WO2004070041A1

WO2004070041A1 - Vectors for expression of toxins

Info

Publication number: WO2004070041A1
Application number: PCT/GB2004/000441
Authority: WO
Inventors: Kai Stefan Lipinski
Original assignee: ML Laboratories PLC
Current assignee: Innovata Ltd
Priority date: 2003-02-06
Filing date: 2004-02-05
Publication date: 2004-08-19
Anticipated expiration: 2005-08-06

Abstract

The invention relates to a method of manufacturing vectors encoding recombinant proteins or glycoproteins that are toxic to cells in which they are expressed. As a result, expression of these proteins must be tightly controlled to allow, for example, production of viral vectors for gene therapy in cells that would otherwise be killed by the toxin encoded by the vectors. Vectors comprising highly tissue-specific synthetic promoter constructs are described, which allow expression of the toxin only in cells having a nuclear accumulation of β-catenin.

Description

VECTORS FOR EXPRESSION OF TOXINS

Background of the Invention

Recombinant proteins are routinely expressed in living cells by means of introducing expression vectors encoding the proteins under control of appropriate control elements. Such control elements include promoters, which are c/s-acting elements controlling expression of open reading frames to which they are operably linked. In eukaryotic cells such promoters may be constitutive and non-tissue-specific, or they may be tissue-specific, allowing expression of the product encoded by the open reading frame only in cells or tissues of the appropriate type. However, although many such tissue-specific promoters are known, few are highly stringent (or 'tight') in their control of expression.

Some recombinant proteins are highly toxic in the cells usually used for manufacture of expression vectors. So, where vectors encoding highly toxic molecules are required, for example, for the ablation of cancer cells by gene therapy, it is currently very difficult to grow cells containing and synthesising such vectors because of the lethal effects of even low-level expression of the toxins in the cells. In the case of toxins such as the diphtheria toxin, expression of a few molecules, possibly even one, is fatal to the cell (Yamaizumi et a), 1978, Cell 15: 245-250). Although, in principle, use of a tissue-specific promoter that is not permissive for expression in a suitable host cell could be used for production of such vectors, in practice none is known with sufficient stringency to allow this approach.

Currently, a number of cumbersome methods are available that attempt to overcome the shortcomings of 'leaky' expression of toxin genes. One approach relies on a recombination strategy to ensure that an intact toxin gene is only assembled once the eventual target cell is reached. For example, an insert or 'stuffer' sequence, flanked by recombination sites, may be used to interrupt the toxin coding sequence or to separate the coding sequence from its promoter. A transgene encoding the recombinase required to excise the stuffer must also be delivered to the target cell, and the overall size of the elements required exceeds the capacity of viral vectors, necessitating the co-infection of two viral vectors. Such a system using a Cre/Lox recombination system to express diphtheria toxin A is described by Lee and Jameson (2002, Human Gene Therapy 13: 533-542). Peng et al (2002, Mol Therapy 6: 537-545) describe a system that is similar in principle, but with the added sophistication of a tamoxifen-activated Flp recombinase, expression of which is controlled by a tumour-selective Bcl-2 regulatory region. Such systems suffer from the disadvantages of complexity and the inefficiency of requiring coinfection of target cells by both vectors of the binary system,

Another approach is to attempt to control leaky transcription in producer cells by post-transcriptional blocking with specific antisense oligonucleotides (Raykov et al, 2002, Gene Therapy 9: 358-362). This approach requires efficient penetration of producer cells with the oligonucleotides and involves an extra and expensive intervention.

We have previously disclosed (WO 01/64739, the contents of which are herein incorporated in their entirety) constructs and vectors comprising the highly active, and highly selective β-catenin-dependent promoters, CTP1 , CTP2 and CTP3, comprising tandem repeats of T cell factor (TCF)-binding elements (in this case, CCTTTGATC), and described their application for the treatment of secondary colorectal liver metastases and other tumour types using a gene-directed enzyme prodrug therapy approach with adenoviral (Ad) vectors.

TCFs are a family of transcription factors within the High Mobility Group (HMG) of DNA-binding proteins (Love et al., Nature, 376, 791-795, 1995). The family includes TCF-1 , TCF-3 and TCF-4 which are described in van der Wetering et. al, (EMBO J., 10, 123-132, 1991), EP-A-0 939 122 and Korinek et al. (Science, 275, 1784-1787, 1997). TCF-4 has been shown to be involved in tumorigenesis related to Wnt Wingless signalling. TCF and LEF-1 (lymphoid enhancer factor-1) are considered to mediate a nuclear response to Wnt signals by interacting with β- catenin. Wnt signalling and other cellular events that increase the stability of β- catenin are considered to result in transcriptional activation of genes by LEF-1 and TCF proteins in association with β-catenin. In the absence of Wnt signalling, LEF- 1/TCF proteins repress transcription in association with Groucho and CBP (CREB binding protein).

In the absence of Wnt signalling, β-catenin is found in two distinct multiprotein complexes. One complex, located at the plasma membrane, couples cadherins (calcium dependent adhesion molecules) with the actin cytoskeleton whereas the other complex (containing the proteins adenomatous polyposis coli protein (APC), axin and glycogen synthase kinase 3β (GSK3β)) targets β-catenin for degradation. Wnt signalling antagonises the APC-axin-GSK3β complex, resulting in an increase in the pool of free cytoplasmic β-catenin. The free cytoplasmic β-catenin can translocate to the nucleus where it binds LEF-1/TCF factors and activates Wnt target genes. The regulation of LEF-1 /TCF transcription factors by Wnt and other signals is discussed in Eastman et al, (Current Opin. Cell Biology, U_, 233-240, 1999).

The APC gene is a tumour supressor gene that is inactivated in most colorectal cancers. Mutations of APC are considered to cause the accumulation of free β- catenin, which then binds TCF causing increased transcriptional activation of genes including genes important for cell proliferation (e.g. cyclin D1 (Tetsu er a/., Nature 398, 422-426, 1999 and Shtutman et al., PNAS USA, 96, 5522-5527, 1999) and c- myc (He et al., Science, 28_1, 1509-1512, 1998)). The involvement of APC in tumour development is discussed in He et al, (supra).

TCFs are known to recognise and bind TCF binding elements which have the nucleotide sequence CTTTGNN, wherein N indicates A or T (van der Wetering et al, supra).

TCF reporter genes have been constructed and are described in Korinek et al,

(Science, 275, 1784-1787, 1997), Morin et al, (Science, 275, 1787-1790, 1997), EP-

A- 0 939 122 and WO 98/41631. The TCF reporter gene is said to comprise three

TCF binding elements upstream of either a minimal c-Fos promoter driving luciferase expression or a minimal herpes virus thymidine kinase promoter driving chloramphenicol acetyl-transferase expression. He et al (supra) discloses TCF reporter gene constructs comprising four TCF binding elements inserted into pBV-

Luc.

Promoters containing TCF-binding elements are therefore useful for obtaining selective expression of operably-linked genes in cells in which there is an accumulation of nuclear β-catenin, which co-operates with TCF and activates expression in combination with other transcription factors. In addition to the use of such promoters allowing tumour-specific expression of prodrug-converting enzymes, as we have disclosed previously (WO 01/64739), alternative known therapeutic approaches include tumour-specific, replication-competent adenoviral vectors (Iggo et al, WO 00/56909). However, there still exists a need for effective means of selectively killing specific tumour cells and for a simple method to allow manufacture of the required expression vectors.

Statement of Invention

It is well-established that the multifunctional protein β-catenin is constitutively activated in the majority of colorectal cancers and many more tumour types (e.g. tumours of the liver and prostate, glioma, melanoma, various adenocarcinomas, osteosarcoma, and others). A targeted strategy to treat these tumours with viral vectors would be highly valuable.

We have developed a selective cytotoxic gene therapy strategy for the treatment of such tumours, in which the therapeutic gene product is intrinsically highly toxic. This demands a highly selective and stringent ('tight') promoter preventing breakthrough expression of toxin in non-tumour tissue in adenoviral producer cell lines. Insufficient specificity and stringency may result in undesirable killing of non- tumour cells, in a therapeutic application, and killing of the cells in which vectors are grown, complicating the manufacturing process.

Therefore, we optimised CTPI's activity/specificity profile as follows : (1) by replacing the SV40 promoter from CTP1 with a TATA box from the adenovirus E1A promoter; (2) by varying both the space between the TCF binding sites and the distance between the TCF sites and the TATA box, resulting in CTP3; and (3) by inserting 5 more TCF binding sites, resulting in our optimal promoter variant, CTP4. In summary, therefore:

CTP1 comprises 5 copies of the TCF-binding element CCTTTGATC (SEQ ID NO:1 ) separated by gaps of 10, 12, 10 and 10 nucleotides, respectively. The proximal binding element is separated from the TATA box of the SV40 promoter by 140 nucleotides.

CTP2 comprises the same arrangement of TCF-binding elements, but the proximal element is situated 46 nucleotides upstream of the TATA box of the adenovirus E1 A promoter. CTP3 comprises 5 copies of the TCF-binding element separated by 4, 3, 4 and 3 nucleotides, respectively with the proximal element 25 nucleotides upstream of the E1A TATA box.

CTP4 comprises 10 copies of the TCF-binding element, arranged as 2 sets of the CTP3 arrangement separated by 40 nucleotides (that is spaced by 4, 3, 4, 3, 40, 4, 3, 4, 3) again situated 25 nucleotides upstream of the E1A TATA box.

This improved promoter construct was designed for selective expression of a variety of highly toxic proteins and peptides, in particular diphtheria toxin A (DTA) and the gibbon ape leukaemia virus fusogenic membrane glycoprotein [GaLV FMG] (Fielding et al, 2000, Hum Gene Ther 11: 817-826; Johnson et al, 2003, Gene Ther 10: 725- 732).

Accordingly, the invention provides a method of manufacturing an expression vector encoding one or more gene products that are toxic to the host cell. Preferably the host cell is a eukaryotic cell. More preferably the method comprises introducing into a cell having no nuclear accumulation of β-catenin a vector encoding said toxic product, wherein expression of said toxic product is controlled by a control element comprising one or more TCF-binding elements.

As used herein, a toxic product is one that is intrinsically toxic to cells, requiring no further activation or extrinsically added substrate or prodrug.

More preferably said cell is a producer cell suitable for the production of viral vectors, still more preferably, adenoviral vectors. Most preferably it is a cell line that complements the growth of replication-deficient adenoviral vectors, such as PerC6.

In one embodiment of the invention, the vector comprises at least one copy of a TCF- binding element. Preferably this element comprises the nudeotide sequence 5'- CTTTGNN-3' (SEQ ID NO:2), wherein N indicates A or T. More preferably it is 5'- CCTTTGATC-3' (SEQ ID NO:1)

In a preferred embodiment, the vector comprises up to 10 copies of a TCF-binding element. Preferably the vector comprises 5 to 10 copies. Preferably the copies are spaced by 10 to 12 nucleotides. Alternatively, they are spaced by 3 or 4 nucleotides. In a most preferred embodiment, the vector comprises 10 copies of a TCF-binding element arranged with spaces of 4, 3, 4, 3, 40, 4, 3, 4, and 3 intervening nucleotides respectively. It is further preferred that the spacing between the proximal TCF- binding element and the promoter TATA box is between 10 and 100 nucleotides. More preferably it is between 10 and 50 nucleotides. Further preferably it is between 20 and 40 nucleotides and, most preferably, it is 24 nucleotides.

Highly preferred embodiments comprise constructs CTP1 , CTP2, CTP3 and CTP4 as disclosed above and in Figure 1.

In one aspect of the invention, the vectors encode toxic molecules. The toxic molecules may be proteins such as enzymes, enzyme inhibitors, fusogenic peptides, antibodies or functional fragments thereof, or may be ribonucleotides such as antisense RNA or ribozymes. Preferably such toxic molecules are proteins selected from the list consisting of ricin, abrin, diphtheria toxin, botulinum toxin, gibbon ape leukaemia virus (GaLV) fusogenic membrane protein or cytotoxic peptides derived therefrom. Most preferably the toxic protein molecule if diphtheria toxin A.

In another aspect the invention provides a vector comprising up to 10 copies of a TCF-binding element, operably linked to an expressible gene encoding a toxic protein molecule. Preferably the vector comprises 5 to 10 copies of a TCF-binding element. Preferably the copies are spaced by 10 to 12 nucleotides. Alternatively, they are spaced by 3 or 4 nucleotides. In a most preferred embodiment, the vector comprises 10 copies of a TCF-binding element arranged with spaces of 4, 3, 4, 3, 40, 4, 3, 4, and 3 intervening nucleotides respectively. The toxic molecules may be proteins such as enzymes, enzyme inhibitors, fusogenic peptides, antibodies or functional fragments thereof, or ribonucleotides such as antisense RNA or ribozymes. Preferably such toxic molecules are selected from the list consisting of ricin, abrin, diphtheria toxin, botulinum toxin, gibbon ape leukaemia virus (GaLV) fusogenic membrane protein or cytotoxic peptides derived therefrom. An especially preferred embodiment comprises an expressible gene encoding diphtheria toxin A (DTA).

As used herein 'gene' means a nucleic acid sequence comprising an expressible open reading frame, with or without introns or other non-coding or regulatory sequences, encoding a desirable protein, peptide or RNA. The vector is preferably a viral vector, more preferably an adenoviral vector. Also disclosed are nucleic acid constructs, including plasmids, encoding all or part of such a viral vector, such as those used to manipulate the viral genome prior to packaging. Most preferably, the vector comprises the control element CTP4, as shown in Figure 2.

Also disclosed are host cells containing any such nucleic acid construct, plasmid, vector, or adenovirus.

In another aspect of the invention, the cytotoxic effect is derived from another mechanism such as conditional replication of a virus, preferably an adenovirus.

The present invention also provides the nucleic acid construct, vector or host cell of the present invention for use in therapy.

Preferably, the nucleic acid construct, vector or host cell is used in the treatment of cancer.

The present invention also provides the use of the nucleic acid construct, vector or host cell of the present invention in the manufacture of a composition for use in the treatment of cancer.

The present invention also provides a method of treatment, comprising administering to a patient in need of such treatment an effective dose of the nucleic acid construct, vector or host cell of the present invention. Preferably, the patient is suffering from cancer.

Preferably, the cancer is any cancer associated a nuclear accumulation of β-catenin, as occurs when the Wnt signalling pathway is deregulated, such as colorectal cancer, melanomas, breast, prostate and hepatocellular carcinomas, glioma, various adenocarcinomas or osteosarcoma.

The present invention also provides a pharmaceutical composition comprising the nucleic acid construct, vector or host cell of the present invention in combination with a pharmaceutically acceptable recipient. The pharmaceutical compositions of the present invention may comprise the nucleic acid construct, vector or host cell of the present invention, if desired, in admixture with a pharmaceutically-acceptable excipient, buffer, carrier or diluent, for therapy to treat a disease.

The nucleic acid construct, vector or host cell of the invention or the pharmaceutical composition may be administered via a route which includes systemic, intramuscular, subcutaneous, intradermal, intravenous, aerosol, oral (solid or liquid form), topical, ocular, as a suppository, intraperitoneal and/or intrathecal and local direct injection.

The exact dosage regime will, of course, need to be determined by individual clinicians for individual patients and this, in turn, will be controlled by the exact nature of the protein expressed by the therapeutic gene and the type of tissue that is being targeted for treatment.

The dosage also will depend upon the disease indication and the route of administration.

The amount of nucleic acid construct or vector delivered for effective treatment according to the invention will preferably be in the range of between about 50 ng - 1000 μg of vector DNA kg body weight; and more preferably in the range of between about 1-100 μg vector DNA/kg.

Although it is preferred according to the invention to administer the nucleic acid construct, vector or host cell to a mammal for in vivo cell uptake, an ex vivo approach may be utilised whereby cells are removed from an animal, transduced with the nucleic acid construct or vector, and then re-implanted into the animal. The liver, for example, can be accessed by an ex vivo approach by removing hepatocytes from an animal, transducing the hepatocytes in vitro and re-implanting the transduced hepatocytes into the animal (e.g., as described for rabbits by Chowdhury et al., Science 254:1802-1805, 1991 , or in humans by Wilson, Hum. Gene Ther. 3:179-222, 1992). Such methods also may be effective for delivery to various populations of cells in the circulatory or lymphatic systems, such as erythrocytes, T cells, B cells and haematopoietic stem cells. The present invention also provides a composition for delivering the nucleic acid construct or vector of the present, to a cell.

Also provided is a toxin expressed by means of the method described above, or using the nucleic acid constructs or expression vectors herein provided.

Detailed Description of the Invention

The invention is described in more detail with reference to the Figures, as briefly described below.

Figure 1 depicts four β-catenin-dependent promoter constructs used to express toxic products such as diphtheria toxin-A in cells permissible for vector propagation but not for β-catenin -dependent gene expression. For the avoidance of doubt, the numbers below each diagram refer to the number of nucleotides (10, 12, 4 or 3) between the CCTTTGATC repeats. In the case of the gap between the proximal repeat and the E1A TATA box '140 to...', '46 to...' and '25 to...' is inclusive of the first nudeotide of the TATA box sequence (as made clear in Figure 2), and so the number of intervening nucleotides is 139, 45 and 24, respectively.

Figure 2 shows the nudeotide sequence of construct CTP4. The TCF-binding elements are underlined and the TATA box from the adenoviral serotype 5 E1A gene is in bold.

Figure 3. Shown is the DTA sequence as sequenced from the Ad.CTP4-DTA preparations LSP1 and LSP3 (see also Figure 4). "Query" is the sequence obtained by sequencing adenoviral DNA which was BLASTed using the NCBI homepage showing the result "Sbjct". The results indicate wild-type DTA DNA-sequence.

Figure 4. Three independent viral preparations of Ad.CTP4-DTA were generated indicating that the virus can be grown to normal particle/ml titres.

Figure 5. Ad.CTP4-DTA kills permissive, β-catenin-deregulated tumour cell lines. SW480 (mutant APC) and HepG2 (mutant β-catenin), were infected with Ad.CTP4- DTA (hatched) or Ad.CTP4-LacZ control virus (black) at the indicated MOIs. After four days cell viability was analysed using the MTS substrate assay (Promega) and reading the OD450. Infections were done in triplicates and shown is the mean OD450 value. One out of at least two to three representative experiments is shown.

Figure 6. Ad.CTP4-DTA shows no killing of non-permissive cells. HeLa, HNX14C and HUVEC (Human Umbilical Vein Endothelial Cells) were infected with Ad.CTP4- DTA (hatched) or Ad.CTP4-LacZ control virus (black) at the indicated MOIs. After four days cell viability was analysed using the MTS substrate assay (Promega) and reading the OD450. Infections were done in triplicates and shown is the mean OD450 value. One out of at least two to three representative experiments is shown.

Example 1

Materials and methods

Tissue Culture

SW480, HeLa, HNX14C and HepG2 were obtained from ATCC (Manassas, VA, USA). HUVEC were obtained from Promocell (Heidelberg, Germany) and cultured as recommended by the supplier. PER.C6 were obtained from Crucell (Leiden, The Netherlands). The adenoviral vector producer cell line 911 was used for plaque titering adenoviruses and was kindly provided by Prof. L. S. Young (Cancer Research UK Institute for Cancer Studies, University of Birmingham, Birmingham, UK). 911 were cultured in DMEM containing 10% FCS, 10 mM MgCI₂ and antibiotics. All purchased cell lines were cultured as recommended by the supplier.

Plasmid construction

To generate CTP4-DTA a cDNA coding for diphtheria toxin A (DTA) was amplified by PCR (oligol : 5'-CATGCCATGGGCGCTGATGATGTTGTTG (SEQ IS NO:3); oligo2: δ'-GCTCTA GACTATCGCCTGACACGATTTCCT) (SEQ ID NO:4) using the plasmid pDT201 (ATCC, Manassas, VA, USA) as template. The resulting PCR product comprises amino acids 26-193 of wild-type DT protein with an additional codon for methionine in front of amino acid 26. The PCR product was digested with A/col and Xba\ and then cloned into Nco\IXba\ digested CTP4-Luc thereby exchanging luciferase against DTA. The correctness of the DTA open reading frame was confirmed by DNA sequencing the plasmid DNA (Seqlab GmbH, Goettingen, Germany).

Adenoviral vector constructions

To generate pPS1128/CTP4-DTA-p(A) the CTP4-DTA-p(A) expression cassette (1.1 kb) was cut out by Nhe\/BamH\, T4 blunted and then cloned into blunted pPS1128 (lit). The virus Ad.CTP4-DTA were constructed by homologous recombination in PER.C6 cells using pPS1128/CTP4-DTA-p(A) and the overlapping adenoviral backbone vector pPS1 160 (lit), scaled-up and CsCI-purified and titred as described elsewhere using standard methods (lit).

Preparation of adenoviral DNA for DNA-sequencing

Viral DNA was prepared from about 2-4x10¹¹ CsCI-banded virus particles by incubation with 100 μg/ml proteinase K in 20 mM Tris/HCI pH7.5, 5 mM EDTA pH8.0, 0.1% SDS for 3-4 hours at 37°C. The crude DNA preparation was then extracted two times with an equal volume of Phenol:Chloroform:lsoamylalcohol (25:24:1), once with chloroform only and then precipitated with 1/10 Vol. 3.0 M NaAcetate and 2 Vol. 100% Ethanol. After centrifugation, (10 min at 13K at RT) the resultant DNA pellet was washed with 70% Ethanol, air-dried and then resuspended in water. The CTP4 promoter and the coding sequence for DTA of LSP1 and LSP3 of Ad.CTP4-DTA were sequenced by Seqlab GmbH, Germany.

In vitro cell killing assay

To study cell viability in vitro 10⁴ cells were infected with the indicated viruses and MOIs for 3-4 h and then seeded into 96-well plates with 200 μl medium containing 5% FCS. After four days 40 μl of CellTiter 96^® AQueous One Solution Cell Proliferation Assay MTS substrate (Promega, Madison, Wl, USA) was added and cell viability was calculated by reading the OD at 450 nm after subtraction of the medium background. For each data point triplicates were performed.

Results

We used the highly potent cytotoxin diphtheria toxin-A (DTA) to evaluate the usefulness of the CTP4 promoter, This allowed the successful construction and propagation of an Ad5-based vector capable of direct expression of DTA (Ad.CTP4- DTA). Ad.CTP4-DTA kills permissive SW480 cells in vitro but does not affect the viability of non-permissive HeLa cells. Ad.CTP4-DTA is genetically stable and can be propagated to normal high titres.

From the results obtained we conclude that the activity and selectivity profile of the CTP4 promoter enables the development of Ad vectors expressing extremely potent cytotoxins for treatment of β-catenin deregulated tumours.

Three independent preparations of Ad.CTP4-DTA were generated so far. The vector seems to be genetically stable and can be grown to high titres (LSP3=1.8x10¹² particles/ml; see Figure 4). DNA-sequencing of the DTA coding sequence and the CTP4 promoter did not indicate for any changes which could have favoured the growth of the virus by reducing the killing effect (Figure 3; e.g. point mutations within the coding sequence or mutations of the Tcf-binding sites).

Permissive tumour cell lines SW480 and HepG2 were infected with increasing MOIs of Ad.CTP4-DTA and control virus Ad.CTP4-LacZ. After four days cell killing was evaluated with the MTS solution assay (Promega). Both cell lines could be killed with relatively low MOIs (25-50; See Figure 5). In contrast, infection of the non-permissive tumour cell lines HeLa and HNX14C did not show more toxicity than the infection with the control virus Ads.CTP4-LacZ (Figure 6) confirming again the high level of specificity of the CTP4 promoter.

Claims

1. A method of manufacturing an expression vector encoding a toxic product, comprising introducing into a cell having no nuclear accumulation of β-catenin a vector encoding said toxic product, wherein expression of said toxic product is controlled by a control element comprising one or more TCF-binding elements.

2. The method of claim 1 , wherein the expression vector is an adenoviral expression vector.

3. The method of either of claims 1 or 2, wherein the control element comprises 5 to 10 copies of a TCF-binding element.

4. The method of claim 3, wherein the control element comprises 10 copies of a TCF-binding element.

5. The method of claim 4, wherein the 10 copies of the TCF-binding element are have intervening spaces of 4, 3, 4, 3, 40, 4, 3, 4, 3 nucleotides, respectively.

6. The method of claim 5, wherein the vector comprises the control element CTP4 as depicted in Figure 2 (SEQ ID NO: 5).

7. The method of any one of the above claims, wherein the TCF-binding element has the sequence CTTTGNN, wherein N is A or T.

8. The method of any of claims 1 to 7, wherein the toxic product is selected from the list consisting of ricin, abrin, diphtheria toxin, botulinum toxin, gibbon ape leukaemia virus (GaLV) fusogenic membrane protein or cytotoxic peptides derived therefrom.

9. An expression vector comprising a control element comprising one or more TCF- binding elements and an operably-linked expressible gene encoding a toxic product.

10. The expression vector of claim 9, wherein said vector is an adenoviral expression vector.

11. The vector of either of claims 9 or 10, wherein the control element comprises 5 to 10 copies of a TCF-binding element.

12. The vector of claim 11 , wherein the control element comprises 10 copies of a TCF-binding element.

13. The vector of claim 12, wherein the 10 copies of the TCF-binding element are have intervening spaces of 4, 3, 4, 3, 40, 4, 3, 4, 3 nucleotides, respectively.

14. The vector of claim 13, comprising the control element CTP4 as depicted in Figure 2.

15. The vector of any one of claims 9 to 14, wherein the TCF-binding element has the sequence CTTTGNN, wherein N is A or T.

16. The vector of any of claims 9 to 15, wherein the toxic product is selected from the list consisting of ricin, abrin, diphtheria toxin, botulinum toxin, gibbon ape leukaemia virus (GaLV) fusogenic membrane protein or cytotoxic peptides derived therefrom,

17. A host cell containing the vector of any one of claims 9 to 16.

18. The vector of any one of claims 9 to 16, or the host cell of claim 17 for use in therapy.

19. Use of the vector of any one of claims 9 to 16 or the host cell of claim 17 in the manufacture of a composition for use in the treatment of cancer.

20. A method of treatment, comprising administering to a patient in need of such a treatment an effective dose of the vector of any one claims 9 to 16 or the host cell of claim 17.

21.A pharmaceutical composition comprising the vector of one of claims 1 to 16, or the host cell of claim 17, in combination with a pharmaceutically-acceptable excipient, buffer, carrier or diluent.

22. A composition for delivering the vector of any of claims 9 to 16, or the host cell of claim 17, to a cell.

23. A toxin expressed by means of the method of any one of claims 1 to 8.