A new conditional Apc-mutant mouse model for colorectal cancer

Carcinogenesis vol.31 no.5 pp.946–952, 2010 doi:10.1093/carcin/bgq046 Advance Access publication February 22, 2010

A new conditional Apc-mutant mouse model for colorectal cancer Els C.Robanus-Maandag1, Pim J.Koelink2, Cor Breukel1, Daniela C.F.Salvatori3,4, Shantie C.Jagmohan-Changur1, Cathy A.J.Bosch1, Hein W.Verspaget2, Peter Devilee1, Riccardo Fodde1,5 and Ron Smits1,6, 1 Department of Human Genetics, 2Department of GastroenterologyHepatology, 3Department of Anatomy and Embryology and 4Central Animal Facility, Leiden University Medical Center, 2300 RC Leiden, The Netherlands, 5Department of Pathology, Josephine Nefkens Institute and 6 Department of Gastroenterology and Hepatology, Erasmus MC University Medical Center, ‘s Gravendijkwal 230, 3015 CE Rotterdam, The Netherlands To whom correspondence should be addressed. Tel: þ31 10 703 5944; Fax: þ31 10 703 2793; Email: m.j.m.smits@erasmusmc.nl

Mutations of the adenomatous polyposis coli (APC) gene predispose individuals to familial adenomatous polyposis (FAP), characterized by multiple tumours in the large intestine. Most mouse models heterozygous for truncating mutant Apc alleles mimic FAP, however, the intestinal tumours occur mainly in the small intestine. To model large intestinal tumours, we generated a new conditional Apc-mutant allele, Apc15lox, with exon 15 flanked by loxP sites. Similar survival of Apc1638N/15lox and Apc1638N/1 mice indicated that the normal function of Apc was not impaired by the loxP sites. Deletion of exon 15, encoding nearly all functional Apc domains and containing the polyadenylation signal, resulted in a mutant allele expressing low levels of a 74 kDa truncated Apc protein. Germ line Cre-mediated deletion of exon 15 resulted in ApcD15/1 mice, showing a severe ApcMin/1-like phenotype characterized by multiple tumours in the small intestine and early lethality. In contrast, conditional Cre-mediated deletion of exon 15 specifically directed to the epithelia of distal small and large intestine of FabplCre;Apc15lox/1 mice led to longer survival and to tumours that developed predominantly in the large intestine, mimicking human FAP-associated colorectal cancer and sporadic colorectal cancer. We conclude that the FabplCre;Apc15lox/1 mouse should be an attractive model for studies on prevention and treatment of colorectal cancer.

Introduction The adenomatous polyposis coli (APC) tumour suppressor gene encodes the adenomatous polyposis coli protein. Individuals with germ line inactivating mutations of this gene are predisposed to develop hundreds to thousands of adenomatous polyps in the colon and rectum, the hallmark of the hereditary cancer syndrome familial adenomatous polyposis (FAP) (1). Moreover, the majority of sporadic colorectal tumours is initiated by mutations in APC (2,3). The APC protein has a broad spectrum of functions, ranging from control of the WNT signal transduction pathway, to functions in cell adhesion, migration, apoptosis and chromosomal segregation at mitosis (4). Each of these roles is potentially linked with cancer development. However, the main tumour-suppressing function of APC resides in its role in the Wnt signal transduction pathway. This pathway induces the nuclear translocation of b-catenin and has a key role in cell-fate determination (2). b-Catenin is found in a membrane-bound pool where it provides the essential link between E-cadherin and a-catenin in the adherens junctions (thus enhancing cell adhesion) and a more soluble pool that shuttles between cytoplasm and nucleus (thus transducing the canonical Wnt signal) (1,5). Cytoplasmic APC binds to and downregulates Abbreviations: aa, amino acid; APC, adenomatous polyposis coli; ES, embryonic stem; FAP, familial adenomatous polyposis; IP, immunoprecipitation; PBS, phosphate-buffered saline; PCR, polymerase chain reaction.

b-catenin, thereby preventing its nuclear activity. Tight regulation of Wnt/b-catenin signalling is essential, as uncontrolled nuclear accumulation of b-catenin can cause developmental defects and tumourigenesis in the adult organism (3). In humans, most APC mutations causing colorectal tumours are inactivating mutations resulting in loss of functional APC protein and nuclear accumulation of b-catenin. To study the role of APC in development and tumourigenesis, many inactivating mutant alleles of the mouse Apc gene have been generated. Several of these mutant alleles demonstrated gastrointestinal tumourigenesis in the heterozygous state, thereby mimicking to a certain extent the tumour phenotype of FAP patients (6–15). However, mice with these constitutional Apc mutations developed mainly tumours in the small intestine. They succumbed to these tumours at a rather young age, thereby preventing studies on the long-term effect of loss of the wild-type Apc function in other tissues. Moreover, nearly all homozygous Apc-mutant mouse models were embryonic lethal. To direct the development of intestinal tumours more specifically to the large intestine and to circumvent embryonic lethality associated with homozygosity, conditional Apc-mutant alleles have been created using the Cre/loxP conditional gene targeting system (16). Indeed, mice homozygous for the Apc580S allele, with loxP sites flanking exon 14, developed colorectal adenomas upon local infection with a Cre recombinase-encoding adenovirus (17). However, Apc580S/580S mice showed strongly reduced Apc expression (30% of wild-type RNA expression) in the intestinal tract, indicating that the loxP sites and/or the positive selection cassette had an adverse effect on the expression level of the Apc gene. Inactivation of the Apc580S allele by Cre expression preferentially in the large intestinal epithelium resulted in a shift of the distribution of the tumours to this part of the intestine. However, the total number of intestinal tumours was low (18). In a second conditional Apc-mutant allele, Apc2lox14, again with exon 14 flanked by loxP sites, the selection cassette had been removed by Cre-mediated recombination (12). Early studies using mice homozygous for these conditional Apc-mutant alleles and with an inducible Cre transgene showed that conditional deletion of Apc within the intestinal epithelium leads to a ‘crypt progenitor cell-like’ phenotype, with intestinal enterocytes showing proliferation, perturbed differentiation and migration (19,20). We report here the construction and characterization of a new conditional Apc-mutant allele, Apc15lox, with loxP sites flanking exon 15 and its germ line and distal intestine-specific knockout by Cre-mediated deletion of this exon. Materials and methods Targeting The targeting construct was generated in two steps (Figure 1a). First, a 10.3 kb SphI–SacI fragment, containing the 3#part of exon 15 and 3 kb of the 3# flanking region, was derived from a 129/J mouse genomic cosmid library and subcloned into the pBluescript vector. Next, a loxP-flanked PGKtk/PGKneo cassette was introduced into the ApaI site 350 bp downstream of the Apc polyadenylation signal. Both selectable markers were subsequently removed by introducing this vector into a bacterial strain, expressing Cre recombinase (kindly provided by F.Stewart, University of Technology, Dresden, Germany), leaving behind a single loxP site. For the second step, an 11.6 kb EcoRI fragment containing exons 11–14 and the 5# region of exon 15 was cloned into the pGEM7 vector. A loxP-flanked PGKneo cassette was cloned into the BglII site of intron 14. Subsequently, both vectors were combined to generate a 22.5 kb targeting vector which was linearized at the KpnI site before electroporation. Electroporation and G418 selection of IB10 embryonic stem (ES) cells was carried out as described before (21). Southern blot analysis of BamH1-digested DNA from 350 selected ES cell clones with a 3# external probe (1.1 kb SphI– ApaI genomic Apc fragment) showed in 20 clones a 5.7 kb band, indicative of a correctly recombined fragment (Figure 1a and b). Sixteen of these 20 ES cell clones showed correct recombination at the 5# end of the targeting construct (12.0 kb band) as determined by Southern blot analysis of KpnI/SalI-digested

Ó The Author 2010. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

946

New conditional Apc colorectal mouse model

Fig. 1. Generation and characterization of the conditional mutant Apc15lox allele and the truncated mutant ApcD15 protein upon Cre recombination. (a) Schematic drawing of exons 10–15 of the mouse Apcþ allele, the targeting construct, the resulting targeted Apcneo-15lox allele before and Apc15lox allele after transient in vitro Cre expression in ES cells and the ApcD15 allele after tissue-specific Cre expression in the mouse. In the targeting construct, a loxP site (black triangle) was introduced in the ApaI site 350 bp downstream of the polyadenylation signal and a loxP-flanked neomycin selection cassette into the BglII site in intron 14. The neomycin selectable marker was deleted in vitro by transient co-transfection of a Cre-encoding plasmid and hygromycin selectable marker. Some of the restriction sites relevant for cloning or Southern blot analysis are depicted. Positions of external and internal probes used for Southern blot analysis are also indicated. (b) Southern blot analysis of BamHI and KpnI/SalI-digested DNA isolated from a wild-type (lane 1) and properly targeted (lane 2) ES cell clone using a 3# and 5# external probe, respectively. Molecular sizes of wild-type and mutant fragments are depicted for BamHI (42.7 and 5.7 kb, respectively) and KpnI/SalIdigested (21.8 and 12.0 kb, respectively) DNA. (c) Schematic representation of the full-length Apc protein (2842 aa) and the truncated ApcD15 polypeptide (650 aa) lacking all domains involved in regulation of Wnt/b-catenin signaling as well as 2.5 of 7 N-terminal Armadillo domains and the C-terminal domains that associate with the microtubular cytoskeleton. (d) Western blot analysis of total lysates and IPs of ES cells using the rabbit polyclonal antibody AFPN. Western blots were developed with the mouse monoclonal antibody Ab-1. (Lane 1) Apcþ/þ, (lane 2) ApcMin/þ, (lane 3) ApcD15/þ and (lane 4) Apc15lox/þ ES cells. Upon Cre-mediated deletion of loxP-flanked exon 15, the resulting ApcD15 allele encoded a truncated Apc protein in diminished amounts. (e) Kaplan–Meier survival plot of Apc15lox/þ, Apc15lox/15lox, Apc1638N/þ and Apc15lox/1638N mice. Homozygosity for the Apc15lox allele and compound heterozygosity for the Apc15lox and Apc1638N alleles showed no significant different survival compared with heterozygosity for the Apc15lox (P 5 0.454) and Apc1638N (P 5 0.408) allele, respectively.

947

E.C.Robanus-Maandag et al.

DNA with a 5# external probe (0.5 kb SacI–EcoRI genomic Apc fragment) (Figure 1a and b). PGKneo was deleted in vitro by transient Cre expression after transfection. A single ES cell clone with a correct deletion of PGKneo and with two loxP sites flanking exon 15 of Apc was selected by a combination of Southern blot and polymerase chain reaction (PCR) strategies. ApcD15/þ ES cells, lacking the loxP-flanked exon 15, were generated by transient Cre expression in Apc15lox/þ ES cells, as described above. ApcMin/þ ES cells have previously been described (22). Immunoprecipitation and immunoblotting Immunoprecipitation (IP) and western blot analysis of Apc were essentially performed as described (21). For detection of Apc protein by IP, 2 107 ES cells were rinsed twice in phosphate-buffered saline (PBS) and lysed for 15 min on ice in 800 ll of Triton X-100 IP buffer (30 mM Tris–HCl pH 7.4, 250 mM NaCl, 0.1% Triton X-100, 5 mM ethylenediaminetetraacetic acid, 50 mM NaF, 0.1 mM sodium vanadate and a cocktail of protease inhibitors (Complete Protease Inhibitor Cocktail tablets, Roche Diagnostics GmbH, Mannheim, Germany). To 400 ll of cleared lysate, obtained after centrifugation for 10 min at 11 000g, a preformed complex of AFPN rabbit polyclonal antibody (21) and protein A–Sepharose beads (Pharmacia, Roosendaal, The Netherlands) was added. After incubation at 4°C for 1 h under continuous mixing, the beads were washed 3–4 times with 1 ml of IP buffer. After the final wash, the pellet was resuspended in 50 ll of Laemmli sample buffer (120 mM Tris–HCl pH 6.8, 20% glycerol, 200 mM dithiothreitol, 4% sodium dodecyl sulphate and 0.02% bromophenol blue) and heated at 70°C for 10 min. Total lysates were obtained by washing the culture plates twice with PBS, followed by direct lysis on the plates using Laemmli sample buffer and heating at 70°C for 10 min. For detection of Apc protein by western blot analysis, lysates were resolved on 8% sodium dodecyl sulphate–polyacrylamide gels and transferred onto polyvinylidene difluoride membranes (Immobilon-P, Millipore Corporation, Billerica, MA) by overnight electroblotting at 80 mA constant current at 4°C in 50 mM Tris–HCl pH 8.3, 384 mM glycine and 0.01% sodium dodecyl sulphate. Following transfer, the membranes were blocked with 5% nonfat dry milk in TBST (10 mM Tris–HCl pH 8.0, 150 mM NaCl and 0.05% Tween 20) for at least 1 h. After incubation for 1 h with primary Apc mouse monoclonal antibody Ab-1 (Calbiochem, Merck, Darmstadt, Germany) at a dilution of 1:100, the blot was washed 2–3 times with TBST and subsequently incubated with horseradish peroxidase-conjugated secondary antibodies for 1 h. The peroxidase was visualized by enhanced chemiluminescence. Mice All experiments on mice were performed in accordance with institutional and national guidelines and regulations. Mice were housed under conventional conditions. Germ line chimeras were generated by injection of 12–15 Apc15lox/þ ES cells into C57Bl/6 blastocysts and crossed with C57Bl/6 mice to produce outbred heterozygous offspring. Apc15lox/15lox mice were bred with homozygous EIIaCre mice (kind gift of H.Westphal, National Institutes of Health, Bethesda, MD) to generate EIIaCre;Apc15lox/þ offspring. These double heterozygous mice were backcrossed to C57Bl/6 mice to generate ApcD15/þ progeny, which were maintained in a C57Bl/6 genetic background. Heterozygous FabplCre mice [Tg-Fabpl4x at –132Cre, kind gift of J.Gordon (Washington University School of Medicine, St Louis, MO) via National Cancer Institute-Frederick (23)] were bred with Apc15lox/15lox mice to generate FabplCre;Apc15lox/þ progeny. Genotyping The genotypes of all offspring were analysed by PCR on tail-tip DNA. The Apcþ and Apc15lox alleles were detected in a single reaction with a combination of two forward primers, Apc-Lox3FW2 (5#-TAGGCACTGGACATAAGGGC-3#) and Apc-LoxNotBW3 (5#-CTTCGAGGGACCTAATAAC-3#), and one reverse primer, Apc-Lox3R2 (5#-GTAACTGTCAAGAATCAATGG-3#), yielding a 87 bp wild-type product (Apc-Lox3FW2/Apc-Lox3R2) and a 243 bp loxP product (Apc-LoxNotBW3/Apc-Lox3R2), respectively, plus under optimal conditions an extra 352 bp loxP product (Apc-Lox3FW2/Apc-Lox3R2). The ApcD15 allele was detected with primers Apc-plseq5#B (5#-GCAGGAATTCGATATCAAGC-3#) and Apc-Lox3R2 yielding a 259 bp product. The Apcþ and Apc1638N alleles were detected in a single reaction with one forward primer, Apc-A3 (5#-CTAGCCCAGACTGCTTCAAAAT-3#), and two reverse primers, Apc-C2 (5#-GGAAAAGTTTATAGGTGTCCCTTCT-3#) and PN3 (5#-GCCAGCTCATTCCTCCACTC-3#), yielding a 178 bp and 347 bp product, respectively. The EIIaCre and FabplCre transgenes were assayed with the forward primer Cre-F (5#-ATGTCCAATTTACTGACCGTACAC-3#) and reverse primer Cre-R (5#-GCCGCATAACCAGTGAAACA-3#) yielding a 300 bp product. Allelic imbalance analysis of Apc Tumour DNA isolation and allelic imbalance analysis were performed as described before (24). In a single PCR reaction, both the Apcþ and ApcD15 allele

948

were amplified using four allele-specific primers as follows: Apc-Lox3FW2 and Apc-Lox3R2, yielding a product of 87 bp product diagnostic of the Apcþ allele and Apc-plseq5’lox (5#-CGATAAGCTTCGAGGGACCT-3#) and Apcplseq3’lox (5#-AAATAGGCGTATCACGAGGC-3#), yielding a product of 119 bp diagnostic of the ApcD15 allele. Extent of Cre-mediated exon 15 deletion in the intestine PCR analysis of exon 15 deletion was performed using primers Apc-plseq5#B and Apc-Lox3R2 as described for the genotyping. Southern blot analysis of exon 15 deletion was performed according to standard procedures. Six micrograms of DNA were digested by SacI and hybridized with the 2.3 kb EcoRV–SacI genomic Apc fragment probe. In case of Cre-mediated recombination, a band of 6.0 kb was detected corresponding to the ApcD15 allele, besides the bands of 3.0 and 3.2 kb corresponding to the Apcþ and Apc15lox alleles, respectively. Image quant analysis was used to quantify the relative intensity of the ApcD15 band of each DNA sample. To verify the efficacy of Cre-mediated recombination of the FabplCre mice in situ, these mice were bred with ROSA26-LacZ reporter mice (25) to generate bi-transgenic mice. The intestinal tract of the bi-transgenic mouse was opened along the cephalocaudal axis, flushed with PBS and spread out flat on filter paper. These whole mount preparations were fixed in fixative (1% formaldehyde, 0.2% glutaraldehyde, 2 mM MgCl2 and 5 mM ethyleneglycol-bis (aminoethylether)-tetraacetic acid in PBS) at room temperature for 45 min, washed three times in PBS, equilibrated in wash buffer (0.1 M sodium phosphate buffer pH 7.8, 2 mM MgCl2, 0.01% deoxycholate and 0.02% NP40). Subsequently, the Escherichia coli b-galactosidase activity was detected by staining of the intestinal tissue in wash buffer containing 1 mg/ml 5-bromo4-chloro-3-indolyl b-D-galactoside (X-Gal), 5 mM K3Fe(CN)6 and 5 mM K4Fe(CN)6, until blue staining was clearly visible. Histological and immunohistochemical analysis Tissues were fixed in NotoxÒ (Earth Safe Industries, Bellemead, NJ) or 4% neutral buffered formalin, embedded in paraffin, sectioned at 5 lm and stained with haematoxylin and eosin, according to standard procedures. Immunohistochemical analysis for b-catenin and Ki-67 was performed as described in Koelink et al. (26). Immunohistochemical analysis for c-Myc was performed as described in Sansom et al. (27). Histopathological analysis of the tumours was performed following the recommendations reported in the consensus report on pathology of mouse models of intestinal cancer (28).

Results Conditional Apc15lox/15lox mice are normal A conditional mutant allele of Apc, referred to as Apc15lox, was generated by gene targeting in ES cells using the Cre/loxP system (Figure 1a and b). Cre-mediated recombination leads to deletion of the loxP-flanked exon 15, resulting in the ApcD15 allele (Figure 1a). This 6.6 kb deleted exon 15 represents the last exon of Apc and encodes all domains involved in regulation of Wnt/b-catenin signalling as well as the last 2.5 of 7 N-terminal Armadillo domains and the C-terminal domains that associate with the microtubular cytoskeleton (Figure 1c). Since the polyadenylation site is also included in exon 15, the recombination was predicted to lead to a hypomorphic mutant allele, expressing an unstable messenger RNA and a truncated Apc protein in diminished amounts (29). To analyse the expression of the mutant Apc protein, we performed western blot analysis of ApcD15/þ ES cells, carrying a heterozygous deletion of exon 15. As it can be seen in Figure 1d, we barely detected the predicted 74 kDa ApcD15 protein in total protein lysates, whereas the similar-sized 95 kDa ApcMin protein was easily detectable. Enrichment of the Apc protein by IP made the truncated protein more discernable. Based on quantification relative to the ApcMin protein, we estimated that expression of the ApcD15 protein was 5% of normal Apcþ protein. After chimera formation and germ line transmission, Apc15lox/þ mice were intercrossed to generate Apc15lox/15lox mice. Twenty-five of the 100 offspring had the Apc15lox/15lox genotype, which is according to the expected Mendelian ratio. These Apc15lox/15lox mice were viable, healthy and fertile and developed no intestinal tumours over a 2 year period, suggesting that the introduction of loxP sites in the genomic regions flanking exon 15 did not hamper normal Apc expression. Kaplan–Meier analysis revealed no significant difference between the survival of Apc15lox/15lox and Apc15lox/þ mice (P 5 0.454, log-rank test) (Figure 1e). A second control on the non-hypomorphic


nature of the Apc15lox allele was performed by interbreeding the Apc15lox/15lox mice with Apc1638N/þ mice, which show an intermediate intestinal phenotype (five to six intestinal tumours at 1 year of age) and lifespan (8–16 months) (8). Apc15lox/1638N offspring were born at the expected Mendelian ratio and succumbed at a comparable age to intestinal tumours as their control Apc1638N/þ littermates (mean age of 13.1 and 11.1 months, respectively). Again, Kaplan–Meier analysis revealed no significant difference in survival between Apc15lox/1638N and Apc1638N/þ mice (P 5 0.408, log-rank test) (Figure 1e). We therefore concluded that the insertion of the loxP sites into the Apc introns did not impair the function of Apc in these mice. Germ line mutant ApcD15/þ mice develop small intestinal tumours To determine the phenotypic consequences of Cre-mediated deletion of Apc exon 15, we generated germ line heterozygous knockout mice carrying the ApcD15 allele in all cells. To this aim, Apc15lox/15lox mice were bred with ‘deleter’ EIIaCre mice with mosaic Cre expression in germ cells (30). EIIaCre;Apc15lox/þ progeny, with germ cells mosaic for the EIIaCre;ApcD15/þ genotype, were backcrossed to C57Bl/6 mice to generate progeny heterozygous for the ApcD15 allele and lacking the EIIaCre transgene. The age at death of these ApcD15/þ mice (n 5 33) inversely correlated with the backcross generation: on average 4.5 (N2) to 2.5 (N6) months of age (Figure 2a). The main reason of death was intestinal obstruction. Macroscopic analysis of the ApcD15/þ mice (n 5 7) revealed an average of 185 intestinal tumours (Table I), 1–2 cutaneous cysts, infrequently desmoid tumours and severe anaemia. Nearly all intestinal tumours were found in the small intestine (95.5%), and the size was predominately small (,2 mm). Within the small intestine, the majority of tumours were located in the ileum (77.1%), whereas the remaining tumours were seen in the jejunum (13.0%), duodenum (7.2%) and periampullar region (2.7%). A high number of polyp adenomas was seen (Figure 2b), as well as a few adenocarcinomas. Macroscopically, most adenomas had either a polypoid shape or flat appearance. Microscopically, tumours consisted of columnar epithelial cells with elongated, hyperchromatic, crowded nuclei, with loss of the normal maturation gradient of the crypt. DNAs of intestinal tumours of ApcD15/þ mice were isolated and screened for allelic imbalance of the Apcþ allele. We selectively amplified in a single PCR both the Apcþ and ApcD15 allele. Allelic imbalance analysis of eight tumour and two control samples is shown in Figure 2c. A significant loss of the Apcþ allele (normalized ApcD15:Apcþ ratio 1.5) could be detected in seven of eight (88%) tumours. Thus, ApcD15/þ mice, generated by Cre-mediated germ line conversion of the normal Apc15lox allele into an ApcD15 allele, showed a severe ApcMin-like phenotype, characterized by a high multiplicity of tumours in the small intestine and early lethality. Conditional FabplCre;Apc15lox/þ mice develop large intestinal tumours To direct Cre-mediated deletion of exon 15 more specifically to the large intestinal epithelium, we bred Apc15lox/15lox mice with FabplCre transgenic mice (line Fabpl4x at –132Cre) (23). The transgene in these mice expresses the Cre recombinase driven by the enhancer–promoter region of the rat liver fatty acid-binding protein gene Fabpl, with greatest expression occurring in the distal small intestinal, caecal and large intestinal epithelia (31). First, we investigated the efficiency of Cre-mediated deletion in total intestinal DNA of FabplCre;Apc15lox/þ mice at 4 weeks of age (Figure 3a and b). Based on a comparison of the relative intensity of the ApcD15 allele versus that of the Apc15lox allele, we estimated that the deletion occurred in 30, 52 and 20% of the cells in distal ileum, colon and rectum, respectively, which is in accordance with previously published data (31). Although these data most probably represented an underestimation of the true frequency due to dilution of the intestinal epithelial DNA with that of stromal cells lacking the exon 15 deletion, we concluded that exon 15 was deleted in a significant pro-

Fig. 2. Characterization of ApcD15/þ mice and related intestinal tumours. (a) Kaplan–Meier survival plot of Apc15lox/þ, ApcD15/þ and FabplCre;Apc15lox/þ mice. Both ApcD15/þ and FabplCre;Apc15lox/þ mice showed a significantly different survival compared with Apc15lox/þ mice (P , 0.0001). (b) Histological section, stained with haematoxylin–eosin, of a small high-grade adenoma in the distal small intestine of a 2.5-month-old female (magnification 100). (c) Allelic imbalance analysis for Apc of eight intestinal tumours (T1–T4) of two ApcD15/þ mice. In control ApcD15/þ DNA (N), the 87 bp sequence of the Apcþ allele amplifies in the single PCR reaction more efficiently than the 119 bp sequence of the ApcD15 allele, this allelic ratio was set to 1.0. The depicted comparative ratios ApcD15:Apcþ of seven of eight intestinal tumour DNAs (related to the allelic ratios in the control DNAs) are indicative of allelic imbalance (comparative ratios 1.5). The remaining tumour showed a tendency of allelic imbalance (comparative ratio 1.4).

portion of the Cre-expressing intestinal tract. At this early age, we observed no intestinal tumour development microscopically (data not shown). Cre recombination in these regions was heterogeneous, as determined by b-galactosidase staining of intestinal tissues of young adult FabplCre;ROSA26-LacZ reporter bi-transgenic mice (Figure 3c). All FabplCre;Apc15lox/þ mice developed intestinal tumours. Compared with ApcD15/þ mice, they succumbed later to these tumours, i.e. at a mean age of 7.6 versus 3.4 months (Figure 2a). Moreover, the total number of tumours was lower in the conditional knockout mice: on average 41 intestinal tumours compared with 185 in the ApcD15/þ mice (Table I). The intestinal tumours developed predominantly in

949


Table I. Numbers and sizes of intestinal tumours of ApcD15/þ and FabplCre;Apc15lox/þ mice Genotype

Number

Age at death (months)

Mean tumour number per mouse Total

Small intestine Size (mm)

D15/þ

Apc FabplCre;Apc15lox/þ

7 14

3.4 ± 0.8 7.6 ± 2.4

184.7 40.6

Large intestine Size (mm)

,2

2–5

.5

142.3 10.6

42.3 26.8

0.1 3.2

176.4 (95.5%) 14.9 (36.9%)

Size (mm)

,2

2–5

.5

138.4 2.4

38.0 10.2

0 2.3

8.3 (4.5%) 25.6 (63.1%)

,2

2–5

.5

3.9 8.1

4.3 16.6

0.1 0.9

adenomas (Figure 4a–d). The remaining 18% of tumours were welldifferentiated adenocarcinomas (Figure 4e–f). Ninety-one per cent of mice developed low-grade adenomas, 50% high-grade adenomas and 50% adenocarcinomas. Hyperplasia was very uncommon. The wellcircumscribed adenomas were mainly composed of tubular structures lined by dysplastic epithelium. The low-grade adenomas showed branching or elongation of crypts with some reduction of interglandular stroma. The high-grade adenomas were characterized by marked reduction of interglandular stroma and complex irregularity of glands, with cribriform (sieve-like) structures and back-to-back glands. Cytologic atypia was more pronounced. Numerous mitoses with abnormal mitotic figures were present. Immunohistochemistry for b-catenin showed widespread expression throughout the tumour epithelium (Figure 4b and f). Since c-Myc is required for the majority of Wnt target gene activation following Apc loss (27), we examined the expression of c-Myc in the tumours. Immunostaining clearly showed increased c-Myc expression in the tumour cells (Figure 4c). Consistent with the driving role of c-Myc in proliferation, we observed increased Ki-67 immunostaining in these cells (Figure 4d). Discussion

Fig. 3. Fabpl4x at –132Cre transgenic mice express Cre in the distal small and large intestine. (a) Qualitative detection of Cre-mediated deletion of exon 15 by PCR analysis (259 bp amplified product) in the duodenum (duo), jejunum (jej) and ileum (ile) of the small intestine and in the colonic (col) and rectal (rec) region of the large intestine of FabplCre;Apc15lox/þ mice. (b) Quantitative detection of Cre-mediated deletion of exon 15 by Southern blot analysis of SacI-digested DNA of the 5 regions of the intestine of FabplCre;Apc15lox/þ mice. The sizes of the fragments detected by the 2.3 kb EcoRV–SacI probe are indicated. The extent of deletion is depicted, based on comparison of the relative intensity of the ApcD15/þ allele versus that of the Apc15lox allele. (c) In situ detection of Cre-mediated deletion in the intestine of FabplCre;ROSA26-LacZ mice. Whole mount intestines, prepared from bitransgenic mice, were stained with X-Gal to visualize LacZ as blue.

the large intestine (63%), and to a larger size (2 mm) compared with those of ApcD15/þ mice (,2 mm) (Table I, Figure 4e). Macroscopically, the growth pattern was pedunculated (polypoid) or sessile. Microscopically, the tumours were mainly adenomas (82%), 65% of which were low-grade adenomas and 17% were high-grade

950

APC mutations play a prominent role in the pathogenesis of colorectal adenomas and carcinomas in man. Germ line mutations of APC lead to colorectal polyposis, the hallmark of the hereditary syndrome FAP (1), and somatic mutations of APC are associated with the majority of sporadic colorectal tumours (2,3). To study the role of Apc in development and tumourigenesis, different mouse models carrying targeted germ line Apc mutations have been generated. However, the tumours in these mice mainly develop in the small intestine, instead of the large intestine. To model human colorectal tumourigenesis more faithfully, new Apc-mutant mice with tumours developing preferentially in the large intestine are urgently needed. To create such a mouse model, we generated a new conditional Apc15lox-mutant allele, with exon 15 flanked by loxP sites. Cre-induced deletion of exon 15 in Apc15lox/þ ES cells resulted in synthesis of a mutant protein of 74 kDa (Figure 1d). Deletion of exon 15 would prevent normal splicing at the donor splice site of exon 14, resulting in retention of intron 14 in the transcript and generation of an in-frame stop codon. This would result in the synthesis of a truncated protein of 667 amino acids (aa), including 650 aa of the normal Apc protein, which is in accordance with the observed molecular weight of the truncated protein. The strongly reduced expression of the ApcD15 protein (5% of normal Apcþ protein) implicates that the ApcD15 allele is a hypomorphic mutant allele, comparable with the Apc1638N allele that expresses the mutant protein at only 2% of the normal level (8,22). The Apc15lox/þ and Apc15lox/15lox mice were viable and tumour free, whereas compound heterozygous Apc15lox/1638N mice were viable and survived like Apc1638N/þ mice (Figure 1e). These data strongly suggest that the introduced loxP sites did not impair the normal function of Apc. Germ line Cre-mediated deletion of exon 15 resulted in ApcD15/þ mice with an intestinal tumour burden (mean number of 185 tumours per mouse) and distribution (95% small intestine and 5% large intestine) comparable with that of most other constitutional


Fig. 4. Characterization of the intestinal tumours of FabplCre;Apc15lox/þ mice. (a) Macroscopic view of tumours in the large intestine of a 3-month-old FabplCre;Apc15lox/þ mouse. (b–d) Immunostaining for b-catenin (b), c-Myc (c) and Ki-67 (d) of a low-grade adenoma in the large intestine of the 3-month-old FabplCre;Apc15lox/þ mouse (a). Note the increased levels of b-catenin and c-Myc proteins and the increased level of proliferation based on increased Ki-67 immunostaining. (e) Macroscopic view of tumours in the gastrointestinal tract of a 7.5-month-old FabplCre;Apc15lox/þ mouse. (f) Immunostaining for b-catenin of an adenocarcinoma in the large intestine of the 7.5-month-old FabplCre;Apc15lox/þ mouse (e). Magnification (b–d, f) 200; (b–d squared regions) 400.

mutant mouse models (Table I) (6,7,9–15). A notable exception is the Apc1638N/þ mouse, which develops only five to six tumours mainly in small intestine (8). The ApcD15 protein lacks all Wnt pathwaydownregulating domains, whereas the Apc1638N protein comprises several of the b-catenin-binding and -downregulating domains and one axin-binding domain. Although both are synthesized at a low level, apparently the presence of these domains in the Apc1638N protein suffices to strongly reduce the tumour burden in Apc1638N/þ mice compared with ApcD15/þ mice. We found loss of the wild-type Apc allele in most of the intestinal tumours of ApcD15/þ mice, which is in accordance with Knudson’s two-hit model (Figure 2c). Compared with ApcD15/þ mice generated by germ line Cremediated deletion of exon 15, conditional deletion specifically directed to the distal intestinal epithelium of FabplCre;Apc15lox/þ mice resulted in a significantly longer survival and in the majority of tumours developing in the large intestine (Table I). In addition, FabplCre;Apc15lox/þ mice developed more tumours of larger size than ApcD15/þ mice and also adenocarcinomas at considerable frequency (18%), thereby mimicking human FAP-associated as well as sporadic colorectal tumour formation. FabplCre;Apc15lox/þ mice developed on average 4-fold more intestinal tumours, compared with a different conditional Apc-mutant mouse model, CDX2P-NLSCre;Apc580S/þ, carrying a CDX2PNLSCre transgene and the impaired exon 14-floxed Apc580S allele (18). Expression of the CDX2P-NLSCre transgene in the intestine was shown to be restricted to distal ileum, caecum and colorectum during late gestation and in adult tissues. Apart from the substantial difference in total tumour burden, the preferential colorectal accumulation of tumours, 63% in FabplCre;Apc15lox/þ mice and 71% in CDX2P-NLSCre;Apc580S/þ mice, and the frequency of adenocarcinomas (18% and 15–20%) in both models was remarkably similar. In addition, the median survival of each mouse model was comparable (7.6 months and 200 days). Loss of exon 14 would result in a shift of the reading frame and synthesis of a truncated protein of 605 aa, of which the first 580 aa correspond to the normal Apc protein (17). A mutant Apc product with the predicted size of 70 kDa has been found in brain lysates of Apc580D/þ mice carrying a constitutional

deletion of exon 14 of Apc and indirect evidence suggests that this protein is synthesised in low amounts (14,17). If so, in both mouse models a low level of truncated Apc protein of comparable length is expressed (650 aa in FabplCre;Apc15lox/þ mice and 580 aa in CDX2PNLSCre;Apc580S/þ mice). The 4-fold difference in tumour burden between the two models remains to be explained. Among others, it might be caused by differences in the promoters driving the Cre transgene expression (promoters of the rat liver fatty acid binding protein (Fabpl) gene and the human CDX2 homeobox gene) or by differences in exogenous factors, such as unidentified environmental parameters in which the mice were raised. Both promoters are known to differ in their extraintestinal activity. Besides the specific Cre expression of the Fapbl enhancer–promoter in the epithelial cells, including the stem cells, of the distal small intestine, caecum and large intestine, only one extraintestinal site of Cre expression has been detected: all layers of the urothelium, from the renal calyces to the ureters and bladder (31). In contrast, although the CDX2 promoter and upstream flanking elements show restricted Cre expression in the distal small intestine, caecum and colon during late gestation and in adult tissues, they lead to broad Cre expression and function in the caudal half of mice during early development (18). This may limit the use of this CDX2P-NLSCre transgene, especially if other loxP alleles are introduced into Apc-floxed backgrounds. More recently, the same authors reported an interesting approach to generate a mosaic Apc mutant genotype in the intestinal tract using a functionally inactive CDX2P9.5-G22Cre recombinase transgene with a long mononucleotide tract (32). Stochastic activation of the Cre transgene by spontaneous frameshift mutation induced a severe colonic polyposis phenotype in CDX2P9.5-G22Cre;Apc580S/580S mice, leading to the death of the animals within the first month of life, whereas heterozygous CDX2P9.5-G22Cre;Apc580S/þ mice showed no overt signs of disease through 200 days of age (32). FabplCre;Apc2lox14/þ mice, carrying the FabplCre transgene and another exon 14-floxed Apc allele, also developed large intestinal tumours upon Cre expression directed to this part of the intestine (33). However, from the available data it could not be deduced whether these mice developed an increased number of large intestinal tumours compared with mice with a constitutional deletion of exon 14 (12).

951


In conclusion, we generated and characterized a new conditional Apc15lox-mutant allele in the mouse, which can be converted by tissuespecific Cre expression into a hypomorphic mutant ApcD15 allele, encoding a truncated protein expressed in low amounts. The conditional Apc15lox allele is well suited to study the lack of functional Apc in specific tissues in embryonic and adult stage. For instance, conditional inactivation of Apc in chondrocytes of Col2a1Cre;Apc15lox/15lox embryos enabled us to demonstrate that Apc-mediated control of b-catenin is essential for both chondrogenic and osteogenic differentiation of skeletal precursors (34). In addition, our conditional FabplCre;Apc15lox/þ mouse model, because of its extended lifespan compared with most other Apc-mutant models, and the development of a significant number of adenomas and adenocarcinomas in the large intestine, is useful to study the genetic alterations associated with the adenoma– carcinoma sequence in the mouse. Finally, the FabplCre;Apc15lox/þ mouse is an attractive model for studies on prevention and treatment of sporadic colorectal cancer. Using this model, we have been able to show that 5-aminosalicylic acid (mesalazine) inhibits colitis-associated colorectal neoplasia but not sporadic colorectal neoplasia (26). Funding Association for International Cancer Research (04-354) to E.R.-M., R.F.; The Netherlands Organisation for Scientific Research (NWO/ Vidi 917.56.353 to R.S., NWO/Vici 016.036.636) to R.F.; Besluit Subsidies Investeringen Kennisinfrastructuur program of the Dutch Government (BSIK 03038) to R.F.; EU FP6 ‘Migrating Cancer Stem Cells’ (LSHC-CT-2006-037297) to R.F.; Dutch Cancer Society (DDHK 2005-3299) to R.S. Acknowledgements We thank F.Stewart for the Cre-expressing bacterial strain; H.Westphal for the EIIaCre mice; J.Gordon for the Fabpl4x at –132Cre mice; Chris van der Zwan and Michel Mulders for animal care; Vanesa Muncan for immunostaining for c-Myc, and Annemieke van der Wal and Saskia Maas for histotechnical assistance. Conflict of Interest Statement: None declared.

References 1. Kinzler,K.W. et al. (1996) Lessons from hereditary colorectal cancer. Cell, 87, 159–170. 2. Polakis,P. (2000) Wnt signaling and cancer. Genes Dev., 14, 1837–1851. 3. Fodde,R. et al. (2001) APC, signal transduction and genetic instability in colorectal cancer. Nat. Rev. Cancer, 1, 55–67. 4. Fodde,R. (2003) The multiple functions of tumour suppressors: it’s all in APC. Nat. Cell Biol., 5, 190–192. 5. Fearnhead,N.S. et al. (2001) The ABC of APC. Hum. Mol. Genet., 10, 721– 733. 6. Moser,A.R. et al. (1990) A dominant mutation that predisposes to multiple intestinal neoplasia in the mouse. Science, 247, 322–324. 7. Su,L.K. et al. (1992) Multiple intestinal neoplasia caused by a mutation in the murine homolog of the APC gene. Science, 256, 668–670. 8. Fodde,R. et al. (1994) A targeted chain-termination mutation in the mouse Apc gene results in multiple intestinal tumors. Proc. Natl Acad. Sci. USA, 91, 8969–8973. 9. Oshima,M. et al. (1995) Loss of Apc heterozygosity and abnormal tissue building in nascent intestinal polyps in mice carrying a truncated Apc gene. Proc. Natl Acad. Sci. USA, 92, 4482–4486. 10. Quesada,C.F. et al. (1998) Piroxicam and acarbose as chemopreventive agents for spontaneous intestinal adenomas in APC gene 1309 knockout mice. Jpn. J. Cancer Res., 89, 392–396.

952

11. Sasai,H. et al. (2000) Suppression of polypogenesis in a new mouse strain with a truncated ApcD474 by a novel COX-2 inhibitor, JTE-522. Carcinogenesis, 21, 953–958. 12. Colnot,S. et al. (2004) Colorectal cancers in a new mouse model of familial adenomatous polyposis: influence of genetic and environmental modifiers. Lab. Invest., 84, 1619–1630. 13. Kuraguchi,M. et al. (2006) Adenomatous polyposis coli (APC) is required for normal development of skin and thymus. PloS Genet., 2, e146. 14. Shibata,H. et al. (2007) Alpha-catenin is essential in intestinal adenoma formation. Proc. Natl Acad. Sci. USA, 104, 18199–18204. 15. Pollard,P. et al. (2009) The Apc1322T mouse develops severe polyposis associated with submaximal nuclear b-catenin expression. Gastroenterology, 136, 2204–2213. 16. Gu,H. et al. (1993) Independent control of immunoglobulin switch recombination at individual switch regions evidenced through Cre-loxP-mediated gene targeting. Cell, 73, 1155–1164. 17. Shibata,H. et al. (1997) Rapid colorectal adenoma formation initiated by conditional targeting of the Apc gene. Science, 278, 120–123. 18. Hinoi,T. et al. (2007) Mouse model of colonic adenoma-carcinoma progression based on somatic Apc inactivation. Cancer Res., 67, 9721– 9730. 19. Smits,R. et al. (1999) Apc1638T: a mouse model delineating critical domains of the adenomatous polyposis coli protein involved in tumorigenesis and development. Genes Dev., 13, 1309–1321. 20. Sansom,O.J. et al. (2004) Loss of Apc in vivo immediately perturbs Wnt signaling, differentiation, and migration. Genes Dev., 18, 1385–1390. 21. Andreu,P. et al. (2005) Crypt-restricted proliferation and commitment to the Paneth cell lineage following Apc loss in the mouse intestine. Development, 132, 1443–1451. 22. Kielman,M.F. et al. (2002) Apc modulates embryonic stem-cell differentiation by controlling the dosage of b-catenin signaling. Nat. Genet., 32, 594–605. 23. Saam,J.R. et al. (1999) Inducible gene knockouts in the small intestinal and colonic epithelium. J. Biol. Chem., 274, 38071–38082. 24. Smits,R. et al. (1997) Loss of Apc and the entire chromosome 18 but absence of mutations at the Ras and Tp53 genes in intestinal tumors from Apc1638N, a mouse model for Apc-driven carcinogenesis. Carcinogenesis, 18, 321–327. 25. Soriano,P. (1999) Generalized lacZ expression with the ROSA26 Cre reporter strain. Nat. Genet., 21, 70–71. 26. Koelink,P.J. et al. (2009) 5-Aminosalicylic acid inhibits colitis-associated but not sporadic colorectal neoplasia in a novel conditional Apc mouse model. Carcinogenesis, 30, 1217–1224. 27. Sansom,O.J. et al. (2007) Myc deletion rescues Apc deficiency in the small intestine. Nature, 446, 676–679. 28. Boivin,G.P. et al. (2003) Pathology of mouse models of intestinal cancer: consensus report and recommendations. Gastroenterology, 124, 762–777. 29. Mandel,C.R. et al. (2008) Protein factors in pre-mRNA 3’-end processing. Cell. Mol. Life Sci., 65, 1099–1122. 30. Lakso,M. et al. (1996) Efficient in vivo manipulation of mouse genomic sequences at the zygote stage. Proc. Natl Acad. Sci. USA, 93, 5860–8565. 31. Wong,M.H. et al. (2000) Genetic mosaic analysis based on Cre recombinase and navigated laser capture microdissection. Proc. Natl Acad. Sci. USA, 97, 12601–12606. 32. Akyol,A. et al. (2008) Generating somatic mosaicism with a Cre recombinase-microsatellite sequence transgene. Nat. Methods, 5, 231– 233. 33. Haigis,K.M. et al. (2008) Differential effects of oncogenic K-Ras and N-Ras on proliferation, differentiation and tumor progression in the colon. Nat. Genet., 40, 600–608. 34. Miclea,R.L. et al. (2009) Adenomatous polyposis coli-mediated control of b-catenin is essential for both chondrogenic and osteogenic differentiation of skeletal precursors. BMC Dev. Biol., 9, 26. Received December 29, 2009; revised February 15, 2010; accepted February 17, 2010