR.

22. A viral vector encoding or comprising a TrCytR.

23. A product containing

(a) a vector that displays, encodes or upregulates the expression of a TrCytR and is capable of selectively introducing said

TrCytR to the surface of a target epithelial cell layer or selectively increasing the display of said TrCytR on the surface of a target epithelial cell layer

(b) a therapeutic macromolecule or aggregate thereof wherein said macromolecule or aggregate thereof comprises at least one region which is capable of inducing transcytosis of said macromolecules or aggregates thereof upon binding to said TrCytR when said TrCytR is present on the surface of the target epithelial cell layer as a combined preparation for simultaneous, separate or sequential use

24. The method, VLP, in vitro host cell line, viral vector or product of any preceding claim wherein the TrCytR is selected from the polymeric immunoglobulin receptor, the melanotransferrin receptor, gp60, the IgG receptor, the transferrin receptor, LDL receptor, insulin receptor, megalin and functional derivatives or fragments thereof.

25. The method, VLP, in vitro host cell line, viral vector or product of any one of claims 1 to 23 wherein the TrCytR is an artificial TrCytR

26. The method, VLP, in vitro host cell line, viral vector or product of claim 25 wherein the artificial TrCytR comprises a natural binding partner of a macromolecule or aggregate thereof, or an antibody or a functional fragment or derivative thereof directed to a macromolecule or aggregate thereof; or an affibody molecule or a functional fragment or derivative thereof, directed to a macromolecule or aggregate thereof on a transmembrane scaffold that has a region capable of interaction with a caveolae protein, preferably a caveolin, gp60 or aminopeptidase P.

27. The method, VLP, in vitro host cell line, viral vector or product of claim 25 wherein the artificial TrCytR is a fusion protein comprising (i) a caveolae protein, preferably a caveolin, gp60 or aminopeptidase

P, or a functional fragment or derivative thereof; and

(ii) a protein that can be bound by a macromolecule or aggregate thereof, preferably a natural binding partner of the macromolecule or aggregate thereof, or an antibody or a functional fragment or derivative thereof directed to the macromolecule or aggregate thereof; or an affibody molecule or a functional fragment or derivative thereof, directed to the macromolecule or aggregate thereof.

28. A pharmaceutical composition comprising a VLP or a viral vector as defined in any one of claims 19, 22 and 24 to 27 together with at least one pharmaceutically acceptable carrier, diluent or excipient.

29. A fusion protein comprising the amino acid sequence of a caveolae protein or a functional derivative or fragment thereof and another protein sequence wherein said another protein sequence is capable of binding to a macromolecule or an aggregate thereof.

30. The fusion protein of claim 29 wherein the caveolae protein is selected from caveolin-1 , -2 and -3 and functional fragments and derivatives thereof.

31. The fusion protein of claim 29 or claim 30 wherein the macromolecule or aggregate thereof is as defined in any one of claims 12 to 18.

32. The fusion protein of any one of claims 29 to 31 wherein the other protein sequence is a natural binding partner of a macromolecule or aggregate thereof, or an antibody or a functional fragment or derivative thereof directed to a macromolecule or aggregate thereof; or an affibody molecule or a functional fragment or derivative thereof, directed to a macromolecule or aggregate thereof.

33. The VLP, viral vector, product, composition or fusion protein of any one of claims 19 and 22 to 32 for use in therapy.