Several important molecules implicated in the tumorigenic process act on the cell cycle, resulting in a disturbed homeostasis between cell proliferation and apoptosis. Although complex genetic alterations accumulate along the adenoma-carcinoma sequence, mutations involving components of the Wnt-Wingless signalling cascade play a key role in the early transformation of colonic epithelium (reviewed in Bienz and Clevers, Cell 103 p 311-20,2000, Giles et al. , Biochim Biophys Acta 1653 (1) p 1-24,2003).

Familial adenomatous polyposis (FAP) patients are individuals who inherit mutations in the adenomatous polyposis coli (APC) tumor suppressor gene. FAP individuals develop thousands of colorectal tumors, consistent with a gatekeeping role of the APC protein in colorectal tumorigenesis (Nagase et al. , Hum. Mutat. 1993, 2: 2425-34 and Kinzler et al., Cell 1996 87: 159-170). The APC protein has been observed to interact with another gene product, B-catenin (Vogelstein et al., N Engl J Med 1988,319 : 525-532, Rubinfeld et al. , Science 1993,262 : 1731-1734), originally identified on the basis of its association with cadherin adhesion molecules but now recognized as an essential component of the Wnt-Wingless cascade (Gumbiner, Curr Opin Cell Biol 1990,2 : 881-887). In this cascade, extracellular Wnt proteins signal through binding receptors of the"frizzled"family of transmembrane proteins, phosphorylate and thus activate a homologue of"dishevelled". This phosphoprotein in turn phosphorylates and inactivates a serine/threonine kinase, glycogen synthase kinase GSK-3B. APC and GSK-3B proteins complex together and inhibit B-catenin degradation, resulting in an accumulation of free B-catenin. B-catenin functions as a transcriptional co-activator when complexed with members of the T Cell Factor (TCF) family of DNA-binding proteins. The main function of APC is thought to be the regulation of free B-catenin in concert with the glycogen synthase kinase GSK-3B and Axin proteins. Loss of APC function, inactivation of Axin or activating B-catenin mutations all result in the cellular accumulation of B-catenin, which is normally rapidly degraded. Upon translocation to the nucleus, B-catenin serves as an activator of T-cell

factor (TCF) -dependent transcription, leading to an increased expression of several specific target genes. hTCF-4 is a TCF family member that is expressed in colonic epithelium, and transactivates transcription only when associated with B-catenin. Constitutive transcription of TCF target genes, caused by loss of APC function, or activation of p-catenin, is a crucial event in the early transformation of colonic epithelium. hTCF-4 consensus sequences have been demonstrated within the promoters of a variety of oncogenes and cell cycle regulators, such as c-MYC, a potent cellular oncogene, which is upregulated by constitutive active B-catenin/TCF mutants and could be suppressed by wildtype APC (He TC, Science 281, pl509-12, 1998). Other targets of TCF-4 include cyclin D1, PPAR delta, MDR-1, MMP-7, ITF-2 and CD44 (Brabletz et al. , Am J Path 155 p 1033-8,1999 ; He et al., Cell 99 p 335-45,1999 ; Kolligs et al. , Cancer Cell 1 p 145-55,2002 ; Tetsu and McCormick, Nature 398, p 422-6,1999 ; Van de Wetering et al., Cell 111 p 241-50,2002 ; Wielenga et al., Am J Path 154 p 515-23,1999 ; Yamada et al., Cancer Res 60 p 4761-6,2000). Also hTCF-4 has been shown to be a target for mutations, leading to uncontrolled downstream transcription (Duval, Cancer Res.

59 (17) p 4213-5m 1999, Saeki et al. , Oncology 61 (2) pl56-61, 2001).

Inherited or germline mutations in APC give rise to the forementioned autosomal dominantly inherited, pre-malignant condition FAP, which is characterized by the development of hundreds to thousands of colorectal adenomatous polyps. Over 300 different germline mutations have been identified in FAP kindred's to date (Beroud C, et al, NAR 24 (1) pl21-4, 1996). Study of the genetics of FAP simultaneously established the basis for'classical'sporadic colorectal cancer (CRC). In sporadic cases, APC mutations are again seen in small adenomas at the earliest stage of neoplasia, i. e. in aberrant crypt foci. This strongly implicates APC as an initiator of sporadic colorectal cancer as well (Powel SM, Nature 359 p235-7, 1992).

While APC mutations are almost exclusively found in colorectal cancers, deregulation of Wntlß-catenin/Tcf signaling is also common in other gastrointestinal and extra-gastrointestinal human cancers. In a fraction of hepatocellular carcinomas the Wnt pathway is deregulated by inactivation of Axin or stabilizing mutations of P-catenin. The majority of hepatoblastomas and a group of gastric cancers also carry P-catenin mutations (Blaker et al. , Genes Chromosomes Cancer 25 p 399-402,1999 ;

Koch et al. , Cancer Res 59 p 269-73,1999 ; Park et al. , Cancer Res 59 p 4257-60, 1999).

In APe-or deficient colon carcinoma cell lines, transcriptionally active nuclear B-catenin/Tcf-4 complexes are constitutively present (Korinek, Science 1997 275: 1784- 7). Comparable complexes between B-catenin and Tcf/Lef proteins exist inAPC+/+ colon carcinoma (Morin et al., Science 1997 275: 1787-90) and melanoma cells (Rubinfeld et al., Science 275: 1790-2) as a result of dominant mutations affecting the amino terminus of B-catenin. Thus, mutation in either APC or inß-catenin can lead to constitutive nuclear complexes between co-activator B-catenin and Tcf-4 in intestinal epithelium. This will result in activated transcription of Tcf-4 target genes in such cells.

The inventors previously found that in FAP patients and inApcmi"/+ mice the facultative heparan sulphate proteoglycan (HSPG) CD44 is upregulated in the earliest lesions, the aberrant crypt foci (ACF) (Wielenga et al., Am J Path 1999 154 : 515-23).

Furthermore, the inventors reported that CD44 expression is absent in small intestinal crypt cells of Tcf4 null mice. These combined findings implied that CD44 was a target of the Wnt signalling cascade. Very recently CD44 has been unequivocally identified as one of the targets of the Wnt/p-catenin signalling cascade in colorectal cancer cell lines (Van de Wetering et al. , Cell 2002 111 : 241-50). The array of genes whose expression is shut off or turned on as a result of Wnt signalling imposes a non- differentiated (crypt) stem cell phenotype on colorectal cancer cells.

CD44 is a type I transmembrane glycoprotein and is expressed by virtually every cell in the vertebrate body. Depending on the species, the CD44 locus contains about 20 coding exons, roughly half of which encode the so-called constant region of CD44 (Screaton et al. , PNAS 1992 89: 12160-4, Tölg et al, NAR 1993 21: 1225-9, Screaton et al., JBC 1993 268: 12235-8). Retention of different combinations of variable exons in the mRNA results in a myriad of CD44 splice variants (reviewed in Gunthert, Curr. Top Microbiol Immunol 1993 184: 47-63). Although theoretically more than 1,000 individual splice variants may be produced in this way, the main molecular species expressed is the standard form of CD44 (CD44s), which is the shortest form and is encoded by an mRNA consisting exclusively of constant exons. Even in cases where multiple splice variants are co-expressed in one tissue or cell type, CD44s

remains the main isoform (see e. g. Ni et al. , J Lab Clin Med 2002 139 : 59-65, Bell et al. , MCB 1998 18: 5930-41). The potentially tremendous variation in CD44 species is furthermore enhanced by differential N-and O-glycosylation and glycosaminoglycanation (reviewed inNaor et al. , Adv Cancer Res 1997 71: 241-319).

The biological function of most of these splice variants remains to be elucidated.

CD44 functions as a receptor for hyaluronic acid (Miyake et al. , J Exp Med 1990 172 : 69-75, Aruffo et al., Cell 1990 61: 1303-13) a glycoprotein found in abundance in extracellular matrices. The interaction between CD44 and HA has been mapped to conserved basic residues in the extracellular constant region (Peach et al. , J Cell Biol 1993 122: 257-64), while the cell's choice of the CD44 splice variant is the main determinant for the binding affinity (Lesley et al. , J Exp Med 1995 182 : 431-7, Stamenkovic et al. , Embo J 1991 10: 343-8, Van der Voort et al. , Biochem Biophys Res Commun 1995 214: 135-144). CD44 may enrol in adhesive interactions with many more partners, including fibronectin (Jalkanen et al. , J Cell Biol 1992 116 : 817-25), collagen (Faassen et al. , J Cell Biol 1992 116 : 521-31), E-and L-selectin (Dimitroff et al. , PNAS 2000 97: 13841-6 and Dimitroff et al. , J Cell Biol 2001 153 : 1277-86) and aggrecan (Fujimoto et al., Int Immunol 2001 13: 359-66), together classifying CD44 as an adhesion molecule. The juxtamembrane portion of the cytoplasmic tail of CD44 binds to members of the ezrin-radixin-moesin (ERM) family of actin linker molecules, thus providing a connection between cell surface bound CD44 and the actin cytoskeleton (Tsukita et al. , J Cell Biol 1994126 : 391-401) and establishing the basis for CD44-dependent cellular motility. More recently, CD44 has been implicated in the amplification of growth factor signalling. The exon v3 (exon 8) encoded protein moiety contains the amino acid sequence Ser-Gly-Ser-Gly that serves as a recognition site for heparan sulphate side chain addition (Greenfield et al. , JBC 1999 274: 2511-7). Heparan sulphated forms of CD44 can act as a high-capacity, low-affinity binding sites for heparin-binding growth factors such as MIP-1 (3 (Tanaka et al. , Nature 1993 361: 79-82) FGF-2 (Bennett et al. , J Cell Biol 1995 128: 687-98) and HGF (Van der Voort et al., JBC 1999 274: 6499-506) and can amplify signalling processes through their cognate receptors. Currently, CD44 is also considered as an activator of TGFß signalling by virtue of its ability to recruit matrix metalloproteinases to the plasma membrane, which in turn leads to cleavage of inactive TGF p into active TGF (3 (Yu et al. , Genes Dev 1999 13 : 35-48, Herrlich et al., Immunol Today 1993 14: 395-9).

CD44 has originally been described as lymphocyte homing receptor, CD44- dependent cellular adhesion and trafficking being the underlying biological processes.

Apart from its role in physiology, CD44 is strongly linked to a multitude of pathological conditions, including inflammation and cancer. Under these circumstances splicing patterns are often radically altered (Gunthert et al. , Cell 1991 65 : 13-24, Heider et al. , J Cell Biol 1993 120: 227-33, Wielenga et al., Cancer Res 1993 53: 4754-6).

Gunthert et al. have implicated the variable exon v6 in metastatic propensity of rat pancreatic carcinoma cells.

Membrane-anchored CD44 can be cleaved by MT1-MMP at the cell surface and is subject to subsequent dual intramembranous cleavage by presenilin-1.

Presenilin-1-dependent processing of CD44 results in the liberation of a cytosolic fragment, CD44-ICD, that can translocate to the nucleus to control gene transcription.

Proteolysis of membrane-anchored CD44 results in the release of CD44 preassembled into complexes with matrix components or the release of the ectodomain that then can accumulate as an integral component of the matrix due to association with other matrix components. Alternatively, transmembrane CD44 may be proteolytically released from the cell surface or synthesized de novo in soluble form. The released ectodomain of CD44 can be retained in the ECM by establishing physical associations with other matrix components such as fibronectin, HA, and collagen (reviewed by Cichy and Pure, J. Cell Biol. 2003,161 (5): 839-43).

CD44 undergoes sequential proteolytic cleavages in the extracellular and transmembrane domains. The ectodomain cleavage is mediated by membrane- associated metalloproteases (MMPs) under the regulation of multiple signaling pathways such as the activation of PKC or the influx of calcium. The first cleavage process generates a soluble NH2-terminal fragment released into the cellular exterior (or culture supernatant) and the membrane-bound COOH-terminal cleavage product (CD44EXT, CD44 extracellular truncation). Following the ectodomain cleavage, required for the next step, the intramembranous cleavage of CD44EXT occurs through presenilin gamma secretase (PSy), resulting in the release of a CD44 intracellular domain (CD44ICD) into the cytoplasm. CD44ICD translocates to the nucleus and enhances transcription that is mediated through TPA responsive elements (TREs). The proteolytic cleavage of CD44 at each domain regulates CD44-mediated cell behavior (Murakami et al. , Oncogene 2003,22 : 1511-6 and Lammich et al. , JBC 2002,277 (47):

44754-9). Enhanced CD44 cleavage was observed in gliomas, breast carcinomas, non- small cell lung carcinomas, colon carcinomas, and ovarian carcinomas (Okamoto et al., Am J Pathol. 2002 ; 160 (2): 441-7).

One determining factor whether a protein is a substrate for presenilin dependent cleavage appears to be the size of the extracellular domain, rather than the recognition of specific target sequences (Struhl et al. , Mol. Cell, 2000 6: 625-36). This suggests that for the variability of the extracellular domain of CD44 (with its vl-v10 variable exons) not only the difference in sequence, but also (perhaps to a larger extent) differences in size may influence or interfere with the cleavage of CD44. Potentially the variable array of exons vl-v10 may be replaced by a specific splice variant, a mutated splice variant and/or even unrelated sequences, such as heterologous sequences from different proteins from (preferably) the same species or different species.

It was known in the art before the present invention that expression of CD44 in malignant cells may be an important factor in primary tumor growth, local invasiveness and metastatic potential. Numerous studies have found associations between malignant transformation and cancer metastasis and the expression of CD44 variant isoforms, for example in metastasis of rat pancreatic carcinoma cells (Gunthert et al. , Cell 65 p 13- 24, 1991), human glioma adhesion and invasion (Li H. Cancer Res, Cancer Res.

1993 ; 53 (22): 5345-9 and Merzak A, Cancer Res. 1994; 54 (15): 3988-92) uterine cervical carcinomas (exons v7 and v8, Dall P, Cancer Res. 1994; 54 (13): 3337-41) and colorectal tumor progression and unfavorable prognosis (Wielenga et al. , Cancer Res.

1993; 53 (20): 4754-6, Finn L et al. , Biochem Biophys Res Commun. 1994; 200 (2): 1015- 22) gastric carcinomas (exons v5 and v6, Heider et al., Cancer Res. 1993 ; 53 (18): 4197-203) breast tumors (Joensuu et al., Am J Pathol. 1993; 143 (3): 867-74, Kaufmann et al., Lancet. 1995; 345 (8950): 615-9) Because of their association with malignancies, expression of CD44 and its variants also serve as useful prognostic markers for various malignant or pre-malignant cells (Mulder et al., Lancet 344, p 1470-2,1994 ; Wielenga et al. , Cancer Res 53, p 4754-6,1993 ; Wielenga et al. , Scand J Gastroenterol 33 p 82-7 1998; Wielenga et al., Adv. Cancer Res 77 p 169-187, 2000 and US patents: US 5616468, US 5830646, US 5879898, US 5885575, US 5916561,6010865, US 6372441)

CD44 expression is also a known target for anti-tumour and anti inflammatory therapies. Experiments in animals have shown that targeting of CD44 by antibodies, antisense oligos and CD44-soluble proteins markedly reduces the malignant activities of various neoplasms, stressing the therapeutic potential of anti-CD44 agents.

Antisense strategies and various oligonucleotide based (improved) therapies directed against CD44 expression have been developed (US 6150162 and US 5990299).

However, the current invention demonstrates for the first time a tumor suppressing activity by CD44 molecules.

Description of the invention Definitions Per definition a CD44 splice variant protein is a CD44 protein comprising in addition to the constant exons at least one of the variable exons vl to v10 (for reference see sequence listing). A tumor suppressing CD44 protein is determined by the herein disclosed bio-assay or functional assay, which is defined as the tumor suppressing activity by a CD44 splice variant comprising at least one, preferably at least 5 of the variable exons, when expressed in the colon of Apcmin/+ mice.

The term"CD44 splice variant protein"is herein understood to refer to a polypeptide that comprises, preferably in an N-terminus to C-terminus order: (a) an amino acid sequence that is substantially similar to the amino acid sequence encoded by (the constant) exons 1 to 5 of a mammalian CD44 gene ; (b) an amino acid sequence that is substantially similar to an amino acid sequence encoded by a nucleotide sequence comprising at least 5 variable exons selected from the variable exons vl to v10 of a mammalian CD44 gene ; and, (c) an amino acid sequence that is substantially similar to the amino acid sequence encoded by (the constant) exons 15 to 19 of a mammalian CD44 gene.

Amino acid sequences that are substantially similar to amino acid sequences encoded by the constant exons of a mammalian CD44 gene, preferably have at least 50, 60,70, 80, 90,95 or 99% amino acid identity with an amino acid sequence as depicted in SEQ ID NO: 1 or SEQ ID NO: 2 (human and mouse CD44s sequences respectively which are 78% identical and 85% similar at amino acids level).

Amino acid sequences that are substantially similar to amino acid sequences encoded by variable exons vl to v10 of a mammalian CD44 protein preferably have at

least 50,60, 70,80, 90,95 or 99% amino acid identity with an amino acid sequence as depicted in SEQ ID NO: 1 or SEQ ID NO: 2 (human and mouse protein sequences respectively, comprising all variable exons). The human CD44 sequence: SwissProt ID : P16070, murine Cd44 SwissProt ID: P15379, human CD44v2-v10 SwissProt ID : P16070, murine Cd44vl-v10 PIR ID: S30397. The average amino acid identity (including all variable exons): between mouse-human 64%; mouse-rat 91% ; rat-human 67%, calculated as specified in the following section.

Preferably the CD44 splice variant protein is a naturally occurring splice variant of a mammalian CD44 protein. Alternatively, a nucleotide sequence encoding a CD44 splice variant protein may be reconstructed from the various CD44 exons in vitro, whereby preferably the natural occurring order of the exons is maintained, as well as a correct reading frame for proper translation. Such reconstructed proteins may contain additional amino acid sequences linking the various exon-encoded peptides or they may lack short amino acid sequences. The cDNA sequences of the variable exons of human and murine CD44 can be found in the sequence listing.

Particularly preferred CD44 splice variant proteins are the mammalian CD44 splice variants comprising at least the variable exons v4 to v10, more preferably the variable exons v3 to v10. These CD44 splice variant proteins are referred to herein as CD44v4 and CD44v3. Preferred CD44v4-v10 and CD44v3-v10 proteins have an amino acid sequence that has at least 50,60, 70,80, 90,95 or 99% amino acid identity with the amino acid sequences as depicted from SEQ ID NO. 's : 5 and 6, respectively, up to SEQ ID NO 12.

Included in the term"CD44 splice variant protein"are mutant proteins having amino acid sequences as defined above but comprising insertion, deletions and/or substitutions. Such mutant proteins include both naturally occurring allelic variants of CD44 splice variants as well as engineered mutants. Preferably the mutant CD44 splice variant protein retain at least one activity of a CD44 protein.

Preferred CD44 splice variant proteins as described above have at least one biological activity of a native CD44 splice variant protein such as e. g. CD44v3 or CD44v4. Preferably, the biological activity of the CD44 proteins for use in the present invention at least comprises one or more of the ability to bind hyaluronic acid; the ability to bind fibronectin ; the ability to bind collagen ; the ability to bind E-and/or L- selectin ; the ability to bind aggrecan ; the ability to maintain homeostasis of colonic

epithelial polyps; the ability to restore differentiation of colonic epithelial polyps; the ability to prevent, inhibit and/or abrogate the initiation of colonic epithelial polyps; and, the ability to inhibit, prevent or reverse progression of colonic epithelial polyps to malignant carcinomas and metastases.

A suitable assay for the biological tumour suppressing activity of a CD44 splice variant protein is to test its ability to reduce the number of polyps and tumours in ApcMin/+mice in vivo as the assays described in the examples, provided in this application. Preferably, the ability of a CD44 splice variant protein to reduce the number of polyps and tumours means that the ApcMin/+ mice in vivo in which the CD44 splice variant protein is expressed have develop at least 50,60, 70,80, or 90% less polyps and/or tumours as compared to the same ApcMin/+ mice expressing the CD44s protein. SEQ ID NO: Description 1 Human CD44 protein sequence, 742 AA, 81553 Da. 2 Murine CD44 protein sequence, 778 AA, 85617 Da. Human Cd44s cDNA (all constant exons) 4 Human Cd44 v2 (variable exon v2 cDNA) 5 Human Cd44 v3 Human Cd44 v4 7 Human Cd44 v5 8 Human Cd44 v6 9 Human Cd44 v7 10 Human Cd44 v8 11 Human Cd44 v9 12 Human Cd44 vlO 13 Murine Cd44s cDNA (all constant exons) 14 Murine Cd44 vl (variable exon vl cDNA) 15 Murine Cd44 v2 16 Murine Cd44 v3 17 Murine Cd44 v4 18 Murine Cd44 v5 19 Murine Cd44 v6 20 Murine Cd44 v7 21 Murine Cd44 v8 22 Murine Cd44 v9 23 Murine Cd44 vlO 24 Primers HygroRes (forward) 25 Primer HygroRes (reverse) 26 Primer NeomycinRes (forward) 27 Primer NeomycinRes (reverse) 28 Primer Murine Cd44 exon2 (forward-1) 29 Primer Murine Cd44 exon2 (forward-2) 30 Primer Murine Cd44 exon3 (reverse) 31 Primer Murine Cd44 exon8 (forward) 32 Primer Murine Cd44 exon9 (reverse) 33 Primer Murine Cd44 exon20 (reverse)

Sequence identity The amino acid sequence identity between the various polypeptides comprised in the term"CD44 splice variant protein"may be readily calculated by known methods, including but not limited to those described in (Computational Molecular Biology, Lesk, A. M., ed., Oxford University Press, New York, 1988; Biocomputing: Infomatics and Genome Projects, Smith, D. W., ed., Academic Press, New York, 1993; Computer Analysis of Sequence Data, Part I, Griffin, A. M. , and Griffin, H. G. , eds. , Humana Press, New Jersey, 1994; Sequence Analysis in Molecular Biology, von Heine, G., Academic Press, 1987; and Sequence Analysis Primer, Gribskov, M. and Devereux, J., eds. , M Stockton Press, New York, 1991; and Carillo, H. , and Lipman, D., SIAM J.

Applied Math. , 48: 1073 (1988). Preferred methods to determine identity are designed to give the largest match between the sequences tested. Methods to determine identity are codified in publicly available computer programs. Preferred computer program methods to determine identity between two sequences include, but are not limited to, the GCG program package (Devereux, J. , et al., Nucleic Acids Research 12 (1): 387 (1984)), BestFit, BLASTP, BLASTN, and FASTA (Altschul, S. F. et al. , J. Mol. Biol.

215: 403-410 (1990). The BLAST X program is publicly available from NCBI and other sources (BLAST Manual, Altschul, S. , et al. , NCBI NLM NIH Bethesda, MD 20894; Altschul, S. , et al. , J. Mol. Biol. 215: 403-410 (1990). The well-known Smith Waterman algorithm may also be used to determine identity. Preferred parameters for polypeptide sequence comparison include the following: 1) Algorithm: Needleman and Wunsch, J. Mol. Biol. 48: 443-453 (1970) Comparison matrix: BLOSSUM62 from Hentikoff and Hentikoff, Proc. Natl. Acad. Sci. USA. 89 : 10915-10919 (1992); Gap Penalty: 12; and Gap Length Penalty : 4. A program useful with these parameters is publicly available as the"Ogap"program from Genetics Computer Group, located in Madison, WI. The aforementioned parameters are the default parameters for peptide comparisons (along with no penalty for end gaps).

Operably linked As used herein, the term"operably linked"refers to a linkage of polynucleotide elements in a functional relationship. A nucleic acid is"operably linked"when it is placed into a functional relationship with another nucleic acid sequence. For instance, a promoter or enhancer is operably linked to a coding sequence if it affects the transcription of the coding sequence. Operably linked means that the DNA sequences being linked are typically contiguous and, where necessary to join two protein coding regions, contiguous and in reading frame.

Promoter As used herein, the term"promoter"refers to a nucleic acid fragment that functions to control the transcription of one or more genes, located upstream with respect to the direction of transcription of the transcription initiation site of the gene, and is structurally identified by the presence of a binding site for DNA-dependent RNA polymerase, transcription initiation sites and any other DNA sequences, including, but not limited to transcription factor binding sites, repressor and activator protein binding sites, and any other sequences of nucleotides known to one of skill in the art to act directly or indirectly to regulate the amount of transcription from the promoter. A "constitutive"promoter is a promoter that is active under most physiological and developmental conditions. An"inducible"promoter is a promoter that is regulated depending on physiological or developmental conditions. A"tissue specific"promoter is only active in specific types of differentiated cells/tissues.

Detailed description of the invention.

The invention is based on the unexpected discovery that CD44 splicing can be a determinant of intestinal polyp formation in ApcMin/+ mice. A comparative analysis of the function of three different CD44 variants in ApcMin/+ mice revealed striking differences in their biological effects. Apem"/+ mice contain one defective and one functional allele of the Apc gene and are a well accepted experimental laboratory animal model (Luongo et al., Cancer Res 54 p 5947-52,1994 ; Moser et al. , Science 247 p 322-4,1990), mimicking the adenoma-carcinoma sequence as observed in human FAP patients. The unexpected outcome of CD44 knock-out (Schmits et al., Blood 90 p 2217-33,1997) experiments and knock-in experiments of CD44 splice variants, which have been bred into an ApcMin/+ genetic background, show that CD44 splicing is a

major determinant of polyp initiation that governs the progression and outcome of malignant disease in these mice.

ApcMin/+ mice expressing tumor suppressing CD44 splice variant proteins of the invention develop a markedly reduced number of colonic polyps of a smaller diameter compared to the polyps of ApcMin/+ mice expressing the short form CD44s, i. e. the wild type CD44, or mice lacking a functional CD44 gene (CD44 knock out mice). Mice expressing tumor suppressing CD44 splice variants also have a markedly reduced tumour initiation in their small intestines (figure 4). These surprising effects are highly significant for the survival of ApcMin/+ mice. Figure 5 shows that ApcMin/+ mice expressing wildtype or no CD44 exhibit almost identical survival rates, whereas mice expressing exclusively the CD44 splice variants CD44v4-10 or CD44v3-10 have a significantly enhanced, almost doubled life span. Moreover, this effect of the tumor suppressing CD44 splice variants is dominant over the CD44s allele, as indicated by the heterozygous ApcMitl/+ [CD44v3-10/CD44s] mice, which have a comparable survival rate to the homozygous CD44v3-10 and Cd44v4-10 ApcMin/+ mice.

Clearly these results are in sharp contrast with the prevailing view in the art which identifies CD44 and CD44 splice variants as molecules which are upregulated in malignant cells and expression of which is associated with metastasis and poor prognosis. As described above, various methods have been developed which aim at the use of CD44 as a diagnostic or therapeutic target and methods are known in the art which seek to antagonise, reduce of abolish the expression of CD44s or its splice variants. None of these studies however have identified the tumor suppressor activity encoded by the CD44 locus that forms the basis of the present invention.

In the broadest sense, the present invention thus relates to CD44 proteins, in particular CD44 splice variant proteins and nucleic acids encoding such variants and their use in inhibition of initiation of malignancies and inhibition of progression of malignant transformation in mammalian cells. In particular the invention aims at prophylaxis or therapy of neoplastic conditions that involve a defective Wnt signalling in mammalian cells. In a particular embodiment, the invention is aimed at the use of tumor suppressing CD44 splice variant proteins in the prophylaxis or therapy of neoplastic conditions that involve a defective Ape tumour suppressor gene or an otherwise aberrant Wnt signalling (such as oncogenic mutations in p-catenin or TCF) in mammals. In a more particular embodiment, the invention relates to tumor

suppressing CD44 splice variant proteins and their use in maintaining intestinal homeostasis, preferably in adult mammals. The invention thus relates to methods in which a source of tumor suppressing CD44 splice variant proteins is provided to tissues along the intestinal tract, preferably to stem cells in the crypts of the villi in the intestinal mucosa, whereby levels, location and timing of the provision of the source of the CD44 splice variant protein molecules results in a delay, reduction or inhibition of carcinogenesis in these tissues.

While it was known in the art before the present invention that CD44 is involved in the onset and progression of various types of malignancies in various types of tissues, no understanding was available in the art regarding a role for CD44 splice variants in the inhibition of tumorigenesis in tissues, in particular no information was available as to the role of CD44 splice variants in suppressing tumorigenesis in intestinal tissues. We have now found that expression of tumor suppressing CD44 splice variants prevents, as well as provides for a treatment for carcinogenesis in the adult intestinal tract and colonic tissues, in particular the colonic mucosa.

The present invention thus provides for methods of treatment, and compositions for use in methods of diagnosis, prevention and therapy of intestinal epithelial tumorigenesis, in particular carcinogenesis of gastric and colonic tissues, more in particular the intestinal tract mucosa, using compositions comprising tumor suppressing CD44 splice variant proteins or nucleic acids coding therefor, or compositions for the detection of these splice variant molecules, or compositions regulating the expression of endogenous tumor suppressing CD44 splice variants.

Thus in a first aspect, the present invention relates to a method of treating or preventing a neoplastic condition resulting from an altered regulation of the Wnt signaling pathway, wherein the method comprises providing a source of a mammalian CD44 splice variant protein, whereby a tumor suppressing CD44 splice variant protein comprises: (a) an amino acid sequence that is substantially similar to the amino acid sequence encoded by exons 1 to 5 of a mammalian CD44 gene; (b) an amino acid sequence that is substantially similar to an amino acid sequence encoded by a nucleotide sequence comprising at least 5 variable exons selected from the variable exons vl to v10 of a mammalian CD44 gene; and, (c) an amino acid sequence that is substantially similar to the amino acid sequence encoded by exons 15-19 of a mammalian CD44, whereby the source of the tumor suppressing CD44 splice variant

protein is provided in an amount sufficient to prevent or inhibit the initiation or progression of a neoplasm The altered regulation of the Wnt signalling is usually caused by a congenital or acquired mutation in an APC gene or in a p-catenin gene.

The congenital or acquired deficiency of Wnt signalling may result in enhanced transcriptional activation by TCF transcription factors such as TCF-4. Enhanced transcriptional activation by TCF transcription factors through Wnt-> (3-catenin/Apc-> TCF signalling pathway may be caused by congenital or acquired mutations in the Apc gene or in the p-catenin gene. In the method, the source of the tumor suppressing CD44 splice variant protein may be provided to the GI tract of a subject suffering from the defective Wnt signalling for the prophylaxis of carcinogenesis in the GI tract.

Preferably the source of tumor suppressing CD44 splice variant protein is provided for the prophylaxis of small intestinal or colorectal cancer. The method comprises providing a source of tumor suppressing CD44 splice variant protein to the GI tract of a subject suffering from the acquired deficiency in the Wnt signalling pathway, preferably by congenital or acquired mutations in the Apc gene or the (3 catenin gene, for the treatment of a GI tract neoplasia. Preferably the source of tumor suppressing CD44 splice variant protein is provided for the treatment of cancers of the intestinal tract or colon, however, it is clear for a skilled person that any tissue or tumour type in which the Wnt signalling pathway is perturbed by for instance mutations in Apc, p-catenin or Tcf genes, may be treated by the methods disclosed herein. Other embodiments apart from treating intestinal tract malignancies with CD44 splice variant proteins are readily envisaged by a person skilled in the art.

Preferably in the method of the invention, wherein the tumor suppressing CD44 splice variant protein comprises the amino acids that are encoded by at least 5,6, 7 or 8 variable exons selected from the variable exons vl to v10. More preferably, the amino acids encoded by the exons are present in a naturally occurring order. Most preferably, the tumor suppressing CD44 splice variant protein is a naturally occurring splice variant protein. Particularly preferred tumor suppressing CD44 splice variant proteins are CD44v4-10 or CD44v3-10 comprising the variable exons v4 to v10 (7 variable exons) or v3 to v10 (8 variable exons). Preferably the CD44 splice variant protein for use in the methods of the invention is a human CD44 splice variant protein.

In a preferred embodiment of the invention, the invention relates to methods for treating or preventing a neoplasm in the gastrointestinal (GI) tract. Neoplasms of the GI

tract that may be treated by the methods of the invention include e. g. a colonic adenomatous polyp, an invasive adenocarcinoma, a small intestinal adenoma, a small intestinal carcinoma, a desmoid tumour or a colorectal cancer. The methods of the invention are particularly suited for treating or preventing a neoplasm in a subject diagnosed with familial adenomatous polyposis coli (FAP). In subjects diagnosed with FAP a total colectomy is performed, generally before the age of 20 years in order to prevent colonic cancers. However, at a later age colectomised FAP patients frequently develop tumours in the small intestine. Unfortunately, preventive surgical removal of the small intestine is not possible. Therefore, the methods of the present invention are particularly suited for prevention and therapy of neoplasms in the small intestine in colectomised FAP patients.

The methods comprises administering to a subject diagnosed with FAP a source of tumor suppressing CD44 splice variant protein to the GI tract of the subject. The method preferably is a method that prevents or reverses tumorigenesis in the subject diagnosed with FAP. Preferably the method is a method that prevents the onset or treats GI tract tumours, in particular, adenomatous polyps in the colon (if still present) and in the small intestine and delays, inhibits or abrogates the onset of polyps and the progression of polyps into invasive adenocarcinomas, small intestinal adenomas and cancers, desmoid tumors and metastasis to surrounding tissues and organs in the body.

In these prophylactic and therapeutic methods, the source of tumor suppressing CD44 splice variant proteins is administered in such an amount that functional levels of CD44 splice variant protein is maintained at a sufficient level in the subject's relevant tissues, such as the GI tract, to inhibit the onset of polyp formation and the progression to adenomatous polyps and carcinomas. The functional level of tumor suppressing CD44 splice variant protein achieves the desired prophylactic or therapeutic effects.

Such a functional level preferably is a level that maintains homeostasis of gastric, (small) intestinal and/or colonic epithelia, or a level that restores differentiation of tumorigenic cells in these tissues and prevents uncontrolled proliferation and subsequent invasive or metastasising behaviour. The functional level may be determined by the use of the experimental model provided in the current invention, which demonstrates the use of the ApcM"'+ mice as a model to test the effectiveness of different CD44 splice variants for the inhibition of colonic tumorigenesis. Expression constructs or delivery systems, expression of different splice variants and/or mutant

CD44 proteins, expression levels, distribution and timing may be assayed using this experimental setup of transgenic knock-in mice in an ApcMin/+ genetic background.

During the course of the prophylaxis or therapy the administered amount of the source of tumor suppressing CD44 splice variant protein may be adjusted based on the levels of malignancies, polyps or tumours, with respect to number, size and grade or stage, measured in the relevant tissues. The norm for a functional level in a given intestinal tissue in a given physiological condition may be established by determining the status of corresponding tissues under comparable conditions in healthy individuals by methods known in the art per se. However, the administered amount of the source of the CD44 splice variant protein may be therapeutic amount that effects a supranormal level of the CD44 splice variant protein in GI tract tissues. Such a supranormal level may be a factor 2, 10,100, 1000 or higher than the norm for a normal physiological level of specific CD44 splice variant proteins in the GI tract tissues.

In the methods of the invention, the source of tumor suppressing CD44 splice variant protein may be any composition that may be administered to a subject, or to organs, tissues or cells, in a method of delivery that is capable of effecting a functional protective level of tumor suppressing CD44 splice variant protein in the intestinal epithelium or other tissues at risk or affected by the consequences of altered Wnt signalling, alterations as caused by for instance, but not limiting to, mutations of the Ape, (3-catenin or Tcf genes. Thus, the source of CD44 splice variant protein may be a pharmaceutical composition comprising a tumor suppressing CD44 splice variant protein, preferably a pharmaceutical composition that is suitable for oral administration; vesicles or other release factors capable of delivering tumor suppressing CD44 splice variant proteins to the desired target cells, tissues or organs; a gene therapy vector comprising a nucleotide sequence encoding a CD44 splice variant protein and capable targeting expression of that sequence in the relevant tissues; an (enteric) bacterium capable of colonising (parts of) the GI tract, wherein the bacterium comprises a nucleotide sequence encoding a tumor suppressing CD44 splice variant protein, that confers to the bacterium the ability to secrete the tumor suppressing CD44 splice variant protein in the lumen or deliver the tumor suppressing CD44 splice variant protein directly to cells; a virus or viral vector capable of delivering tumor suppressing CD44 splice variant expression to cells and/or tissues; a (stem) cell, preferably an autologous stem cell, preferably an epithelial stem cell or a peripheral mononuclear

blood cell that has been transformed ex vivo with a nucleotide sequence that is capable of expressing a tumor suppressing CD44 splice variant protein; a (small) molecule that (up) regulates expression of endogenous tumor suppressing CD44 splice variant protein by affecting the level of expression or influencing the nature of the splicing process ; or a molecule that alters CD44 splice variant protein activity, such as e. g. an antibody against an CD44 splice variant protein.

In another aspect the invention relates to the use of a tumor suppressing CD44 splice variant protein for the manufacture of a pharmaceutical composition for the treatment or prevention of a neoplastic condition resulting from a deficiency of a Wnt signalling in the human body, preferably the human GI tract. The invention also relates to the use of a gene therapy vector comprising a nucleotide sequence encoding a tumor suppressing CD44 splice variant protein, for the manufacture of a pharmaceutical composition for the treatment or prevention of a neoplastic condition resulting from aberrant Wnt signalling in the GI tract and other parts of the body.

In a further aspect, the invention relates to a gene therapy vector comprising a nucleotide sequence encoding a tumor suppressing CD44 splice variant protein.

Nucleotide sequences encoding CD44 splice variant proteins, gene therapy vectors and methods for their construction and use are as described below. In particular a vector capable of targeting/transforming stem cells in the crypts of villi in the intestinal mucosa are preferred. In such a method, the intestines of individuals to be treated may be clamped en cleaned, the mucosa may be stripped by various methods known in the art per se (see e. g. Canto et al. , 2000, Gastrointest. Endosc. 51 : 560-568), after which the crypt cells are brought into contact with a transforming vector or virus, capable of transducing expression of a tumor suppressing CD44 splice variant proteins to the stem cells in the crypts of the villi.

Pharmaceutical compositions In some methods, tumor suppressing CD44 splice variant protein purified from mammalian, insect or microbial cell cultures, from milk of transgenic mammals or other source is administered in purified form together with a pharmaceutical carrier as a pharmaceutical composition. The preferred form depends on the intended mode of administration and therapeutic application. The pharmaceutical carrier can be any compatible, nontoxic substance suitable to deliver the polypeptides to the patient.

Sterile water, alcohol, fats, waxes, and inert solids may be used as the carrier.

Pharmaceutically acceptable adjuvants, buffering agents, dispersing agents, and the like, may also be incorporated into the pharmaceutical compositions.

The concentration of the tumor suppressing CD44 splice variant protein in the pharmaceutical composition can vary widely, i. e. , from less than about 0. 1% by weight, usually being at least about 1 % by weight to as much as 20% by weight or more.

For oral administration, the active ingredient can be administered in solid dosage forms, such as capsules, tablets, and powders, or in liquid dosage forms, such as elixirs, syrups, and suspensions. Active component (s) can be encapsulated in gelatin capsules together with inactive ingredients and powdered carriers, such as glucose, lactose, sucrose, mannitol, starch, cellulose or cellulose derivatives, magnesium stearate, stearic acid, sodium saccharin, talcum, magnesium carbonate and the like. Examples of additional inactive ingredients that may be added to provide desirable colour, taste, stability, buffering capacity, dispersion or other known desirable features are red iron oxide, silica gel, sodium lauryl sulfate, titanium dioxide, edible white ink and the like.

Similar diluents can be used to make compressed tablets. Both tablets and capsules can be manufactured as sustained release products to provide for continuous release of medication over a period of hours. Compressed tablets can be sugar coated or film coated to mask any unpleasant taste and protect the tablet from the atmosphere, or enteric-coated for selective disintegration in the gastrointestinal tract. Liquid dosage forms for oral administration can contain colouring and flavouring to increase patient acceptance.

The tumor suppressing CD44 splice variant protein composition may be administered parentally. CD44 splice variant protein for preparations for parental administration is preferably sterile. Sterilisation is readily accomplished by filtration through sterile filtration membranes, prior to or following lyophilisation and reconstitution. The parental route for tumor suppressing CD44 splice variant protein administration is in accord with known methods, e. g. injection or infusion by intravenous, intraperitoneal, intramuscular, intraarterial or intralesional routes. Tumor suppressing CD44 splice variant protein is administered continuously by infusion or by bolus injection. A typical composition for intravenous infusion could be made up to contain 10 to 50 ml of sterile 0.9% NaCl or 5% glucose optionally supplemented with a 20% albumin solution and 1 to 50 ig of the CD44 splice variant protein. A typical

pharmaceutical composition for intramuscular injection would be made up to contain, for example, 1-10 ml of sterile buffered water and 1 to 100 zg of a tumor suppressing CD44 splice variant protein of the present invention. Methods for preparing parenterally administrable compositions are well known in the art and described in more detail in various sources, including, for example, Remington's Pharmaceutical Science (15th ed. , Mack Publishing, Easton, PA, 1980) (incorporated by reference in its entirety for all purposes).

The pharmaceutical compositions of the present invention may be administered orally. Intradermal, intramuscular or intravenous administration is also possible in appropriate circumstances. The compositions can be administered for prophylactic treatment of individuals suffering from, or susceptible to, carcinogenesis of the GI tract in an amount sufficient to prevent, delay or reduce the severity of subsequent disease.

For therapeutic applications, the pharmaceutical compositions are administered to a patient suffering from established disease, carcinogenesis of the GI tract, in an amount sufficient to reduce the severity of symptoms and/or prevent or arrest further development of symptoms. An amount adequate to accomplish this is defined as a "therapeutically-"or"prophylactically-effective dose. "Such effective dosages will depend on the severity of the condition and on the general state of the patient's health.

In the present methods, tumor suppressing CD44 splice variant protein is usually administered at a dosage of about 1 llg/kg patient body weight or more per week to a patient. Often dosages are greater than 10 ug/kg per week. Dosage regimes can range from 10 pg/kg per week to at least 1 mg/kg per week. Typically dosage regimes are 10 Fg/kg per week, 20, ug/kg per week, 30 pg/kg per week, 40 pg/kg week, 60 ig/kg week, 80 pg/kg per week and 120 ! lg/kg per week. In preferred regimes 10 u. g/kg, 20 jug/kg or 40 ug/kg is administered once, twice or three times weekly. Treatment is typically continued for at least 4 weeks, sometimes 24 weeks, and sometimes for the life of the patient. Treatment is preferably administered by oral route. Alternatively, in some conditions it may be desirable to achieve higher than normal levels, e. g. 150% of normal levels, 200% of normal levels or even 300% of normal levels.

Nucleotide sequences encoding CD44 proteins The species from which the DNA segment encoding a tumor suppressing CD44 splice variant sequence is obtained is not necessarily human. Due to the high

percentage of identity between the CD44 orthologues (TheiBen, Nature 415 p 741, 2002) other mammalian CD44 sequences may be used (sequences available at the NCBI website: www. ncbi. nhn. nih. gov, under accession numbers: Anas platyrynchos AF332869 Bos taurus NM 174013 Bos taurus (2) X62881 Canis familiars Z27115 Ceratotherium simum simum AF045939 Cricetulus migratorius M33827 Equus caballus X66862 Gallus gallus AF153205 Homo sapiens NM 000610 Homo sapiens (2) AY101193 Mus musculus BC005676 Mus musculus (2) BC051388 Rattus norvegicus NM 012924 The use of mammalian and highly homologous CD44 sequences can be envisaged as they provide for the same functionality and are to a large extent interchangeable. Expression and gene therapy vectors or transgenic mammals expressing allelic, cognate and induced variants of any of the prototypical sequence described in these references are included in the invention. Such variants usually show substantial sequence identity at the amino acid level with other CD44 genes. Such variants usually hybridise to a known gene under stringent conditions or cross-react with antibodies to a polypeptide encoded by one of the known genes. Other examples of genomic and cDNA sequences are available from GenBank or the NCBI website. To the extent that additional cloned sequences of CD44 genes are required, they may be obtained from genomic or cDNA libraries (preferably human) using known CD44 sequences. In genomic constructs, it is not necessary to retain all intronic sequences and these may or may not be removed.

For example, some intronic sequences can be removed to obtain a smaller transgene facilitating DNA manipulations and subsequent microinjection. See Archibald et al. , WO 90/05188 (incorporated by reference in its entirety for all purposes). Removal of some introns is also useful in some instances to enhance expression levels. Removal of one or more introns to reduce expression levels to ensure that posttranslational modification is substantially complete may also be desirable. It is also possible to delete some or all of the non-coding exons. In some transgenes, selected nucleotides in CD44 encoding sequences may be mutated to remove

proteolytic cleavage sites or otherwise enhance stability. The sequence encoding a CD44 splice variant protein or any functional homologue thereof may be introduced in an enteric bacterium as described below, or may be incorporated in a viral or non-viral gene therapy vector as described below or introduced as naked DNA expression construct.

Functional CD44 proteins are membrane bound proteins and post translationally modified. CD44s and CD44 splice variant proteins are heavily glycosylated. The most preferred embodiment of the current invention is to provide for expression of tumor suppressing CD44 splice variant coding sequences in situ. Preferably the tumor suppressing CD44 splice variant contains all of the exons v4 to v10, and more preferably the CD44 splice variant protein contains all the variant exons v3 to v10.

However, within the sequence stretch encoded by exons v3 to v10, parts of the sequence may be deleted, mutated or replaced with heterologous (human or other) sequences without significantly altering the functionality of the tumor suppressing CD44 splice variant protein ; the functionality being defined as the tumor suppressor activity by CD44 splice variants described in this application and the examples provided herein.

Nucleotide sequences encoding CD44 and CD44 splice variant proteins may also be defined by their capability to hybridise with the nucleotide sequences encoding the amino acid sequences of SEQ ID NO. 1 to SEQ ID NO. 2, under moderate, or preferably under stringent hybridisation conditions. Stringent hybridisation conditions are herein defined as conditions that allow a nucleic acid sequence of at least about 25, preferably about 50 nucleotides, 75 or 100 and most preferably of about 200 or more nucleotides, to hybridise at a temperature of about 65°C in a solution comprising about 1 M salt, preferably 6 x SSC or any other solution having a comparable ionic strength, and washing at 65°C in a solution comprising about 0.1 M salt, or less, preferably 0.2 x SSC or any other solution having a comparable ionic strength. Preferably, the hybridisation is performed overnight, i. e. at least for 10 hours and preferably washing is performed for at least one hour with at least two changes of the washing solution.

These conditions will usually allow the specific hybridisation of sequences having about 90% or more sequence identity.

Moderate conditions are herein defined as conditions that allow a nucleic acid sequences of at least 50 nucleotides, preferably of about 200 or more nucleotides, to hybridise at a temperature of about 45°C in a solution comprising about 1 M salt, preferably 6 x SSC or any other solution having a comparable ionic strength, and washing at room temperature in a solution comprising about 1 M salt, preferably 6 x SSC or any other solution having a comparable ionic strength. Preferably, the hybridisation is performed overnight, i. e. at least for 10 hours, and preferably washing is performed for at least one hour with at least two changes of the washing solution.

These conditions will usually allow the specific hybridisation of sequences having up to 50% sequence identity. The person skilled in the art will be able to modify these hybridisation conditions in order to specifically identify sequences varying in identity between 50% and 90%.

Recombinant techniques and methods for recombinant production of CD44 splice variant (poly) peptides Peptides and polypeptides for use in the present invention, such e. g. the tumor suppressing CD44 splice variant proteins, can be prepared using recombinant techniques in which a nucleotide sequence encoding the polypeptide of interest is expressed in cultured cells such as described in Ausubel et al. ,"Current Protocols in Molecular Biology", Greene Publishing and Wiley-Interscience, New York (1987) and in Sambrook and Russel (2001) "Molecular Cloning: A Laboratory Manual (3rd edition), Cold Spring Harbor Laboratory, Cold Spring Harbor Laboratory Press, New York ; both of which are incorporated herein by reference in their entirety. Also see, Kunkel (1985) Proc. Natl. Acad. Sci. 82: 488 (describing site directed mutagenesis) and Roberts et al. (1987) Nature 328: 731-734 or Wells, J. A. , et al. (1985) Gene 34: 315 (describing cassette mutagenesis).

Typically, nucleic acids encoding the desired polypeptides are used in expression vectors. The phrase"expression vector"generally refers to nucleotide sequences that are capable of affecting expression of a gene in hosts compatible with such sequences.

These expression vectors typically include at least suitable promoter sequences and optionally, transcription termination signals. Additional factors necessary or helpful in effecting expression can also be used as described herein. DNA encoding a polypeptide is incorporated into DNA constructs capable of introduction into and

expression in an in vitro cell culture. Specifically, DNA constructs are suitable for replication in a prokaryotic host, such as bacteria, e. g., E. coli, or can be introduced into a cultured mammalian, plant, insect, e. g., Sf9 or Sf21, yeast, fungi or other eukaryotic cell lines. Expression constructs may also be used to generate transgenic plant or transgenic animals capable of producing the protein of interest. Alternatively, both viral and non-viral expression constructs may be employed for gene therapy as outlined below.

DNA constructs prepared for introduction into a particular host typically include a replication system recognized by the host, the intended DNA segment encoding the desired polypeptide, and transcriptional and translational initiation and termination regulatory sequences operably linked to the polypeptide encoding segment. A DNA segment is"operably linked"when it is placed into a functional relationship with another DNA segment. For example, a promoter or enhancer is operably linked to a coding sequence if it stimulates the transcription of the sequence. DNA for a signal sequence is operably linked to DNA encoding a polypeptide if it is expressed as a preprotein that participates in the secretion of the polypeptide. Generally, DNA sequences that are operably linked are contiguous, and, in the case of a signal sequence, both contiguous and in reading phase. However, enhancers need not be contiguous with the coding sequences whose transcription they control. Linking is accomplished by ligation at convenient restriction sites or at adapters or linkers inserted in lieu thereof.

The selection of an appropriate promoter sequence generally depends upon the host cell selected for the expression of the DNA segment. Examples of suitable promoter sequences include prokaryotic, and eukaryotic promoters well-known in the art (see, e. g. Sambrook and Russel, 2001, supra). The transcriptional regulatory sequences typically include a heterologous enhancer or promoter that is recognized by the host. The selection of an appropriate promoter depends upon the host, but promoters such as the trp, lac and phage promoters, tRNA promoters and glycolytic enzyme promoters are known and available (see, e. g. Sambrook and Russel, 2001, supra). Expression vectors include the replication system and transcriptional and translational regulatory sequences together with the insertion site for the polypeptide encoding segment can be employed. Examples of workable combinations of cell lines and expression vectors are described in Sambrook and Russel (2001, supra) and in

Metzger et al. (1988) Nature 334: 31-36. For example, suitable expression vectors can be expressed in, e. g. , insect cells, e. g., Sf9 or Sf21 cells, mammalian cells, e. g. , CHO or Hela cells and bacterial cells, e. g. , E. coli.

In vitro mutagenesis and expression of mutant proteins are described generally in Ausubel et al. (1987, supra) and in Sambrook and Russel (2001, supra). Also see, Kunkel (1985, supra; describing site directed mutagenesis) and Roberts et al. (1987, supra; describing cassette mutagenesis).

Another method for preparing polypeptides is to employ an in vitro transcription/translation system. DNA encoding a polypeptide is cloned into an expression vector as described supra. The expression vector is then transcribed and translated in vitro. The translation product can be used directly or first purified.

Polypeptides resulting from in vitro translation typically do not contain the post- translation modifications present on polypeptides synthesized in vivo. Methods for synthesis of polypeptides by in vitro translation are described by, for example, Berger & Kimmel, Methods in Enzymology, Volume 152, Guide to Molecular Cloning Techniques, Academic Press, Inc. , San Diego, CA, 1987 (incorporated herein by reference in its entirety) Gene therapy constructs Viral and or non-viral vectors (or constructs) are used for transfecting the targeted GI tract tissue in particular the gastric, small intestinal and or colonic tissues, more preferably the stem cells in these tissues. The term vector refers to a nucleic acid, protein, lipid or other molecule capable of transporting a nucleic acid to which it has been operatively linked. Vectors may include circular double stranded DNA plasmids and viral vectors. Some vectors are capable of autonomous replication in a host cell into which they are introduced (such as bacterial vectors having a bacterial origin of replication and episomal mammalian vectors). Other vectors (such as non-episomal mammalian vectors) may be integrated into the genome of a host upon introduction into the host cell, and thereby may be replicated along with the host genome. Certain vectors may be capable of directing the expression of genes to which they are operatively linked. In recombinant vectors of the invention, the nucleotide sequences encoding a peptide may be operatively linked to one or more regulatory sequences, selected on the basis of the host cells to be used for expression. The terms operatively

or operably linked mean that sequences encoding the peptide are linked to the regulatory sequence (s) in a manner that allows for expression of the peptide compound.

The term regulatory sequence includes promoters, enhancers, polyadenylation signals and other expression control elements. Such regulatory elements are described in for example in Goeddel ; Gene Expression Technology: Methods in Enzymology 185, Academic Press, San Diego, Calif. (1990). Recombinant expression vectors of the invention may be designed for expression of the tumor suppressing CD44 splice variant peptides in prokaryotic or eukaryotic cells. For example, tumor suppressing CD44 splice variant proteins or peptides may be expressed in bacterial cells such as E. coli, insect cells (using baculovirus expression vectors) yeast cells or mammalian cells.

Suitable host cells are discussed further in Goeddel ; Gene Expression Technology: Methods in Enzymology 185, Academic Press, San Diego, Calif. (1990). Alternatively, the recombinant CD44 expression vector may be transcribed and translated in vivo, for example using T7 promoter regulatory sequences and T7 polymerase. Examples of vectors for expression in yeast S. cerevisiae include pYepSecl, pMFa and pYES2.

Examples of Baculovirus include pAc series and the pVL series. Mammalian expression vectors include pCDM8, often control functions are provided by viral regulatory elements. For example commonly used promoters are derived from polyoma, Adenovirus 2, cytomegalovirus and Simian Virus 40. Vector DNA can be introduced into the prokaryotic or eukaryotic cells via conventional transformation or transfection techniques, including introducing foreign nucleic acid into a host cell using calcium phosphate or calcium chloride co-precipitation, DEA-dextran mediated transfection, lipofection, electroporation, microinjection and viral mediated transfection of which examples can be found in Sambrook et al. (Molecular Cloning: A Laboratory Manual, 2nd Edition, Cold Spring Harbor Laboratory press (1989), and other laboratory manuals. Methods for introducing DNA into mammalian cells in vivo are also known, and may be used to deliver the vector DNA of the invention to a patient for gene therapy. A nucleic acid sequence of the invention may be delivered to cells in vivo using methods such as direct injection of DNA, receptor-mediated transfection or non- viral transfection and lipid based transfection, all of which may involve the use a gene therapy vectors. Defective retroviruses are well characterised for use as gene therapy vectors (Review by Miller, A. D. (1990) Blood 76: 271). Protocols for producing recombinant retroviruses and for infecting cells in vitro or in vivo with such viruses can

be found in Current Protocols in Molecular Biology, Ausubel, F. M. et al (eds. ) Greene Publishing Associates, (1989), Sections 9.10-9. 14 and other standard laboratory manuals. Examples include pLJ, pZIP, pWE, pLZRS and pEM, which are well known to those skilled in the art. (See for example Patent applications W094/26914 and W095/02697). For use as a gene therapy vector, the genome of an adenovirus may be manipulated so that it encodes and expresses a tumor suppressing CD44 splice variant peptide of the invention, but is inactivated in terms of its availability to replicate in a normal lytic viral life cycle. See for example Berkner et al (1988) Biotechniques 6: 616, Rosenfeld et al. (1991) Science 252: 431-434 and Rosenfeld et al (1992) Cell 68: 143- 155. Suitable adenoviral vectors derived from the adenovirus strain Ad type 5 dl234 or other strains of adenovirus (e. g. Ad2, Ad3, Ad7 etc. ) are well known to those skilled in the art. Recombinant adenoviruses are advantageous, as they do not require dividing cells to be effective gene delivery vehicles and can be used to infect a wide variety of cell types, including epithelium, endothelial, hepatocytes and muscle cells. Adeno- associated virus (AAV) may be used as a gene therapy vector for delivery of DNA for gene therapy purposes. AAV may be used to integrate DNA into non-dividing cells (see for example Flotte et al. (1992) Am. J. Respir. Cell. Mol. Bio. 7: 349-356; Samulski et al. (1989) J. Virol. 63: 3822-3828 and Patent applications W091/18088 ; W093/09239; US 4,797, 368; US 5,139, 941 and EP 488 528. An AAV vector may be used to introduce DNA into cells (see for example Hermonat et al. (1985) Mol. Cell.

Biol. 4: 2072-2081. Herpes viral and or Lenti viral gene therapy may also be adapted for use in the invention. General methods for gene therapy are known in the art. See for example US. Pat. No. 5, 399, 346 by Anderson et al (incorporated herein by reference).

In a preferred embodiment the intestines of the subject to be treated can be clamped, opened and the mucosa can be stripped and/or cleaned by methods known in the art per se (see e. g. Canto et al. , 2000, Gastrointest. Endosc. 51: 560-568), after which the intestine is incubated with liposomes, vectors, viral vectors, or a combination thereof, said vectors being capable of transducing stem cells in the crypts of the villi with sequences encoding tumor suppressing CD44 splice variants proteins. During colonoscopy the colonic mucosa may be removed from the colonic epithelium using a water jet and subsequently stained with indigo carmine, which will aid the identification of abberrant crypts. Individuals may be screened using a magnifying colonoscope for the presence of normal crypts or the presence of early or later stage

dysplastic crypts and may subsequently be treated with therapeutic vectors or pharmaceutical preparations providing the tumor suppressive CD44 splice variants according to the current invention.

Figure Legends Figure 1. Targeting constructs and analyses of CD44 knock-in mice. a, Targeting constructs for CD44 variant knock-in mice. H denotes a graphical discontinuity; IRES, independent ribosomal entry site; lacZ, ß-galactosidase encoding gene; Hyg, hygromycin resistance gene; Neo, neomycin resistance gene; Tk, thymidine kinase gene; loxP, bacteriopliage Pl recombinase Cre specific locus of crossing-over. CD44 exons are designated C for constant exon and V for variable exon and numbered consecutively. Genomic fragments and cassettes are not drawn to scale. From top to bottom: targeting construct for CD44s, CD44v4and CD44v3knock-in mice. After transient expression of the Cre-encoding plasmid pOG231, the Neo-Tk cassette is excised from the CD44v3 ES cells'genome. b, Southern blot analysis of ES cells that have undergone a homologous recombination event. Molecular sizes are in kb. Lane 1, wild type ES cells; lane 2, CD44s; lane 3, CD44v4; lane 4, CD44v3; lane 5, an ES cell clone that has undergone 2 recombination events and that is homozygous CD44v3/v3. c, Examples of a PCR analysis on genomic (tail or ear punch) DNA. Molecular sizes are in kb. Lane 1, CD44 exon v3; lane 2, hygromycin; lane 3 = neomycin; lane 4 = ApcMin allele. d, RT-PCR analysis of CD44 knock-out and knock-in mice. Odd lanes depict reactions without, even lanes with reverse transcription. Molecular sizes are in kb. Lanes 1&2, CD44wt; lanes 3&4, CD44-/- ; lanes 5&6, CD44s; lanes 7&8, CD44v4; lanes 9&10, CD44v3. e, Western blot analysis of CD44 knock-out and knock-in mice.

Molecular sizes are in kDa. Lane 1, CD44wt; lane 2, CD44-/-, lane 3, CD44s, lane 4, CD44v4; lane 5, CD44v3.

Figure 2. Polyp number and diameter in opt7/+ mice expressing different CD44 isoforms. Polyp number (a) and diameter in mm (b) in the small intestines of 16- weeks-old Apc"+ mice with different CD44 backgrounds. Numbers of mice per experimental group: wt, CD44 wild type (n=20) ;-/-, CD44-/- (n=10) ; s, CD44s (n=10) ;

v4, CD44v4 (n=ll) ; v3, CD44v3 (n=l 1) ; v3/s, CD44v3/s (n=8). Results are shown as mean i s. e. m.

Figure 3. Expression of CD44 and pKi-67 in the duodenum of CD44 variant mice.

Bars are abc (a), klm (b), or xyz (c-1) um, respectively. a, overview of"Swiss roll"of wild type duodenum, stained with H&E. b, Large polyp in wild type duodenum, stained with H&E. c-e, Sections of wild type duodenum, stained for total CD44 (c), CD44 exon v6 (d), or pKi-67 (e), respectively. f-h, Sections of CD44v3 duodenum, stained for total CD44 (f), CD44 exon v6 (g), or pKi-67 (h), respectively. i, CD44 can not be detected in duodenum of CD44-/-mouse with a pan-CD44 antibody. j, CD44s duodenum stained with antibody 9A4 recognising exon v6 of mouse CD44. Note absence of signal. k-1, pKi-67 staining of CD44-/- (k), or CD44s (1) duodenum, respectively.

Figure 4. Aberrant crypt formation in ApcMin/+ mice expressing different CD44 isoforms. a, Number of aberrant crypt foci (ACF), minipolyps (MP; <0.25 mm) and polyps (P; >0. 25 mm) in the small intestine of 8-weeks-old ApcM"'+ mice with different CD44 backgrounds. b, Total number of anomalies (=ACF+MP+P) specified according to location: D, duodenum; J, jejunum; I, ileum. Open bars, CD44wt; dark grey bars, CD44-/- ; solid bars, CD44v4; light grey bars, CD44v3. Results are shown as mean s. e. m. of 3 mice per experimental group.

Figure 5. Survival of ApcMin'+ mice is dependent on CD44 variant expressed.

ApcMil+ mice with different CD44 backgrounds were sacrificed when first signs of disease were observed, or when becoming moribund. Numbers of mice per experimental group: CD44 wild type (n=44), CD44-/- (n=33), CD44s (n=19), CD44v4 (n=37), CD44v3 (n=52), CD44v3/s (n=37).

Figure 6. Model for redundancy and dominant negative effects of CD44 splice variants. a, Under normal conditions, signalling events downstream of CD44 are required for the formation of intestinal polyps. b, In the presence of dominant negative variants, such as CD44v4-10 and CD44v3-10, these signals are no longer generated.

Alternatively, negative signals are generated. c, In the absence of CD44, an unknown

molecule serves as a molecular stand-in, and signalling events required for polyp formation are not affected.

Table legends Table I. Total number of included and censored animals per genotypic group in the survival study. obs, observation.

Table II. Overview of primers used in this study. ori, orientation of primer : F, forward primer; R, reverse primer; position, exact nucleotide position of primer in gene, or position in specific exon. Amplicon length are indicated in base pairs.

Examples A. Materials and Methods Mice. All mice were kept in conventional housing, subject to a 12 hour dark/light cycle with ad libitum access to water and food. A priori approval for the experiments had been obtained from the University's animal experimentation committee. Mice were inspected daily by 1 or 2 trained observers.

Targeting constructs, embryonic stem cells and blastocysts. A 14 kb clone from a XF1X-II 129/Sv (Stratagene, La Jolla, CA) phage library containing the 5'region of the CD44 locus was isolated and subcloned. For CD44s and CD44v4 knock-in mice a replacement vector was constructed that contained 7.5 kb of genomic DNA upstream from the translation initiation codon and the genomic CD44 portion encompassing constant exon 1 through the ClaI site in constant exon 5. This fragment was fused to a partial cDNA of either CD44s or CD44v4, followed by an IRES, a lacZ gene and a hygromycin cassette (Fig. 1 a). For the CD44v3 knock-in mice a slightly different method was used. The same genomic fragment was fused to a partial CD44v3 cDNA, followed by a floxed neomycine-thymidine kinase cassette (Fig. la). ES cells were cultured as described. (Te Riele et al. , Prox. Natl. Acad. Sci. USA 89 p 5128-32,1992 and Robanus Maandag et al. , Embo J 13 p 4260-68, 1994). Two hundred million 129/Ola ES cells (clone E14) (Hooper et al., Nature 326 p 292-5,1987) were electroporated with 80 u@ of linearised targeting vector. Clones were screened by Southern blot analysis and 2 positive clones were expanded, re-analysed by Southern blot analysis, karyotyped and injected into C57BL6/J blastocysts. The CD44v3 knock- in ES clones were before injection into blastocysts transiently transfected with the Cre- encoding plasmid pOG231 (gift of Dr. Stephen O'Gorman, Salk Institute, San Diego, CA) to obtain ES cells devoid of foreign DNA (Fig. la). Male chimerical mice (for all constructs mentioned above) were crossed with C57BL/6J*CD44 null females (Schmits et al. , Blood 90 p 2217-33,1997) to obtain hemizygous offspring on a mixed 129/Ola x C57BL/6J background. Germ line transmission was determined with the polymerase chain reaction on genomic DNA (see below). Positive animals were crossed back at least 4 times on a C57BL/6JCD44 null background and only hemizygous animals (i. e. CD44s/-, CD44""', or CD44V3/). were retained. By subsequent

interbreeding, homozygous offspring was obtained (96.9% C57BL/6J; balance 129/Ola). Two independent founders per construct were used throughout this study.

Generation of Apc mice expressing specific CD44 splice variants. Initially, C57BL/6J#CD44-/- females (Schmits et al. , Blood 90 p 2217-33,1997) were crossed with C67BL/6J#ApcMin/+ males (Moser et al., Science 247 p 322-4, 1990, purchased from The Jackson Laboratory, Bar Harbor, ME). After a second crossing with CD44 null animals, the offspring was monitored by Southern blot for the loss of the wild type CD44 allele and the retention of the Min allele. The resulting animals, with genotype [CD44-/-]#[ApcMin/+], were either crossed with CD'animals to propagate the [CD44-/- ]#[ApcMin/+] genotype, or crossed with hemizygous CD44 splice variant knock-in animals (see above). After an additional crossing with CD44-1-mice and a final interbreeding round animals were obtained with genotypes [CD4e]. [Apcminl+], [CD44v4/v4]#[ApcMin/+], or [CD44v3/v3]#[ApcMin/+] (99.2% C57BL/6J; balance 129/Ola).

[CD44v3/s]#[ApcMin/+] animals were obtained by a single breeding of [CD44v3/v3]#[ApcMin/+] males with CD44s/s females. To rule out any putative genetic background heterogeneity, (Dietrich et al., Cell 75 p 631-9,1993, MacPhee et al. , Cell 81 p 957-66,1995, van Wezel et al. , Nat Genet 14 p 468-70, 1996, van Wezel et al., Cancer Res 59 p4216-8, 1999, Ruivenkamp et al. , Nat Genet 31 p295-300,2002) we alternatively interbred CD44v3 expressing ApcMin/+ mice and CD44s expressing ApcMin/+ mice to obtain ApcMin'+ mice expressing both isoforms of CD44 at equal levels.

The transheterozygote animals were interbred, giving rise to the original homozygous animals (25% each) and transheterozygote mice (50%). The phenotypes of the resulting homozygous animals were identical to those of the original breeds, pointing to the homogeneity of the genetic background.

Genomic DNA preparation and PCR analysis. Genomic DNA preparation from ear punches or tails and PCR conditions were as described before (Van der Neut et al. , Nat Genet 13 p366-9, 1996). The primers used for the screening of CD44 genotype were as described in Table II. Presence of the Min allele was detected by a PCR described elsewhere (Dietrich et al., Cell 75 p631-9,1993). The status of the Moml locus was analysed by PCR as described before (MacPhee et al., Cell 81 p957-66,1995). PCR

reaction products were analysed on a 0.8 or 2 % agarose gel, visualised with ethidium bromide and the results recorded with an ImagowReader.

Detection of Cd44 mRNA. RNA was prepared from 10 mg of spleen with Trizol (Gibco Life Technologies) according to the manufacturer's recommendation. Five pg of total RNA was reverse transcribed in a total volume of 50 Ill using 0.18 Fg of pdN6 primers, 50 mM Tris-HCl (pH 8.3), 75 mM KC1, 3 mM MgCl2, 8 mM DTT, 1.0 mM of each Na-dNTP, 60 U RNase inhibitor and 400 U of M-MLV reverse transcriptase (Gibco Life Technologies). For the second strand 1 ul of this reaction mixture was used in a PCR reaction with primers flanking the CD44 variable exons (see Table I). As positive controls we used a plasmid encoding mouse CD44 and negative controls were 5 ug of the total RNA preparation without reverse transcriptase added. PCR products were run on a 1% agarose gel and visualised with ethidium bromide.

Western blot analysis. Tissue samples were homogenised in RIPA buffer with a douncer, centrifuged to remove insoluble material, the protein concentration determined and the lysates separated on a 10% polyacrylamide gel. Proteins were blotted on PVDF membrane (Immobilon) by tank blotting, and the blots blocked, incubated with the rat anti-mouse CD44 monoclonal antibody KM114 (Pharmingen), washed, incubated with HRP-conjugated goat anti-rat IgGs (Dako, Glostrup, Denmark).

Immunocomplexes were visualised with a standard ECL protocol (Amersham) according to the manufacturer's instructions.

Immunohistochemical analysis. Tissue sections were deparaffinised and rehydrated before the staining reactions. In some cases, tissue sections had to be treated for 10 min. with citrate buffer at 95°C. All immune reactions were preceded by a blocking step with phosphate-buffered saline containing 2% w/v bovine serum albumin and were carried out for 45 min. at room temperature. The following monoclonal antibodies were used: monoclonal rat anti-mouse CD44 antibody IM7 (Pharmingen), monoclonal rat anti-mouse anti-CD44v6 antibody 9A4 (Weiss et al. , J Cell Biol 137 p1137-47, 1997) and monoclonal mouse anti-Ki-67 antibody MIB-5 (Dako). IM7 and 9A4 staining were followed by biotinylated rabbit anti-rat IgG (Dako) and streptavidin ABC complex/HRP (Dako) for IM7, or by biotinylated rabbit anti-rat IgG and streptavidin-

HRP (Dako) for 9A4. For pKi-67 staining, the monoclonal antibody MIB-5 was pre- biotinylated with the Animal Research Kit (Dako) and staining was developed with streptavidin-HRP. Preparations were recorded with a digital camera.

Determination of ACF and polyp numbers. For the determination of polyp numbers 16- weeks-old animals (10-20 per experimental group) were sacrificed and a sample of the tail, spleen, and liver retained. The entire small intestines were taken out, divided in 3 equal parts (denominated duodenum, jejunum and ileum), cut open longitudinally with fine scissors, and spread inside up. With an operation binocular (10 x magnification) the complete length of the intestines was inspected and the polyps classified, counted and measured according to standard criteria as described before (Dietrich et al., Cell 75 p 631-9,1993, Fodde et al., Proc Natl Acad Sci USA 91 p8969-73,1994, Smits et al., Carcinogenesis 18 p 321-7,1997). The smallest lesions observed with this method were 0.25 mm. Lesions with surrounding normal tissue were sampled for routine processing and either fixed in buffered formalin, or frozen in TissueTek O. C. T. compound. For the ACF determination 8-weeks-old animals (3 per experimental group) were sacrificed, the small intestines taken out and divided in 3 equal parts. From each part a Swiss roll was prepared, which was fixed in 4% formalin for no longer than 48 h, embedded in paraffin and cut into 7 um sections. H&E stained tissue sections were analysed by 2 independent and trained observers. All anomalies were counted and categorized into 3 classes: ACF, minipolyps (<0.25 mm), or polyps (>0. 25 mm).

Survival. Animals were inspected daily by 2 independent, trained observers and allowed to stay alive until the first signs of illness appeared. These signs varied from anal prolapses, weight loss, anaemia (white feet) to general discomfort. Animals were immediately sacrificed if any of the above-mentioned signs were detected. Animals that did not develop any of the signs were kept alive up to 70 weeks of age. Ages of animals were expressed in days and survival analysed with a censored Kaplan-Meier cumulative survival test, followed by a Logrank (Mantel-Cox) test.

Example 1 Knock-in mice expressing a single tumor suppressing CD44 variant

Three separate CD44 knock-in mouse lines were generated by homologous recombination in embryonic stem cells, each expressing a single splice variant of CD44, viz. CD44s, CD44v4-v10 (CD44v4) and CD44v3-v10 (CD44v3) (see Fig. la).

We used minigene replacement-type targeting vectors, in which a partial genomic CD44 clone was fused to a partial CD44 cDNA, giving rise to mice expressing a sole CD44 variant (Fig. 1d-e). As judged from histological staining of different tissues, the spatiotemporal distribution pattern of the CD44 variants was indistinguishable from that in wild type mice (not shown). All mice were fertile, had a comparable life expectancy and showed no gross aberrations (not shown).

Example 2 Choice of CD44 splice variant affect polyp number, not size.

Three different CD44 knock-in and CD44 null mice were crossed with C57BL61JApcMin/+ mice and interbred to homozygosity with respect to the CD44 allele.

After 16 weeks, or earlier when moribund, the mutant animals were sacrificed and the number and sizes of the polyps in the entire small intestine determined (Fig. 2a-b). The smallest polyps detected with an operation binocular (lOx magnification) was 0.25 mm.

All adenomas in the small intestine were of the villous and tubulovillous type. The mean number of polyps in [CD44+/+]#[ApcMin/+] (i.e. "normal" ApcMin/+) mice was 79~ 12 (mean s. e. m. , n = 20) and compared well to those found in earlier studies (Moser et al., Science 247 p322-4,1990, Su et al., Science 256 p668-70,1992, Dietrich et al., Cell 75 p 631-9,1993). Targeted disruption of the CD44 locus and the consequential absence of detectable mRNA and protein levels (Fig. 1 d-e) had no significant effect on the development of polyps. Thus, in [CD44-/-]#[ApcMin/+] mice 74 : : L 14 (mean s. e. m., n = 10) polyps were detected. Mice expressing the shortest CD44 variant, CD44s, had 64 11 (mean i s. e. m. , n = 10) polyps throughout the small intestines. The differences between these 3 groups were statistically not different. By contrast, mice expressing long variants of CD44 had dramatically decreased polyp numbers.

[CD44v4/v4]#[ApcMin/+] mice had developed 15 ~ 7 (mean ~ s. e. m. , n = 11) polyps and in [CD44v3/v3]#[ApcMin/+] mice polyps were virtually absent: 1.2 : : L 0.5 (mean s. e. m. , n = 11). To study whether the polyp inhibiting effect of CD44 long variants was dominant we bred CD44V3/v3 females with [CD44s/s]#[ApcMin/+] male mice and found that

[CD44v3/s]#[ApcMin/+] animals harboured 7 zt 3 (mean s. e. m. , n = 8) polyps, i. e. the presence of one CD44v3 allele was sufficient to inhibit polyposis.

The polyp diameters in all experimental groups, except for [CD44v3/v3]#[ApcMin/+] mice, were statistically not different and ranged from 1.5 A 0.2 mm (mean A s. e. m. , n = 644) in [CD44s/s]#[ApcMin/+] mice to 1.69 i 0.14 mm (mean i s. e. m., n= 1579) in [CD44+/+]#[ApcMin/+] animals (Fig. 2b). The [CD44v3/v3]#[ApcMin/+] mice had marginally smaller polyps: 1.2 0.5 mm (mean s. e. m., n=13) (Fig. 2b).

Example 3 CD44 and pKi-67 expression in small intestines.

One of the decisive clues in intestinal cell fate is the balance between apoptosis and proliferation. To study whether the effects of differential variant expression in the small intestines on polyp burden that we observed could be attributed to differences in proliferation, we stained Swiss rolls of 8-weeks-old small intestines with an antibody against the proliferation marker pKi-67 (Gerdes et al., Int J Cancer 31 pl3-20, 1983).

At the same time we verified that the expression pattern of CD44 had not been influenced as a result of the knock-in procedure.

Small intestines were divided in 3 equal parts of ~ 10 cm, designated duodenum, jejunum and ileum. Of all parts Swiss rolls were prepared as shown in Fig.

3a. At this stage polyps had already visibly developed (Fig. 3b). Serial sections were stained with the panning anti-mouse CD44 antibody IM7 (Fig. 3c, f, i), the anti-mouse CD44v6 antibody 9A4 (Fig. 3d, gj), or an antibody against pKi-67 (Fig. 3e, h, k, 1). In duodenums of [CD]- [c] and [CD44v3/v3]#[ApcMin/+] mice, CD44 staining detected with IM7 was found in the crypts, the polyps (if present) and intravillous lymphocytes, but not in villous epithelium (Fig. 3c, f). Staining of [CD44v4/v4]#[ApcMin/+] and [CD44s/s]#[ApcMin/+] duodenums revealed the same pattern (not shown). With 9A4 a similar staining pattern was found in [CD44v3/v3] #[ApcMin/+] (Fig. 3g) and [CD44v4/v4]#[ApcMin/+] mice (not shown), but not in [CD44+/+]#[ApcMin/+] mice. Here, intravillous lymphocytes were CD44v6 negative (Fig 3d). Obviously, in [CD44-/- ]#[ApcMin/+] mice no CD44 was detected (Fig. 3i, see also Fig. le), whereas in [CD44s/s]#[ApcMin/+] duodenums CD44v6 staining was absent (Fig. 3j).

Proliferative patterns, as revealed by pKi-67 staining, were indistinguishable between all experimental groups studied (Fig. 3e, h, k, 1). Thus, strongly pKi-67-positive

nuclei were found in the crypts and adenomas (when present), but not in normal villous epithelium. When comparing the different variant mice, we observed no differences in the numbers of proliferating (pKi-67 positive) cells per crypt. Occasional intra- epithelial and intravillous lymphocytes were positive. We studied intestinal epithelial cell apoptosis by staining the sections with an antibody directed against active caspase- 3. No differences were apparent between all genotypes studied (not shown).

Example 4 Tumor suppressing CD44 splice variants and tumour initiation.

One of the earliest morphological changes along the adenoma-carcinoma sequence is the appearance of ACF in the intestinal epithelium. These aberrant crypts are precursor lesions that will progress to large (r) adenomas and eventually evolve into carcinomas.

We had observed that the forced expression of either CD44v3 or CD44v4 inhibited the number, but not the size, of polyps in the small intestine of ApcMin/+ mice (see Fig. 2a and b). This pointed to an inhibitory effect of these two CD44 variants on tumour initiation rather than progression. To study whether the lower number of polyps in [CD44v4/v4]#[ApcMin/+] and [CD]- [c] mice was indeed due to a lower number of precursor lesions or to diminished progression rate, we determined the numbers of aberrant crypts in the small intestines on an earlier time point, viz. in 8- weeks-old apcMin/+ mice on a CD44 wild type, null or variant knock-in background.

Swiss rolls were prepared and the number and type of anomalies in the small intestine were monitored. Anomalies were divided into 3 classes: ACFs, minipolyps (<0. 25 mm) and polyps (>0. 25 mm) (see also Methods section for classification). Whereas in the small intestines of ApcMi77/+ mice on a CD44+/+ or CD44-/- or CD44-/- background the numbers of all types of anomalies were comparable, mice expressing the CD44v4 or CD44v3 variant forms had significantly lower or no aberrations (Fig. 4a). All anomalies were evenly distributed along the small intestines, although there was a moderate shift towards duodenal localisation of lesions in the ApcMin/+ intestines (Fig. 4b).

Example 5 Survival of ApcMi7ff+ mice is enhanced by long CD44 variants.

To analyse whether differences in polyp take were reflected in actual differences in the survival rate, we used a separate group of mice (see Table I). Animals were inspected

and scored daily by 2 independent, trained observers. Close attention was paid to well known symptoms, such as anal prolapses, weight loss, anaemia (white feet) and general discomfort. Animals were immediately sacrificed if any of the above-mentioned signs were detected, or around 60-70 weeks of age if animals stayed clinically unaffected. In the sacrificed animals we observed polyps up to 6 mm in diameter, which occasionally led to twisting of the small intestines. Especially in cases where a large number of polyps were adjacently positioned, intestinal flow had been impeded, or even discontinued, leading to diminished nutritional uptake. Various groups have reported thatApcMi"/+ mice die around 16 weeks of age, at which time they have accumulated 70-80 intestinal polyps (Moser et al., Science 247 p322-4,1990, Su et al. , Science 256 p668-70,1992). These figures were corroborated by our study (Fig. 5), but we also found that, in line with their comparable polyp numbers, [CD44-/-]#[ApcMin/+] and [CD44s/s]#[ApcMin/+] mice had almost identical survival rates as ApcMin/+ mice (Fig. 5).

In marked contrast, survival rates of ApcMi'+ mice expressing long CD44 variants were greatly enhanced (Fig. 5). Clearly, cumulative survival is directly linked to the mice's propensity to develop polyps.