WO2001016167A2 - Beta-catenin, tcf-4, and apc interact to prevent cancer - Google Patents

Abstract

A recombinant adenovirus (Ad-Mini-ME) which constitutively expresses the central third of APC includes all of the known β-catenin binding repeats. When expressed in colon cancer cells, Ad-Mini-ME blocked the nuclear translocation of β-catenin and inhibited β-catenin/Tcf-4-mediated transactivation. Accordingly, expression of endogenous targets of the APC/β-catenin/Tcf-4 pathway were down-regulated. Ad-Mini-ME infection of colorectal cancer cell lines with mutant APC but wild-type β-catenin resulted in substantial growth arrest followed by apoptosis. These findings suggest that the β-catenin binding domain in the central third of APC is sufficient for its tumor suppresion activity.

Description

β-CATENIN, TCF-4, AND APC INTERACT TO PREVENT CANCER

The U.S. Government has a paid-up license in this invention and the right in limited circumstances to require the patent owner to license others on reasonable terms as provided for by the terms of grant CA57345 awarded by the National Institutes of Health.

This application is a continuation-in-part of application Serial No. 09/003,687 filed January 6, 1998, which is a division of application 08/821,355 filed March 20, 1997, now U.S. Patent No. 5,851,775. TECHNICAL FIELD OF THE INVENTION

This invention is related to the field of cancer diagnostics and therapeutics. More particularly it relates to methods for diagnosing and treating cancers associated with APC or β-catenin mutations. BACKGROUND OF THE INVENTION Mutations of the adenomatous polyposis coli (APC) gene are the most common disease-causing genetic events in humans; approximately 50% of the population will develop colorectal polyps initiated by such mutations during a normal life span (14). Individuals who inherit APC mutations develop thousands of colorectal tumors, consistent with APC's tumor suppressor or "gatekeeping" role in colorectal tumorigenesis (15,16). APC homodimerizes through its amino-terminus (17), and interacts with at least six other proteins: β-catenin (18), γ-catenin (plakoglobin) (19), tubulin (20), EB1 (21), hDLG, a homologue of a Drosophϊla tumor suppressor protein (22), and ZW3/GSK3β kinase (23). Whether any of these interacting proteins communicate APC growth-controlling signals is unknown. Thus there is a need in the art for a fuller understanding of how the tumor suppressor gene APC functions in cells.

SUMMARY OF THE INVENTION

It is an object of the present invention to provide human nucleotide sequences encoding transcriptional activation proteins.

It is another object of the present invention to provide isolated preparations of transcriptional activation proteins.

It is an object of the present invention to provide methods of determining the presence or absence in a cell of wild-type APC or a downstream protein in the APC transcription regulatory pathway.

It is an object of the present invention to provide a fusion protein useful for binding to β-catenin.

It is an object of the invention to provide a polypeptide useful for binding to β-catenin. It is an object of the invention to provide a method for inhibiting the growth of a tumor cell.

It is an object of the invention to provide a method for inducing apoptosis of a tumor cell.

It is an object of the invention to provide an adenovirus vector useful for binding to β-catenin.

Another object of the invention is to provide methods of identifying candidate drugs for use in Familial Adenomatous Polyposis (FAP) patients or patients with increased risk of developing cancer.

It is yet another object of the invention to provide methods of identifying candidate drugs for the treatment of cancer patients, in particular those with APC or β-catenin mutations.

Another object of the invention is to provide a method for diagnosing cancer in a sample suspected of being neoplastic. Another object of the invention is to provide a method for treating a patient with colorectal cancer or other cancer associated with FAP.

These and other objects of the invention are achieved by providing one or more of the embodiments described below. In one embodiment of the invention an intron-free DNA molecule is provided which encodes Tcf-4 protein as shown in SEQ ID NO: 2 or 4.

According to another embodiment of the invention an isolated Tcf-4 protein is provided. The protein is substantially free of other human proteins, and has a sequence as shown in SEQ ID NO: 2 or 4. In another embodiment of the invention a method is provided for determining the presence or absence in a cell of wild-type APC or a downstream protein in the APC transcription regulatory pathway. The method comprises the steps of: introducing a Tcf-responsive reporter gene into the cell; and measuring transcription of said reporter gene; wherein a cell which supports active transcription of said reporter gene does not have wild- type APC or does not have a wild-type downstream protein in the APC transcription regulatory pathway.

According to yet another embodiment of the invention a method is provided for determining the presence or absence in a cell of wild-type APC.

The method comprises the steps of: contacting a Tcf-responsive reporter gene with a lysate of the cell; and measuring transcription of said reporter gene; wherein a lysate which inhibits said transcription has wild-type APC.

In still another embodiment of the invention a method of identifying candidate drugs is provided. The drugs may be useful for treatment of FAP or other cancer patients or patients with increased risk of developing cancer. The method comprises the steps of: contacting a cell having no wild-type APC or a mutant β- catenin with a test compound; measuring transcription of a Tcf-responsive reporter gene, wherein a test compound which inhibits the transcription of the reporter gene is a candidate drug for cancer therapy.

According to yet another aspect of the invention another method is provided for identifying candidate drugs for use in for use in FAP patients, colon cancer patients, patients with mutations in β-catenin or APC, or patients with increased risk of developing cancer. The method, comprises the steps of: contacting a Tcf-responsive reporter gene with a test compound under conditions in which the reporter gene is transcribed in the absence of the test compound; and measuring transcription of the Tcf-responsive reporter gene; wherein a test compound which inhibits said transcription is a candidate drug for cancer therapy.

According to another aspect of the invention a method is provided for identifying candidate drugs for use in FAP patients or patients with increased risk of developing cancer. The method comprises the steps of: contacting a test compound with β-catenin and Tcf-4 under conditions in which β-catenin and Tcf-4 bind to each other; and determining whether the test compound inhibits the binding of β-catenin and Tcf-4, a test compound which inhibits the binding being a candidate for cancer therapy or prophylaxis.

According to still another embodiment of the invention a method is provided for diagnosing cancer in a sample suspected of being neoplastic. the method comprises the steps of: comparing a CTNNB sequence found in the sample to a second CTNNB sequence found in a normal tissue, wherein a difference between the first and second sequence is an indicator of cancer. According to one aspect of the invention a fusion protein is provided. It comprises an enzyme covalently linked to a portion of APC which comprises its β-catenin binding domain. A polynucleotide encoding the fusion protein is also provided. According to another embodiment a polypeptide is provided. The polypeptide comprises a portion of APC which comprises its β-catenin binding domain. An isolated and purified polynucleotide encoding the polypeptide is also provided.

According to another aspect of the invention a method is provided for inhibiting the growth of a tumor cell. A polynucleotide which encodes a polypeptide comprising the β-catenin binding domain of APC is administered to the tumor cell, whereby growth of the tumor cell is inhibited.

According to another aspect of the invention a method is provided for inducing apoptosis of a tumor cell. A polynucleotide which encodes a polypeptide comprising the β-catenin binding domain of APC is administered to the tumor cell, whereby apoptosis of tumor cells is induced.

Also provided by the present invention is an adenovirus vector which comprises a polynucleotide encoding the β-catenin binding domain of APC.

According to another aspect of the invention a method is provided for treating a patient with colorectal cancer or other cancer associated with FAP.

The method comprises the step of: administering to the patient a nucleotide sequence comprising a portion of the APC coding sequence, said portion consisting of the β-catenin binding site. According to another aspect of the invention a method is provided for treating a patient with colorectal cancer or other cancer associated with FAP. The method comprises the step of: administering to the patient a polypeptide comprising a portion of the APC coding sequence, said portion consisting of the β-catenin binding site. The present invention thus provides the art with diagnostic, therapeutic and drug discovery methods especially useful for FAP and other cancers with APC or β-catenin mutations.

BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1. Sequence comparison of hTcf-4 and hTcf-1.

Two alternative splice forms of hTcf-4 were identified, each encoding a different COOH-terminus. One form (hTcf-4E; SEQ ID NO:6) was homologous to hTCF-lE (Fig. 1A) (7); the other form (hTcf-4B; SEQ ID NO: 5) was homologous to hTcf-lB (Fig. IB). The highly conserved NH₂- terminal interaction domain and the High Mobility Group (HMG) box DN A- binding region are boxed. Abbreviations for the amino acids are: A. Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, He; IC, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gin; P, Ar g; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. The nucleotide sequence has been deposited in GenBank (accession number:)

Figs. 2A, 2B, and 2C. Analysis oϊhTcf-4 expression in colonic epithelium.

(Fig. 2A) Northern blot analysis of hTcf-4, hTcf-1, hLef-I expression in Jurkat T cells (lane 1); colonic mucosa (lane 2); colon carcinoma cell lines DLD-1 (lane 3), HCT116 (lane 4); SW480 (lane 5); SW620 (lane 6);

HT29 (lane 7). Lane 2 contains 5 μg total RNA; all others contain 15 μg total RNA. The positions of 18S and 28S ribosomal RNAs are shown. EtBr, ethidium bromide stain. (Fig. 2B) In situ hybridization of healthy human colon tissue to an hTcf-4 probe. (Fig. 2C) In situ hybridization to a negative control probe (a fragment of the E. coli neomycin resistance gene). Figs. 3A, 3B. Transactivational properties of β-catenin/hTcf-4.

All reporter assays were performed as duplicate transfections. For each condition, both values are shown. (Fig. 3 A) Reporter gene assays in HA 1.6 B cells. Cells were transfected by electroporation with 1 μg luciferase reporter plasmid, 5 μg β-catenin expression plasmid, and 3 II- hTcf-4 expression plasmids. Empty pCDNA was added to a total of 10 μg, plasmid DNA. (Fig. 3B) Reporter gene assays in SW480 colon carcinoma cells. Cells were transfected with 0.3 μg, of the indicated luciferase reporter gene, 0.7 μg pCATCONTROL as internal control, the indicated amounts of pCMVNeo- APC, and empty PCDNA to a total of 2.5 μg plasmid DNA. Control CAT values are given in the right panel.

Fig. 4. Constitutive presence of β-catenin-hTcf-4 complexes in APC^"' cells. Gel retardation assays were performed on nuclear extracts from the indicated cell lines before and after a 20-hour exposure to Zή^. Samples in lanes 1, 4, 7, 10 were incubated under standard conditions. To the samples in lanes 2, 5, 8, 11, 0.25 μg, anti β-catenin was added. To the samples in lanes 3, 6, 9, 12, 0.25 μg of a control (human CD4) antibody was added. N.S., nonspecific band also observed with mutant (nonbinding) probe (lane Mt).

Figs. 5A and 5B. Effects of APC mutations on catenin regulated transcription (CRT). (Fig. 5A) Schematics of wild-type (WT) and mutant

APC. APC is a 2843-amino-acid (AA) protein (32) with contains armadillo (ARM) repeats in the amino-terminus (33), 15 and 20 AA β-catenin-binding repeats in the central region (18,19), and a basic region in the carboxyl-terminus (32). The carboxyl-terminus also contains a TXV sequence which mediates DLG binding (22). (Fig. 5B) Effects of WT and mutant

APC on CRT. SW480 cells containing endogenous mutant APC were transfected with the APC expression vectors shown in (Fig. 5A) and CRT was measured. Cells were transfected with increasing amounts of WT APC (0, 0.15 and 0.5 μg) or 0.5 μg mutant APC. CRT reporter activities are expressed relative to assays containing no WT APC and are the means of three replicates. Error bars represent standard deviations.

Lipofectamine was used to cotransfect SW480 cells with an internal control (0.5 μg pCMV-βgal), a reporter construct (0.5 μg pTOPFLASH or pFOPFLASH) and the indicated amount of the various APC expression vectors. The pTOPFLASH reporter contained an optimized Tcf-binding site 5' of a luciferase reporter gene, whereas pFOPFLASH contained a mutated site that does not bind Tcf. The amount of DNA in each transfection was kept constant by addition of an appropriate amount of empty expression vector (pCEP4). Luciferase and β-galactosidase activities were determined 16 hours after transfection. Luciferase activity was corrected for transfection efficiency (using the control β-galactosidase activity) and nonspecific transcription (using the pFOPFLASH control).

Figs. 6A and 6B. Evaluation of CRT in colorectal cancer cell lines with WT APC. (Fig. 6A) Immunoblot of endogenous APC in the DLD 1 , S W480,

HCT116, SW48 and 293 cell lines, developed with APC monoclonal antibody FE9 (34). (Fig. 6B) Effects of exogenous WT APC on CRT in cell lines with endogenous mutated or WT APC. Cells were transfected with increasing amounts (0, 0.15 μg, 0.5 μg for DLD1 and SW48; 0, 0.5 μg, 5 μg for HCT116) of WT APC or APC1309Δ mutant (0.5 μg for DLD1 and SW48; 5 μg for HCT116) and CRT was assessed as in Fig. 5. CRT reporter activities are expressed relative to activity in extracts without exogenous APC and are the means of three replicates. Error bars represent standard deviations.

Figs. 7 A, 7B, and 7C. Evaluation of β-catenin in colorectal cancer cell lines with WT APC. (Fig. 7A) Immunoblot of the cell lines used in this study, developed with β-catenin monoclonal C 19220 (Transduction Laboratories, Lexington, KY)(57). (Fig. 7B) Sequence of CTNNBl in HCT116 and SW48. Overlapping segments constituting the entire CTNNBl were amplified by RT-PCR from SW480, DLD1, HCT116, and SW48 cells, and sequenced directly with ThermoSequenase (Amersham). In the case of HCT1 16, a PCR product containing the deleted region was also cloned into pCI-neo (Promega, Madison) and multiple clones corresponding to each allele were individually sequenced.

The left panel (nts 121 to 143 from HCT116) reveals the presence of a deletion in addition to the WT sequence. The middle panel (antisense strand

156 to 113 of the WT and deleted alleles of HCT116) reveals the 3-bp deletion (ΔTCT) that removed codon 45 in half the clones. The right panel (nts 80 to 113 from SW48) reveals a C to A transition affecting codon 33 (TCT to TAT). (Fig. 7C) Schematic of β-catenin illustrating the armadillo repeats (33) in human (SEQ ID NO: 10), xenopus (SEQ ID NO: 10), and drosophila (SEQ ID

NO: 1 1), and negative regulatory domain. The residues in larger type fit the consensus sequence for GSK3β phosphorylation (29) and those in bold have been demonstrated to affect down regulation of β-catenin through GSK3β phosphorylation in Xenopus embryos (27). The five mutations found in human colon cancers are indicated at the top.

Figs. 8A and 8B. Functional evaluation of β-catenin mutants. (Fig. 8A) Constitutive nuclear complex of β-catenin and Tcf in HCT116 cells. The presence of nuclear β-catenin-Tcf complexes was assessed by gel shift assays. Lanes 1 to 3, optimal Tcf retardation probe shifted with nuclear extract from HCT116 cells with addition of no antibody (lane 1), anti β-catenin (0.25 μg, lane 2), or an irrelevant antibody (0.25 μg, lane 3). Lane 4, mutant Tcf retardation probe shifted with nuclear extract from HCT116 cells, n.s., nonspecific shifting seen with the mutant probe. (Fig. 8B) Effects of the β-catenin mutations on CRT. 293 cells were transfected with WT (WT) or mutant (Δ45, S33Y) β-catenin and CRT was assessed. CRT reporter activities are expressed relative to WT β-catenin and are the means of three replicates. Error bars represent standard deviations, β-catenin expression constructs were prepared as follows. WT CTNNBl was amplified by RT-PCR from SW480 cells and cloned into the mammalian expression vector pCI-neo (Promega) to produce pCI-neo-β-cat. The pCI-neo-β-cat Δ45 and S33Y were generated by replacing codons 1 to 89 in pCI-neo-β-cat with a PCR product encoding the equivalent region from HCT116 or SW48 cDNA, respectively. The structures of all constructs were verified by sequence analysis. Lipofectamine was used to cotransfect 293 cells with an internal control (0.1 μg CMV-βgal), a reporter (0.5 μg pTOPFLASH or pFOPFLASH), a Tcf-4 expression vector (0.5 μg pCDNA-TCF4), and β-catenin (0.5 μg) or dominant negative hTcf-4 1.0 μg) expression vectors. CRT was determined as described above. Figs. 9A, 9B and 9C. Characterization of the Ad-Mini-ME. Fig. 9A.

Map of GFP/cAPC incorporated into Ad-Mini-ME. Linear representation of different domains in APC including oligomerization domains, armadillo repeats, 15-aa repeats, 20-aa repeats, basic domain, and EB-1 binding site are shown. The β-catenin binding and degradation domain which comprises 15-aa repeats, 20-aa repeats and SAMP repeats was fused with the carboxyl-terminal of GFP. Expression of this cassette is driven by a CMV promoter in Ad-Mini-ME adenovirus vector. Fig. 9B. Western blot analysis with an anti-GFP antibody. Infection of SW480 cells with Ad-Mini-ME resulted in production of a fusion protein of the expected size (150 kD) whereas the Ad-GFP infected cells generated the expected 17 kD GFP. Fig. 9C.

Fluorescence microscopy revealed that the GFP/cAPC (green fluorescence protein/central region of APC) fusion protein was diffusely localized to the cytoplasm in DLD1 and HCT116 cells.

Figs. 10A and 10B. Ad-Mini-ME suppresses CRT. Fig. 10A. The luciferase activity was dramatically inhibited in Ad-Mini-ME infected DOT cells as compared with control and Ad-GFP infected DOT cells. Fig. 10B. No significant differences in luciferase activity among control, Ad-GFP infected, and Ad-Mini-ME infected Dluc cells were observed at day 1 and 2. Values are the average of three experiments. Fig. 11. Western blot analysis of c-MYC and cyclin Dl in DLD1, SW480, and HCTl 16. As compared with the control and Ad-GFP infected cells, c-MYC expression is strongly repressed in both Ad-Mini-ME infected DLD1 and SW480 cells but is only partially inhibited in Ad-Mini-ME infected HCTl 16. Similarly, repression of cyclin Dl expression by Ad-Mini-ME is observed in DLDland SW480 but not in HCTl 16. Similar expression levels of α-tubulin in each lane indicate that similar amounts of protein are loaded.

Figs. 12A and 12B. Immunofluorescence staining of β-catenin in DLD1 (Fig. 12A) and SW480 (Fig. 12B). Cells were infected with Ad-Mini-ME or Ad-GFP as indicated. Cells were stained for β-catenin and counterstained with DAPI as indicated. Ad-GFP infected cells demonstrate a predominant nuclear staining of β-catenin. In contrast, almost all the Ad-Mini-ME infected cells exhibited cytoplasmic and membrane β-catenin staining with minimal nuclear staining.

Figs. 13A and 13B. Cell growth and colony formation assays. Fig. 13 A. Growth kinetics in DLD1, SW480, and HCTl 16 cells after mock-, Ad-GFP- or Ad-Mini-ME infections, as indicated. Fig. 13B. Colony formation in collagen gel after mock-, Ad-GFP- or Ad-Mini-ME infections, as indicated. The number of colonies is indicated below each well.

Fig. 14. Apoptosis in virus-infected cells. Cells were stained with Hoechst33258 dye and at least 500 cells were counted and the results are expressed as the fold increase in the percentage of apoptotic cells in each sample. Fig. 15. Proliferation and apoptosis in DLD1 induced tumors.

Subcutaenous tumors in nude mice were generated by injecting Ad-Mini-ME or Ad-GFP infected DLD cells as indicated. Hematoxylen and eosion (H&E) staining revealed extensive apoptosis in Ad-Mini-ME infected cells. Immunoperoxidase staining with the M30 antibody (M30) which recognizes apoptotic cells revealed that the majority of Ad-Mini-ME infected cells were positive while only a few Ad-GFP infected cells were positive for this antibody. Immunoperoxidase staining with an antibody specific for the proliferation marker Ki-67 (Ki-67) revealed a remarkable decrease in Ki-67 nuclear labeling in Ad-Mini-ME infected cells. In contrast, the Ad-GFP infected cells were diffusely positive.

DETAILED DESCRIPTION It is a discovery of the present invention that hTcf-4 binds to β-catenin and activates transcription in colorectal epithelial cells. Moreover, it has now been found that APC regulates this transcriptional activation, at least in part by binding to β-catenin. In colorectal cancer cells this regulation is frequently abrogated, either by mutation of APC or by mutation of β-catenin. Two alternative splice forms of human Tcf-4 have been found. One form (hTcf-4E) is homologous to hTcf-lE and the other (hTcf-4B) is homologous to hTcf-lB. The sequence of the nucleotide and amino acid sequences are shown in SEQ ID NOs: 1-4. The coding sequences and proteins can be used in assays as described below. Intron-free DNA molecules are provided which are originally made by reverse transcription of a mRNA molecule. They can be propagated in cells or amplified as is desired. Isolated Tcf-4 proteins can be provided substantially free of other human proteins if, for example, the nucleotide sequences are expressed in non-human cells. Methods and vectors for achieving such expression are well known in the art. Choice of such expression means is made by the skilled artisan according to the desired usage and convenience.

Cells can be tested to determine if they have a wild-type APC or a wild- type downstream protein in the APC transcription regulatory pathway, called herein the CRT pathway ( β-catenin/Tcf-regulated transcription). One protein within the CRT pathway which has been identified as a target of mutations in human cancers is β-catenin (encoded by the CTNNBl gene). Other parts of the pathway are also likely to be targets. Although the target genes of the CRT pathway have not been identified, they can be readily identified using the system disclosed here. Genes which are differentially transcribed in the presence of wild-type and mutant CTNNBl, for example, can be identified. Tcf-responsive reporter genes are those constructs which comprise a readily detectable or assayable gene (such as luciferase, β- galactosidase, chloramphenicol acetyltransferase) linked in cis to a Tcf-responsive element.

Such responsive elements are known in the art (7) and any such elements can be used. An optimal Tcf motif contains the sequence CCTTTGATC. From one to twenty copies, and preferably from three to six copies, of the motif may be used. Mutation of the sequence to CCTTTGGCC abrogates responsiveness. Another necessary part of such constructs is a minimal promoter, such as the c-Fos or the Herpes virus thymidine kinase promoter. Transcription of the reporter gene may be performed by any means known in the art, usually by assaying for the activity of the encoded gene, although immunological detection methods can also be used. In addition, transcription can be monitored by measuring the transcribed mRNA directly, typically using oligonucleotide probes.

As shown below, a cell which has a wild-type APC protein will inhibit CRT. However, most mutations in APC render APC unable to inhibit CRT. Similarly, certain mutations in CTNNBl render β-catenin super-active and/or refractory to the inhibition by APC. Thus measuring Tcf-responsive reporter gene transcription is an indication of the status of APC and CTNNBl. Mutations in both of these genes are associated with cancers and therefore provides diagnostic and prognostic information. Assays for CRT can be accomplished in vitro or in cells. If the assay is to be accomplished in cells, then a Tcf-responsive reporter gene must be introduced into the cell. Any means for introducing genetic material into cells can be used, including but not limited to infection, transfection, electroporation. If the assay is to be performed in vitro then the components for transcription must be present. These include suitable buffers, RNA polymerase, as well as ribonucleotides. If the protein product is to be assayed, then the components for translation must also be present, such as ribosomes, and amino acids. These assays can also be used to screen compounds for potential as anti- cancer therapeutic agents. Using either the in vitro or cell form of the assay, test compounds can be introduced to determine whether they are able to mimic the effect of wild-type APC or to convert a mutant APC into a form which is able to inhibit CRT or a mutant β-catenin into a form which is regulatable by APC. In addition, compounds can be tested for the ability to inhibit the binding of β-catenin and Tcf-4, thus mimicking the action of APC. Such a test can be conducted in vitro or in vivo, for example using a two hybrid assay.

A means for diagnosis of cancers is the result of the observation that CTNNBl mutations are found in tumor cells, especially those which have wild-type APC. Such mutations can be found, inter alia, by sequencing either the gene or the protein found in a sample. Functional assays can also be used, such as whether β-catenin binds to APC or Tcf-4, or whether it is capable of mediating CRT. Sequences can be compared to those found in a normal tissue of a human, especially the same human who provided the sample being tested. Suitable tumors for testing include, but are not limited to those which are associated with FAP. Suitable tumors include colorectal cancer, thyroid cancer, brain cancer, medulloblastoma, desmoid tumor, osteoma, breast cancer, and head and neck cancer. Because APC mutations are so frequent, and because it appears that APC mutations do not occur in the same tumors as CTNNBl mutations, one can prescreen samples for APC mutations before performing a CTNNBl determination.

Fusion proteins are known in the art and are the product of recombinant technology. At least two polypeptides or proteins which derive from different naturally occurring proteins are joined together as a single polyamino acid sequence. Typically this is accomplished by engineering two polynucleotides which encode the polypeptides or proteins so that they form a single message- encoding unit which contains a single open reading frame. One of the polypeptides can be an enzyme or other readily detectable protein. One of the polypeptides is a portion of APC which comprises the β-catenin binding site. The APC-derived polypeptide may be the N-terminal or the C-terminal portion of the fusion protein, or if more than two polypeptides are joined together, the APC-derived polypeptide can be internal. One particularly useful fusion partner is green fluorescent protein (GFP) which is readily detectable. Other useful enzymes, such as luciferase, β-galactosidase, and chloramphenicol acetyltransferase can also be used as is convenient. Other polypeptide moieties, such as epitope tags, can also be engineered into the fusion protein. These can be used, for example, to facilitate detection and purification of the fusion protein. The fusion protein is advantageous because it permits an independent means of verification that the APC-derived polypeptide is being expressed in the cell. However, if such verification is not desired or required, a non-fused polypeptide portion of APC can be used. Such portion must be functional in binding to β-catenin. Preferably the portion will not contain at least one of the following domains: oligomerization domain, armadillo repeat domain, human homolog of Drosophila tumor suppressor gene discs large domain, and EB1 binding domain. These domains are apparently not required to mediate growth inhibition and apoptosis. Polynucleotides which encode the APC-derived portion, whether in fusion protein form or as a polypeptide, can be regulated using any of the promoters, transcription terminators, enhancers, and transregulators known in the art. Expression can be regulated or constitutive as is desired. The regulation may be by an exogenous agent, such as a drug which can be administered, or by endogenous agents, such as developmentally or environmentally regulated factors. The polynucleotide may be isolated and purified away from other molecules, and in particular isolated and purified from full-length APC-encoding polynucleotides. Methods fof isolating and purifying polynucleotides of known sequence and/or size are known in the art, and any such method can be used. Any vector known in the art can be used, including viral vectors, plasmids, mini-chromosomes, and yeast artificial chromosomes. One particularly preferred vector is an adenovirus vector.

Any method known in the art can be used for administering polynucleotides to tumor cells or subjects. These include without limitation, transfection, liposomes, viral infection, particle bombardment, nanoparticles, microparticles, compacted nucleic acid complexes, intravenous, intradermal, subcutaneous, intramuscular, direct injection into a tumor. Combinations of these methods may also be used. The portion of the APC gene which encodes the β-catenin binding site can be used in a gene delivery format. This portion comprises approximately amino acid residues 958 to 2075, although fewer residues may be sufficient to bind to β-catenin. For example, residues 1000-2000, 1100-1900, 1200- 1800, or 1300-1700, may be sufficient to bind to β-catenin. Suitable techniques are known in the art for administering genes to tumors and tumor cells, and any such technique can be used. Suitable expression vectors are also known in the art and it is within the skill of the artisan to select an appropriate one. Upon expression in a tumor cell of the β-catenin binding portion of APC, β-catenin will be bound and titrated away from binding to Tcf-4, thus preventing unregulated expression of the CRT target genes. Similarly, a polypeptide portion of APC containing the β-catenin binding site can be administered to cells to perform a titration of β-catenin. Techniques for such administration to cells is well known in the art. Cells which are treated with either the polynucleotide or the polypeptide can be used to study the interaction between APC and β-catenin, and for developing drugs which interfere with such binding. Such treatment can also induce apoptosis and inhibit cell growth.

Polynucleotides may be administered to cells, subjects, or patients according to the present invention for the purpose of screening for agents which enhance polynucleotide transfer to cells or enhance subsequent biological effects of the polynucleotides or their encoded proteins. Such biological effects include expression of the polynucleotide to form mRNA and/or protein, replication of the polynucleotide, hybridization to complementary messenger RNA and inhibition of its translation, and integration of the polynucleotide. According to the present invention, expression of a polypeptide or fusion protein encoded by the administered polynucleotide is preferred so that the polypeptide or fusion protein can bind to β-catenin. The result of such binding is inhibition of tumor cell growth and/or induction of apoptosis. Agents can also be screened for the ability to enhance the binding of the encoded polypeptide or fusion protein to β-catenin. Agents which can be screened include any test compounds or substances, whether natural products or synthetic, which can be administered to the cells, subject, or patient. Libraries or mixtures of compounds can be tested. The compounds or substances may be those for which a pharmaceutical effect is previously known or unknown. The compounds or substances may be delivered before, after, or concomitantly with the polynucleotides. They may be administered separately or in admixture with the polynucleotides. Integration of delivered DNA can be monitored by any means known in the art. For example, Southern blotting of the delivered DNA can be performed. A change in the size of the fragments of the delivered polynucleotide indicates integration. Replication of the delivered polynucleotide can be monitored inter alia by monitoring incorporation of labeled nucleotides combined with hybridization to a probe for the delivered polynucleotides. Expression of the polynucleotide can be monitored by detecting production of RNA which hybridizes to the delivered polynucleotide or by detecting protein encoded by the delivered polynucleotide. The protein can be detected immunologically or by activity, for example. The expression of the protein can also be monitored by detecting the effects of β-catenin binding. Thus the administration of the polynucleotides according to the present invention provides an excellent system for screening agents for their ability to promote delivery, integration, hybridization, expression, replication or integration in cells or in an animal, preferably a mammal, more preferably a human. Tumor cell inhibition can be measured or assessed according to any method known in the art. Tumor mass may be observed and measured. Time until palpable tumor development after administration of tumor cells to a nude mouse can be assessed. Survival time of animals with tumors can be measured. Incoφoration of tritiated thymidine can be measured. Focus formation can be measured. Any method known in the art can be used, without limitation. Monitoring of apoptosis can also be done as is known in the art. DNA fragmentation, a hallmark of apoptosis, can be assayed by any means known in the art, including but not limited to electrophoresis on a gel, staining nuclei with TUNEL, and photometric enzyme immunoassay for the determination of cytoplasmic histone-associated DNA fragments. The assays can be performed in vitro or in vivo.

The above disclosure generally describes the present invention. A more complete understanding can be obtained by reference to the following specific examples which are provided herein for puφoses of illustration only, and are not intended to limit the scope of the invention.

EXAMPLE 1 This example identifies Tcf-4 as the expressed family member in colorectal epithelial cells and provides the complete sequence of the cloned cDNA.

There are four known members of the Tcf/Lef family in mammals: the lymphoid-specific factors Tcf- I and Lef- 1 (7,8), and the less well characterized Tcf-3 and 4(9). We performed a qualitative Reverse Transcriptase-Polymerase Chain Reaction assay for expression of the four

Tcf/Lef genes on 43 colon tumor cell lines. While most colon cell lines expressed more than one of the genes, only hTcf-4 mRNA was expressed in essentially all lines.

We then screened a human fetal CDNA library and retrieved clones encoding full-length hTcf-4 (Fig. 1). A genomic fragment encoding, the

HMG box region of hTcf-4 (7) was used to probe a human 12 week-fetal cDNA library in Lambda GT-11. Positive clones were subcloned into pBluescriptSK and sequenced. See SEQ ID NOs: 1 and 3. The predicted sequence of hTcf-4 was most similar to that of hTcf-1. Alternative splicing yielded two COOH-termini that were conserved between hTcf- 1 and hTcf-4.

The NH₂-terminus, which in hTcf- 1, mLef-1 and Xenopus TCF-3 mediates binding to β-catenin (6), was also conserved in hTcf-4. Northern blot analysis of selected colon carcinoma cell lines revealed high-level expression of hTcf-4 (Fig. 2A). Northern blot hybridizations (7) were performed with full-length hTcf-1, hLef-Iαnd hTcf-4 cDNA. Colon epithelial cells were freshly prepared from a mucosal preparation dissected from a healthy surgical colon sample. The sample was minced, and incubated with 1 mM dithiothreitol (DTT) in Hanks' medium to remove mucus. Single-cell suspensions were prepared by incubation at RT in 0.75 mM EDTA in Hanks' medium. Epithelial cells were separated from lymphocytes by Percoll gradient centrifugation.

As evidenced by in situ hybridization (Fig. 2, B and C) and Northern blotting (Fig. 2A), hTcf-4 MRNA was readily detectable in normal colonic epithelium, whereas hTcf-I and hLeϊ-\ were not detectable. In situ hybridization of 6 μ frozen sections of healthy colon biopsy samples were performed as described(lθ). hTcf-4 cDNA encoding amino acids 200 to 310 was amplified and labeled with Dig-11-dUTP (Boehringer Mannheim, Germany) by PCR. After hybridization and washing, the sections were sequentially incubated with mouse anti-Dig antibody (Boehringer) and a horseradish peroxidase conjugated rabbit antibody to mouse immunoglobulin (Dako, Glostrup, Denmark). The signal was visualized with diaminobenzidine, which produces a reddish-brown precipitate. Blue counterstaining was performed with hematoxylin.

EXAMPLE 2

This example demonstrates the interaction of Tcf-4 and β-catenin and their function as a transcriptional activating factor.

To investigate whether hTcf-4 functionally interacts with β-catenin, we used two sets of reporter constructs in a β -catenin-Tcf reporter gene assay (7). One contained three copies of the optimal Tcf motif CCTTTGATC, or three copies of the mutant motif CCTTTGGCC, upstream of a minimal c-Fos promoter driven-luciferase expression (PTOPFLASH and PFOPFLASH). The second set contained three copies of the optimal motif, or three copies of the mutant motif, upstream of a minimal Heφes virus thymidine kinase promoter driven-Chloramphenicol Acetyl Transferase (CAT) expression (PTOPCAT and PFOPCAT, respectively). Reporter gene assays were performed as in (7). In brief, 2 x 10⁶ cells were transfected with plasmids by electroporation. After 24 hours, cells were harvested and lysed in 1 mM DTT, 1 % Triton X- 100, 15 % glycerol, 25 mM Tris pH 7.8 and 8 mM MgCl₂. cDNAs encoding Myc-tagged versions of β-catenin and hTcf-4 were inserted into the mammalian expression vector pCDNA (Invitrogen). PCATCONTROL, encoding the CAT enzyme under the control of the SV40 promoter, was purchased from Promega.

Epitope-tagged hTcf-4 and a deletion mutant lacking, the NH₂-terminal 30 amino acids (ΔNhTcf-4) were cloned into the expression vector pCDNA. Transient transfections were performed in a murine B cell line (IIA1.6), that does not express any of the Tcf genes (6). The TOPFLASH reporter was strongly transcribed upon cotransfection with the combination of β-catenin and hTcf-4 plasmids, but not with the individual plasmids or with the combination of β-catenin and ΔNhTcf-4 plasmids. No enhanced transcription was detected in cells transfected with the negative control PFOPFLASH (Fig. 3A). These results show that interaction of the NH₂-terminus of hTcf-4 with β-catenin results in transcriptional activation.

EXAMPLE 3 This example demonstrates the functional regulation of CRT transcriptional activation by wild-type APC. In three APC^"'^" carcinoma cell lines, SW480, SW620 and DLD- 1 (Fig.

3B), the PTOPFLASH reporter was 5-20 fold more actively transcribed than PFOPFLASH. Importantly, transfection of SW480 cells with the reporter gene and an APC-expression vector abrogated the transcriptional activity in a dose-dependent manner (Fig. 3B). In contrast APC had no effect on a cotransfected internal control (pCATCONTROL), or on the basal transcription of PFOPFLASH (Fig. 3B). The use of PTOPCAT and PFOPCAT instead of PTOPFLASH and PFOPFLASH led to comparable observations. The constitutive transcriptional activity of Tcf reporter genes in APC^"'^" colon carcinoma cells was in stark contrast to the inactivity of these genes in non- colonic cell lines, including IIA1.6 B cells (Fig. 3 A), the C57MG breast carcinoma cell line; the Jurkat and BW5147 T cell lines; the Daudi and NS1 B cell lines; the K562 erythromyeloid cell line; the HeLa cervical carcinoma line; the HepG2 hepatoma cell line; 3T3, 3T6, and Rat-I fibroblasts; and the kidney derived SV40-transformed COS cell line (7,16).

EXAMPLE 4 This example demonstrates that a functional β-catenin-hTcf-4 complex exists constitutively in APC^"'^" cells.

We used HT29-APC^"'^" colon carcinoma cells (12), in which APC is controlled by a metallothionein promoter. Induction by Zn^"1^ restores wild- type levels of APC, and leads to apoptosis (12). HT29-Gal cells which carry a Zn^-inducible LacZ gene were used as a control. The only Tcf family member expressed in HT29 is hTcf-4 (Fig. 2C). In nuclear extracts from uninduced HT29 derived transfectants, we readily detected hTcf-4 by gel retardation (Fig. 4). An additional band of slightly slower mobility was also observed. The addition of a β-catenin antibody resulted in the specific retardation of the latter band, indicating that it represented a β-catenin-hTcf-4 complex (Fig. 4) (12). After Zn^"^ induction for 20 hours, the β-catenin- hTcf-4 complex was diminished sixfold relative to uncomplexed hTcf-4 in HT29-APC1, while no significant change was observed in HT29-Gal cells

(Fig. 4). Importantly, the overall levels of cellular β-catenin do not change during the induction period in HT29-APC1 cells (12).

Gel retardation assays were performed as described elsewhere (7). Extracts were prepared from intact nuclei that were washed four times to avoid contamination with cytoplasmic β-catenin. As the optimal Tcf/Lef probe, we used a double-stranded 15-mer CCCTTTGATCTTACC (SEQ ID NO: 3); the control probe was CCCTTTGGCCTTACC (SEQ ID NO: 4). (All oligonucleotides were from Isogene, Holland). The β-catenin antibody was purchased from Transduction Laboratories (Lexington, KY). A typical binding reaction contained 3 μg nuclear protein, 0. 1 ng radiolabeled probe, 100 ng of dldC, in 25 μl of binding buffer (60 mm KC1, 1 mM EDTA, 1 mM DTT, 10% glycerol). Samples were incubated for 20 min at room temperature, antibody was added, and the samples incubated 20 min further. On the basis of these data, we propose the following model. In normal colonic epithelium hTcf-4 is the only expressed member of the Tcf family. The interaction of β-catenin with hTcf-4 is regulated by APC. When appropriate extracellular signals are delivered to an epithelial cell, β-catenin accumulates in a form that is not complexed with GSK3β-APC, and that enables its nuclear transport and association with hTcf-4. The HMG domain of hTcf-4 binds in a sequence-specific fashion to the regulatory sequences of specific target genes; β-catenin supplies a transactivation domain. Thus, transcriptional activation of target genes occurs only when hTcf-4 is associated with β-catenin. The hTcf-4 target genes remain to be identified. However, the link with APC and catenin suggests that these genes may participate in the generation and turnover of epithelial cells. Upon loss of wild-type APC, monomeric β-catenin accumulates in the absence of extracellular stimuli, leading to uncontrolled transcription of the hTcf-4 target genes. The apparent de novo expression of other members of the Tcf family in some colon carcinoma cell lines might lead to a further deregulation of Tcf target gene expression by the same mechanism. The control of β-catenin -Tcf signaling is likely to be an important part of the gatekeeper function of APC (19), and its disruption an early step in malignant transformation.

EXAMPLE 5 This example demonstrates that mutant APC protein does not regulate

CRT and that a complete set of 20-AA repeats in APC is required to mediate inhibition of CRT. We tested four APC mutants (Fig. 5A) for their ability to inhibit β-catenin/Tcf-regulated transcription (CRT) in transfection assays. The first mutant, APC331Δ represents a type of mutation found in the germline of Familial Adenomatous Polyposis (FAP) patients as well as in sporadic tumors (15). The APC331 Δ protein is truncated at codon 331, amino-terminal to the three 15-amino-acid (AA) β-catenin-binding repeats between codons 1020 and 1169. The second mutant, APC1309Δ, is the most common germline APC mutation (15), a 5-bp deletion that produces a frameshift at codon 1309 and truncation of the protein. The APC1309Δ protein retains the 15-AA β-catenin repeats but lacks the seven 20-AA repeats between codons 1323 and 2075 that have been implicated in binding and phosphorylation of β-catenin (18). The third mutant, APC1941Δ, represents one of the most distal somatic mutations observed in colorectal tumors (25). The APC1941Δ protein is truncated at codon 1941 and therefore contains the 15-AA repeats and all but the last two 20-AA repeats. Finally, APC2644Δ represents a germline mutation resulting from a 4-bp deletion in codon 2644. Patients with this type of unusual carboxyl-terminal mutation develop few polyps (attenuated polyposis) but have pronounced extracolonic disease, particularly desmoid tumors (26).

Each of the APC mutants was cotransfected with a CRT reporter into the SW480 colorectal cancer cell line. SW480 cells have truncated APC and constitutively active CRT which can be suppressed by exogenous WT APC. Although all four mutants produced comparable levels of APC protein after transfection, they varied in their CRT inhibitory activity. The three mutants found in patients with typical polyposis or cancer were markedly deficient in inhibition of CRT (Fig. 5B). The reduced activity of APC1309Δ and

APC1941Δ suggests that β-catenin binding is not sufficient for APC-mediated inhibition of CRT and that the complete set of 20-AA repeats is required. Interestingly, the inhibitory activity of the APC2644Δ mutant associated with attenuated polyposis was comparable to that of WT APC (Fig. 5B), suggesting that the DLG-binding domain at the carboxyl-terminus of APC is not required for down-regulation of CRT.

WT and mutant APC constructs (2 μg) were transfected into 293, SW480, and HCTl 16 cells using Lipofectamine (GIBCO/BRL, Gaithersburg). Protein was harvested 24 hours later and subjected to immunoblot analysis with APC monoclonal antibody FE9 (23). In HCTl 16 and 293 cells, exogenous WT APC comigrated with the endogenous APC. In SW480 cells, APC1309Δ comigrated with the endogenous mutant APC. In all other cases, the nonfunctional APC constructs (APC331Δ, APC 1309Δ, and APC1941Δ) produced as much or more protein than the CRT-functional forms of APC

(APC WT and APC 2644Δ).

EXAMPLE 6 This example demonstrates that other components of the APC- regulatory pathway are affected in some cancer cells. We evaluated CRT in two colorectal tumor cell lines (HCTl 16 and

SW48) that express full-length APC (Fig. 6A). Both HCTl 16 and SW48 displayed constitutively active CRT and, in contrast to cell lines with truncated APC (DLDl and SW480), this activity was not inhibited by exogenous WT APC (Fig. 5B, 6B). Other (noncolorectal cancer) cell lines expressing WT APC do not display constitutive CRT activity. These transfection results suggested that the constitutive CRT in HCTl 16 and SW48 might be due to an altered downstream component of the APC tumor suppressor pathway.

EXAMPLE 7 This example demonstrates a defect in the gene encoding β-catenin in some cancer cells, which affects CRT.

We evaluated the status of a likely candidate for a downstream component of the APC tumor suppressor pathway, β-catenin, in the same four lines. All four lines expressed similar amounts of apparently intact β-catenin, as assessed by immunoblots (Fig. 7A). However, sequence analysis revealed that both HCTl 16 and SW48 harbored mutations in the β-catenin gene (CTNNBl) (Fig. 7B). HCTl 16 had a 3-bp deletion that removed one AA (Ser-45), and SW48 had a C to A missense mutation that changed Ser-33 to Tyr. Analysis of paraffin-embedded archival tissue from the HCTl 16 patient confirmed the somatic nature of this mutation and its presence in the primary tumor prior to culture. Interestingly, both mutations affected serines that have been implicated in the down regulation of β-catenin through phosphorylation by the ZW3/GSK3β kinase in Xenopus embryos (Fig. 7C) (27,28). Genomic DNA was isolated from paraffin-embedded normal and tumor tissue from the patient from whom the HCTl 16 cell line was derived. A 95 bp PCR product encompassing the mutation was then amplified by PCR and directly sequenced using THERMOSEQUENASE (Amersham). The 3 bp deletion was observed in tumor but not in normal tissue. To test the generality of this mutational mechanism, we evaluated five primary colorectal cancers in which sequencing of the entire coding region of APC revealed no mutations (25). Three of these five tumors were found to contain CTNNBl mutations (S45F, S45F, and T44A) that altered potential ZW3/GSK3β phosphorylation sites (Fig. 7C). Each mutation appeared to affect only one of the two CTNNBl alleles and to be somatic.

Genomic DNA was isolated from frozen-sectioned colorectal cancers and a 1001 bp PCR product containing exon 3 of CTNNBl was then amplified by PCR and directly sequenced using ThermoSequenase (Amersham). An ACC to GCC change at codon 41 (T41 A) and a TCT to TTT at codon 45 (S45F) was observed in one and two tumors, respectively.

EXAMPLE 8 This example demonstrates dominant mutations of CTNNBl that render CRT insensitive to the effects of WT APC.

Because the β-catenin mutations were heterozygous, we hypothesized that the mutations might exert a dominant effect, rendering a fraction of cellular β-catenin insensitive to APC-mediated down regulation. To test this notion, we performed gel shift analyses with nuclear extracts from untransfected HCTl 16 cells. In contrast to noncolorectal cancer cell lines with intact APC, HCTl 16 cells contained a β-catenin/Tcf complex that gel-shifted an optimized Tcf-binding oligonucleotide, and this complex supershifted with anti-β-catenin (Fig. 8A). We also constructed β-catenin expression vectors and compared the biologic activity of the mutant β-catenin from HCTl 16 (β-Cat Δ45) and SW48 (β-Cat S33Y) with that of their WT counteφart. For these experiments, we used the 293 kidney epithelial cell line as it is highly transfectable, exhibits low endogenous CRT, and contains a high level of endogenous APC (Fig. 6A). In the presence of endogenous APC, both mutant β-catenins were at least 6-fold more active than the WT protein and this activity was inhibited by dominant-negative hTcf-4 (Fig. 8B).

Together, these results indicate that disruption of APC-mediated regulation of CRT is critical for colorectal tumorigenesis. This is most commonly achieved by recessive inactivating mutations of both APC alleles but, as shown here, can also be achieved by dominant mutations of CTNNBl that render CRT insensitive to the effects of WT APC. Our results suggest that APC inhibition of CRT requires phosphorylation of β-catenin at multiple sites. These potential phosphorylation sites are consistent with the known specificity of ZW3/GSK3β (29) a serine kinase that negatively regulates β-catenin in Xenopus and Drosophila cells (27) and that interacts with APC and β-catenin in mammalian cells (23). These results also suggest a functional basis for the occasional CTNNBl mutations observed in other tumor types (30) and illustrate how a critical pathway in human disease can be illuminated by the discovery of mutations in different components of the pathway. The next step in understanding APC function will be the identification of the genes that are activated by hTcf-4/β-catenin complexes and inhibited by WT APC. These genes are likely to be related to APC's ability to induce apoptosis in colorectal cancer cells (31).

EXAMPLE 9

Characterization of the Ad-Mini-ME recombinant adenovirus.

Understanding the structure-function relationships of APC has been hindered by the inability to readily restore APC tumor suppressor activity to human cells. To address this problem, we developed an adenovirus system that would allow the relatively facile evaluation of APC effects in a variety of cell lines. Three features of the adenovirus construction are of particular interest. First, to facilitate the actual construction of the adenoviral vectors, we used the AdEasy adenovirus system, which employs recombination in bacteria rather than in mammalian cells to generate recombinant adenovirus.

Second, we included a GFP marker to allow easy identification of APC expressing cells. This eliminates problems related to differences in infectivity and allows the use of viral titers well below levels that result in virus-induced cytopathic effect (CPE). Avoiding such nonspecific CPE is especially important for assessing tumor suppressive effects. Finally, we generated a virus that expressed a fusion protein (GFP/cAPC) containing GFP fused to the central third of APC (Fig. 9A). The employment of a fusion protein ensured that expression of GFP was coupled with APC and allowed positive verification of cAPC expression. The growth suppressive effects of tumor suppressor genes impose a powerful negative selection force that can result in loss of expression of the tumor suppressor gene even in the presence of a positive selection marker.

The central portion of APC was chosen for the following experiments because it contains all of the known β-catenin and Axin/Conductin binding domains (Fig. 9A) and is sufficient for promoting β-catenin degradation. In keeping with precedent, the clone expressing this smaller version of APC Minus its amino- and carboxyl-terminal Ends was designated Ad-Mini-ME. (37) Infection of 293 and SW480 cells with Ad-Mini-ME resulted in the production of a fusion protein of the expected size (150 kD fusion) as judged by Western blot analysis with an anti-GFP antibody (Fig. 9B). Fluorescence microscopy revealed that the GFP/cAPC fusion protein was diffusely present in the cytoplasm (Fig. 9C).

Ad-Mini-ME inhibits Tcf4/β-catenin mediated transactivation. One of the best-characterized functions of APC is its ability to inhibit CRT. To investigate the effects of GFP/cAPC on this function, we generated stable DLDl cell lines with either an integrated Tcf-4 responsive luciferase reporter (DOT cells) or a reporter driven by the SV40 promoter (Dluc cells).

Infection of DOT cells with Ad-Mini-ME markedly suppressed luciferase activity, whereas a virus expressing GFP alone (Ad-GFP) had no effect (Fig 10A). This inhibition appeared to be specific as infection of Dluc cells with Ad-Mini-ME had no effect on luciferase activity (Fig. 10B). To determine whether the Ad-Mini-ME suppression could be extended to endogenous targets of CRT, we evaluated the expression of c-MYC and cyclin D 1 , two recently described targets of the APC/β-catenin/Tcf-4 pathway. Expression of c-MYC and cyclin Dl was examined in three human colorectal cancer cell lines. In DLDl and SW480 CRT was constitutively activated because of APC mutations, while in HCTl 16 CRT activation was due to a β-catenin mutation.

Infection of SW480 and DLDl cells with Ad-Mini-ME resulted in a marked reduction of c-MYC protein levels, and a minor reduction of cyclin

Dl levels (Fig. 11). Ad-GFP infection had no inhibitory effect on either c-MYC or cyclin Dl protein levels in these cells. As expected, HCTl 16 cells, which possess a stabilizing β-catenin mutation, were relatively resistant to the effects of Ad-Mini-ME (Fig. 11).

The subcellular localization of β-catenin was dramatically altered by Ad-Mini-ME infection. Whereas mock-infected or Ad-GFP-infected DLDl cells displayed predominantly nuclear β-catenin staining, cells infected with Ad-Mini-ME showed cytoplasmic and membrane β-catenin staining with minimal nuclear staining (Fig. 12A). A similar alteration in the subcellular distribution of β-catenin was observed in SW480 cells (Fig. 12B). The selective depletion of nuclear β-catenin by Ad-Mini-ME is consistent with the model for APC action proposed previously.

Ad-Mini-ME suppresses cell growth and colony formation in colorectal cell lines.

In both DLDl and SW480, Ad-Mini-ME infection resulted in a clear growth inhibition by day 2 (Fig. 13 A). This growth inhibition was persistent, with Ad-Mini-ME infected DLDl and SW480 cells failing to reach confluence even after 2 weeks of culture, long after mock- and Ad-GFP infected cells became superconfluent. Consistent with the effects of Ad-Mini-ME on CRT in HCTl 16 cells, the growth of Ad-Mini-ME infected HCT 116 cells was only partially inhibited (Fig. 13 A) .

We next examined the effects of Ad-Mini-ME mediated CRT inhibition on colony formation in collagen gels using a series of eight colorectal cancer cell lines. The cells were flow sorted to select virally-infected cells, as described in Materials and Methods. Infection with Ad-Mini-ME resulted in a marked suppression of colony formation in SW480, DLDl, LoVo, HT29,

SW837, and SW1417 cells, all of which possess mutated APC (Fig. 13B). Colony numbers were reduced by at least 96% in each of these lines compared with cells mock-infected or infected with Ad-GFP virus. Examination under phase-contrast microscopy revealed that most of the Ad-Mini-ME infected cells were pyknotic. A small number of growth-arrested single cells remained in the gels, and these continued to express APC as judged by fluorescence. In the few colonies that did form after Ad-Mini-ME infection, APC did not appear to be expressed. In contrast, cells with intact APC but mutated β-catenin were able to form a significant number of colonies following Ad-Mini-ME infection (HCTl 16 and SW48 in Fig. 13B).

Ad-Mini-ME induces apoptosis in colorectal cell lines.

Ad-Mini -ME-expressing cells revealed a gradual loss of the Gl peak and an accumulation of cells in the S and G₂ phases of the cell cycle. Five days after infection, all six colorectal cell lines with APC mutations demonstrated significantly increased apoptosis following Ad-Mini-ME infection (Fig. 14). In line with the effects of Ad-Mini-ME on CRT and growth, the mutant β-catenin containing cell lines, HCTl 16 and SW48, exhibited little increase in apoptosis in response to Ad-Mini-ME infection. Time course studies revealed that the first moφhological signs of apoptosis were not evident in DLDl and SW480 cells until 72 hours after plating. The induction of apoptosis was confirmed by Annexin V, which has been shown to bind to phosphatidylserine exposed on the outer leaflet of apoptotic cell membranes. The proportion of DLDl and SW480 cells staining with Alexa568-labeled annexin V was in good agreement with the fraction of cells displaying moφhological signs of apoptosis (>90% of annexin V-labeled cells displayed apoptotic nuclei). Tumor growth in nude mice. To investigate the effects of Ad-Mini-ME on colorectal cancer cells in vivo, Ad-GFP and Ad-Mini-ME-infected DLDl cells were selected by flow sorting and injected subcutaneously into nude mice. Mice sacrificed three days after injection showed a large number of apoptotic tumor cells at the injection site (Fig. 15, H&E). Such changes were not evident after injection of mock-infected or Ad-GFP-infected cells. Apoptosis was further evaluated by immunoreactivity with the M30 antibody, which recognizes a caspase-cleaved epitope of cytokeratin 18. Ad-Mini-ME infected cells were diffusely positive for the M30 antibody, while only a small fraction of Ad-GFP-infected cells stained with this antibody (Fig. 15, M30). Immunoperoxidase staining with an antibody specific for the proliferation marker Ki-67 showed a significant decrease in nuclear labeling of Ad-Mini-ME infected tumors, with only 10% of cells positive (Fig. 15, Ki- 67). In contrast, over 90% of Ad-GFP-infected cells were Ki-67 positive. Tumors derived from Ad-GFP infected cells continued to grow in the nude mouse, eventually forming tumors larger than 1 cm in diameter, while there was no visible tumor formation after injection of Ad-Mini-ME infected cells after three weeks.

Materials and Methods

Cell Culture, Medium, and Reagents. 293 cells were purchased from

Microbix Biosystems (Toronto, Canada) and 911 cells were kindly provided by Alex J. Van der Eb of the University of Leiden. Both lines were maintained in DMEM growth medium (Life Technologies,

Gaithersburg, MD) supplemented with 10% fetal bovine serum (HyClone, Logan, UT), 100 units/ml of penicillin and 100 μg/ml of streptomycin. The human colon carcinoma cell lines, SW480, DLDl, HCTl 16, LoVo, SW48, HT29, SW837, and SW 1417 were obtained from the American Type Culture Collection and cultivated in McCoy's 5A media (GIBCO/BRL,

Gaithersburg, MD) supplemented with 10% fetal bovine serum (HyClone), 100 units/ml of penicillin and 100 μg/ml of streptomycin. The DOT and Dluc cell lines were derived from DLDl cells as described below. Both cell lines were maintained in McCoy's 5A media (GIBCO/BRL) supplemented with 10% fetal bovine serum (HyClone) and 0.4 mg/ml Hygromycin B

(Calbiochem).

Generation of Recombinant Adenovirus Expressing Central Third of APC (Ad-Mini-ME). The recombinant adenovirus, Ad-Mini-ME, which carried the central third of APC gene, was generated using a modified system as previously described. The central third of APC containing amino acids 958-2075 (nucleotides 2890-6240) was isolated from pCMV-APC by Bgl II digestion. This fragment was subcloned into the pEGFP- C 1 (Clontech, Palo Alto, CA). The cassette containing the EGFP-tagged central third of APC gene (cAPC) was further subcloned into the shuttle vector (pShuttle) using Apal I and Mlu I digestion. Recombinant adenoviral plasmid was generated by homologous recombination in E. coli (BJ5183). B J5183 cells were transformed using electroporation with pAdEasy-1 and pShuttle/EGFP-cAPC linearized with Pmel. Successful recombinants were identified by restriction endonuclease mapping. The recombinant EGFP-cAPC virus (Ad-Mini-ME) was produced in the 911 and 293 adenovirus packaging lines and the viral particles were purified by CsCl banding. The control virus (Ad-EGFP) with EGFP alone was also prepared and purified side by side. Viral titer was determined by a modified CPE³ endpoint assay. A series of Ad-GFP infections was performed on HCTl 16 cells to determine the optimal multiplicity of infection (MOI) to avoid adenovirus-associated CPE. Typically, viral CPE could be observed at an MOI of above 100, which resulted in more than 80% of cells becoming fluorescent 18 hours after Ad-GFP infection. To avoid any cytopathic effects of viruses, we infected cells with a minimal MOI, generating fluorescence in 20-30% of the cells (MOI=5-l 1), then flow-sorted infected cells to obtain homogeneous populations.

Viral Infection and Cell Sorting. Viral stocks were predialyzed using 1% agarose in microcentrifuge tube. Three million cells were infected with either Ad-Mini-ME or Ad-GFP in a 75 cm² flask. After 18 hours of incubation at 37°C, cells were washed, trypsinized and subjected to fluorescence activated cell sorting. Cells with green fluorescence were collected for experiments or were replated in culture flasks immediately after sorting.

Reporter Assay. DOT and Dluc cells were generated from DLD 1 cells by cotransfection of pTK-hygro (Clontech) and a Tcf-4 responsive luciferase plasmid (pGL3-OT, Yu, J. unpublished data) or a constitutive luciferase plasmid (pGL3 -control, Promega, Madison, WI), respectively. Clones were isolated and the sensitivity to CRT determined using a dominant negative Tcf-4 adenovirus (Yu, J. unpublished data). Luciferase reporter activity in the DOT clone was constitutively high as expected for a CRT responsive reporter in a colorectal cancer cell line with mutated APC. This constitutive activity was inhibited by dominant negative Tcf-4. In contrast, the luciferase activity in the Dluc clone was unaffected by dominant negative Tcf-4 as expected for expression driven by the SV40 promoter. To assess the effects of cAPC on CRT, DOT and Dluc cells were infected with Ad-GFP and Ad-Mini-ME. Eighteen hours after viral infection, equal numbers of GFP positive cells were pelleted, lysed, and collected for luciferase assays using Luciferase Assay Reagents (Promega).

Western Blot Analysis. Whole cell lysates were prepared in a solution containing 62.5 mM Tris-HCl (pH 6.8), 2% SDS, and 0.5% β-mecaptoethanol. Equal amounts of total protein from each lysate were loaded and separated on 4-12% Tris-Glycine-SDS polyacrylamide gels (Novex, San Diego, CA) and electroblotted to Millipore Immobilon-P polyvinylidene difluoride membranes. Western blots were developed by chemiluminescence (NEN Life Science, Boston, MA), detected by Kodak Image Station 440CF, and analyzed by ID Image Analysis software (NEN

Life Science). Primary antibodies included anti-GFP polyclonal antibody from Clontech (Palo Alto, CA), anti-c-MYC monoclonal antibody (9E10) from Santa Cruz (Santa Cruz, CA), anti-cyclin Dl monoclonal antibody (A- 12) from Santa Cruz, and anti-α-tubulin monoclonal antibody (TU-02) from Santa Cruz. Secondary peroxidase-conjugated antibodies were goat anti-mouse IgG and goat anti-rabbit IgG from Pierce (Rockford, IL).

Immunofluorescence Staining. Cells were infected with Ad-GFP or

Ad-Mini-ME for 18 hours and sorted. Fluorescent cells were cultured on 8-well chamber CultureSlides (Becton Dickinson, Bedford, MA). After 8 hours, cells were fixed in 3% paraformaldehyde in PBS at room temperature for 8 min, then permeabilized with 0.3% NP-40 in PBS for another 8 min. After washing in PBS, the cells were incubated with primary mouse anti-β-catenin monoclonal antibody (1 μg/ml; Transduction Laboratories, Lexington, KY) at 4°C overnight. After washing, cells were incubated with biotinylated goat anti-mouse IgG (Pierce, Rockford, IL) at room temperature for 1 hour. The immunoreactivity was revealed using Alexa568-conjugated streptavidin (Molecular Probes, Eugene, OR) and cells were counterstained with 10 μg/ml Hoechst 33258. The cells were examined under a Nikon fluorescence microscope (Image Systems, Columbia, MD).

Cell Growth and Colony Formation Assay. 10⁵ cells were plated in one well of a 24 well-plate. Cells were counted using a hemocytometer after trypsinization on days 1, 2, 3 and 5. For colony formation assays, each well of 24-well plates was precoated with 100 μl of collagen gel containing 50% type I collagen (Collaborative Biomedical Science, Bedford, MA), 40% culture medium, and 0.75% NaHCO₃ (Halttunen). 160 μl of collagen gel-cell suspension containing 10,000 cells, 45% type I collagen, 40% culture medium, and 0.075% NaHCO₃ was added to the wells. After solidification, each well was covered with 1 ml of culture medium and the plates were incubated at 37°C. Twelve days after seeding, cells were stained with 0.05% crystal violet (Sigma, St. Louis, MO) containing 10% buffered formalin (Sigma).

Hoechst Staining and Annexin V Staining for Apoptosis Detection. Both attached and floating cells were harvested for staining. For Hoechst staining, 3x10⁵ cells were resuspended in 50 μl PBS and 350 μl of staining solution containing 0.6% Nonidet P-40, 3% paraformaldehyde, and 10 μg/ml Hoechst 33258. For annexin V staining, 10⁵ cells were suspended in 100 ul annexin-binding buffer containing lOmM HEPES, 140 mM NaCl and 2.5 mM CaCl₂. 5 μl of Alexa568-conjugated annexin V (Molecular

Probes) was added and incubated at room temperature for 15 min, at which point an additional 400 μl of annexin-binding buffer was added to each sample. Apoptotic cells were defined as those cells containing condensed and/or fragmented nuclei after Hoechst staining or were fluorescent after annexin V staining. At least 500 cells were counted and the results were expressed as the percentage of apoptotic cells in each sample.

Experimental Animals and Immunoperoxidase. Eight pathogen-free female athymic nude mice (Harlan, Indianapolis, IN), 6-8 weeks old were injected with 3xl0⁶ DLDl cells infected either with Ad-GFP or Ad-Mini-ME in the subcutaneous tissues adjacent to the lower spine.

Before injection, these cells were sorted and green fluorescent cells were collected and cultured for 24 hours. Three days after injection, two of the mice from each group were sacrificed and the tumors were formalin-fixed, paraffin-embedded, and processed for routine histological examination. Tissue sections of 5 micron thickness were deparaffinized and incubated with a mouse anti-Ki-67 antibody (AMAC Inc., Westbrook, ME) or with a mouse antibody specific for a caspase-cleaved epitope of cytokeratin 18 (M30, Boehringer Mannheim, Mannheim, Germany). Cells were incubated with antibodies at 4°C overnight, then washed and developed with an avidin-biotin complex peroxidase method (Biogenex, San Ramon, CA). Immunoreactivity was detected using the 3,3 '-diaminobenzi dine chromagen and counterstained with 0.1% hematoxylin (Sigma, St Louis,

MO). DISCUSSION

While the ability of APC to suppress intestinal tumorigenesis has been known for several years, the sequences required for this inhibition have not been well defined. In this study, we define a minimal portion of

APC that is sufficient for its growth suppressive effects. Our results suggest that expression of the central third of APC is sufficient to inhibit cellular proliferation and induce apoptosis in colorectal cancer cells. The biological effects of the central third of APC are likely related to abrogation of the APC/β-catenin/Tcf-4 signaling pathway. This conclusion is based on the fact that Ad-Mini-ME inhibits β-catenin nuclear translocation (Fig. 12), suppresses β-catenin/Tcf-4 mediated transcription in reporter assays (Fig. 10), and downregulates the expression of targets of the APC/β-catenin/Tcf-4 pathway (Fig. 11). Cellular proliferation and colony formation are dramatically suppressed by Ad-Mini-ME in cell lines which contain mutations in the APC gene, but are only partially inhibited in lines containing mutations of β-catenin that render it resistant to APC degradation.

At the cellular level, expression of the central third of APC eventually results in apoptosis of colorectal cancer cells containing APC mutations. This observation is consistent with those in a previous report, demonstrating apoptosis 60 hours after induction of full length APC expression. In both cases, the delay in appearance of apoptotic cells suggests that APC-induced apoptosis may not be a direct result of suppression of CRT.

While the β-catenin binding domain in the central third of APC is sufficient for growth suppression by APC, it may not recapitulate all of the functions of this gene. For example, the carboxyl-terminal third of APC can associate with the human homo log of the Drosophila tumor suppressor gene discs large (hDLG) and EB-1. The latter has recently been implicated in the spindle checkpoint. In addition, a carboxyl-terminal fragment of APC has been shown to induce assembly and bundling of micro tubules in vitro and has a role in directed cell migration. Like other canonical tumor suppressor genes, it is likely that APC functions at several levels to regulate cell growth and suppress neoplastic transformation. However, the finding that the middle third of APC is sufficient to inhibit tumor cell growth focuses further attention on the APC/β-catenin interaction. Future experiments to understand the upstream regulators and downstream transducers of this interaction should shed further light on tumorigenesis associated with defects in the APC pathway.

REFERENCES

1. B. Rubinfeld et al Science, 262, 1731 (1993); L. K. Su, B. Vogelstein, K.W. Kinzler, ibid 262, 1734 (1993).

2. B. Gumbiner, Curr. Opin. Cell Biol. 7, 634 (1995).

3. B. Rubinfeld et al, Science 272, 1023 (1996).

4. J. Papkoff, B. Rubinfeld, B. Schryver, P. Polakis, Mol. Cell. BioL 16, 2128 (1996). 5. S. Munemitsa, B. Souza, I. Albert, B. Rubinfeld, P. Polakis, Proc.

Natl. Acad Sci. U.S.A. 92, 3046 (1995); B. Rubinfeld, B. Souza, I. Albert, S. Muneinitsa, P. Polakis, J Biol Chem. 270, 5549 (1995). 6. M. Molenaar et al, Cell 86, 396 (1996); J. Behrens et al, Nature 382, 638 (1996); 0. Huber et al, Mech. Dev. 59, 3 (1996).

7. M. van de Weterinc,, M. Oosterwegel, D. Dooijes, H. Clevers, EMBO J 10, 123 (1991); M. van de Wetering, J. Castrop, V. Korinek, Mol Cell Biol, 16, 745 ( 1996).

8. A. Travis et al. Genes Dev. 5, 880 (199 1); M.L. Waterman, W.H. Fischer, K.A. Jones ibid. p. 6562. H. Clevers and R. Grosschedl, Immunol. Today 17, 336 (1996).

9. J. Castrop, K. van Norren, H.C. Clevers. Nucleic Acids Res. 20, 611 (1992).

10. E. van Hoffen et al, Am. J Pathol 149, 1991 (1996).

11. M. van de Wetering, M. Oosterwegel, K. van Norren, H. Clevers, EMBO J. 12, 3847 (1993)

12. P. Morin, B. Vogelstein, K.W. Kinzler, Proc. Natl. Acad. Sci. U.S.A. 93, 7950 (1996).

13. K. W. Kinzler and B. Vogelstein, Cell 87, 159 (1996).

14. About 50% of the Western population develop colorectal adenomas by the age of 70 [D. Ransohoff and C. Lang, N. Eng.U. Med. 325, 37 (1991)] and at least 85% of these tumors contain APC mutations ; Y. Miyoshi et al, Hum Mol Genet 1, 229-33 (1992); J. Jen et al, Cancer

Res. 54, 5523 (1994).

15. H. Nagase and Y. Nakamura, Hum. Mutation 2, 425 (1993).

16. K. W. Kinzler and B. Vogelstein, Cell 87, 159 (1996); S. M. Prescott and R. L. White, ibid., p. 783. 17. G. Joslyn, D. S. Richardson, R. White, T. Alber, Proc. Natl. Acad.

Sci. U. S. A. 90, 11109 (1993); L. K. Su et al., Cancer Res. 53, 2728 (1993). 18. B. Rubinfeld et al., Science 262, 1731 (1993); L. K. Su, B. Vogelstein, K. W. Kinzler, ibid., p. 1734. 19. J. Hulsken, J. Behrens, W. Birchmeier, Curr. Opin. Cell. Biol. 6, 711 (1994); B. Rubinfeld, B. Souza, I. Albert, S. Munemitsu, P. Polakis, J. Biol. Chem. 270, 5549 (1995).

20. S. Munemitsu et al, Cancer Res. 54, 3676 (1994); K. J. Smith et al, ibid. p. 3672.

21. L. K. Su et al, Cancer Res. 55, 2972 (1995).

22. A. Matsumine et al. , Science 272, 1020 ( 1996).

23. B. Rubinfeld et al. , Science 272, 1023 ( 1996).

24. M. Molenaar et al. , Cell 86, 391 (1996); J. Behrens et al. , Nature 382, 638 (1996).

25. S. M. Powell et al, Nature 359, 235 (1992).

26. D. M. Eccles et al. , Am. J. of Hum. Genet. 59, 1 193 ( 1996); W. Friedl et al, Hum Genet 97, 579 (1996); R. J. Scott et al, Human Molecular Genetics 5, 1921 (1996). 27. C. Yost et al, Genes Dev. 10, 1443 (1996).

28. S. Munemitsu, I. Albert, B. Rubinfeld, P. Polakis, Mol Cell Biol 16, 4088 (1996).

29. M. Peifer, L. M. Pai, M. Casey, Dev. Biol 166, 543 (1994).

30. D J. Kawanishi, et al., Mol. Cell Biol. 15, 1175 ( 1995); P. F. Robbins, et al. , J. Exp. Med. 183, 1185 (1996).

31. P. J. Morin, B. Vogelstein, K. W. Kinzler, Proc. Natl. Acad. Sci. U. S. A. 93, 7950 (1996).

32. J. Groden et al, Cell 66, 589 (1991); G. Joslyn et al, ibid., p. 601; K. W. Kinzler et al, Science 253, 661 (1991); I. Nishisho et al, ibid., p. 665.

33. M. Peifer, S. Berg, A. B. Reynolds, Cell 76, 789 (1994).

34. K. J. Smith et al, Proc. Natl. Acad. Sci. U. S. A. 90, 2846 (1993).

35. S. E. Goelz, S. R. Hamilton, B. Vogelstein, Biochem. Biophys. Res. Commun. 130, 118 (1985) 6. M. Myers, Austin Powers: The Spy Who Shagged Me. Los Angeles, CA: New Line Cinema, Inc. (1998)

Claims

1. A fusion protein which comprises an enzyme covalently linked to a portion of APC which comprises its β-catenin binding domain.

2. The fusion protein of claim 1 wherein the β-catenin binding domain consists of at least amino acids 958 to 2075.

3. The fusion protein of claim 1 wherein the enzyme is GFP.

4. An isolated and purified polynucleotide which encodes the fusion protein of claim 1.

5. An isolated and purified polynucleotide which encodes the fusion protein of claim 2.

6. An isolated and purified polypeptide comprising a portion of APC which comprises its β-catenin binding domain.

7. The isolated and purified polypeptide of claim 6 wherein the binding domain consists of at least amino acids 958 to 2075.

8. An isolated and purified polynucleotide which encodes the polypeptide of claim 6.

9. An isolated and purified polynucleotide which encodes the polypeptide of claim 7.

10. A method for inhibiting the growth of a tumor cell comprising: administering to a tumor cell a polynucleotide according to claim 8, whereby growth of the tumor cell is inhibited.

11. The method of claim 10 wherein the tumor cell is in vitro.

12. The method of claim 10 wherein the tumor cell is in a human body.

13. The method of claim 10 wherein the polynucleotide is part of an adenovirus.

14. A method for inducing apoptosis of a tumor cell comprising: administering to a tumor cell a polynucleotide according to claim 8, whereby apoptosis of tumor cells is induced.

15. The method of claim 14 wherein the tumor cell is in vitro.

16. The method of claim 14 wherein the tumor cell is in a human body.

17. The method of claim 14 wherein the polynucleotide is part of an adenovirus.

18. An adenovirus vector which comprises the polynucleotide of claim 8.

19. An adenovirus vector which comprises the polynucleotide of claim 9.

20. The adenovirus vector of claim 19 which is Ad-mini-ME.

21. The fusion protein of claim 1 which lacks at last one APC domain selected from the group consisting of: oligomerization domain, armadillo repeat domain, human homolog of Drosophila tumor suppressor gene discs large domain, and EB 1 binding domain.

22. The polynucleotide of claim 4 wherein the fusion protein lacks at least one APC domain selected from the group consisting of: oligomerization domain, armadillo repeat domain, human homolog of Drosophila tumor suppressor gene discs large domain, and EB1 binding domain.

23. The isolated and purified polypeptide of claim 6 which lacks at least one APC domain selected from the group consisting of: oligomerization domain, armadillo repeat domain, human homolog of Drosophila tumor suppressor gene discs large domain, and EB1 binding domain.

24. The isolated and purified polynucleotide of claim 8 wherein the polypeptide lacks at lest one APC domain selected from the group consisting of: oligomerization domain, armadillo repeat domain, human homolog of Drosophila tumor suppressor gene discs large domain, and EB 1 binding domain.

25. A method for treating a patient with colorectal cancer or other cancer associated with FAP, the method comprising the step of: administering to the patient a nucleotide sequence comprising a portion of the APC coding sequence, said portion consisting of the β-catenin binding site.

26. A method for treating a patient with colorectal cancer or other cancer associated with FAP, the method comprising the step of: administering to the patient a polypeptide comprising a portion of the APC coding sequence, said portion consisting of the β-catenin binding site.