Field of the Invention

The invention relates to recombinant viral particles incorporating protease cleavable proteins and to various applications of the recombinant particles.

Retroviral envelopes

Retroviral envelope glycoproteins mediate specific viral attachment to cell surface receptors and subsequently trigger fusion between the viral envelope and die target cell membrane. All retroviral envelope spike glycoproteins examined to date are homooligomers containing two to four heterodimeric subunits (Doms et al. 1993 Virology 193. 545). Each subunit comprises a large extraviral glycoprotein moiety (SU) noncovalently attached at its C-terminus to a smaller transmembrane polypeptide (TM) that anchors the complex in the viral membrane. In the case of murine C-type retroviral vectors, SU comprises two domains connected by a proline-rich hinge, the N-teπninal domain conferring receptor specificity and exhibiting a high degree of conservation between murine leukemia viruses (MLVs) with different host ranges (Battini et al. 1992 J. Virol 66, 1468-1475). Moloney MLV envelopes confer an ecotropic host range because they attach selectively to a peptide loop in the murine cationic amino acid transporter (CAT-1), found only on cells of mouse and rat origin (Albritton et al. 1989 Cell 57, 659- 666). 4070A MLV envelopes attach to an epitope on the ubiquitous RAM-1 phosphate symporter that is conserved throughout many mammalian species, and confer an amphotropic host range (Miller et al. PNAS 91,, 78-82; VanZeijl et al. 1994 PNAS 9_i, 1168-1172). Thus, retroviral vectors with 4070A envelopes infect human cells promiscuously, whereas vectors wid Moloney envelopes fail to infect human cells.

Proteolytic activation

In all retroviruses that have been studied to date, the SU and TM polypeptides are derived from a single chain precursor glycoprotein that undergoes proteolytic maturation in the

Gogi compartment during its transport to the cell surface. Uncleaved envelope precursor glycoproteins can be incorporated into viruses but are unable to trigger membrane fusion. The requirement for proteolytic maturation/activation is a feature common to the fusogenic membrane glycoproteins of many virus families and is most commonly mediated by the ubiquitous Golgi compartment serine protease, furin. However, there are well- documented examples of viral membrane glycoproteins that resist cleavage by ubiquitous intracellular proteases and instead are cleaved by secreted proteases available only in a few host systems (Klenk & Garten 1994 Trends Microbiol. 2, 39). Moreover, there is at least one example of an influenza virus strain whose haemagglutinin is activated by a target cell protease at the stage of virus entry (Boycott et al. 1994 Virology 203, 313). In these instances, the binding reactions of the viral membrane glycoproteins are unaffected by dieir proteolytic cleavage - only their ability to trigger membrane fusion is affected.

Retroviral display of nonviral polypeptides as N-terminal extensions of SU

A general method has been disclosed which allows the display of a (glyco)polypeptide on the surface of a retroviral vector as a genetically encoded extension of the SU glycoprotein (WO 94/06920, Medical Research Council). The polypeptide is fused (by genetic engineering) to the N-terminal part of the SU glycoprotein such that the envelope protein to which it has been grafted remains substantially intact and the fused nonviral polypeptide ligand is displayed on die viral surface. To date, die approach has been used to display many different polypeptide ligands on MLV - based retroviral vectors, including single chain antibodies, cellular growth factors and immunoglobulin binding domains (WO 94/06920, Medical Research Council; WO 96/00294 Medical Research Council; Cosset et al. 1994 Gene Therapy 1 , SI; and Nilson et al. 1994 Gene Therapy J,, S17). In contrast to other chimaeric retroviral envelope proteins that have been described (CD4 chimaera, Kasahara et al. 1994 Science 266, 1373; Chu & Dornburg 1995 J. Virol. 69, 2659; Somia et al. 1995 PNAS 92, 7570) viral incorporation of N-terminally extended SU glycoproteins does not require the presence of unmodified envelope glycoproteins.

In principle, a virus displaying such a chimaeric envelope protein might be capable of multivalent attachment both to the natural virus receptor (via the N-terminal domain of SU) and to the cognate receptor for the displayed polypeptide. We have found that this

holds true for retroviral vectors displaying epidermal growth factor (EGF). However, depending on its precise nature, its propensity to oligomerise and its mode of linkage to the SU glycoprotein, the displayed polypeptide may sterically hinder the interaction between the N-terminal domain of SU and the natural virus receptor.

Modification of retrovirus tropism by N-terminally extended SU glycoproteins

When different receptor-binding domains were displayed on MLV retroviral vectors as N- terminal extensions of their intact SU glycoproteins, it was found that host range could be extended or restricted by die displayed ligand (Cosset et al. 1995 J. Virol. 69, 6314- 6322). Thus, as a demonstration of host range extension, murine ecotropic vectors displaying the RAM-1 receptor-binding domain from 4070A SU were able to bind and infect RAM-1 -positive human cells. In contrast, as a demonstration of host range restriction, ecotropic and amphotropic vectors displaying EGF could bind to EGF receptors but were thereafter sequestered into a non-infectious entry pathway, giving greatly reduced titres on EGF receptor-positive cells, but normal titres on EGF receptor- negative cells. EGF receptor-negative cells, which were fully susceptible to the engineered retroviral vector, showed reduced susceptibility when they were genetically modified to express EGF receptors. The reduction in susceptibility was in proportion to the level of EGF receptor expression. Moreover, when soluble EGF was added to competitively inhibit virus capture by the EGF receptors, gene transfer was restored. In diis latter example, the engineered vector is capable of binding to the natural virus receptor or to the receptor for EGF; attachment to the natural virus receptor leads to infection of the target cell, whereas the attachment to the EGF receptor does not lead to infection of the target cell. Where the target cell expresses both species of receptor, the two binding reactions (4070A envelope protein to RAM-1, and EGF to EGF receptor) proceed in competition and the infectivity of the virus for the target cells is reduced in proportion to the efficiency with which the EGF-EGF receptor binding reaction competes virus away from RAM-1.

The degree to which gene transfer can be inhibited by this mechanism depends on die relative affinities of the two binding reactions (envelope protein to natural receptor and non-viral ligand to its cognate receptor), die relative densities of die two receptors on the

target cell surface, and die relative densities of ie nonviral ligand and the intact envelope protein on the viral surface. Inhibition of gene transfer is additionally influenced by intrinsic properties of die receptor for the non-viral ligand, such as d e distance it projects from die target cell membrane, its mobility widiin die target cell membrane and its half life on the cell surface after engagement of ligand.

Steric hindrance of the interaction between the N-terminal domain of SU and the natural virus receptor provides an alternative mechanism whereby polypeptides displayed as N- terminal extensions of SU can restrict retroviral host range. For example, chimaeric envelopes displaying die N-terminal domain from 4070A MLV SU as an N-terminal extension of Moloney MLV SU can apparently bind to RAM-1 (the receptor for 4070A SU) but not to ecoR (the receptor for Moloney SU); it may be possible that the displayed domains from 4070A SU may form a trimeric cap over the Moloney SU trimer, completely masking its receptor binding sites. If this model is correct, then it should also be possible to generate chimaeric envelopes in which the receptor binding sites of the intact 4070A SU glycoprotein (through which the virus attaches to human cells) are masked by a displayed polypeptide, such as the N-terminal domain of Moloney MLV SU, iat does not bind to human cells.

Phage display of cleavable domains

There are certain similarities between retroviral vectors displaying polypeptide ligands as N-terminal extensions of their envelope glycoproteins and filamentous bacteriophage displaying polypeptide ligands as N-terminal extensions of the gene III protein. Libraries of filamentous "substrate phage" displaying cleavable binding domains have recently been used to identify optimal substrates for known proteases (Matthews & Wells 1993 Science 260. 1113; Matthews et al. 1994 Protein Science 3, 1197; Smith et al. 1995 J. Biol. Chem. 270, 6440). However, filamentous phages do not naturally infect mammalian cells and there has been no demonstration that cleavable domains fused to die gene III protein can influence the tropism of the phages on which they are displayed.

Summary of the Invention

In a first aspect the invention provides a recombinant viral particle capable of infecting a

eukaryotic cell, the viral particle comprising: a substantially intact viral glycoprotein fused, via an intervening linker region, to a heterologous polypeptide displayed on the surface of the particle, which heterologous polypeptide modulates the ability of the viral panicle to infect one or more eukaryotic cell types and is cleavable from the viral glycoprotein by a protease acting selectively on a specific protease cleavage site present in the linker region, such that cleavage of the heterologous polypeptide from the viral glycoprotein allows the glycoprotein to interact normally with its cognate receptor on the surface of a target cell.

Such a panicle is of considerable benefit in die targeted delivery of nucleic acid sequences, which may be present within the panicle, to specific desired target cells, such as is required for gene dierapy.

In another aspect the invention provides a nucleic acid construct, comprising a sequence encoding a fusion protein, the fusion comprising a substantially intact viral glycoprotein fused, via an intervening linker region, to a heterologous polypeptide, wherein the fusion protein is capable of being incorporated into a viral particle capable of infecting an eukaryotic cell, and further wherein the heterologous polypeptide modulates the ability of the viral particle to infect one or more eukaryotic cell types, but cleavage of the heterologous polypeptide from the fusion protein allows the viral glycoprotein to interact normally with its cognate receptor on the surface of the eukaryotic cell.

In a further aspect the invention provides a nucleic acid sequence library comprising a plurality of the nucleic acid constructs defined above, wherein at least part of the sequence encoding the intervening linker region is randomised in each construct, such diat each construct comprises one of a plurality of different linker regions which are represented in the library.

The invention also provides a library of the viral panicles defined above, each panicle comprising a single nucleic acid construct from the nucleic acid library defined above.

The term "substantially intact" as used herein is intended to refer to a viral glycoprotein

which retains all of its domains so as to conserve post-translational processing, oligomerisation (if any), viral incorporation and fusogenic properties. However, certain alterations (e.g. point mutations, deletions, additions) can be made to the glycoprotein without significantly affecting these functions, and glycoproteins containing such minor modifications are considered substantially intact for present purposes. In particular, the gluycoprotein may lack a few (e.g. about 1 to 10) amino acid residues, especially at the N terminus, but will otherwise be generally the same size as the wild-type protein and possess substantially the same biological properties as the wild-type protein.

The intervening linker region will preferably be quite short, typically comprising from 4 to 30 amino acid residues, more typically 5 to 10 residues. A short linker is preferred, because this will tend to maximise the modulation of infection effected by the heterologous polypeptide. In certain embodiments, a suitable linker region may be present as a natural part of the heterologous displayed polypeptide.

The viral particle may be any virus capable of infecting one or more eukaryotic cell types, but conveniently will be a viral particle suitable for use in gene therapy, such as an adenovirus or a retrovirus (especially a C-type retrovirus).

The viral glycoprotein will typically comprise a viral envelope glycoprotein, or may be a chimeric polypeptide comprising sequences conesponding to different viral glycoproteins but which, in total, consitute a substantially intact, functional protein.

The heterologous polypeptide may be a short amino acid sequence (say, a peptide of about 10-20 residues, especially if d e sequence undergoes oligomerisation, e.g. a leucine zipper peptide sequence) but more typically will comprise 30 or more amino acid residues. Generally, but not essentially, the polypeptide will comprise a functional binding domain. The heterologous polypeptide, when fused to die viral glycoprotein via the linker region, modulates die ability of the viral particle to infect one or more eukaryotic cell types. Specifically, the presence of the heterologous polypeptide serves to inhibit the process of infection of a eukaryotic target cell mediated by the viral glycoprotein. The term "heterologous" is intended to refer to any polypeptide which is not naturally fused or

otherwise bound to d e viral glycoprotein.

The heterologous polypeptide may or may not possess specific binding affinity for a surface component of a target cell. In one embodiment, die heterologous polypeptide has affinity for a cell surface component, binding to which will not lead to infection of the cell by the virus. Within this general embodiment, a variety of different examples (each with different properties) can be envisaged. In one example, a eukaryotic cell expresses a receptor for the viral glycoprotein (binding to which allows the virus to infect the cell) and a non-permissive receptor for the heterologous polypeptide, with inhibition of infection resulting simply from competition between the viral glycoprotein and the heterologous polypeptide for binding to their respective receptors on the target cell. In a different example, the conformational arrangement of the respective receptors and their ligands is such that binding of the heterologous polypeptide to its receptor causes steric hindrance, such that binding of the viral glycoprotein to its receptor, or fusion of the virus and die cell, is blocked.

In a second embodiment, the heterologous polypeptide does not bind to a non-permissive receptor on the target cell, but the presence of the heterologous polypeptide serves to create steric hindrance sufficient to prevent binding of the viral glycoprotein to its receptor, or may allow binding to occur but inhibits subsequent fusion of the viral particle with the target cell, such that infection of the cell by the viral particle is inhibited at the binding and/or fusion stage.

In a particular embodiment the heterologous polypeptide is capable of forming oligomers when displayed on the surface of the viral particle. Typically the oligomer will be a dimer or, more preferably, a trimer. Such oligomerisation may allow for efficient inhibition of the interaction between the substantially intact viral glycoprotein and its receptor, which inhibition may be removed by proteolytic cleavage of the oligomerised heterologous polypeptide from the viral glycoprotein. The intervening linker may also undergo oligomerisation.

Where the viral glycoprotein is itself capable of forming oligomers (e.g. retroviral env

protein), it is preferred that the heterologous polypeptide oligomerises with the same stoichiometry as that of the viral glycoprotein. Vascular endothelial growth factor (VEGF) and rumour necrosis factor (TNF) are both proteins which are known to oligomerise and have high affinity for cell surface ligands. Effective (oligomer-forming, and preferably ligand-binding) portions of these proteins may be particularly suitable for use as heterologous polypeptides in accordance with the present invention.

In the present invention the heterologous polypeptide is cleavable from the viral glycoprotein by the selective action of a protease (i.e. a molecule capable of cleaving a peptide bond) which cleaves the linker region at a protease cleavage site. The cleavage site represents a unique peptide sequence not present, or at least not accessible to the protease, in the viral glycoprotein, although a similar site may be present in the heterologous polypeptide (this is generally preferably avoided, as proteolytic attack on the heterologous polypeptide may affect its functioning). The size, and number, of the protease cleavage sites in the linker region may be varied with advantage. Thus, for example, the presence of two or more cleavage sites, recognised by the same or by respective proteases could facilitate cleavage, whilst the use of one long cleavage site will tend to enhance specificity of cleavage.

Large numbers of specific proteases, and die cleavage sites d ey recognise, are known to those skilled in the art (see, for example, Vassalli & Pepper 1994 Nature 370, 14-15, and references cited dierein). Proteases are involved in a number of physiological and/or pathological processes, such as tissue remodelling, wound healing, inflammation and tumour invasion, and such proteases would be of use in the present invention. Specific classes of protease which would be of use include: serine proteases (such as plasminogen/plasmin enzymes); cysteine proteases; and matrix metalloproteinases (MMPs) of various types, (such as Gelatinase A and membrane-type MMP [or MT-MMP]).

The protease which serves to cleave the heterologous polypeptide from the viral glycoprotein is preferably selectively secreted by die cell to which it is desired to target the viral particle, or at least the tissue in which the target cell is located. It is preferred that die protease will be secreted only by cells of the target cell type or, less preferably,

only by cells (other than the target cells) remote from die tissue containing die target cell. This confers an extra degree of specificity, which is desirable when die particle is used for targeted gene delivery. Thus die present invention allows for two-step targeting, in which a first level of specificity may be imposed by the heterologous polypeptide (e.g. with specific affinity for a ligand on d e surface of the target cell), and a second level of specificity may be imposed by selective cleavage of d e heterologous polypeptide by proteases secreted by, or in the same tissue as, the target cell. Alternatively, the relevant protease may be added exogenously, such d at if the viral particle is used for targeted gene delivery in a patient, the protease may be administered (e.g. by injection) to die tissue in which the target cell is located.

Accessibility of the protease cleavage site to the relevant protease (i.e. that which recognises and cleaves the site) may also be varied. It has been found by the present inventors that use of a short intervening linker region (e.g. 5 amino acid residues) tends to restrict accessibility of the cleavage site, and use of a larger linker region (e.g. 15 to 20 residues) tends to increase accessibility of the cleavage site. This phenomenon is presumably due to seric hindrance of the cleavage site due to die proximity of the viral glycoprotein and or die heterologous polypeptide. Accordingly, it should also be possible to modify accessibility of the cleavage site, as desired, by varying the size of the heterologous polypeptide.

In one embodiment, the cleavage site is accessible to die relevant protease before the viral particle becomes bound to an eukaryotic cell, whilst in an alternative embodiment die cleavage site is inaccessible to the protease until the viral particle has become bound to a eukaryotic cell. In this latter embodiment, die cleavage site may be made accessible by a conformational change occurring as a result of binding of the heterologous polypeptide to its cognate receptor. Alternatively, the viral glycoprotein binding to its cognate receptor may make die cleavage site accessible, cleavage of the heterologous polypeptide then allowing fusion of the viral particle to the eukaryotic target cell.

As indicated above, in another aspect the invention provides for a method of selectively delivering a nucleic acid to a target eukaryotic cell present among non-target cells,

comprising: administering to die target and non-target cells a recombinant viral particle capable of infecting eukaryotic cells, the particle comprising die nucleic acid to be delivered, and a fusion protein comprising a substantially intact viral glycoprotein fused, via an intervening linker region, to a heterologous polypeptide displayed on the surface of the particle, which heterologous polypeptide modulates die ability of the particle to infect one or more eukaryotic cell types and being cleavable from die glycoprotein by a protease acting selectively on a specific protease cleavage site present in the linker region, such that cleavage of the heterologous polypeptide from the glycoprotein occurs preferentially at, or in the vicinity of, the target cell and allows d e viral glycoprotein to interact normally wid its cognate receptor on the surface of the target cell.

The method may be performed in vitro, for example to deliver a ledial nucleic acid to fibroblasts in tissue culture, which cells often outgrow a slower-growing, more differentiated cell type in culture. Alternatively, the mediod may be performed as a method of gene therapy, in vivo or may be performed ex vivo, on cells which are then re- introduced into a human or animal subject. Preferential cleavage of the protease cleavage site may occur only when the viral particle is bound to die target cell, or when the viral particle is adjacent to die target cell and dius exposed to a protease secreted by die target cell. It may well be preferred to add the relevant protease exogenously, after administration of die viral particle, so as to ensure sufficient concentration of the protease and as anodier aid to specificity of delivery (by local administration of the protease). It is already known that some proteases may be safely given in vivo (e.g. those enzymes, such as urokinase, streptokinase and tPA, given to patients with myocardial infarcts).

The invention also provides, in a further aspect, a method of screening nucleic acid sequences for diose which encode an amino acid sequence which may or may not be cleaved by a protease. As already mentioned, many viral envelope glycoproteins are processed through d e cellular export pathway of the eukaryotic cell in which they are synthesised, generally leading to cleavage, which cleavge is essential for production of an infectious viral particle.

The invention d erefore provides a method of screening nucleic acid sequences for those

which encode an amino acid sequence which may or may not be cleaved by a protease present in the export pathway of an eukaryotic cell, comprising: causing die expression of a plurality of nucleic acid sequences in eukaryotic cells, each sequence encoding a substantially intact viral glycoprotein fused to a heterologous polypeptide via a randomised intervening linker region, the presence of the heterologous polypeptide serving to inhibit die (binding or fusion) interaction of the viral glycoprotein with its cognate receptor, and wherein each nucleic acid sequence further comprises a packaging signal allowing for viral incorporation, such that those intervening linkers which are recognised by a protease present in die export padiway of the eukaryotic cells will allow for cleavage of the heterologous polypeptide from die viral glycoprotein, resulting in the production of an infectious viral particle; and recovering those nucleic acid sequences directing the expression of such cleavable linker regions from an infected cell.

Nucleic acid sequence determination may optionally be performed, to deduce diose amino acid sequences which are recognised by an export protease.

A modification of the above method will allow for die screening of nucleic acid sequences for diose which encode an amino acid sequence which may or may not be cleaved by a protease present in die eukaryotic cell import pathway. As explained above, the presence of a heterologous polypeptide may, in some embodiments, still allow for binding of the viral glycoprotein to its cognate receptor, but will prevent fusion of the viral particle with d e eukaryotic cell to which it is bound. Cleavage of die heterologous polypeptide by a protease in the cellular import pathway will then allow infection of the cell.

Thus, in a further aspect the invention provides for a method of screening nucleic acid sequences for those which encode an amino acid sequence which may or may not be cleaved by a protease, comprising: causing the expression of a plurality of nucleic acid sequences in eukaryotic cells, each sequence encoding a substantially intact viral glycoprotein fused to a heterologous polypeptide via a randomised intervening linker region, the presence of the heterologous polypeptide serving to inhibit the fusion of a viral particle with a eukaryotic cell to which it is bound, and wherein each nucleic acid sequence further comprises a packaging signal allowing for viral incorporation; enriching

the viral particles so produced for those which retain the heterologous polypeptide (and so are non-infectious); and contacting the enriched particles with a susceptible eukaryotic cell comprising, or in the presence of, a protease such that diose intervening linkers which are recognised by the protease will allow for cleavage of the heterologous polypeptide from the viral glycoprotein, resulting in productive infection of the eukaryotic cell; and recovering those nucleic acid sequences directing die expression of such cleavable linker regions from the infected cell. As above nucleic acid sequence determination may optionally be performed to allow deduction of the conesponding amino acid sequences.

The enrichment step is required because of die possibility that die heterologous polypeptide may be cleaved from d e viral glycoprotein by an export pathway protease during syndiesis of the panicles. A number of possible enrichment techniques will be readily apparennt to those skiled in the an widi die benefit of the present teaching. For example, prior to infection of the susceptible cells, die viral particles could be subjected to an affinity enrichment technique - die particles could be passed dirough an antibody affinity column, wherein the antibody has affinity for the heterologous polypeptide. Those particles which retain die heterologous polypeptide will be bound to ie column, whilst those in which the heterologous polypeptide was cleaved during export from die producing cell will pass straight through the column. After washing, the bound particles may be eluted (e.g. by competition widi free heterologous polypeptide, or the part d ereof recognised by die antibody, or by alteration of pH or other factors) and then used to infect d e susceptible "indicator" cells.

The invention will now be further described by way of illustrative examples, and widi reference to the accompanying figures, in which:

Figure 1 is a schematic representation of retroviral vector constructs coding for chimeric envelopes;

Figures 2 and 3 are photographs of Western blots demonstrating viral incorporation of certain chimeric polypeptides and dieir sensitivity to Factor Xa protease;

Figure 4 is a photograph showing d e infectivity of various /3-galactosidase transducing viruses on target cells with or without Factor xa treatment, as judged by assay on X-gal containing plates;

Figure 5 is a schematic representation of how two-step targeting of gene delivery might be achieved using die present invention;

Figure 6A is a photograph of a Western blot demonstrating viral incorporation of certain chimeric polypeptides and dieir sensitivity to Factor Xa protease;

Figure 6B is a bar chart illustrating die infectivity of certain recombinant viruses in the presence or absence of Factor Xa;

Figure 7 is a schematic representation of retroviral vector constructs coding for chimeric envelopes;

Figure 8A is a photograph of two Western blots, the upper one comparing electrophoretic mobility of various chimeric polypeptides, die lower one comparing the amount of protein present;

Figure 8B is a photograph of a Western blot comparing the sensitivity to Factor Xa protease of various chimeric polypeptides;

Figure 8C is a photograph of a Western blot comparing processing of certain chimeric polypeptides;

Figure 9 is a panel of photographs comparing the growth of of a recombinant virus on NIH 3T3 and A431 cells, widi or widiout Factor Xa treatment;

Figure 10 is a schematic representation of retroviral vector constructs coding for chimeric envelopes;

Figure 11 shows three Tables, A, B and C, illustrating die titre (in enzyme forming units, "e.f.u. ") of various recombinant viruses on NIH 3T3 or A431 cells in the absence (-) or presence ( +) of Factor Xa protease;

Figure 12 is a schematic representation of retroviral vector constructs coding for chimeric envelopes;

Figure 13 A is a photograph of a Western blot demonstrating viral incorporation of various chimeric polypeptides;

Figure 13B is a photograph of a Western blot comparing the sensitivity of various chimeric polypeptides in die presence (+) or absence (-) of pro-gelatinase A, with (+) or without (-) pre-activation of die protease by p-aminopheny .mercuric acetate (APMA);

Figure 14 is a bar chart showing how infectivity of a recombinant virus is dependent upon concentration of pro-gelatinase A;

Figure 15 is a bar chart comparing die infectivity of three different recombinant viruses on HT 1080 or A431 cells;

Figure 15 A is a photograph comparing the growth of a recombinant virus on HT 1080 or A431 cells;

Figure 16 is a panel of four photographs (I, II, III and IV) comparing the infectivity of various viruses on HT 1080 (H) or A431 (A) cells;

Figure 17 is a photograph of a gel for detection of gelatinolytic activity; and

Figures 18 and 19 are schematic representations of retroviral vector constructs coding for chimeric envelopes.

EXAMPLES

Example 1 Summary

Tropism-modifying binding domains were anchored to murine leukaemia virus (MLV) envelopes via factor Xa-cleavable linkers to generate retroviral vectors whose tropism could be regulated by factor Xa protease. The binding domains could not be cleaved from vector particles by factor Xa when the linker was fused to amino acid +7 of Moloney MLV SU but could be efficiently cleaved when fused to amino acid + 1 of Moloney or 4070 A MLV SU glycoproteins. Vectors displaying a cleavable EGF domain were selectively sequestered on EGF receptor-expressing cells, but their infectivity was fully restored when the EGF domain was cleaved from die vector particles widi factor Xa. Partial restoration of infectivity was observed when only a fraction of the envelope proteins were cleaved. Conversely, vectors that displayed a cleavable RAM-1 binding domain fused to Moloney MLV SU had an expanded host range that was reversible upon treatment with factor Xa. It is suggested diat retroviral vectors with engineered binding specificities whose tropism is regulated by exposure to specific proteases may facilitate novel strategies for targeting retroviral gene delivery.

Introduction, results, and discussion

MLV-derived retroviral vectors are versatile gene delivery vehicles whose host range can be varied by incorporation of different envelope spike glycoproteins (Miller, 1992 Curr. Top. Microbiol. Immunol. 158, 1; Vile & Russell, 1995 British Medical Bulletin. 51, 12; Weiss, in Retroviridae, J. Levy, Ed. (Plenum Press, 1993), pp. 1-108). Retroviral envelope spike glycoproteins mediate virus attachment to specific receptors on the target cell surface and subsequently trigger fusion between the lipid membranes of virus and host cell. The envelope spike glycoproteins of murine leukaemia viruses (MLVs) are homotrimers in which each of the three heterodimeric subunits comprises a large extraviral glycoprotein moiety (SU) attached at its C-terminus to a smaller transmembrane polypeptide (TM) that anchors the complex in the viral membrane (August et al. , 1974 Virology 60, 595; Ikeda et al. , 1975 J. Virol. 16, 53; Kamps et al., 1991 Virology 184, 687). SU consists of two domains connected by a proline-rich hinge, the N-terminal domain conferring receptor specificity and exhibiting a high degree of conservation between MLVs with different host ranges (Battini et al. , 1992 J. Virol. 66, 1468; Morgan

et al. , 1993 J. Virol. 67, 4712; Battini et al. , 1995 J. Virol. 69, 713). Moloney MLV envelopes confer an ecotropic host range because they attach selectively to a peptide loop in d e murine cationic amino acid transporter (CAT-1), found only on cells of mouse and rat origin (Albritton et al. , 1989 Cell 57, 659; Albritton et al., 1993 J. Virol. 67, 2091). 4070A MLV envelopes attach to an epitope on the ubiquitous RAM-1 phosphate symporter that is conserved diroughout many mammalian species and confer an amphotropic host range (Miller et al. , 1994 Proc. Natl. Acad. Sci. U.S.A. 91, 78; VanZeijl et al. , 1994 Proc. Natl. Acad. Sci. U.S.A. 91, 1168; Kavanaugh et al. , 1994 Proc. Natl. Acad. Sci. U.S.A. 91, 7071). Thus, retroviral vectors with 4070A envelopes infect human cells promiscuously whereas vectors with Moloney envelopes fail completely to infect human cells.

Cell-selective retroviral gene delivery has recently been achieved by engineering new binding domains into die envelope glycoproteins of retroviral vectors (Valsesia-Wittmann et al., 1994 J. Virol. 68, 4609; Kasahara et al. , 1994 Science 266, 1373; Chu & Dornburg, 1995 J Virol. 69, 2659; Nikunj et al. , 1995 Proc. Nad. Acad. Sci. USA 92, 7570; Cosset et al., 1995 J. Virol. 69, 6314). When different receptor-binding domains were displayed on MLV retroviral vectors as N-terminal extensions of their intact SU glycoproteins (Russell et al. , 1993 Nucl. Acids Res. 21, 1081), it was found diat host range could be extended or restricted by die displayed ligand (Cosset et al., 1995 J. Virol. 69, 6314). Thus, ecotropic vectors displaying a RAM-1 receptor-binding domain from 4070 A SU were able to infect RAM-1 -positive human cells whereas amphotropic vectors displaying epidermal growth factor (EGF) could bind to EGF receptors but were thereafter sequestered into a noninfectious entry pathway, giving greatly reduced titres on EGF receptor-positive cells, but normal titres on EGF receptor- negative cells. In the current study, we have explored d e possibility of generating retroviral vectors whose engineered tropism can be regulated by specific proteases.

Initially, we inserted a short factor Xa protease-sensitive linker (amino acid sequence IEGR), (Lottenberg et al. , 1981 Methods Enzymol. 80, 341), into a previously described EGF-MLV envelope chimaera (EMo7) in which the EGF domain was fused to amino acid + 7 in Moloney MLV SU by a short linker containing three alanines. The Xa cleavage

signal was inserted between the alanine linker and amino acid +7 of Moloney SU to give die construct EXMo7, described below and illustrated in Figure 1. In Figure 1, the general format for all of the constructs is shown diagrammatically and the sequence surrounding die site of fusion between the displayed ligand (EGF or N-terminal binding domain of 4070A-MLV) and die MLV envelope protein (Moloney or 4070 A) is shown in detail for each of the constructs. Beside each construct is a schematic representation of the N-terminal region of the expressed envelope glycoprotein monomer; Open circles indicate N-terminal receptor-binding domain of die (ecotropic) Moloney MLV SU glycoprotein, filled squares indicate die N-terminal receptor binding domain of die (amphotropic) 4070 A MLV SU glycoprotein, grey triangles represent EGF, and factor Xa cleavage sites are denoted with arrows. LTR: long terminal repeat, L: envelope signal peptide, p: polyadenylation sequence. The Notl cloning site is also shown.

The chimaeric envelopes and a control ecotropic (Moloney) envelope were expressed in TELCeBό cells which express MLV gag-pol core particles and an nlsLacZ retroviral vector (Cosset et al., 1995 J. Virol. 69, 7430-7436). Virus-containing supernatants from the transfected TELCeBό cells were harvested, filtered (0.45μm), digested widi 0 or 4 μg/ml factor Xa protease for 90 minutes and ultracentrifuged to pellet the viral particles. Retroviral particles incorporating chimaeric envelopes were analyzed by Western immunoblotting (Figure 2) before (-) or after (+) treatment with factor Xa protease. Lanes A, B and C were loaded widi pelleted retroviral vectors incorporating Mo, EXMol, and EXMo7 envelopes, respectively. The different envelope expression constructs were transfected (as described in Sambrook et al., Molecular cloning, A laboratory manual, (Cold Spring Habour, N.Y., 1989) pp. 16.33-16.36) into TELCeBό packaging cells and stable phleomycin (50 μg/ml) resistant colonies were expanded and pooled. Cells were grown in DMEM supplemented with 10% fetal calf serum and when confluent transferred from 37°C to 32 °C and incubated for 72 hrs. Supernatants containing retroviral particles were harvested after overnight (16 hrs) incubation in 10 mis serum-free DMEM at 32°C and filtered (0.45μm) before being incubated with 0 or 4 μg/ml of factor Xa (Promega) for 90 minutes at 37°C in the presence of 2.5 mM CaC The supernatants were centrifuged at 30 000 rpm in a SW40 rotor (Beckman) for 1 hour at 4°C and the pelleted viral particles were resuspended in lOOμl phosphate buffered saline. 20μl of each sample

was separated on a 10% polyacrylamide gel under reducing conditions (Laemmli, Nature (London, 277, 680 (1970)) followed by transfer of the proteins onto nitrocellulose paper. The SU proteins were detected as previously described (Cosset et al. , 1995 J. Virol. 69, 6314) using specific goat antibodies raised against Rausher murine leukaemia virus envelope glycoproteins (Quality Biotech Inc, USA) followed by Horseradish peroxidase-conjugated rabbit anti-goat antibodies (DAKO Denmark) and developed using an enhanced chemiluminescence kit (Amersham Life Science).

EGF was not cleaved from the EXMo7 envelope by factor Xa (Fig. 2, lane C), suggesting diat die cleavage site was not accessible to the protease when inserted in diis position.

We d erefore made new constructs, EMol and EXMol (described below), coding for chimaeric envelopes in which EGF is fused to amino acid + 1 (rather dian +7) of Moloney SU by a linker comprising 3 alanines, or 3 alanines and the IEGR factor Xa cleavage site (see Figure 1). EMol and EXMol chimaeric envelopes were incorporated into virions and analysed on immunoblots after treatment with 0 or 4 μg/ml factor Xa protease for 90 minutes. Figure 2 shows that EXMol envelopes were cleaved by factor Xa to yield an SU cleavage product whose mobility was indistinguishable from unmodified Moloney SU. Control EMol envelopes which lack the factor Xa cleavage site were not cleaved. These results indicate diat die precise positioning of the IEGR peptide in die chimaeric envelopes is important for its optimal recognition and cleavage by factor Xa.

We have previously demonstrated EGF receptor-mediated host range restriction of retroviral vectors displaying chimaeric envelopes in which EGF was fused to amino acid +5 of 4070A SU by a short (AAA) linker (Cosset et al., J. Virol. 69, 1995, cited above). Retroviruses displaying these chimaeric envelopes could bind to EGF receptors but were thereafter sequestered into a noninfectious entry pathway, giving greatly reduced titres on EGF receptor-positive cells, but near-normal titres on EGF receptor- negative cells.

We therefore constructed plasmids EA1 and EXA1 (described below) coding for chimaeric envelopes in which EGF is fused to amino acid -I- 1 of 4070A SU by AAA or AAAIEGR linkers respectively (see Figure 1).

DNA Constructs

The expression plasmids FBMoS ALF and FB4070ASALF (described by Cosset et al. , 1995 J. Virol. 69, cited above) coding for unmodified Moloney and 4070A MLV envelopes are referred to in the text as Mo and A respectively. Construction of EA, EMo7 (previously called EMO) and AMO expression plasmids was also described by Cosset et al., (cited above).

To generate EXo7, EMol and EXMol, PCR primers NotXMo7Back, NotMolBack and NotXMolBack (respectively) were used widi primer envseq7 to amplify modified envelope fragments from Mo (FBMoSALF) which were digested widi Notl and BamHI and cloned into the N_>fI/_3αmHI-digested backbone of EMo7.

To generate EA1 and EXA1, PCR primers NotAlBack and NotXAlBack (respectively) were used widi primer 4070Afor to amplify modified envelope fragments from A (FB4070ASALF) which were digested widi Notl and BamHI and cloned into die Notl/BamHl-digested backbone of EA.

Finally, the AMol and AXMol constructs (referred to below) were generated by cloning the Ndel-Notl fragment from AMO into the N_f__/N_».l-d_gested backbones of EMol and EXMol, respectively. The correctness of all constructs were confirmed by DNA sequencing.

Oliogonucleotides used (with restriction sites underline) were:

NotXMo7Back, 5'-GCA AAT CTG CGG CCG CAA TCG AGG GAA GGC CTC ATC AAG TCT ATA ATA TCA CC (Seq ID No. 1);

NotMolBack, 5' -GCA AAT CTG CGG CCG CAG CTT CGC CCG GCT CCA GTC C

(Seq ID No. 2);

NotXMolBack, 5'-GCA AAT CTG CGG CCG CAA TCG AGG GAA GGG CTT CGC CCG GCT CCA GTC C-3' (Seq ID No. 3);

NotAlBack 5'GCA AAT CTG CGG CCG CAA TGG CAG AGA GCC CCC ATC-3' (Seq ID No. 4);

NotXAlBack 5'-GCA AAT CTG CGG CCG CAA TCG AGG GAA GGA TGG CAG AGA GCC CCC ATC-3' (Seq ID No. 5);

envseq7, 5 ^* -GCC AGA ACG GGG TTT GGC C-3' (Seq ID No. 6);

4070Afor, 5'-CTG CAA GCC CAC ATT GTT CC-3' (Seq ID No. 7).

Figure 3 shows diat the IEGR sequence in die interdomain linker of the expressed EXA1 envelopes was correctly recognized and cleaved by factor Xa whereas there was no cleavage of control EAl envelopes. Referring to Figure 3 (an immunoblot of the recombinant amphotropic retroviral particles before (-) or after (+) treatment with factor Xa protease): lanes A, B and C were loaded widi pelleted retroviral vectors incorporating A, EAl and EXA1 envelopes, respectively. The analysis was performed as described above for Figure 2.

We dien titrated vectors incorporating EAl and EXA1 chimaeric envelopes on EGF receptor-negative and EGF receptor-positive human cell lines as follows: EGF receptor-expressing cell lines A431 (ATCC CRL1555), HT1080 (ATCC CCL121), and EJ (Bubenik, et al., 1973 Int. J. Cancer 11, 765) were grown in DMEM supplemented with 10% fetal calf serum (Gibco-BRL) at 37°C in an atmosphere of 5 % CO.. Jurkat T cells (ATCC CRL8805) were grown in RPMI supplemented widi 10% fetal calf serum at 37°C in an atmosphere of 5% CO ₂ . For infections, target cells were seeded at 2 x 10 ⁵ cells/well in six-well plates and incubated at 37 °C overnight. Producer cell supernatants containing _/ 3-galactosidase-transducing retroviruses were filtered (0.45 μm) after overnight incubation at 32 ^C C in serum free medium. Supernatant dilutions in 2.5 ml serum-free medium were incubated with target cells for 2 hours in die presence of 8 μg/ml polybrene. The retroviral supernatant was then removed and the cells were incubated with regular medium for 48-72 hours. X-Gal staining for detection of j3-galactosidase activity was performed as previously described (Takeuchi et al. , 1994 J. Virol. 68, 8001). Viral titre

(enzyme forming units/ml) was calculated by counting blue stained colonies microscopically wid die use of a grid place underneath the 6 well plates.

Botii vectors incorporating EAl or EXAl envelopes could infect EGF receptor-negative Jurkat cells but were selectively sequestered on EGF receptor-expressing human cells, although EXAl was sequestered less completely than EAl (Table 1). When soluble EGF was added as competitor to prevent the vectors from binding to EGF receptors their infectivity on EGF receptor positive cells could be fully restored (Table 1), confirming that sequestration was mediated specifically dirough binding of the engineered envelopes to EGF receptors.

We then tested whedier die restricted host range conferred by EXA 1 envelopes could be extended (i.e. revert to amphotropic) upon cleavage of die displayed EGF domain. Vectors incorporating EAl or EXAl envelopes were treated widi increasing doses of factor Xa and titrated on EGF receptor-expressing A431 cells (Table 2). Complete cleavage of the fused EGF domain widi 4 μg/ml factor Xa for 90 minutes (Figure 3) completely restored the infectivity of vectors with EXAl envelopes but had no effect on the infectivity of vectors carrying EAl envelopes. Partial restoration of vector titre was seen at lower concentrations of factor Xa indicating diat the vector particles could recover a low level of infectivity when only a fraction of their envelope proteins were cleaved. These data provide further evidence d at retroviral vectors displaying EGF are competitively sequestered by EGF receptors, and show diat dieir tropism can be regulated by a specific protease that cleaves die EGF domain from the viral surface.

Factor Xa protease is capable of binding directly to procoagulant phospholipid on die surface of an enveloped virus (Pryzdial & Wright, 1994 Blood 84, 3749-3757) and might therefore go on to become stably associated widi phospholipid in the engineered vector particles after cleaving their EXAl envelopes. A control experiment was therefore performed to confirm that the restoration of infectivity of vectors incorporating EXAl envelopes on A431 cells was due to cleavage of EGF, and not mediated by panicle-associated factor Xa protease. We therefore constructed plasmid AXMol and a control plasmid AMol (described above), coding for chimaeric envelopes in which the

RAM-1 receptor binding domain is fused to amino acid + 1 of Moloney SU by a factor Xa protease-cleavable (AAAIEGR) or non-cleavable (AAA) linker (Figure 1). As expected, vectors incorporating AMol and AXMol chimaeric envelopes could bind to RAM-1 allowing targeted infection of a variety of human cell lines, and die IEGR sequence in die interdomain linker of the AXMol envelope was correctly recognized and cleaved by factor Xa (data not shown). We dierefore tested whedier the extended host range conferred by AXMol envelopes could be restricted (i.e. revert to ecotropic) upon cleavage of the displayed 4070A domain. Treatment with 4 μg/ml factor Xa for 90 minutes selectively destroyed die infectivity of vectors with AXMol envelopes on human cells but did not reduce dieir infectivity on mouse cells (Table 2). Vectors carrying .AMol envelopes were unaffected by die protease treatment. These data confirm that the extended tropism of retroviral vectors displaying AMo envelopes is due to d e displayed 4070A domain and show diat die restoration of infectivity of vectors incorporating EXAl envelopes on A431 cells was due to cleavage of EGF, and not mediated by particle-associated factor Xa protease.

In summary, we have generated retroviral vectors displaying cleavable binding domains diat are anchored to die viral envelope glycoprotein by a linker that acts as a substrate for factor Xa protease. The displayed binding domains confer novel host range properties upon the vectors and these host range alterations are reversible upon treating the vectors wid factor Xa. In principle it should be possible to use linkers that are substrates for proteases other dian factor Xa in conjunction with binding domains at recognise cell

Table 1.

Infection of human cell lines with retroviral vector particles displaying EGF-4070A chimaeric envelopes.

* EGF-receptor status determined by FACS analysis.

t + indicates incubation of cells with retroviral vectors in the presence of 1 μM human EGF (R&D systems, UK).

Table 2. Regulation of vector tropism by factor Xa protease.

* Filtered supernatants containing /J-galactosidase-transducing retroviruses were preincubated with various concentrations (0, 0.0156, 0.25 and 4 μg/ml) of factor Xa (Promega) for 90 minutes at 37°C with 2.5 mM added CaCl ₂ . The treated supernatants were then added to the target cells and the viral titres were determined as described in (12).

surface receptors other than those described above. Retroviral vectors with engineered binding specificity whose tropism is regulated by exposure to specific proteases may facilitate novel strategies for targeting retroviral gene delivery.

Example 2

The inventors sought to establish whether vectors incorporating EXAl envelopes would recover dieir infectivity on EGF receptor-positive cells upon cleavage of dieir displayed EGF domain. Vectors incorporating EAl or EXAl envelopes were therefore treated widi factor Xa and titrated on EGF receptor-expressing A431 cells. Complete cleavage of the fused EGF domain with 4 μg/ml factor Xa for 90 minutes completely restored die infectivity of vectors with EXAl envelopes but had no effect on the infectivity of vectors carrying EAl envelopes (Figure 4). Figure 4 illustrates factor Xa-mediated infection of A431 cells with chimaeric EGF-4070A MLV vector particles. Filtered supernatants containing /3-galactosidase-transducing retroviruses (A, EAl, or EXAl) were preincubated with 0 (-) or 4 (+) μg/ml concentrations of factor Xa (Promega) for 90 minutes at 37°C with 2.5 mM added CaCl ₂ . The treated supernatants were then used for target cell transduction, as described above. X-gal-stained plates were photographed widiout magnification.

Partial restoration of vector titre was seen a lower concentrations of factor Xa (Table 2) indicating diat die vector particles could recover a low level of infectivity when only a fraction of their envelope proteins were cleaved. These data provide further evidence diat retroviral vectors displaying EGF are competitively sequestered by EGF receptors, and show diat dieir tropism can be regulated by a specific protease that cleaves die EGF domain from the viral surface.

In the two step targeting strategy outlined above, cleavage of the chimaeric envelope is preceded by its attachment to the target cell via the engineered binding domain. Therefore, to determine whedier EGF receptor-bound vector particles that were cleaved at die cell surface could go on to infect their target cells, we loaded die vectors onto EGF receptor-positive A431 cells and EJ cells, washed die cells, and then treated diem widi factor Xa protease. Table 3 shows that, when sequestered onto EGF receptors and dien

cleaved by factor Xa protease, die vectors incorporating EXAl, but not EAl envelopes, proceeded to infect their target cells.

The availability of a targetable, iηjectable vector would greatly facilitate the development of gene tiierapy approaches requiring direct in vivo gene delivery to selected target tissues. In this report we have demonstrated the feasibility of a novel two step-targeting strategy which may allow the generation of retroviral vectors engineered to infect cells expressing specific receptor/protease combinations. There are many membrane-associated proteases

Table 3. Factor Xa protease triggering retroviral infection on the cell-surface of human A431 and EJ cells.

Titre (CFU/ml) against cell line:

A431 A431 EJ EJ

FXa (μg ml): ^* 0 4 0 4

Constructs*

A ψ 3 x 10 ⁴ 3 x 10 ⁴ 1 x 10 ⁴ 1 x 10 ⁴

EA1 ψ <5 <5 0 0

EXA1 ψ 4 x 102 1 x 1O ⁴ <5 3 x 102

^* A431 and EJ cells were incubated with 2 ml of filtered superna¬ tant containing β-galactosidase-transducing retroviruses for 1 hr at 4°C. Cells were then washed two times with cold serum-free medium and incubated with 0 or 4 μg/ml of factor Xa (Promega) for 2 hrs at 37°C in serum-free medium. After incubation for 48 hrs with medium supplemented with 10% fetal calf serum the viral titres were determined as described in Table 1.

that may be of interest in this respect such as the proteases that co-operate in degrading die extracellular matrix during tumour invasion (Poustis-Delpont et al., 1992 Cancer Research 52, 3622-3628; Vassalli & Pepper, 1994 Nature 370, 14-15; Sato et al. , 1994 Nature 370, 61-65; and Chen et al., 1995 Breast Cancer Res. Treat. 31, 217-226); haematopoietic differentiation antigens that are also membrane proteases (Shipp & Look, 1993 Blood 82, 1058-1070) or the membrane protease that has been implicated in the entry pathway of HIV (Murakami et al., 1991 Biochim. Biophys. Acta 1079, 79-284).

Libraries of filamentous "substrate phage" displaying cleavable binding domains have recently been used to identify optimal substrates for known proteases (Matthews & Wells, 1993 Science 260, 1113-1117; Matthews e al., 1994 Protein Science 3, 1197-1205; and Smith et al. , 1995 J. Biol. Chem. 270, 6440-6449). In principle, it should be possible to generate similar retrovirus display libraries expressing N-terminally extended envelopes with randomised linker sequences. Such libraries might provide the basis for selection strategies designed to identify novel intracellular or membrane-associated proteases or to isolate chimaeric envelopes that target novel cell-specific receptor-protease combinations.

Example 3

Summary

As described above, several polypeptides have now been displayed on retroviral vector particles as N-terminal extensions of their envelope spike glycoproteins. Folding, assembly, transport, viral incorporation, receptor attachment and fusion triggering by the chimaeric envelopes can be variably influenced by die N-terminal polypeptides, depending on dieir unique structural and functional characteristics. In this example the inventors demonstrate that die RAM-1 binding domain from the homotrimeric 4070A SU glycoprotein can strongly inhibit Rec-1 mediated infection by die homotrimeric Moloney SU glycoprotein when grafted to its N-terminus. It is also shown that short trimeric leucine zipper peptides, but not a monomeric helical peptide, can inhibit RAM-1 mediated infection by the 4070A envelope when fused to its N-terminus. Cleavage signals were engineered into the chimaeric envelopes such that the displayed polypeptides could be

cleaved from the vector panicles by addition of factor Xa protease. In all of the envelopes displaying trimeric polypeptides, the steric block to Rec-1 or Ram-1 mediated infection was reversed when the trimeric N-terminal extensions were cleaved from the virally incorporated envelopes. These data suggest that the masking of envelope functions by the inhibitory N-terminal extensions is a consequence of their assembly into a trimeric complex at the tip of d e SU glycoprotein trimer to which they were grafted. The implications for retroviral vector targeting are discussed.

MLV-derived retroviral vectors are versatile gene delivery vehicles whose host range properties are determined by membrane glycoproteins which mediate dieir attachment to specific receptors and subsequently trigger fusion. The envelope glycoproteins of the murine leukaemia virus (MLV) are displayed as a homotrimeric complex on the surface of the virus (Fass et al. , Nature Structural Biology 5:465-469; Kamps et al., Virology 754:687-694). Each subunit of the trimer consists of two parts, SU and TM. The SU (surface) component is entirely extraviral and is attached to die retrovirus via die smaller TM component, which anchors the complex in die viral membrane (Pinter et al., Virology 9 :345-351). The N-teπninal domain of the SU glycoprotein confers receptor specificity and exhibits a high degree of conservation between MLVs with different host ranges (Battini et al., J. Virol. 69:713-719). Moloney MLV envelopes confer an ecotropic host range because they bind to a murine cationic amino acid transporter (Albritton et al., J. Virol. 67:2091-2096; Albritton et al. , Cell 57:659-666). 4070A MLV envelopes attach to the RAM-1 phosphate transporter which is conserved diroughout many mammalian species, to confer an amphotropic host range (Kavanaugh et al. , Proc. Nad. Acad. Sci. USA 97:7071-7075). After binding to target cell receptors has occurred, die trimeric SU-TM complex is thought to undergo a large conformational rearrangement which triggers the process of fusion between the viral and target cell membranes.

The inventors and dieir colleagues have been exploring different strategies for targeting the entry of retroviral vectors into selected target cells by engineering new determinants into their SU glycoproteins (Cosset et al., J. Virol. 69:6314-6322; Nilson et al., Gene Ther. 5:280-286; Russell et al. , Cold Spring Harbour Laboratory, Cold Spring Harbour, N.Y.; Valsesia-Wittmann et al. , J. Virol. 65:4609-4619; Valsesia-Wittmann et al., J

Virol. 70:2059-2064). In the preceding examples is described a novel two-step strategy that allows die targeting of retroviral vectors through protease-substrate interactions, in which the retroviral vector attaches to the target cell via an engineered binding domain (step one), whereupon the engineered linker that tethers the virus to the binding domain is cleaved by a specific protease (step two), allowing the virus to go on and infect die target cell. A disadvantage of diis two-step targeting strategy, as set forth above, is diat whilst dominating d e specificity of vector attachment, the cleavable binding domain does not completely block the ability of the SU trimer to attach to its natural receptor on non target cells. The uncleaved vector therefore retains the ability to infect non target cells through die Ram-1 receptor. To overcome this disadvantage, we were interested to develop envelope modifications diat would completely inhibit die infectivity of uncleaved vectors but would permit full restoration of infectivity upon exposure to a selected protease.

In the course of experiments (described in this specification) to characterise the ecotropic infectivity of Moloney MLV envelope chimaeras, it was found that a vector displaying die Ram-1 targeted AXMol envelope (Nilson et al., Gene Ther. 5:280-286), could not efficiently infect cells through the ecotropic receptor (Rec-1) unless it was first cleaved by factor Xa protease. To explain this property of the AXMol envelope we hypothesised diat die displayed RAM-1 binding domain might be forming a trimeric complex at the tip of the Moloney SU glycoprotein trimer to which it was grafted, thereby blocking its Rec-1 binding site. To further test this hypothesis, we grafted oligomerising leucine zipper peptides (Harbury et al., Science 262:1401-1407) onto 4070A SU glycoproteins and characterised die properties of retroviral vectors incorporating the chimaeric envelopes.

MATERIALS AND METHODS

Plasmid Construction

The unmodified envelopes of 4070A MLV and Moloney MLV were encoded by d e expression plasmids FB4070ASALF (A) and FBMoSALF (Mo), respectively (Cosset et al., 1995 J. Virol. 69, 7430-7436)). The constructs AMol and AXMol , which code for chimaeric envelopes in which the RAM-1 receptor binding domain from 4070A SU is

fused to amino acid + 1 of Moloney SU by a factor Xa protease-cleavable (AAAIEGR) or non-cleavable (AAA) linker have been described previously (Nilson et al. , Gene Ther. 5:280-286). Constructs EAl and EXAl, coding for chimaeric envelopes in which EGF is fused to amino acid + 1 of 4070A SU by a linker comprising three alanines, or three alanines and die IEGR factor Xa cleavage site, have also been described (Nilson et al., Gene Ther. 5:280-286).

To construct vectors displaying helical peptides, plasmids pEGSlXAl and pEGS3XAl were first produced in which there is a 12 amino acid (AAAGGGGSIEGR, Seq ID No. 8) or 22 amino acid (AAAGGGGSGGGGSGGGGSIEGR, Seq ID No. 9) linker, respectively, between the 4070A MLV envelope and die displayed EGF domain. PCR primers NotGSlXAlback and NotGS3XAlback (respectively) were used widi primer 4070Afor to amplify modified envelope fragments from EXAl which were digested with Notl and BamHI and cloned into die Notl/BamHI-digested backbone of EAl.

Figure 7 is a diagramatic representation of plasmid constructs coding for chimaeric envelope glycoproteins in which the helical peptides AA, VL and II were fused to residue + 1 of the 4070A MLV SU. The general format is shown diagramatically and die amino acid sequence (single letter code) of die helical peptides and the linkers between these peptides and die SU protein are shown in detail. LTR, long terminal repeat; L, envelope signal peptide. Amino acid residues at die a and d positions of the heptad repeat are shown in bold.

To generate plasmids pVLXAl, pVLGSlXAl and ρVLGS3XAl, PCR primers Gal4 VLback and Gal4 VLfor were used to produce PCR fragments by priming off each other and dien outer primers Gal4back and Gatøfor were used to amplify the fragment further. The PCR products were digested widi Sfil and Notl and cloned into the Sfil/Notl-digested backbones of EXAl, pEGSlXAl and pEGS3XAl.

To generate plasmids pAAXAl and pAAGS3XAl, PCR primers Gal4 AAback and Gal4 AAfor were used to produce PCR fragments by priming off each other and dien outer primers Gal4back and Gal4for were used to amplify the fragments further. The PCR

products were digested widi Sfil and Notl and cloned into die Sfil/Notl-digested backbones of EXAl and pEGS3XAl .

To generate plasmids pIIXAl , pIIGS 1XA1 and pIIGS3XAl , PCR primers Gal4 Ilback and Gal4 Ilfor were used to produce PCR fragments by priming off each other and then outer primers Gal4back and Gal4for were used to amplify die fragments further. The PCR products were digested widi Sfil and Notl and cloned into d e 5/z7/Nσr7-digested backbones of EXAl, pEGSlXAl and pEGS3XAl. The correct sequence of all constructs was verified by DNA sequencing.

The following oligonucleotides (with restriction sites underlined) were used :

NotGSlXAlback, 5'-GCA AAT CTG CGG CCG CAG GTG GAG GCG GTT CAA TCG AGG GAA GGA TGG CAG AG-3' (Seq ID No. 10);

NotGS3XAlback, 5'-GCA AAT CTG CGG CCG CAG GTG GAG GCG GTT CAG GCG GAG GTG GCT CTG GCG GTG GCG GAT CGA TCG AGG GAA GAA TGG CAG AG-3' (Seq ID No. 11);

Gal4 VLback (containing Sfil site), 5'-GGC ATT CAT GCG GCC GCG GCC CAG CCG GCC ATG AAG CAA CTA GAA GAC AAG GTG GAG GAA CTC CTT AGC AAG GTA TAC C-3' (Seq ID No. 12);

Gal4 VLfor (containing Notl site), 5'-GCA AAT CTG CGG CCG CCT CTC CAA CAA GCT TCT TCA GTC GAG CGA CTT CGT TCT CAA GAT GGT ATA CCT TGC TAA GGA G-3' (Seq ID No. 13);

Gal4 AAback (containing Sfil site), 5'-GGC ATT CAT GCG GCC GCG GCC CAG CCG GCC ATG AAG CAA GCA GAA GAC AAG GCA GAG GAA GCT CTT AGC AAG GCT TAC C-3' (Seq ID No. 14);

Gal4 AAfor (containing Notl site), 5'-GCA AAT CTG CGG CCG CCT CTC CAG CAA

GCT TCT TTG CTC GAG CAG CTT CGT TCT CTG CAT GGT AAG CCT TGC TAA GAG C-3' (Seq ID No. 15);

Gal4 Ilback (containing Sfil site), 5'-GGC ATT CAT GCG GCC GCG GCC CAG CCG GCC ATG AAG CAA ATC GAA GAC AAG ATA GAG GAA ATT CTT AGC AAG ATC TAC C-3' (Seq ID No. 16);

Gal4 Ilfor (containing Notl site), 5'-GCA AAT CTG CGG CCG CCT CTC CTA TAA GCT TCT TGA TTC GAG CAA TTT CGT TCT CTA TAT GGT AGA TCT TGC TAA GAA TTT C-3' (Seq ID No. 17);

Gal4 back, 5'-GGC ATT CAT GCG GCC GCG GC-3' (Seq ID No. 18);

Gal4 for, 5'-GCA AAT CTG CGG CCG CCT CTC-3' (Seq ID No. 19); and 4070Afor (described above).

Target cell lines and production of viruses

GP+Env AM12 cells (Markowitz et al., Virology 767:400-406) were derived from the murine cell line NIH 3T3 and express the MLV- A envelope which blocks the RAM-1 receptor by interference. NIH 3T3, GP+Env AM 12 and the human cell line A431 (Giard et al., J. Natl. Cancer Inst. 57, 1417-1421), were grown in DMEM supplemented widi 10% fetal calf serum. The different envelope expression constructs were transfected into TELCeBό packaging cells (Cosset et al., J. Virol. 69:7430-7436) by calcium phosphate precipitation (Takeuchi et al., J. Virol. 65:8001-8007) and stable phleomycin (50mg/ml) resistant colonies were expanded and pooled. Cells were grown in DMEM supplemented widi 10% fetal calf serum and when confluent transferred from 37 ^* C to 32 ^* C and incubated for 72hrs. Supernatants containing retroviral particles were harvested after overnight (lόhrs) incubation at 32 ^* C in lOmls serum-free DMEM for infections, or DMEM supplemented with 2% fetal calf serum for immunoblots. All supernatants were filtered (0.45μm) before use.

Immunoblots

Virus producer cells were lysed in a 20mM Tris-HCl buffer (pH 7.5) containing 1 % Triton X-100, 0.05% SDS, 5mg/ml sodium deoxycholate, 150mM NaCl and ImM PMSF. Lysates were incubated for 10 mins at 4 ^' C and were centrifuged for 10 mins at 10,000 x g to pellet die nuclei. Virus samples were obtained by ultracentrifugation of filtered viral supernatants (10ml) at 30 000 φm in a SW40 rotor (Beckman, USA) for 1 hr at 4 ^* C. The pelleted viral particles were resuspended in lOOμl PBS. Samples (30μl for cell lysates, or lOμl for pelleted virions) were then separated on a 10% polyacrylamide gel under reducing conditions followed by transfer of the proteins onto nitrocellulose paper. For Factor Xa cleavage, lOμl of the pelleted viral particles were incubated with 0 or 4 μg/ml of Factor Xa (Promega, USA) for 90 min at 37'C in the presence of 2.5mM CaCl, before running on the separating gel. The SU proteins were detected as previously described (Cosset et al. , J. Virol. 69:6314-6322) using specific goat antibodies raised against either Rausher leukaemia virus (RLV) gp70 SU or RLV p30 capsid protein (CA) (Quality Biotech Inc, USA) which were diluted 1/1 ,000 and 1/10,000 respectively. Blots were developed with horseradish peroxidase-conjugated rabbit anti-goat antibodies (DAKO, Denmark) and an enhanced chemiluminescence kit (Amersham Life Science, UK).

Target cell Infection

Target cells were seeded at 2 x 10 ⁵ cells/well in six- well plates and incubated at 37 ^* C overnight. Producer cell supernatants containing /3-galactosidase-transducing retroviruses were filtered (0.45μm) after overnight incubation at 32 "C in serum-free medium. The harvested supernatants were incubated with 0 or 4 μg/ml of factor Xa (Promega) for 90 minutes at 37 ' C in the presence of 2.5mM CaCl,. Supernatant dilutions in 2ml serum-free media were incubated with target cells for 6 hrs in the presence of 8μg/ml polybrene. The retroviral supernatant was then removed and the cells were incubated widi regular medium for 48-72 hrs. X-Gal staining for detection of /3-galactosidase activity was performed as previously described (Tatu et al. , EMBO J. 74: 1340-1348). Viral titre (enzyme forming units/ml) was calculated by counting blue stained colonies microscopically with die use of a grid placed underneath the 6 well plates.

RESULTS

Rec-1 mediated infection by envelopes expressing a Ram-1 targeting domain

AMol and AXMol are previously described chimaeric envelopes in which the RAM-1 receptor binding domain from 4070A SU is fused to aminoacid + 1 of Moloney SU by a noncleavable (AAA) or factor Xa-cleavable (AAAIEGR) linker (Nilson et al., Gene Ther. 5:280-286). Viruses incoφorating the AMol and AXMol envelopes were pelleted, cleaved widi 0 or 4μg/ml factor Xa protease and dien analysed on immunoblots using an anti-envelope antiserum as a probe.

The reversible inhibition of infection by retroviral incoφoration and cleavage of chimaeric envelopes expressing a factor Xa-cleavable, N-terminal RAM-1 binding domain is shown in Figure 6 A and 6B. Figure 6 A is an immunoblot of pelleted recombinant retroviral particles incoφorating Mo, AMol or AXMol envelopes before (-) or after (+) treatment with factor Xa protease, probed widi antiserum to the SU glycoprotein. Figure 6B shows die results when the target cell line GP+Env AM12 was infected with harvested producer cell supernatants containing /3-galactosidase-transducing retroviruses (AMol , AXMol , Mo and A) widi or without treatment with factor Xa protease. Detection of /3-galactosidase activity was performed by X-gal staining and titres were expressed as e.f.u./ml.

It is apparent from Figure 6A that the chimaeric envelopes were incoφorated into virions with equal efficiency (although less efficiently dian d e unmodified Moloney SU) and that AXMol , but not AMol envelopes, were cleaved by factor Xa protease to yield an SU cleavage product whose mobility was indistinguishable from unmodified Mo SU.

The infectivity of these Ram-1 targeted vectors was then tested on NIH3T3 cells and on NIH3T3 transfectants (GP+Env AM 12) overexpressing the 4070A envelope which blocks the conesponding Ram-1 receptor by interference. The vectors AMol and AXMol were fully infectious on the unmodified NIH3T3 cells which express both Rec-1 and Ram-1, giving titres in excess of 10 ⁶ efu per ml (not shown), however their infectivity was greatly reduced on the Ram-1 deficient cells, suggesting that they were unable to utilise the ecotropic receptor, Rec-1 (Fig. 6B). This result was unexpected and was in contrast to results obtained wid similar chimaeric Moloney SU glycoproteins displaying monomeric growth factor domains or single chain antibody fragments in which Rec-1 mediated

infection was not seriously compromised by the displayed domains (Ager et al., Human Gene Ther. , in press; Cosset et al. , J. Virol. 69:6314-6322). This led to the proposal that the displayed Ram-1 binding domain might be forming a trimeric complex at the tip of die Moloney SU glycoprotein trimer to which it was grafted, thereby blocking the Rec-1 binding site and/or interfering with Rec-1 mediated fusion triggering. Such a block would be expected to be reversible by cleaving the Ram-1 binding domain from the vector and, in keeping with this prediction, die infectivity of the AXMol vector was fully restored on Rec-1 positive, Ram-1 deficient cells when die Ram-1 targeting domain was cleaved from its surface with factor Xa protease (Fig. 6B).

Construction of chimaeric 4070A envelopes displaying helical peptides

To further test the idea that a trimeric polypeptide could block die functions of a trimeric envelope glycoprotein when fused to its N-terminus, and to determine whedier die concept could be applied to an amphotropic MLV SU glycoprotein, we made a series of constructs coding for chimaeric envelopes in which monomeric or trimerising helical peptides were fused to amino acid + 1 of 4070A SU via Factor Xa-cleavable linkers (Fig. 7). The helical peptides diat were chosen for diese studies were variants of the dimeric GCN4 leucine zipper peptide widi systematic V, L, I or A (single letter aminoacid code) substitutions in the a and d positions of the heptad repeat that are known to force the formation of trimeric coiled coils (VL and II peptides) or to prevent oligomerisation (AA peptide) (Harbury, et al., Science 262: 1401-1407). When designing these constructs, we were concerned diat die oligomerisation of the displayed VL and II peptides might be hindered if they were tethered too closely to the underlying 4070A SU glycoprotein. The spacing between the 4070A SU glycoprotein and the displayed peptide motifs was dierefore varied by insertion of linkers comprising amino acids AAAIEGR, Seq ID No. 20), AAAGGGGSIEGR (Seq ID No. 8) or AAAGGGGSGGGGSGGGGSEEGR (Seq ID No. 9), where the highlighted sequence is known to be recognised and cleaved by Factor Xa (Nilson et al., Gene Ther. 5:280-286).

Expression, viral incorporation and cleavage of chimaeric 4070A envelopes

The AA, VL and II chimaeric envelopes and a control amphotropic (4070A) envelope were stably transfected into TELCeBό cells which express MLV gag-pol core particles and

an nls LacZ retroviral vector (Cosset et al., J. Virol. 69:7430-7436). Virus-containing supernatants were harvested from these stably transfected TELCeBό cells and ultracentrifuged to pellet the viral particles. Pellets were than analysed on immunoblots for the presence of viral core proteins and envelope proteins (Fig. 8A).

Figure 8 illustrates the viral incoφoration and cleavage of chimaeric envelopes expressing factor Xa-cleavable helical peptides as N-terminal extensions of the 4070 A MLV SU. Figure 8A is an immunoblot of pelleted retroviral particles incoφorating chimaeric envelopes. The lane contents are as follows: 1 :VLXA1, 2:VLGS1XA1, 3:VLGS3XA1, 4.AAXA1, 5:AAGS3XA1, 6:IIXA1, 7:IIGS1XA1, 8:IIGS3XA1, and 9:A. The top immunoblot was probed widi an anti-SU antiserum and die lower one widi an anti-p30 antiserum to detect d e p30 CA protein.

Figure 8B shows die Factor Xa-mediated cleavage of chimaeric envelopes and takes the form of an immunoblot of pelleted recombinant amphotropic retroviral particles incoφorating A, VLXA1, AAXA1 , IIXA1 or EXAl envelopes before (-) or after (+) treatment with factor Xa protease, probed widi anti-SU antiserum.

Figure 8C is an immunoblot of cell lysates prepared from the virus producing TELCeBό transfectants A, VLXA1, AAXA1, IIXA1 and the control, untransfected TELCeBό, probed widi anti-SU antiserum.

The number of vector particles present in each sample, determined by staining widi p30 antiserum to detect the p30 CA protein, was found to be approximately equivalent (Fig. 8A). However, when the efficiencies of viral incoφoration of the different chimaeric envelopes were compared, by staining widi an anti-SU antiserum, it was found diat incoφoration is greatly influenced by die presence of the oligomerizing peptide. Envelopes displaying die control monomeric peptide (AA) were incoφorated almost as efficiently as wild type 4070A envelopes whereas envelopes displaying die VL peptide were incoφorated much less efficiently and tiiere was no visible incoφoration of envelopes displaying die II peptide. To determine if the helical peptides could be cleaved from the SU glycoproteins to which they were grafted, viral pellets were digested widi 0

or 4 μg/ml factor Xa protease and then analysed on immunoblots as before.

Figure 8B shows that there is a mobility shift when expressed envelopes VLXA1, AAXA1 and die control EXAl, have been cleaved widi factor Xa protease, indicating diat die helical peptides are indeed cleaved from the SU. Due to d e low levels of incoφoration of the IIXAl chimaeric envelope, cleavage can not be seen for this vector. This immunoblot also indicates diat d e chimaeric envelope AAXA1 was incoφorated 10 times more efficiently than VLXA1.

To further investigate the poor incoφoration of the VL and II chimaeric envelopes we performed immunoblots of cell lysates prepared from the virus producing TELCeBό transfectants. Figure 8C shows that d e unprocessed precursors of all three chimaeric envelopes are detectable in die cell lysates. However, the VL and II envelope precursors are less abundant dian the AA precursor. Also, die processing of the VL and II precursors to mature SU is severely impaired relative to die processing of the AA precursor indicating diat diese chimaeric envelopes are not efficiently transported from the endoplasmic reticulum to die Golgi compartment.

Infectivity of vectors displaying chimaeric envelopes before and after cleavage

To determine whedier die helical peptides were masking the functions of the 4070A envelopes to which they were fused we titrated the vectors on Ram-1 expressing cells, NIH3T3 and A431 before and after they were cleaved widi factor Xa protease (Table 4 and Figure 9). Figure 9 shows the reversible inhibition of infection by cleavage of the chimaeric envelope, VLXA1 , expressing a factor Xa-cleavable, N-terrninal oligomerizing peptide and is a magnified view of virally infected cells after X-gal staining. Chimaeric envelope VLXA1 shows strong inhibition of infectivity on NIH 3T3 and A431 cells, which is reversible on addition of factor Xa.

The control vectors displaying the AA peptide gave titres comparable to that of the wild type amphotropic vector and the titres did not change after factor Xa cleavage indicating diat die AA peptide does not significantly interfere with die functions of the underlying 4070A envelope. Conversely, the vectors displaying die trimerising VL and II helical

peptides gave greatly reduced titres on both cell lines which were enhanced as much as 2000-fold by factor Xa cleavage. In the case of die VLXAl chimaeric envelope, on cleavage with 4 μg/ml factor Xa protease, the titre on NIH3T3 cells increased from 151 efu/ml to 3 x 10 ^s era/ml and on A431 cells from 318 efu ml to 10 ⁵ efu ml (Fig. 9). Vectors displaying die II helical peptide gave generally lower titres dian diose displaying die VL peptide, presumably due to die reduced incoφoration of chimaeric envelopes displaying the II peptide. Interdomain spacing had litde apparent influence on die titre of the uncleaved vectors, nor on die degree of titre enhancement that was observed after exposure to the factor Xa protease.

(0

C O Titre (e.f.u./ml) of harvested JS-galactosidase-transducing retroviruses ω

H

H

C AAXA1 AAGS3XA1 VLXAl VLGS1XA1 VLGS3XA1 I IXAl I IGS1XA1 I IGS3XA1 H m co x Factor Xa ^a + - + — + m m

N I H3T3 10 ⁷ 10 ⁷ 2x10° 2x10 ⁶ 2x10 ⁶ 2x10 ⁶ 151 3x10 ⁵ 5x10 ³ 2x10^ 2x10**3x10 ⁵ 6 10 ³ 34 10 ³ 10 ³ βxlO ³

m A 31 10 ⁷ 10 ⁷ 2xl0 ⁶ 2x10 ⁶ 2x10 ⁶ 2x10 ⁶ 318 10 ⁵ 10 ³ 5x10 10 ³ 3x10 ⁵ 10 10 ³ 16 6x10 ² 162 )0

IN) CO ^a Harvested producer cell supernatants containing jS-galactosidase-transducing retroviruses were preincubated with (-) or without (+) factor Xa protease

Reversible inhibition of infection by cleavage of chimaeric envelopes expressing a factor Xa-cleavable, N-temiinal oligomerizing peptide.

TABLE 4.

DISCUSSION

In the above example we have shown that the Ram-1 binding domain from the homotrimeric 4070A SU glycoprotein can inhibit Rec-1 mediated infection by the homotrimeric Moloney SU glycoprotein when grafted to its N-terminus. We have also shown that short trimeric leucine zipper peptides, but not a monomeric helical peptide, can inhibit Ram-1 mediated infection by the 4070A envelope when fused to its N-terminus. In both cases, by using factor Xa protease to cleave the trimeric N-terminal extensions from the virally incoφorated envelopes, it was possible to reverse the block to Rec-1 or Ram-1 mediated infection. We propose that the masking of envelope functions by these inhibitory N-terminal extensions is a consequence of their assembly into a trimeric complex at the tip of the SU glycoprotein trimer to which they are grafted.

The VL, II and AA peptides diat we fused to the 4070A envelope are mutants of the GCN4 leucine zipper in which the conserved, buried residues diat direct dimer formation have been substituted widi valine, leucine, isoleucine or alanine residues (Harbury et al., Science 262: 1401-1407). The VL mutant oligomerises to form extremely stable (T _m 95 ^* C) two- and three-stranded alpha-helical coiled coil structures whereas the II mutant forms exclusively three-stranded coiled coils which are even more stable (T _ra > 100 ^* C) than the VL structures. In the AA peptide, all of the hydrophobic core residues of the GCN4 leucine zipper were substituted widi alanines to prevent oligomerisation of the mutant peptide whilst preserving its helical structure.

Retroviral incoφoration of chimaeric envelopes displaying die VL and II peptides was significandy impaired relative to chimaeric envelopes displaying die control AA peptide, which showed only a slight reduction in incoφoration compared to unmodified 4070A envelopes. The VL chimaeric envelopes were approximately ten-fold less abundant in viral pellets than the AA chimaeric envelopes, and the II chimaeric envelopes were so poorly incoφorated that they were not visible on immunoblots of pelleted virions. By immunoblotting cell lysates from the transfected TELCeBό cells with anti-envelope antiserum, it was shown that the intracellular abundance of die precursor polypeptides for each of die chimaeric envelopes was closely correlated with their abundance in viral pellets. The low viral incoφoration of die VL and II chimaeric envelopes is therefore a

consequence of their poor expression and/or folding in the virus producing cells.

It is currently unclear what is responsible for he impaired expression of the VL and II chimaeric envelopes. Neither protein appears to be toxic to the virus producing cells since ti ere was no difference in die number or size of stably transduced TELCeBό clones diat were obtained after transfecting the different (oligomerising or control) chimaeric envelope expression plasmids (data not shown). An alternative possibility might be that die low intracellular abundance of the VL and II envelope precursors is due to dieir premature oligomerisation in the endoplasmic reticulum. Premature oligomerisation of the nascent polypeptide chains via their N-terminal VL or II peptides might seriously compromise the folding of individual subunits leading to their aggregation and accelerated proteolytic destruction. In keeping with this idea, in a related system d e chaperone-guided folding of influenza haemagglutinin monomers is known to be completed in die endoplasmic reticulum before die fully folded subunits can be assembled into homotrimers (Valsesia- Wittmann et al., J. Virol. 65:4609-4619). The fact diat die II helical peptide forms very stable trimers and that d e VL peptide forms slightly weaker interactions might then explain why the chimaeric envelopes displaying the VL peptides gave better incoφoration than the chimaeric envelopes displaying d e II peptides. To test this idea, we are planning to generate chimaeric envelopes displaying trimeric leucine zipper peptides widi reduced stability (i.e. lower melting temperatures) compared to the VL and II peptides d at were used in this study.

All vectors carrying the VL or II chimaeric envelopes showed inhibition of infection on NIH3T3 and A431 cells, which was reversible on cleaving the peptides from the vectors with factor Xa protease. Titres were not restored completely to wild type levels due to die reduced levels of incoφoration of these envelopes. The VL and II peptides dierefore function as oligomerising peptide adaptors which mask die functions of the retroviral envelope glycoprotein to which they are fused. The inhibition of infection may be as a result of the oligomerizing peptide blocking binding of the vector to its target cells by masking the underlying binding domain. Alternatively, the presence of an oligomerizing peptide may prevent dissociation of the envelope trimer, blocking fusion. Unfortunately, because of die impaired incoφoration of the VL and II chimaeric envelopes, binding

studies were uninformative so we are unable to determine which of these mechanisms is more dominant.

A low level of background infectivity was consistently observed when uncleaved vectors displaying d e VL and II peptides were used to infect NIH3T3 and A431 cells (Table 4). The background was slightly higher on the NIH3T3 cells than on the A431 cells and tended to increase with increasing length of the linker peptide that was inserted between he 4070 A SU and die oligomerizing peptides. We believe that diis background infectivity occurs because a few of the chimaeric envelopes are cleaved by endogenous proteases derived from die target cells. In previous studies using chimaeric envelopes displaying a cleavable EGF domain we have observed diat the IEGR factor Xa cleavage site can be cleaved to a small extent by proteases released from NIH3T3 and A431 cells. We also found diat increasing die lengtii of the linker sequence between the factor Xa cleavage site and die displayed EGF domain increased me accessibility of the factor Xa site to these endogenous proteases.

In summary, our results demonstrate mat retroviral vector infectivity can be reversibly inhibited by fusing cleavable trimeric peptide adaptors to the N-terminus of the 4070A SU envelope glycoprotein. Infectivity is restored by exposing the vector particles to a protease that cleaves die adaptor from die SU glycoprotein. It is anticipated that diese adaptors will be useful to prevent infection of nontarget cells in the two-step (targeted attachment, targeted cleavage) targeting strategies that we are currently developing. Chimaeric envelopes displaying die VL and II peptide adaptors that were used in diis study were poorly expressed. We are dierefore attempting to identify similar masking adaptors that do not compromise die expression of chimaeric envelopes on which they are displayed.

Example 4

Construction of retroviruses containing cleavable oligomerising adaptors with an EGF binding domain

MATERIALS AND METHODS

Plasmid Construction

Vectors pEGF LVA1 and pEGF LVXA1 display an oligomerising peptide, LV, fused to residue + 1 of 4070A SU widi a non cleavable (SAA) or factor Xa protease-cleavable (SAAIEGR, Seq ID No. 21) linker and also display die EGF binding domain. To generate these vectors, PCR primers Gal4 LV, Gal4 LVbak and Gal4 LVfor were used for assembly of the PCR fragment coding for the oligomerising peptide, LV (Harbury et al., 1993 Science 262, 1401-1407). The PCR product was digested with Notl and Eagl and cloned into die Nøtf-digested backbones of EAl and EXAl. Figure 10 shows a diagramatic representation of the two constructs. The correct sequence of the constructs was verified by DNA sequencing.

+The following oligonucleotides (widi restriction sites underlined) were used:

Gal4 LV, 5'-GAC AAG CTA GAG GAA GTA CTT AGC AAG CTC TAC CAT GTC GAG AAC GAA CTT GCT CGA GTT AAG AAG-3' (Seq ID No. 22);

Gal4 LVback (containing Notl site), 5'-GGC ATT CAT GCG GCC GCA ATG AAG CAA GTG GAA GAC AAG CTA GAG GAA GTA C-3' (Seq ID No. 23);

Gal4 LVfor (containing Eagl site), 5'-GCA AAT CTG CGG CCG ACT CTC CCA GAA GCT TCT TAA CTC GAG CAA GTT C-3' (Seq ID No. 24).

Target cell lines and production of viruses

The murine cell line NIH 3T3, and d e human cell line A431 , were grown in DMEM supplemented widi 10% fetal calf serum. The envelope expression constructs were transfected into TELCeBό packaging cells by calcium phosphate precipitation and stable phleomycin (50mg/ml) resistant colonies were expanded and pooled. Cells were grown in DMEM supplemented with 10% fetal calf serum and when confluent transferred from 37°C to 32°C and incubated for 72hrs. Supernatants containing retroviral panicles were harvested after overnight (lόhrs) incubation at 32 °C in lOmls serum-free DMEM for infections. All supernatants were filtered (0.45μm) before use.

Target cell Infection

Target cells were seeded at 2 x 10' cells/well in six-well plates and incubated at 37 ^* C overnight. The harvested supernatants containing /3-galactosidase-transducing retroviruses were incubated widi 0 or 4 μg/ml of factor Xa (Promega) for 90 minutes at 37' C in die presence of 2.5mM CaCL. Supernatant dilutions in 2ml serum-free media were incubated widi target cells for 6 hrs in the presence of 8μg/ml polybrene. The retroviral supernatant was dien removed and die cells were incubated widi regular medium for 48-72 hrs. X-Gal staining for detection of /3-galactosidase activity was performed and viral titre (enzyme forming units/ml) was calculated by counting blue stained colonies microscopically with the use of a grid placed underneath die 6 well plates.

Host range properties of virus incorporating chimeric envelopes

To determine whether the oligomerising peptide was masking die functions of he 4070A envelope to which it was fused, we titrated die vectors on Ram-1 expressing cells, NIH3T3 and A431 before and after diey were cleaved widi factor Xa protease (Fig. 11). Both vectors gave greatly reduced titres on NIH3T3 and A431 cells compared to wildtype amphotropic vector. On cleavage of EGF LVXA1 with 4 μg/ml factor Xa protease the titre on NIH3T3 cells and A431 cells increased by up to 200 fold. Cleavage of the contol vector EGF LVA1, which does not carry die factor Xa cleavage signal, however, did not result in such an increase in titre.

These data demonstrate dial a monomeric binding domain can be displayed as pan of a trimerising adaptor which blocks the function of die underlying envelope until it is cleaved widi a specific protease.

Example 5

Angiogenesis, inflammation and tumour invasion are linked to die overexpression of matrix metalloproteinases (MMPs) which degrade the extracellular matrix. The MMPs are therefore promising targets for therapy. As an alternative to using MMP inhibitors, we are developing MMP-activatable gene delivery systems. Here, we describe die construction of vectors incoφorating inhibitory adaptors that are efficiently cleaved by activated MMPs. The MMP-sensitive vectors underwent cleavage activation selectively

on target cells expressing endogenous membrane-associated MMPs, and gene delivery was dramatically enhanced. MMP-activatable vectors will offer new opportunities for targeting of therapeutic genes to sites of disease.

Matrix metalloproteinases (MMPs) are important for angiogenesis, tissue remodelling, inflammation and wound healing, and diey play a crucial role in various pathological processes including cancer invasion and metastasis and the destruction of articular cartilage in rheumatoid arthritis (Liotta et al., 1991 Cell 64, 327; Woessner Jr. , 1991 FASEB J.

5, 2145; Ray & Steder-Stevenson 1994 Eur. Respir. J. 7, 2062; Karelina et al, 1995 J.

Invest. Dermatol. 105, 411). The known MMPs include matrilysin, collagenasesl-3, stromelysinsl-3, gelatinases A and B and a group of 4 membrane-type MMPs (MT-MMP) which are anchored to cell membranes (Sato et al. , 1994 Nature 370, 61; Takino et al.,

1995 J. Biol. Chem. 270, 23013; Will & Hinzmann 1995 Eur. J. Biochem. 257, 602;

Puente et al. , 1996 Can. Res. 56, 944). Most of the MMPs are secreted as zymogen forms and require activation before diey can exert dieir proteolytic activities. The net activities of the enzymes are also regulated by die tiiree tissue inhibitors of MMPs (TTMPs

1-3). Once activated, die MMPs co-operate widi one another in a cascade padiway to cause degradation of the extracellular matrix. Gelatinase A (GLA; MMP-2) and die MT-

MMPs are of special interest with respect to tumour invasion. Pro-GLA is secreted by stromal fibroblasts and concentrated on tumour cell membranes, especially at the invasive front of the tumour (Afzal et al., 1996 Lab. Invest. 74, 406; Nomura et al. , 1996 Int. J.

Can. (Pred. Oncol.) 69, 9). It binds as a pro-GLA-TIMP-2 complex to MT1-MMP which then mediates its cleavage activation on the surface of the tumour cell (Strongin et al. ,

1995 J. Biol. Chem. 270, 5331; Emmert-Buck et al., 1995 FEBS Letters 364, 28; Gilles et al. , 1996 Int. J. Can. 65, 209). Indeed, MTl-MMP-mediated activation of pro-GLA is considered to be important for the progression of cancer and the concentration of active

GLA is often found to be elevated in invasive or metastatic tumours. Hence, there is considerable interest in the exploitation of MMPs as promising targets for novel therapeutic agents and there are several general or specific MMP-inhibitors that are currently being tested for their usefulness in treatment of MMP-linked diseases in a number of clinical trials (Hodgson 1995 Biotech. 75, 554; Eccles et al. , 1996 Can. Res.

56, 2815). Here, as an alternative to the use of MMP-inhibitors. we propose the use of

a MMP-activatable gene delivery system.

This example describes die generation of targeted retroviral vectors whose infectivity for human EGF receptor-expressing cancer cells is strongly activated by membrane-associated MMPs.

A series of chimaeric envelope expression constructs was generated in which a cDNA coding for the 53 amino acid receptor binding domain of EGF was linked to die N- terminal codon of the 4070A murine leukemia virus (MLV) SU envelope glycoprotein via short non-cleavable or protease-cleavable linkers. In brief, the chimaeric vectors E. A and E.X.A, have an EGF cDNA, flanked by Sfil and Notl restriction sites, inserted at codon + 1 of die N-terminus of wild type 4070A MLV SU (surface protein gp 70) envelope, widi a linker of either 3 alanines (E.A.) or 3 alanines and die IEGR Factor Xa cleavage sequence (E.X.A.) between die domains. Figure 12 is a schematic representation of the chimaeric envelope expression constructs, E.A, E.G ₄ S.A, E.X.A and E.MMP.A. The envelope constructs were transfected into TELCeBό complementing cells, virus-producing clones were pooled and expanded in 10% FCS-DMEM selection medium containing 50 μg/ml phleomycin. Arrows indicate potential site of cleavage by respective proteases.

To obtain constructs E.MMP.A and E.G ₄ S.A, PCR primers AlGelA Nb (5' GCA AAT CTG CGG CCG CAC CTT TGG GAC TTT GGG CAA TGG CAG AGA GCC CCC ATC, Seq ID No. 27) or NL1A1B (5' GCA AAT CTG CGG CCG CAG GTG GAG GCG GTT CAA TGG CAG AGA GCC CCC ATC, Seq ID No. 28) respectively, were used widi primer 4070 Af or (described above) on E.A to generate iV -tailed PCR fragments of the 4070A SU coding sequence encoding die MMP-cleavable (PLGLWA) or non- cleavable linker (G ₄ S) as a 5' extension. The PCR fragments were digested widi Notl and BamHI and cloned into the Notl-BamHl digested backbone of E.A to generate constructs E.MMP.A and E.G ₄ S.A. The sequences of die constructs were checked and verified by DNA sequencing.

The E.A and E.G ₄ S.A chimaeric envelopes contained non-cleavable linkers AAA and AAAGGGGS (Seq ID No. 25) respectively (single letter amino acid code), the E.X.A

envelope contained a Factor Xa-cleavable linker AAAIEGR and die E.MMP.A envelope contained die linker AAAPLGLWA (Seq ID No. 26) in which the highlighted sequence is known to be recognised and cleaved by GLA and by MT1-MMP (Ye et al. , 1995 Biochem. 34, 4702; Will et al. , J. Biol. Chem. 277, in press) (Figure 12).

The chimaeric envelope constructs and a wild type 4070A envelope expression construct were stably transfected into TELCeBό complementing cells which express Moloney MLV gag-pol proteins and die nlsLacZ retroviral vector, as described in the preceding examples. Upon transfection of these cells with a functional envelope expression plasmid, infectious enveloped vector particles capable of transferring the lacZ marker gene are rescued into die culture supernatant. Viral supernatants were harvested from confluent plates of pooled transfected TELCeBό cells and die viral particles were pelleted by ultracentrifugation and immunoblotted using an anti-envelope antiserum as probe. Immunoblotting was performed as described in die preceding examples. The results are shown in Figure 13 A, B.

Figure 13 A is an immunoblot showing comparative viral incoφoration of the EGF chimaeric vectors (lane 2=E.A; lane 3=E.X.A; lane 4 ^* = E.MMP.A; lane 5 =E.G ₄ S.A) and die wild type 4070A SU (lanes 1 and 6). Figure 13B is an immunoblot demonstrating cleavage of MMP-cleavable linker in E.MMP.A by purified p-aminophenylmercuric acetate (APMA)-activated gelatinase A (GLA). An aliquot of the E.MMP.A viral pellet was incubated, respectively, with PBS only (lane 1), APMA-activated GLA (final concentration 32 μg/ml; lane 2) and APMA at a final concentration of 2 mM (lane 3). Lane 4 shows unmodified wild type 4070A-SU.

It is apparent from Figure 13 A that all four chimaeric envelopes were expressed and incoφorated into virions, as indicated by die decrease in mobility compared to wild type 4070A-SU, and that the relative efficiencies of envelope incoφoration were comparable in the four different recombinant virus stocks.

To determine if GLA could recognise the PLGLWA sequence on die E.MMP.A vector and thus cleave the EGF domain from the chimaeric viral envelope without degrading die underlying 4070A SU glycoprotein, we incubated aliquots of die E.MMP.A and control

viral pellets for 30 min at 37°C widi PBS, p-aminophenylmercuric acetate (APMA) or APMA-activated GLA, after which immunoblots were performed as before. [Gelatinase A (GLA) was purified as a zymogen form and requires activation by incubation with APMA (2 mM) for 1 h at 25°C prior to use. lOμl of the resuspended E.X.A, E.G ₄ S.A or E.MMP.A viral pellets were incubated with PBS, APMA (final concentration 2 mM) or APMA-activated GLA (32 μg/ml) for 30 min at 37°C].

On treatment of E.MMP. A-SU with activated GLA, a band with the same mobility as the wild type 4070A-SU was recovered, indicating diat the EGF domain could be efficiently cleaved from this chimaeric envelope without further GLA-mediated degradation (Fig. 13B). The E.G ₄ S.A and E.X.A chimaeric envelopes were unaffected by treatment with GLA indicating diat cleavage was specific for the MMP-sensitive linker (not shown).

Our previous data (Nilson et al. , 1996 Gene Therapy 5, 280) indicated diat die infectivity of the E.X.A vector was minimal on EGF-receptor positive A431 cells but could be fully and selectively restored by cleaving die chimaeric envelope with Factor Xa protease. This result was confirmed using the E.X.A vector stocks that were generated in the current study which bound strongly to EGF receptors on A431 cells and gave a titre of 10 ³ efu/ml rising to 10 ⁶ efu/ml after the EGF domain was cleaved from dieir surface with Factor Xa protease (data not shown). The infectivities of the E.A, E.G ₄ S.A and E.MMP.A vectors on A431 cells were low between 10 ² -10 ³ efu/ml and were not greatly increased by treatment with Factor Xa protease (not shown).

To determine whedier die infectivity of the MMP-cleavable E.MMP.A vector could be activated by GLA, we performed infections on A431 cells in the presence of increasing concentrations of exogenous pro-GLA. Since A431 cells are known to activate pro-GLA to GLA, pre-activation of the protease with APMA was not necessary.

For the infection assays, A431 cells in 10% FCS-DMEM were seeded, at a density of 3 x 10 ⁴ per individual well, in a 24- well tissue culture plate (Corning, New York) overnight at 37°C. The media were removed the next day and the cells were washed once in serum- free DMEM. Varying amounts (final concentration 2-40 μg/ml) of pro-GLA were mixed

widi 200 μl of filtered E.MMP.A viral supernatant after which the mixture was added to A431 cells and incubated at 37°C for 6 h. At die end of 6 h, die media was removed and cells were washed once in serum-free DMEM. The cells were then incubated in 10% FCS-DMEM for 72 h at 37°C before they were washed once in cold PBS, fixed in 0.5 % gutaldehyde-PBS for 15 min, washed once again widi PBS and incubated widi X-gal overnight at 37°C. The number of colonies transduced widi die vector (blue colonies) were counted and die titre expressed as efii/ml viral supemantant. Figure 14 is a graph showing that increase in titre (efu x lO'Vml) of the E.MMP.A MMP-sensitive vector on A431 cells is correlated widi die amount of pro-gelatinase A (pro-GLA) added onto die cells.

It was found diat as die concentration of exogenous pro-GLA was increased incrementally from 2 to 40 μg/ml, the infectivity of the E.MMP.A vector increased in a dose-dependent manner. From 1.3 x 10 ³ efu/ml in the absence of pro-GLA, the titre increased 50-fold to 6.5 x 10 ⁴ efu/ml in die presence of 40 μg/ml pro-GLA. The titre of die noncleavable E.G S.A vector was relatively unchanged from 1.2 x 10 ² efu/ml in the absence of pro- GLA to 1.4 x 10 ² efu/ml in the presence of 40 μg/ml pro-GLA. Activation of infectivity was specific to the vector with the MMP-cleavable linker as infectivity of E.X.A increased only 3-fold in the presence of 40 μg/ml pro-GLA (not shown).

We next explored die possibility diat endogenous target cell-derived MMPs could activate die E.MMP.A vector in the absence of exogenous MMP. HT1080 is a human fibrosarcoma cell line that constitutively produces MT1-MMP and pro-GLA (Okada et al. , 1995 Proc. Natl. Acad. Sci. 92, 2730). We dierefore incubated the viral supernatants of E.G ₄ S.A, E.X.A and E.MMP.A on HT1080 and A431 cells for 6 h at 37°C in the absence of exogenously added pro-GLA: 200μl of the filtered E.X.A, E.G ₄ S.A or E.MMP.A viral supernatants were added onto A431 or HT1080 cells for 6 h at 37°C with 8 μg/ml polybrene, after which the incubation medium was removed and die cells washed once in serum-free DMEM. The cells were incubated in 10% FCS-DMEM for 72 h at 37°C before they were stained wid X-gal.

Figure 15 is a graph showing die titre of EGF chimaeric vectors on A431 and HT1080

cells. Figure 15A shows the high infectivity of E.MMP.A vector on HT 1080 cells compared to on A431 cells as indicated by the number of blue /3-galactosidase positive colonies. One ml out of 10 ml filtered E.MMP.A viral supernatant was incubated widi

5 mM CaCI ₂ for 30 min at 37°C before it was incubated on die respective cell types for

6 h at 37 ⁰ C. At the end of 6 h, die cells were washed in serum-free DMEM and incubated in 10% FCS-DMEM for 72 h after which diey were stained widi X-gal. The respective titres are expressed as efii/ml viral supernatant.

Consistent wid previous results, the infectivity of the vectors on A431 cells was low in die absence of exogenous pro-GLA. However, on HT1080 cells, the infectivity of the MMP-cleavable vector E.MMP.A was activated by two orders of magnitude compared to the MMP-resistant control vectors E.G ₄ S.A and E.X.A (Fig. 15, 15A). Thus, in die absence of any added exogenous MMP, die higher titre of the MMP-dependent E.MMP.A vector must be due to its cleavage by MMPs produced endogenously by HT1080 cells.

To determine if the MMP-activatable E.MMP.A vector could selectively target the MMP- expressing HT1080 cells in preference over A431 cells, we allowed the vector to infect both cell types on the same petri dish simultaneously. We grew A431 and HT1080 cells separately on coverslips, placed diem in die same petri dish and added supernatant containing the E.MMP.A or control vectors: A431 and HT1080 cells were seeded separately, on 25 mm Theπnanox coverslips (Corning) contained in 6 well plates, in 10% FCS-DMEM overnight after which the media was removed and die cells washed once in serum-free DMEM. The coverslips coated widi die cells were placed in a 10 cm petri dish (Falcon) and E.G ₄ S.A (1: 1.5 dilution), E.MMP.A (1:1.5) or 4070A (1:20) supernatants were added onto die petri dishes widi 8 μg/ml polybrene for 6 h at 37°C. At the end of the incubation period, die media was removed and die cells were incubated in 10% FCS-DMEM for 72 h before X-gal staining. The results are shown in Figure 16: E.MMP.A vector grown on HT1080 (H) cells and A431 (A) cells is shown in I and II, widi die control E.G ₄ S.A vector in III and die wild type 4070 A vector in IV.

When presented widi botii cell types, E.MMP.A infected HT1080 cells preferentially over A431 cells. The wild type 4070 A vector and E.G ₄ S.A vector with die non-cleavable

„„_,. O 97/12048

52 linker showed no such preference (Fig. 16). In diese experiments, the MMP-activatable E.MMP.A vector did not infect A431 cells more efficiently in the presence of HT1080 cells dian in their absence. This suggests that soluble GLA released into the medium from the HT1080 cells does not play a significant role in activation of the vector. Instead, die results strongly indicate that the cleavage activation of die E.MMP.A vector is localised to die surface of the HT1080 cells and that it is mediated by membrane-associated MMPs acting on vector particles that have bound to die EGF receptors on these cells.

The significant role that MT-MMP plays in cleavage activation of the MMP-cleavable vector was supported by results from experiments using natural MMP inhibitors TIMP-1 and TIMP-2 and a synthetic inhibitor, CT 1339. For the inhibition studies, TIMP-1 at a final concentration of 10 μg/ml, TIMP-2 (5 μg/ml) or CT 1339 (1 mM) was used. The inhibitors were added to 200 μl of diluted (1:10) E.MMP.A or undiluted E.G ₄ S.A viral supernatants. The mixture was then added onto A431 or HT1080 cells, which had been washed once in serum free DMEM, and die cells were incubated for 6 h at 37°C. At the end of the incubation period, die cells were washed once in serum free DMEM, incubated for 72 h in 10% FCS-DMEM after which they were stained with X-gal. The E.MMP.A supernatant was diluted to obtain a titre that would allow accurate counting of the number of transduced colonies. Inhibition studies on A431 cells were performed with 200 μl undiluted E.MMP.A or E.G ₄ S.A in presence of 16 μg/ml pro-GLA.

Table 5 : Influence of MMP inhibitors on the titre of vectors on A431 and HT1080 cells.

MMP-dependent E.MMP.A vector was added to A431 cells in presence of 16 μg/ml pro- GLA or to HT1080 cells in the absence of exogenous pro-GLA, with or without the addition of natural MMP inhibitors TIMP-1, TIMP-2 or synthetic inhibitor, CT 1339. Values (means +_ SD, n=3) represent percentage decrease in titre (with inhibitors) compared to d at of the control (without inhibitors).

All three inhibitors have strong activity against GLA and can prevent the activation of E.MMP.A vectors by exogenous GLA (Table 5). However, unlike TIMP-2 and CT 1339, TIMP-1 could not efficiently block the activation of E.MMP.A by endogenous MMPs on HT1080 cells (Table 5). An important difference between TIMP-1 and d e odier inhibitors is that it displays only weak activity against the MT1-MMP expressed on HT1080 cells (Fig. 17. described below). These experiments therefore point to a central role for the MT-MMP in HT1080-mediated activation of the E.MMP.A vector.

Figure 17 is a gelatin zymogram showing the effect of TIMP-1 or a synthetic MMP- inhibitor, CT 1339 on cellular activation of endogenous pro-GLA on HT 1080 cells. The E.MMP.A viral supernatant was incubated on HT 1080 cells for 6 h at 37°C in die absence of any inhibitors (lane 1), in die presence of 10 μg/ml (lane 2) or 30 μg/ml TIMP-1 (lane 3), and 1 μM (lane 4) or 10 μM (lane 5) CT 1339. At the end of the incubation period, an aliquot of die supernatant was loaded onto 7% SDS-PAGE gel containing 0.5 g/ml denatured type I collagen and electrophoresis was carried out at 4°C for 1 h, after which the gel was incubated twice for 15 min each in 2.5% Triton-X 100 to remove the SDS, washed in water and then incubated overnight at room temperature in 100 mM Tris, 30 mM CaCl ₂ 0.0015% Brij and 0.001 % NaN ₃ . The gel was then stained in 0.25% Coomassie Brilliant Blue Green (Sigma). The location of gelatinolytic activity on the gelatin zymogram is detectable as a clear band in the background of blue staining.

There have been many variably successful attempts to target retroviral vectors dirough ligand -receptor interactions (Valsesia-Wittmann et al. , 1994 J. Virol. 65, 4609; Cosset et al. , 1995 J. Virol. 69, 6314; Kasahara et al. , 1994 Science 266, 1373; Matano et al., 1995 J. Gen. Virol. 76, 3165; Somia et al., 1995 Proc. Natl. Acad. Sci. 92, 7570). Here we have adopted a two-step targeting strategy that allows us to utilise the specificity of protease-substrate interactions to activate the infectivity of receptor-targeted retroviral vectors. We previously relied on the addition of exogenous Factor Xa protease for vector activation, an approach that might have rather limited applications for in vivo gene therapy. Here, we have demonstrated for the first time a retroviral vector whose infectivity can be activated by endogenously produced disease-associated proteases. The

vector is optimally cleaved and activated by membrane-associated MMPs on human tumour cell lines.

The targeting strategy diat we have pursued may have interesting parallels with d e mechanism of HIV entry in which primary virus attachment to CD4 leads to a conformational rearrangement or proteolytic cleavage in gpl20, and secondary virus attachment to one of the recently characterised HIV co-receptors (Feng et al. , 1996 Science 272, 872; Deng et al., 1996 Nature 557, 661; Handley et al., 1996 J. Virol. 70, 4451). C-type retroviral vectors with engineered SU glycoproteins could dierefore be developed as model systems to probe die entry mechanisms that are employed by naturally occurring viruses, such as HIV.

It is hoped diat targeted vectors of die type that we have described in this repoπ will open up new possibilities for gene therapy in MMP-associated diseases, for example in cancer, where elevated MMP production in tumour deposits is required for angiogenesis, invasiveness and metastatic potential and is strongly correlated widi poor prognosis (Munay et al., 1996 Nature Med. 2, 461).

Example 6

Retroviral display of trimeric binding domains, TNF alpha and CD40 ligand.

The following experiments demonstrate that chimaeric envelopes bearing TNF alpha or CD40 ligand as an N-terminal extension can be incoφorated into retroviral vector particles where it appears that the trimeric binding domain forms a cap over the envelope glycoprotein to which it is fused. The amphotropic infectivity of the vectors incoφorating diese chimaeric envelopes is therefore low but is greatly enhanced by cleaving the trimeric ligand from their surface.

Cell Lines

The TELCeBό cell line has been described in die preceding examples. The NIH 3T3, A431 (human squamous carcinoma; ATCC CRL1555) and HT1080 (human fibrosarcoma; ATCC CCL121) cell lines were grown in DMEM (Gibco-BRL, UK) supplemented with 10% fetal calf serum (FCS; PAA Biologicals, UK), benzylpenicillin (60 mg/ml) and

streptomycin (100 mg/ml) at 37 ^C C in an atmosphere of 5% CO ₂ . The B cell lines, Daudi (human Burkitt's lymphoma; ATCC CCL 213), Raji (human Burkitt's lympho a; ATCC CCL 86) and K422 (human Non-Hodgkin B cell; Dryer et al., 1990 Blood 75:709-714), and T cell line, Jurkat (human acute T cell leukemia; ATCC TIB 152) were grown in RPMI 1640 (Gibco-BRL) supplemented with 10% FCS, benzylpenicillin (60 mg/ml) an streptomycin (100 mg/ml) at 37°C in an atmosphere of 5 % CO ₂ .

Construction of chimaeric envelope expression vectors

The human tumour necrosis factor-alpha (TNF-a)-4070A SU chimaeric envelope expression vectors TNF-a.A. TNF-a.GS.A, TNF-a.X.A, TNF-a.XA, andTNF-a.MMP.A, have an TNF-a cDNA (Wang et al. , 1995 Science, 225: 149-154), flanked by Sfil and Notl restriction sites, inserted at codon + 1 of the N-terminus of wild type 4070 A MLV SU envelope by different linkers (Fig. 18). The TNF-a.A vector is linked via a 3 alanine (AAA) linker; TNF-a.GS.A via a non-cleavable AAAG ₄ S linker; TNF-a.X.A via Factor Xa protease cleavable linker (AAAIEGR) and TNF-a.MMP.A via an MMP-cleavable linker (AAAPLGLWA) (single letter amino acid code). The Factor Xa protease cleaves IEGR after die arginine residue and die PLGLWA linker is susceptible to gelatinase A (MMP-2) and MT-MMP between the giycine and leucine residues.

The CD40L-4070A SU chimaeric envelope expression vectors have part of the CD40L cDNA, flanked by Sfil and Notl restriction sites, inserted at codon + 1 of the N-terminus of 4070A MLV by the 4 different linkers as mentioned above. The vectors are termed CD40L.A, CD40L.GS.A, CD40L.X.A and CD40L.MMP.A (Fig. 19).

A PCR derived Sfil-Notl DNA fragment encoding the 155 amino acids of d e trimeric human TNF-a was generated using a cDNA template and two primers, sTNFback

(5'> CCG GTACCG GCC CAGCCGGCC TCTTCTTCTCGTACC CCG, SeqID

No. 29) with a Sfil site, and nTNFfor (5' > AAG TCT TAG CGG CCG CCA GAG CGA TGA TAC CGA AG, Seq

ID No. 30) with a No rl site.

The Sfil-Notl PCR fragment encoding d e 145 amino acids of the soluble extracellular

domain of the trimeric CD40L (Gly 116-Leu 261; Kaφusas et al., 1995 Structure, 5: 1031-1039) was generated using a cDNA template (ATCC 79813) and two primers: sCD40Lb (5' > CCG GTA CCG GCC CAG CCG GCC GGT GAT CAG AAT CCT CAA ATT GC, Seq ID No. 31) widi a Sfil site and nCD40Lf (5 ^* > AAG TCT TAG CGG CCG CGA GTT TGA GTA AGC CAA AGG, Seq ID No. 32) with a Norl site. The respective PCR fragments were digested widi Sfil and Norl restriction enzymes and cloned into the Sfil-Notl digested EA.1 backbone or EXA.1 to obtain TΝF-a.A or CD40L.A. and TΝF-a.X.A or CD40L.X.A, respectively (Nilson et al. , 1996 Gene Therapy 5: 280-286).

To obtain TNF-a.GS.A or CD40L.GS.A, and TNF-a.MMP.A or CD40L.MMP.A, the respective Sfil-Notl digested TNF-a or CD40L PCR fragments were cloned into Sfil-Notl digested E.GS.A or E.MMP.A backbones, respectively (Peng et al., A gene delivery system activatable by disease-associated matrix metalloproteinases, submitted). The sequences of the constructs were checked and verified by DNA sequencing.

Production of viruses

The various TNF-a and CD40L envelope expression plasmids were stably transfected by calcium phosphate precipitation (Sambrook et al. , 1989, Molecular cloning: A laboratory manual) into the TELCeBό packaging cells. Transfected cells, grown in 10% FCS- DMEM at 37°C, were selected widi 50 μg/ml phleomycin (Sigma, Poole, Dorset, UK). Resistant colonies were pooled and expanded, and before harvest, the confluent cells were tranfened to 32 °C for 72 h. The viral supernatants were dien harvested and filtered (0.45 μm, Acrodisc, Gelman Sciences MI, USA) after overnight incubation of the confluent cells with serum free DMEM at 32 °C. These filtered supernatants were then used eidier for immunoblotting, binding or infection assays.

Immunoblots

For immunoblotting, the viral particles were pelleted by ultracentrifugation of the filtered viral supernatant (Beckman, USA) at 30,000 φm for 1 h at 4°C in a SW 40 rotor. The pellet was then resuspended in 100 μl cold PBS and stored at -70°C till further analysis. An aliquot (10 μl) of die viral proteins was separated by electrophoresis on a 10% SDS-

PAGE gel, electrotransferred onto nitrocellulose membrane (Hybond ECL, Amersham Life Sciences, UK) and detected by immunostaining with goat antisera raised against Rausher leukemia virus gp 70-SU envelope protein (Quality Biotech, USA), followed by horseradish peroxidase-conjugated rabbit anti-goat immunoglobulins antibodies (DAKO, Denmark) and developed widi an enhanced chemiluminescence kit (Amersham).

To detect the presence of processed (SU) and unprocessed (SU + TM) in the cells, the viral complementing cells were grown to confluency on petri dishes (10 cm in diameter), washed once in cold PBS and dien incubated for 10 min at 4°C widi cell lysis buffer containing 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 % Triton-X, 0.05% sodium dodecyl sulfate, 5 mg/ml sodium doxycholate and 1 mM PMSF. The lysed cells were scraped from die plates and die suspension centrifuged at 10,000 xg for 20 min to pellet the nuclei. Thirty μl of the supernatant was used for electrophoresis and immunoblotting.

Results:

There were no visible bands of envelope proteins seen on the immunoblots for chimaeric TNF-a-4070A SU indicating litde or no envelope expression. When die cell lysates were tested, tiiere was unprocessed (SU + TM) chimaeric envelope proteins but little detectable processed (SU) envelope proteins. Results from the immunoblots for CD40L-4070A SU indicated d at tiiere was chimaeric envelope expression and incoφoration, and expression between d e chimaeric envelopes were comparable except for CD40L.MMP.A-SU which is most highly expressed.

Infection Assays i. Infection of cells by chimaeric vectors-blocking infectivity of 4070A by trimeric ligand

The TNF-a and CD40L chimaeric vectors were tested for infectivity on NIH 3T3, A431 and HT1080 cells. The cells were seeded overnight at 37°C at a density of approximately 1 x lO ⁵ cells per individual well in a 6-well tissue culmre plate (Corning, New York). The medium was removed die next day and cells were washed once in serum- free DMEM. An aliqout (1 ml) of the filtered viral supernatant was used to infect the cells in die presence of 8 μg/ml polybrene. At the end of die 6 h incubation period, die medium was

removed and the cells washed once in serum-free DMEM and 10% FCS-DMEM was added. The cells were then incubated for 72 h at 37° C before they were stained widi X- gal. The cells were washed once in cold PBS, fixed in 0.5% glutaldehyde-PBS for 15 min, washed once in cold PBS and incubated widi X-gal overnight at 37 °C. Number of colonies transduced widi die /3-galactosidase gene (blue colonies) were counted and die titre expressed as enzyme forming units (efu)/ml viral supernatant.

ii. Treatment of viral supernatants with Factor Xa protease: reversal of blockage to infectivity by protease

An aliquot (1 ml) of the filtered supernatant was incubated widi 2.5 mM CaCl, in die presence or absence of 4 μg/ml Factor Xa protease (New England Biolabs, UK) for 1 h at 37 °C. At the end of die incubation period, die supernatant was added onto die NIH 3T3 or HT 1080 cells for 6 h before the supernatant was removed, cells washed and maintained in 10% FCS-DMEM before diey were stained widi X-gal as before.

Results:

TNF-a-4070A chimaeras

The titre of the TNF-a-4070A vectors on NIH3T3 and HT1080 cells were low (Table 6). This low level of infectivity could be due to die low level of chimaeric envelope expression. However, it could also be due to die display of the trimeric TNF-a on the 4070A-SU. The trimer was able to block the infectivity of the amphotropic vector, which would be odierwise be highly infective on the murine NIH 3T3 cells, which do not bear die human TNF-a receptor.

Table 6. Titre of TNF-a-4070A vectors on cell lines in presence of 8 μg/ml polybrene

This blockage of infectivity by the trimeric TNF-a could be reversed by the addition of factor Xa protease to cleave off the TNF-a ligand on die TNF-a.X.A vector and thus,

allowing the 4070A-SU to bind and infect the cells. As a result, the titre of the TNF- a.X.A vector increased 4-fold and 60-fold, respectively, on HT1080 cells and NIH 3T3 cells after treatment with Factor Xa protease (Table 7).

Table 7. Titre (efu/ml) of TNF-a.X.A on cells in absence or presence of Factor Xa protease

CD40L-4070A chimaeras

The infectivity of the CD40L-4070A chimaeras are significandy lower than diat of the wild type on NIH 3T3, A431 and HT1080 cells (Table 8), indicating that the display of CD40L on the envelope is blocking die infectivity of the vector.

Table 8. Titre (efii/ml) of CD40L-4070A vectors on cell lines in presence of 8 μg/ml polybrene

Upon treatment of the CD40L.X. A widi Factor xa protease, the infectivity of the vector is increased dramatically (Table 9).

Table 9. Titre (efu/ml) of CD40L.X.A in absence or presence of Factor-Xa protease

SEQUENCE LISTING

(1) GENERAL INFORMATION:

(i) APPLICANT:

(A) NAME: Medical Research Council

(B) STREET: 20 Park Crescent

(C) CITY: London

(E) COUNTRY: United Kingdom

(F) POSTAL CODE (ZIP): WIN 4AL

(G) TELEPHONE: (0171) 6365422 (H) TELEFAX: (0171) 3231331

(ii ⁾ TITLE OF INVENTION: Recombinant Viruses Incorporating a Protease

Cleavable Protein

(iii) NUMBER OF SEQUENCES: 32

(iv) COMPUTER READABLE FORM:

(A) MEDIUM TYPE: Floppy disk

(B) COMPUTER: IBM PC compatible

(C) OPERATING SYSTEM: PC-DOS/MS-DOS

(D) SOFTWARE: Patentln Release #1.0. Version #1.30 (EPO)

(2) INFORMATION FOR SEQ ID NO: 1:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 53 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: l near

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 1: GCAAATCTGC GGCCGCAATC GAGGGAAGGC CTCATCAAGT CTATAATATC ACC 53

(2) INFORMATION FOR SEQ ID NO: 2:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 37 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 2: GCAAATCTGC GGCCGCAGCT TCGCCCGGCT CCAGTCC 37

(2) INFORMATION FOR SEQ ID NO: 3:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 49 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 3: GCAAATCTGC GGCCGCAATC GAGGGAAGGG CTTCGCCCGG CTCCAGTCC 49

(2) INFORMATION FOR SEQ ID NO: 4:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 36 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 4: GCAAATCTGC GGCCGCAATG GCAGAGAGCC CCCATC 36

(2) INFORMATION FOR SEQ ID NO: 5:

(l) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 48 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 5: GCAAATCTGC GGCCGCAATC GAGGGAAGGA TGGCAGAGAG CCCCCATC 48

(2) INFORMATION FOR SEQ ID NO: 6:

(l) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 19 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 6: GCCAGAACGG GGTTTGGCC 19

(2) INFORMATION FOR SEQ ID NO: 7:

(l) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 20 base pairs

(B) TYPE- nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 7: CTGCAAGCCC ACATTGTTCC 20

(2) INFORMATION FOR SEQ ID NO: 8:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 12 ammo acids

(C) STRANDEDNESS:

(D) TOPOLOGY: unknown

(ii) MOLECULE TYPE: peptide

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 8:

Ala Ala Ala Gly Gly Gly Gly Ser Ile Glu Gly Arg 1 5 10

(2) INFORMATION FOR SEQ ID NO: 9:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 22 amino acids

(B) TYPE: amino acid

(C) STRANDEDNESS:

(D) TOPOLOGY: unknown

(ii) MOLECULE TYPE: peptide

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 9:

Ala Ala Ala Gly Gly Gly Gly Ser Gly Gly Gly Gly Ser Gly Gly Gly 1 5 10 15

Gly Ser Ile Glu Gly Arg 20

(2) INFORMATION FOR SEQ ID NO: 10:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 53 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 10: GCAAATCTGC GGCCGCAGGT GGAGGCGGTT CAATCGAGGG AAGGATGGCA GAG 53

(2) INFORMATION FOR SEQ ID NO: 11:

(l) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 83 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 11: GCAAATCTGC GGCCGCAGGT GGAGGCGGTT CAGGCGGAGG TGGCTCTGGC GGTGGCGGAT 60 CGATCGAGGG AAGAATGGCA GAG 83

(2) INFORMATION FOR SEQ ID NO: 12:

(l) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 79 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 12: GGCAπCATG CGGCCGCGGC CCAGCCGGCC ATGAAGCAAC TAGAAGACAA GGTGGAGGAA 60 CTCCTTAGCA AGGTATACC 79

(2) INFORMATION FOR SEQ ID NO: 13:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 79 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY, linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 13: GCAAATCTGC GGCCGCCTCT CCAACAAGCT TCTTCAGTCG AGCGACTTCG TTCTCAAGAT 60 GGTATACCπ GCTAAGGAG 79

(2) INFORMATION FOR SEQ ID NO: 14:

(ι) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 79 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 14: GGCAπCATG CGGCCGCGGC CCAGCCGGCC ATGAAGCAAG CAGAAGACAA GGCAGAGGAA 60

GCTCπAGCA AGGCπACC 79

(2) INFORMATION FOR SEQ ID NO: 15:

(ι) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 79 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 15: GCAAATCTGC GGCCGCCTCT CCAGCAAGCT TCTTTGCTCG AGCAGCπCG πCTCTGCAT 60 GGTAAGCCπ GCTAAGAGC 79

(2) INFORMATION FOR SEQ ID NO: 16:

(ι) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 79 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 16: GGCAπCATG CGGCCGCGGC CCAGCCGGCC ATGAAGCAAA TCGAAGACAA GATAGAGGAA 60 AπcπAGCA AGATCTACC 79

(2) INFORMATION FOR SEQ ID NO: 17:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 82 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 17: GCAAATCTGC GGCCGCCTCT CCTATAAGCT TCπGAπCG AGCAAπTCG πCTCTATAT 60 GGTAGATCπ GCTAAGAAπ TC 82

(2) INFORMATION FOR SEQ ID NO: 18:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 20 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 18:

GGCAπCATG CGGCCGCGGC 20

(2) INFORMATION FOR SEQ ID NO: 19:

(ι) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 21 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 19: GCAAATCTGC GGCCGCCTCT C 21

(2) INFORMATION FOR SEQ ID NO: 20:

(ι) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 7 amino acids

(B) TYPE: amino acid

(C) STRANDEDNESS:

(D) TOPOLOGY: unknown

(n) MOLECULE TYPE: peptide

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 20:

Ala Ala Ala Ile Glu Gly Arg 1 5

(2) INFORMATION FOR SEQ ID NO: 21:

(l) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 7 ammo acids

(B) TYPE: amino acid

(C) STRANDEDNESS:

(D) TOPOLOGY: unknown

(ii) MOLECULE πPE: peptide

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 21:

Ser Ala Ala Ile Glu Gly Arg 1 5

(2) INFORMATION FOR SEQ ID NO: 22:

(ι) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 66 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 22.

GACAAGCTAG AGGAAGTACT TAGCAAGCTC TACCATGTCG AGAACGAACT TGCTCGAGπ 60 AAGAAG 66

(2) INFORMATION FOR SEQ ID NO: 23:

(ι) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 52 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 23: GGCAπCATG CGGCCGCAAT GAAGCAAGTG GAAGACAAGC TAGAGGAAGT AC 52

(2) INFORMATION FOR SEQ ID NO: 24:

(l) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 49 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 24: GCAAATCTGC GGCCGACTCT CCCAGAAGCT TCπAACTCG AGCAAGπC 49

(2) INFORMATION FOR SEQ ID NO: 25:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 8 amino acids

(B) TYPE: amino acid

(C) STRANDEDNESS:

(D) TOPOLOGY: unknown

(ii) MOLECULE TYPE: peptide

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 25:

Ala Ala Ala Gly Gly Gly Gly Ser 1 5

(2) INFORMATION FOR SEQ ID NO: 26:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 9 ammo acids

n

(ii) MOLECULE TYPE: peptide

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 26:

Ala Ala Ala Pro Leu Gly Leu Trp Ala 1 5

(2) INFORMATION FOR SEQ ID NO: 27:

(l) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 54 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 27: GCAAATCTGC GGCCGCACCT πGGGACπT GGGCAATGGC AGAGAGCCCC CATC 54

(2) INFORMATION FOR SEQ ID NO: 28:

(ι) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 51 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 28: GCAAATCTGC GGCCGCAGGT GGAGGCGGπ CAATGGCAGA GAGCCCCCAT C 51

(2) INFORMATION FOR SEQ ID NO: 29:

(ι) SEQUENCE CHARACTERISTICS:

(A) LENGTH. 39 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 29: CCGGTACCGG CCCAGCCGGC CTCπcπCT CGTACCCCG 39

(2) INFORMATION FOR SEQ ID NO: 30:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH. 35 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 30:

AAGTCπAGC GGCCGCCAGA GCGATGATAC CGAAG 35

(2) INFORMATION FOR SEQ ID NO: 31.

(ι) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 44 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY, linear

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 31: CCGGTACCGG CCCAGCCGGC CGGTGATCAG AATCCTCAAA πGC 44

(2) INFORMATION FOR SEQ ID NO: 32:

(ι) SEQUENCE CHARACTERISTICS.

(A) LENGTH: 36 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(xi ) SEQUENCE DESCRIPTION : SEQ ID NO: 32: AAGTCπAGC GGCCGCGAGT πGAGTAAGC CAAAGG 36