More than two decades of Apc modeling in rodents.

Based in part on the correlation of FAP severity with specific truncating APC mutations, and on the ability of truncated APC to bind to full-length APC, it was proposed that particular APC truncations act in a dominant-negative manner [ 111 ]. However, a direct test of this hypothesis revealed no increased polyp susceptibility in mice carrying a transgene encoding Apc amino acids 1–716, even though the truncated Apc protein was detected in intestinal cells. It is possible that the proposed dominant -negative activity of truncated Apc could not overcome functional Apc from two wild-type Apc alleles that could compensate for any deleterious effect of the truncated allele. To explore this possibility, the transgene for truncated Apc was introduced into Apc Δ716/+ mice [ 90 ]. Because intestinal tumor number, distribution, and morphology were the same in Apc Δ716/+ mice with and without the extra truncated Apc transgene, it was concluded that in this mouse model, truncated Apc does not act in a dominant-negative manner [ 90 ].

A rat model with a germline nonsense mutation at Apc codon 1137 ( Apc am1137 ) was generated to overcome some of the limitations of Apc mouse models [ 134 ]. Rats homozygous for the Apc am1137 allele die as embryos. Apc am1137/+ rats develop both small intestinal and colonic polyps with 100% penetrance, and are called “PIRC” rats for P olyposis I n R at C olons [ 134 ]. The polyps in PIRC rats are adenomas with malignant changes and local invasion seen in old rats. No signs of metastasis have been detected in these rats. As seen in humans with germline Apc mutations, the polyps from PIRC rats show β-catenin nuclear translocation in advanced but not in early adenomas. As with Apc Min/+ mice, most intestinal polyps in PIRC rats show LOH. Because chromosome 18, which carries the Apc gene in rats, is metacentric, pyrosequencing could be used to demonstrate that LOH in PIRC rats predominantly occurs by means of homologous recombination [ 134 ]. The greater width of rat intestines and colons, relative to those of mice, allows for growth of larger intestinal tumors, which facilitates study of tumor progression beyond the early stage. Wider colons and higher colonic tumor multiplicities in rats also supported a longitudinal endoscopic study of tumorigenesis [ 134 ]. Male PIRC rats have more polyps than do females [ 101 ]. Most Apc mouse models do not show a gender bias. However, in Apc Min-FCCC/+ mice, an Apc Min/+ mouse model with a different genetic background, males also develop more colonic polyps than do females [ 135 ]. In contrast, female Apc Δe1–15/+ mice display more severe phenotypes than do males [ 101 ]. In humans, women appear to be slightly less affected by colon cancer than are men [ 136 ]. PIRC rats also show high incidence of jaw tumors, which are the main cause of morbidity in female PIRC rats [ 134 ]. This extra-intestinal phenotype has also been described in patients with FAP syndrome [ 137 ]. A second Apc rat model (Kyoto Apc Delta or KAD rat) was developed with a germline nonsense mutation in the Apc gene, resulting in deletion of the C-terminal 321 amino acids [ 138 ]. This deletion does not appear to affect life expectancy even in homozygous KAD rats, and no spontaneous polyps develop in the KAD rat intestines. However, KAD rats showed enhanced inflammation-mediated colon tumorigenicity, consistent with a Wnt-independent role for the C-terminal domain of Apc in tumor suppression. In summary, rodent models with Apc mutations were first generated more than 2 decades ago. Studies of 45 rodent models with germline and conditional Apc mutations have led to greater understanding of the role of APC in development, differentiation, and homeostasis of intestinal epithelial cells. In addition, these models have allowed exploration of the role of APC in intestinal and extra-intestinal development and tumorigenesis. Mouse and rat models with germline Apc mutations have permitted experimental testing of different molecular pathways and investigation of genetic and environmental contributions to tumor formation, not only in the gastrointestinal tract but also in other tissues. These models have also facilitated testing different preventive and therapeutic agents in preclinical studies. Continued effort should be made to clarify some of the less understood features of the different Apc rodent models. These lingering mysteries include identifying the variables that contribute to differences in extra-intestinal phenotypes, and polyp distribution and number. The full potential of the Apc models has not yet been reached; it is expected that they will continue to provide insight into Apc and cancer biology for decades to come.

Seven mouse models with mutations upstream to that in Apc Min have been described. Apc Δ242 [ 88 ], Apc Δ474 [ 89 ], and Apc Δ716 [ 90 – 92 ] mice have Apc truncation mutations at codons, 242, 474, and 716, respectively, while Apc Δ580 [ 93 ], Apc 580D [ 94 ], and Apc Δ14 [ 95 ] mice each have a deletion of exon 14, resulting in a frameshift and a nonsense mutation at codon 580. Apc Δ15 mice have a deletion of the last Apc exon and the 3’UTR region and have no detectable expression of the mutant allele [ 96 ]. These seven mouse models share many phenotypes with Apc Min/+ mice, including embryonic lethality in the homozygous state, and in heterozygous mice, development of anemia and intestinal polyps predominantly in the small intestine that are indistinguishable at the microscopic level [ 88 – 94 , 96 , 97 ]. Although polyp number varies between these seven models ( Table 2 ), in most cases, direct comparative studies have not been performed. Mammary tumors have been reported for 14.3% of Apc Δ580 18.5% of Apc Δ474 , and 9% of Apc Δ14 mice [ 89 , 93 , 97 ]. Is the variation in polyp number in these mouse models due to the progressive deletion of particular Apc domains (see figure 1 )? The Apc Δ716 protein is 134 a.a. shorter than the Apc Min protein and lacks an additional portion of the armadillo repeat region. Although it is tempting to speculate that the three-fold increase in polyp number seen in Apc Δ716/+ mice compared to Apc Min/+ mice results from interruption of the armadillo repeat region, Apc Δ242/+ mice, which have a truncating Apc mutation that eliminates the entire armadillo repeat region, develop fewer polyps than Apc Δ716/+ mice. Moreover, Apc Δ580/+ , Apc 580D/+ , Apc Δ14/+ and Apc Δ474/+ mice, which have truncating mutations in the middle of the armadillo repeat region, have reported intestinal polyp numbers similar to that seen in Apc Min/+ mice ( Table 2 ).

Two Apc mouse models with reduced Apc expression were generated by inserting a neomycin cassette into Apc intron 13 in either reverse (Apc NeoR ) or forward orientation (Apc NeoF ) [ 109 , 110 ]. The neomycin cassette disrupts an enhancer and reduces the level of full-length Apc expressed from the mutant allele to 20% of normal levels for Apc NeoR , and 10% for Apc NeoF . Each allele produces an embryonic lethal phenotype in the homozygous state. By the age of 15 months, Apc NeoR/+ and Apc NeoF/+ develop intestinal polyps with relatively low incidence (19% and 50%, respectively) and multiplicity (0.26±5.4 and 1.09±8.5 polyps per mouse, respectively). The polyps in Apc NeoR and Apc NeoF mice display loss of the wild-type Apc allele and have less β-catenin stability and accumulation of nuclear β-catenin than do polyps from Apc Δ716/+ mice [ 109 , 110 ]. Thus, in mice, there appears to be a critical threshold level of Apc to support tumor suppression.

In human CRC, APC mutations are predominantly found in a region referred to as the mutation cluster region (MCR), and result in truncation of the C-terminal half of APC [ 98 ]. Complete deletion of APC has been reported in FAP syndrome only rarely [ 99 , 100 ], leading to the hypothesis that truncated APC protein can enhance tumorigenicity in a dominant-negative manner. A mouse model with complete deletion of all 15 Apc exons (Apc Δe1–15 ) was generated to test the requirement of truncated APC for tumor formation [ 101 ]. Apc Δe1–15/+ mice develop intestinal polyps of the same distribution and morphology as those seen in Apc Min/+ mice, but with increased frequency. Polyps from Apc Δe1–15/+ mice had lower levels of Apc + mRNA compared to normal tissue, consistent with a requirement for loss of the wild-type allele for intestinal tumor development, although this was not directly examined. Apc Δe1–15/+ mice also develop more severe anemia than Apc Min/+ mice, and one Apc Δe1–15/+ mouse developed a mammary tumor. Female Apc Δe1–15/+ mice showed more severe phenotypes than did males. Polyps from Apc Δe1–15/+ mice had lower mRNA levels of Wnt target genes Axin2, c-Jun , and β- catenin than polyps from Apc Min/+ mice [ 101 ]. Although puzzling in terms of the underlying mechanism and pathogenesis, this observation is consistent with the hypothesis that there is a level of Wnt signaling optimal for polyp formation, and signaling in excess of this level inhibits polyposis [ 101 ].

Adenomatous polyposis coli (APC) is a critical tumor suppressor gene in the colon. Humans with germline APC mutation develop hundreds to thousands of colon tumors in their first few decades of life, a condition referred to as familial adenomatous polyposis (FAP). These tumors are pre-cancerous, and prophylactic colon removal is recommended to avoid progression to invasive carcinoma that otherwise would occur in FAP patients by age 39, on average [ 1 – 4 ]. Notably, APC mutation is also an early if not the first step in the development of more than 80% of all sporadic colorectal cancers [ 5 , 6 ]. In both inherited and sporadic colorectal cancer, APC mutations result in premature truncation of the large (2843 amino acid) APC protein, eliminating roughly half to three-quarters of the C-terminal portion [ 7 , 8 ]. APC interacts with multiple proteins and participates in diverse cellular processes including proliferation, differentiation, apoptosis, adhesion, and migration. One of the first reported APC functions is as a Wnt-signal antagonist. In this capacity, APC forms a complex with GSK3β, axin, and other proteins to mediate phosphorylation and eventual proteasomal destruction of the oncoprotein β-catenin [ 7 , 9 ]. Animal models have been generated to study APC functions in development and tumorigenesis including Drosophila, C.elegans, zebrafish, mouse, rat and pig.

Mouse models with germline Apc mutations have been useful to probe many aspects of APC biology, especially in intestinal tumorigenesis. However, most of these models are limited by a short life span, the predominance of intestinal phenotypes, and embryonic lethality in the homozygous state. To study functions of APC at different developmental stages and in organs other than the intestine, investigators have developed mice with conditional Apc mutations [ 112 ]. A critical component of most conditional systems is CRE recombinase, which induces recombination between two lox P1 sites, resulting in excision of the DNA between these sites. In conditional Apc mouse models, lox P1 sequences are inserted into introns of the mouse Apc gene flanking particular exon(s). In the presence of Cre, excision of the lox -flanked DNA leads to a frameshift mutation and truncation of Apc. Five different conditional Apc alleles have been made; Apc 580S , Apc CKO , Apc Δex14 , Apc 15flox and Apc lox468 [ 93 – 96 , 113 , 114 ]. The specificity of these Apc mutations is achieved by placing Cre under control of a tissue- or developmental stage-specific promoter or an inducible promoter, or by infecting tissues with Cre-expressing Adenovirus [ 115 ]. Table 3 summarizes different conditional Apc mouse models.