Ectopic expression of certain Wnt genes in mouse mammary tissue is tumorigenic, and mutations that stabilize β‐catenin are found in various human cancers including colorectal cancer. To determine the role of stabilized β‐catenin in intestinal tumorigenesis in mice, we constructed by embryonic stem (ES) cell‐mediated homologous recombination, a mutant β‐catenin allele whose exon 3 was sandwiched by loxP sequences. When the germline heterozygotes were crossed with mice expressing Cre recombinase in the intestines, the serines and threonine encoded by exon 3 and to be phosphorylated by glycogen synthase kinase 3β (GSK3β) were deleted in the offspring intestines, which caused adenomatous intestinal polyps resembling those in ApcΔ716 knockout mice. Some nascent microadenomas were also found in the colon. These results present experimental genetic evidence that activation of the Wnt signaling pathway can cause intestinal and colonic tumors.
β‐catenin is one of the key downstream effectors in the Wnt signaling pathway (reviewed by Cadigan and Nusse, 1997). The Wnt pathway has been implicated in two major biological processes of vertebrates: early embryonic development (reviewed by Wodarz and Nusse, 1998) and tumorigenesis (reviewed by Polakis, 1999). The initial clue for the Wnt signaling involved in tumorigenesis was the discovery of Wnt1 (originally termed int1) as an oncogene that can transform mouse mammary epithelial cells (Nusse and Varmus, 1982). Subsequently, it was found that adenomatous polyposis coli (APC) tumor suppressor protein binds to β‐catenin (Rubinfeld et al., 1993; Su et al., 1993), and that the wild‐type APC but not mutant APC downregulates β‐catenin (Munemitsu et al., 1995). Recently, stabilizing mutations were found in the β‐catenin gene in melanoma and colorectal cancer cell lines (Morin et al., 1997; Rubinfeld et al., 1997), and the list of cancers with such β‐catenin mutations is expanding rapidly (de la Coste et al., 1998; Miyoshi et al., 1998; Voeller et al., 1998; Zurawel et al., 1998; Garcia‐Rostan et al., 1999; Ogawa et al., 1999). These mutations at the serine or threonine residues near the N‐terminus prevent the phosphorylation by the APC–axin/conductin–glycogen synthase kinase 3β (GSK3β) complex, and subsequent downregulation of β‐catenin (Morin et al., 1997; Rubinfeld et al., 1997; Zeng et al., 1997; Behrens et al., 1998; Hart et al., 1998; Sakanaka et al., 1998). Interestingly, such mutations are present only in colorectal cancer cells that retain the wild‐type APC (Sparks et al., 1998). When the APC gene is mutated so that degradation of β‐catenin through phosphorylation by GSK3β is suppressed (Rubinfeld et al., 1996; Yost et al., 1996; Aberle et al., 1997), unphosphorylated and stabilized β‐catenin accumulates and moves into the nucleus where it complexes with Tcf4/Lef transcription factors, which initiate transcription of a new set of genes (Behrens et al., 1996; Huber et al., 1996). While mutations in the β‐catenin or APC genes are found in human cancers, homozygous knockout mutations of either of these genes result in early embryonic lethality at the egg cylinder stages (Haegel et al., 1995; Moser et al., 1995).
To determine the role of β‐catenin in the Wnt signaling pathway regarding intestinal tumorigenesis, we constructed by homologous recombination in mouse ES cells, a novel dominant allele of the β‐catenin gene (Catnb) in which exon 3, which contained all phosphorylation target serine/threonine residues, was sandwiched by the loxP sequences. We then generated germline chimeras, and subsequently crossed their offspring with either of two types of mice that expressed Cre recombinase in the intestinal epithelium. In both types of compound heterozygous offspring, intestinal adenomatous polyps were found at young ages. These results provide direct experimental evidence for the involvement of Wnt signaling in intestinal and colonic tumorigenesis.
Construction of mutant mice expressing stable β‐catenin in the intestine
In some colon cancer cell lines, mutations have been reported that affect the serine/threonine residues of β‐catenin, i.e. S33, S37, T41 and S45, which are targets of phosphorylation by GSK3β (reviewed by Polakis, 1999). In addition, several colorectal cancer cell lines have been reported to carry interstitial deletions in the β‐catenin gene that eliminate the whole exon 3 encoding all these residues (Iwao et al., 1998; Sparks et al., 1998). Accordingly, we constructed a targeting vector, pCatnb‐lox(ex3), in which exon 3 of the mouse β‐catenin gene (Catnb) was sandwiched by two loxP sequences together with a neo selection cassette (Figure 1A, top). In this design, future deletion of exon 3 by Cre recombinase should result in mRNA whose exons 2 and 4 are connected in‐frame, producing a mutant β‐catenin protein in which 76 amino acid residues would be deleted (Figure 1A, bottom). After electroporation into ES cell line RW4, 10 homologous recombinants were identified by PCR followed by Southern hybridization from 672 G418‐resistant candidates (Figure 1B). To ensure that the loxP‐sandwiched exon 3 could be deleted by Cre, we first transfected the targeted ES cells with a cre expression vector (pPGK‐cre), isolated transfectant cell lines, and compared their genomic DNA samples with those from the targeted and parental ES cell lines using another Southern hybridization system (Figure 1C). As shown in Figure 1D, the 3.9 kb XbaI fragment of the targeted allele [Catnblox(ex3)] was converted to a 3.0 kb band (CatnbΔex3) in the transfectant clones, indicating a successful Cre‐mediated deletion of exon 3. We further determined the β‐catenin mRNA and protein levels in the transfectant ES cells. As seen in Figure 1E, the transfectants showed an RT–PCR band deleted in exon 3, in addition to the wild type. The band for the CatnbΔex3 mRNA was weaker than that for the wild type, probably because of the difference in RT–PCR efficiency. In contrast, a Western immunoblot analysis of the transfectant ES cell lysates subfractionated into the cytoplasmic and nuclear fractions showed much more mutated β‐catenin protein than the wild type, some of which was distributed in the nucleus (Figure 1F). These results are consistent with the mutational design that should stabilize β‐catenin, and gave us a reasonable expectation that exon 3 of the mutant β‐catenin allele should be deleted in the mouse tissues where the cre gene would be expressed.
We then injected the targeted ES cell clones into C57BL/6N blastocysts, obtained chimeras, and established two germline‐transmitted mouse lines [Catnb+/lox(ex3)]. On intercrossing, wild‐type, heterozygous and homozygous mutant mice were born in the Mendelian ratio (Figure 2A), and both heterozygous and homozygous offspring appeared normal and were fertile. The levels of β‐catenin in the intestine and liver analyzed by Western immunoblot analysis were essentially the same in the heterozygous and homozygous mutant mice as in the wild type (data not shown). To introduce the Cre‐mediated deletion in the targeted allele [Catnblox(ex3)] of the intestinal epithelium, we constructed two types of Cre‐expressing mice. In one type, the cre coding sequence was placed under the control of the cytokeratin 19 (Ck19) gene (Krt1‐19) by a gene knock‐in strategy in ES cells (Krt1‐19cre; Figure 2B). The Ck19 gene was chosen because it was expressed in the gut epithelium from early stages of embryonic development to adult life (Quaroni et al., 1991; data not shown) and it was dispensable (i.e. the homozygous knockout mice were viable, fertile and appeared normal; data not shown). In the other type, a DNA fragment carrying the cre coding sequence driven by the enhancer‐promoter region (−596 to +21) of the rat liver fatty acid binding protein gene (Fabpl) (Simon et al., 1993) was injected into fertilized eggs and five independent transgenic mouse lines were established (Tg·Fabplcre; Figure 2C).
The targeted mice [Catnb+/lox(ex3) or Catnblox(ex3)/lox(ex3)] were then crossed with either type of Cre‐expressing mice. The results of the crosses are summarized in Table I and a genotype analysis of three offspring from such a cross with the Ck19‐cre mouse strain (Krt1‐19+/cre) is shown in Figure 2D. Compound heterozygous mice were obtained with the Ck19‐cre knock‐in mice [Catnb+/lox(ex3): Krt1‐19+/cre] as well as with the Fabpl‐cre transgenic lines [Catnb+/lox(ex3). Tg·Fabplcre].
Intestinal polyposis in the stable β‐catenin mutant mice
Upon necropsy of the compound heterozygotes with Ck19‐cre [Catnb+/lox(ex3). Krt1‐19+/cre], numerous intestinal polyps in the range of 3000 per mouse were found in all mice at relatively young ages (i.e. by the third week after birth). One of the representative intestinal samples was photographed in Figure 2E, left. Polyps were found at highest densities in the duodenum and proximal jejunum, and at somewhat lower densities in the ileum. A smaller number of polyps were found in the terminal ileum, and essentially none in the cecum or colon, although some nascent microadenomas were found buried in the colonic mucosa (see below). In contrast to the well separated polyps that develop in the ApcΔ716 knockout mice (Oshima et al., 1995), it was difficult to count the polyp numbers precisely in the Ck19‐cre compound mutants because the polyps were often formed in contiguous sheet‐like clusters. We then isolated these polyp clusters and analyzed their DNA by PCR, comparing it with that of the adjacent normal intestinal epithelium, colonic mucosa and liver tissue. As shown in Figure 2G, the deleted β‐catenin allele (CatnbΔex3) was found only in the polyp adenomas of the duodenum, jejunum and ileum, but not in the adjacent normal mucosa. Interestingly, the deleted allele was also detected in the colon (Figure 2G) where no conspicuous polyps were found, but some nascent microadenomas were found buried in the mucosa (see below). Upon Western immunoblot analysis of the polyps as well as the normal mucosa from the compound heterozygotes, the mutant β‐catenin protein was expressed only in the intestinal polyps, but not in the normal epithelium adjacent to the polyps (Figure 2H), and the mutant bands were weaker than those for the wild type. This result is in clear contrast to the much stronger mutant band detected in the targeted ES cells transfected with the pPGK‐cre expression vector (Figure 1F), and is most likely to be due to the presence of non‐epithelial stromal cells in the tumor tissues and/or the Cre‐mediated deletion of the loxP‐sandwiched exon 3 of the β‐catenin gene did not take place in all crypt epithelial cells (see Discussion). As anticipated from the genomic DNA analysis (Figure 2G), the band for mutant β‐catenin protein was also detected in the colon (Figure 2H). On the other hand, neither the mutant β‐catenin allele nor the protein product was detected in the liver (Figure 2G and H). These compound mutants with Ck19‐cre survived longer than those with the Fabpl‐cre transgene (see below) despite their severe polyposis, and died at ∼3 months of age due to anemia and cachexia, as ApcΔ716 heterozygotes do.
In the compound heterozygous mutants carrying the Fabpl‐cre transgene [Catnb+/lox(ex3). Tg·Fabplcre], on the other hand, polyps developed in smaller numbers (∼200–700 per mouse) and were more separated from each other (Figures 2E, right, 3A and B). Accordingly, these latter polyps were investigated histologically in more detail. Many polyps were formed by the fourth week after birth (see below), but most mice died within 4–5 weeks from a still unknown cause(s) (Discussion). In ‘jelly roll’ intestinal samples, polyps were spotted easily as sessile structures (Figure 3C). Upon higher magnifications, the polyp adenomas showed moderate architectural changes of the mucosal epithelium with some budding, branching and resultant crowding patterns that were similar to those in the ApcΔ716 knockout mice (Figure 3D). It is worth noting that the polyp adenomas were always covered with the normal villus epithelium, as in ApcΔ716 polyps (M.Oshima et al., 1995; H.Oshima et al., 1997). Cytologically, rapid cell proliferation was reflected by many mitotic figures and pseudostratified large nuclei (Figure 3F). Upon Alcian blue staining for sulfomucins, no goblet cells or Paneth cells that are stained in the normal intestinal epithelium were found in the polyp adenoma, but it was composed essentially of a single cell type that appeared to be undifferentiated enterocytes (Figure 3E).
To investigate the molecular mechanism of polyp formation, we further analyzed the polyps in the compound heterozygotes [Catnb+/lox(ex3). Tg·Fabplcre] by immunohistochemistry. With an anti‐β‐catenin antibody, prominent staining of the polyp epithelium was found in the nucleus (Figure 4A and B), whereas the basolateral regions were stained in the adjoining normal intestinal epithelium, with a gradient pattern highest at the villus tip and lowest in the crypt bottom (Figure 4C). Interestingly, the extent of nuclear β‐catenin staining was heterogeneous even in a small area of a single polyp. This heterogeneity was also observed in the polyps induced by Ck19‐cre (data not shown) as well as in the ApcΔ716 polyps (Figure 4D). To rule out possible Apc mutations, we then stained the polyp sections with an antibody against the C‐terminal region of APC protein. The mutant polyp adenoma epithelium was stained in the same pattern as the normal epithelium (Figure 4E), unlike the polyp adenoma in ApcΔ716 mice which was not stained (Figure 4F). In the proliferating cell nuclear antigen (PCNA) staining for proliferative cells, the polyp epithelial cells in both types of heterozygotes were stained in a similar manner (Figure 4G). Finally, we stained the polyp tissue with an antibody against cyclooxygenase 2 (COX‐2). Interestingly, abundant COX‐2 expression was found in the stromal cells (Figure 4H) as reported earlier for ApcΔ716 polyps (Oshima et al., 1996).
Early polyp morphogenesis in the stable β‐catenin mutant mouse
To obtain insights into the polyp initiation mechanism, we investigated early histological patterns in the compound heterozygotes carrying the Fabpl‐cre transgene [Catnb+/lox(ex3). Tg·Fabplcre]. Using a combination of hematoxylin and eosin (H & E) staining and PCNA immunohistochemistry on serial sections of 2‐ and 4‐week‐old mouse intestines, we could identify some ‘nascent’ polyps at the very earliest stages of formation. All nascent polyp adenomas were formed inside the normal villus epithelium, emerging near crypts. Compared with the ApcΔ716 nascent polyps, the adenoma architecture in the β‐catenin mutants was more irregular, and heterogeneous in size from the very earliest stages (Figure 5A versus C), which suggests multiple clones initiated concomitantly inside a single polyp. As expected, nuclear staining of β‐catenin was evident from these very early stages (Figure 5D).
In addition to the nascent polyp adenomas near the crypt proliferative zone, clusters of cells were found in the adjoining villus epithelium, which showed some signs of dysplasia such as nuclear pseudostratification and positive staining with the PCNA antibody, although little architectural change was observed except slight constrictions of the villus epithelium (Figure 5A and B). Such ‘dysplastic cells’ have never been observed in ApcΔ716 polyp adenomas. However, these cells did not show nuclear staining with the β‐catenin antibody, although the strong staining in the basolateral regions seen in the normal villus epithelium was lost (Figure 5D). The pathological significance of such ‘dysplastic cells’ remains to be determined.
Although no conspicuous polyps were found in the colon of the compound heterozygotes of either type, ∼25 nascent microadenomas per mouse were found under a dissecting microscope in the colon of the mice carrying Ck19‐cre [Catnb+/lox(ex3). Krt1‐19+/cre] (Figure 5E). As shown in the histological sections in Figure 5F–H, these microadenomas were buried in the mucosa and showed similar architectural and cytological changes to those in the nascent polyp adenomas found in the small intestines.
In our strategy, exon 3 of one of the β‐catenin alleles was deleted by recombination catalyzed by Cre, which was controlled by the regulatory region of either the Ck19 or rat liver fatty acid binding protein (L‐FABP) gene. Recombination took place efficiently even though it was difficult to determine the levels of Cre protein in the targeted mouse tissues (data not shown). Although both endogenous Ck19 (Quaroni et al., 1991) and the Fabpl (−596 to +21) transgene (Sweetser et al., 1988) are expressed in the intestinal crypt epithelium, the cre gene was probably not expressed in all crypt proliferative zone cells of the knock‐in or transgenic mice. Several reasons, as detailed below, may explain our observation that only finite numbers of polyps were formed rather than a contiguous sheet of adenomas all over the intestinal mucosa. In another knock‐in mouse line (Krt1‐19+/lacZ), in which a bacterial β‐galactosidase gene was placed under the control of the same Ck19 promoter, the LacZ staining in the intestine showed numerous white patches where no lacZ was expressed (Y.Tamai and M.M.Taketo, unpublished observation). Although the molecular mechanism remains to be determined, such uneven lacZ expression was not observed in other knock‐in mice with the Apc or Cdx2 gene (Y.Tamai and M.M.Taketo, unpublished observation). Accordingly, it was possible that the uneven expression of cre in the Ck19‐cre compound heterozygotes was partly due to a heterogeneity in the Krt1‐19 enhancer‐promoter activity in vivo. On the other hand, in the Fabpl‐cre transgenic mice (Tg·Fabplcre), the copy number varied widely from one transgenic line to another (data not shown), a phenomenon often observed in transgenic mice including Tg·FabplCatnbΔN89 (Wong et al., 1998; see below). It was also reported that one of two transgenic lines with the same Fabpl enhancer‐promoter (−596 to +21) showed heritable mosaic patterns of reporter expression in the intestinal epithelium (Simon et al., 1993). It is therefore conceivable that such a mosaic pattern of expression was responsible for uneven expression of cre, which resulted in limited numbers of intestinal polyps. In fact, deletion of exon 3 in our compound heterozygotes was found only in the intestine with one transgenic line (Tg·Fabplcre5), whereas a variety of organs including brain, heart, stomach, liver and pancreas showed the PCR band for the deleted allele in the compound heterozygotes with another line (Tg·Fabplcre1; data not shown). While the former lived longer than 13 weeks, the latter died within 4–5 weeks. Accordingly, it is possible that early death of the Fabplcre1 compound heterozygotes was due to some deleterious effects of the β‐catenin mutation in organs other than the intestine.
Recently, a transgenic mouse line was reported in which N‐terminally truncated β‐catenin gene was expressed under the control of the rat Fabpl gene enhancer‐promoter (Tg·FabplCatnbΔN89; Wong et al., 1998). Although the Fabpl regulatory region used in their study was identical to that of ours (i.e. −596 to +21), they found no intestinal polyps or tumors. Instead, increased branching of the intestinal villi was observed at a significantly high frequency. The difference between their results and our present report may be explained as follows. In our design, the β‐catenin mutation [Catnblox(ex3)] was introduced into one of the endogenous alleles by homologous recombination followed by deletion of Catnb exon 3 in the intestine. Accordingly, the mutant β‐catenin gene was driven by its own enhancer‐promoter. On the other hand, Wong et al. (1998) used the Fabpl enhancer‐promoter to drive the mutant β‐catenin gene as a transgene. It is possible that the levels of expression were much lower in the intestine than those of the endogenous β‐catenin gene (see below). Another transgenic mouse line was reported recently in which an N‐terminally truncated β‐catenin gene (Tg·Krt1‐14CatnbΔN87) was expressed in the basal epidermis from the cytokeratin 14 enhancer‐promoter (Gat et al., 1998). The transgenic mice developed benign skin tumors such as trichofolliculomas and pilomatrichomas, which are often found in certain human familial polyposis syndromes. Importantly, they estimated the ΔN87 β‐catenin level to be several‐fold higher than that of endogenous β‐catenin, an observation consistent with our results.
In a morphogenetic analysis of intestinal polyps in the ApcΔ716 knockout mice, we demonstrated earlier that the initial event is formation of an outpocketing pouch by the crypt proliferative zone cells, which then extends into the inner side of adjoining villi (Oshima et al., 1997). In the present study, we have observed essentially the same type of polyp morphogenesis. In contrast to the ApcΔ716 knockout mice in which outpocket formation is initiated by loss of the remaining wild‐type Apc allele in a stochastic manner (Oshima et al., 1995), the β‐catenin mutation (CatnbΔex3) in the present experiment was introduced by intestinal expression of the Cre recombinase, which was expected to be initiated in a synchronized manner upon activation of the endogenous Ck19 or L‐FABP transgene during development. While Ck19 starts to be expressed in the gut epithelium before 10 d.p.c. (Quaroni et al., 1991), a transgene carrying the same Fabpl regulatory region as in our cre expression construct (i.e. −596 to +21) is expressed in the adult intestinal crypt and colon, but not in the embryonic gut (Sweetser et al., 1988). It was therefore anticipated that the exon 3 deletion took place before 10 d.p.c. in the Ck19 compound heterozygotes, but post‐natally in the adult of the Fabpl‐cre offspring. In both types of mutant mice, however, polyps started to form ∼2 weeks after birth. It is more likely that this is because the mutated epithelial cell population did not expand until the crypt–villus unit had been established in the second week after birth (Schmidt et al., 1988).
It is worth noting that the Cre‐recombined allele CatnbΔex3 and its protein product were also detected in the colon of the compound heterozygotes [Catnb+/lox(ex3). Krt1‐19+/cre] by PCR and Western immunoblot analysis, respectively (Figure 2G and H). Although we could not find any conspicuous polyps in the colon, we did find ∼25 nascent microadenomas per mouse buried in the colon of the compound heterozygotes that were 9 weeks old. Such nascent colonic microadenomas were not found in the mice with the Fabplcre transgene [Catnb+/lox(ex3). Tg·Fabplcre] in which deletion of exon 3 was not detected in the colon.
In addition to its role in Wnt signaling, β‐catenin plays a key role in cell adhesion as a major component of adherence junctions linking the actin cytoskeleton to members of the cadherin family of transmembrane cell–cell adhesion receptors (Kemler, 1993; Ben‐Ze'ev and Geiger, 1998). Moreover, APC and β‐catenin are both localized in the adhesion plaques and appear to be involved in cell migration (Näthke et al., 1996; Barth et al., 1997; Neufeld and White, 1997). On the other hand, mammalian APC binds to tubulin through its C‐terminal domain, while such co‐localization with tubulin was not observed with Drosophila E‐APC (Yu and Bienz, 1999). The results we present here, that a stable mutant of β‐catenin can still cause intestinal polyps even though it does not appear to affect the interaction with adherence junctions, strongly suggest that polyposis is caused by transcriptional regulation through the Wnt signaling pathway, although direct changes in the interaction with the adherence proteins may also play an additional role.
In conclusion, we constructed mutant mice that carried a dominant and stable β‐catenin mutation, which was expressed in the intestinal and colonic epithelia. These mice developed numerous intestinal adenomatous polyps and some colonic microadenomas, demonstrating that activation of the Wnt pathway in the gut can lead to tumorigenesis. These results present direct experimental evidence that Wnt signaling is involved in intestinal tumorigenesis mediated by β‐catenin.
Materials and methods
Construction of targeting vectors
β‐catenin (Catnb)‐loxP targeting construct (Figure 1A). Mouse genomic DNA containing exons 2–10 of the β‐catenin gene (Catnb, on chromosome 9) were isolated by screening a 129/Sv mouse genomic library in bacteriophage λ (Stratagene, San Diego, CA) using a 1.3 kb mouse cDNA probe (base Nos 98–1362) isolated by RT–PCR. Primers were designed based on the mouse β‐catenin sequence (DDBJ/EMBL/GenBank accession No. M90364): BCAT‐S1 (5′‐ATGGCTACTCAAGCTGACCTGATGG‐3′) and BCAT‐AS1 (5′‐TTACAGGTCAGTATCAAACCAGGCC‐3′). PCR was for 30 cycles of 94°C for 30 s, 60°C for 90 s and 72°C for 2 min. A 2.3 kb EarI–NheI fragment containing exon 3 of Catnb and PGK‐neo‐bpA cassette (Soriano, 1991) was inserted between the two loxP sequences in pBS246 (Sauer, 1993). For short and long arms of homology, 1.1 kb ApaI–BstXI and 10 kb NheI–MhoI fragments were used, respectively. G418‐resistant ES clones were screened by PCR using primers BCAT‐F1 (5′‐GGTAGTGGTCCCTGCCCTTGACAC‐3′) and P85 (5′‐CTAAGCTTGGCTGGACGTAAACTC‐3′) for 35 cycles of 94°C for 30 s, 60°C for 1 min and 72°C for 1 min. For RT–PCR of the segments shown in Figure 1E, RNA was extracted, reverse‐transcribed as above and amplified using the following primers: BCAT‐F1 and AS5 (5′‐ACGTGTGGCAAGTTCCGCGTCATCC‐3′). Homologous recombination in ES cells was verified by Southern hybridization using probe A shown in Figure 1A, as described previously (Oshima et al., 1995).
For Western blotting analyses, pPGK‐cre‐transfected ES cells were lysed in buffer A [10 mM HEPES pH 7.8, 10 mM KCl, 0.1 mM EDTA, 0.1% NP‐40]. Cytosolic and nuclear/membrane fractions were separated by centrifugation at 2000 g for 1 min at 4°C. Each fraction from 2×104 cells was loaded onto a 7.5% SDS–PAGE gel, blotted and detected by a rabbit polyclonal anti‐β‐catenin antibody (Sigma, MO) coupled with the ECL detection system (Amersham Pharmacia Biotech, Buckinghamshire, UK).
PGK‐cre plasmid. From the PGK‐neo‐bpA cassette (Soriano et al., 1991), the neo coding sequence between the PstI and NotI sites was replaced with the cre coding fragment from pBS39 (Sauer, 1987) with adaptors.
Cytokeratin 19 (Ck19)‐cre knock‐in construct (Figure 2B). A recombinant λ clone was obtained from M.Nozaki, which contained a cytokeratin 19 (Krt1‐19, on chromosome 11) genomic sequence isolated from a 129/Sv genomic DNA library with a 1.4 kb full‐length Ck19 cDNA probe (Ichinose et al., 1989). A 6.5 kb SalI–KpnI fragment containing a part of exon 1 and a 641 bp KpnI–BglII fragment downstream were subcloned for the long and short homology arms, respectively. The ATG translation initiation codon in exon 1 was converted to a KpnI site, and a 365 bp KpnI fragment in exon 1 was replaced with a cassette containing the cre coding sequence of pBS39 (Sauer, 1987) and the polyadenylation signal sequence from pPGKneobpA (Soriano et al., 1991). Thus, the targeting vector pJBX‐cre carried a long homology arm of 6.1 kb (Figure 4B) and was linearized at the unique Sse8387I site. Clones resistant to G418 (Geneticin, Sigma) were screened for recombinants by PCR using the following primers, amplifying a fragment of 739 bp: PGKra (5′‐CTAAAGCGCATGCTCCAGACT‐3′) and Ck19‐R2G (5′‐CCTGACTAGATTCAAGTTAACTG‐3′). Candidates for homologous recombinants were confirmed by Southern hybridization using two probes: a 630 bp PstI–XbaI fragment containing the PGK‐neo‐bpA sequence (not shown); and a 556 bp fragment, located downstream of the vector sequence (Figure 2B). The wild‐type Krt1‐19 allele was identified by PCR using primers Ck19‐R2G (above) and Ck19‐25 (5′‐TGGCGGAGTCCGCGGTGGAAGTT‐3′), amplifying a fragment of 958 bp. Amplification was for 35 cycles; 1 min at 94°C, 2 min at 60°C and 2 min at 72°C.
L‐FABP (Fabpl)‐cre transgenic construct (Figure 2C). The enhancer‐promoter fragment (−596 to +21) of the L‐FABP gene (Fabpl, on chromosome 6; Simon et al., 1993) was amplified from a rat genomic DNA by PCR using primers that carried XhoI and SalI sequences outside of −596 and +21, respectively, and cloned into the respective sites of the pT7BlueT vector (Novagen). In cre‐carrying plasmid pBS39 (Sauer, 1987), the translation initiation signal was altered to give optimal translation (Kozak, 1987). The Fabpl (−596 to +21) fragment and the cre sequence above were cloned into pBluescriptII SK− (Stratagene). From the resulting plasmid, the whole‐insert fragment was excised by XhoI–ScaI digestion, and injected into [C57BL/6×SJL] F2 fertilized eggs at DNX. Offspring mice were genotyped by PCR of the tail DNA samples using primers CreA (5′‐CACGTTCACCGGCATCAACG‐3′) and FABP7 (5′‐GGTATTTTTCTCCACGTAAT‐3′), amplifying a 908 bp fragment. Amplification was for 35 cycles; 30 s at 94°C, 1 min at 55°C and 1 min at 72°C. Five independent transgenic mouse lines were constructed and we analyzed two of them in detail after backcrossing with the C57BL/6N strain for more than three generations.
Generation of knockout mice
ES cell lines D3a2 (Shull et al., 1992) and RW4 were obtained from T.Doetschman and Genome Systems (St Louis, MO), respectively. They were cultured and electroporated, and homologous recombinants were isolated as described (Oshima et al., 1995). The Ck19‐cre mice thus constructed have been backcrossed with the C57BL/6N strain for more than two generations.
Histology and immunohistochemical analyses
Samples were fixed, sectioned and stained with H & E as described previously (Oshima et al., 1996). Staining with Alcian blue was performed as described (Takaku et al., 1998). A rabbit polyclonal antibody against the C‐terminus of human β‐catenin, but identical to the corresponding mouse sequence, was obtained from Sigma. Rabbit polyclonal antibodies against human PCNA that cross‐react with mouse PCNA, and against the C‐terminus of human APC, cross‐reacting with mouse APC, respectively, were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Goat antibody against the C‐terminus of rat COX‐2, cross‐reacting with mouse COX‐2, was also purchased from Santa Cruz Biotechnology.
Note added in proof
We have found fatty degeneration of the liver in the compound mutants carrying the Fabpl‐cre1 transgene [Catnb+/lox(ex3). Tg·Fabplcre], which appears to be the cause of their death within 4–5 weeks after birth.
We thank T.Doetschman for providing the ES cell line, P.Soriano and M.Nozaki for the plasmid constructs, and N.Sugimoto, R.Nakajima and H.Oshima for excellent technical assistance. This work was supported in part by the Joint Research Fund between the University of Tokyo and Banyu Pharmaceutical Co., and grants from Monbusho (MESSC), Japan.
- Copyright © 1999 European Molecular Biology Organization