Mouse Models of Colon Cancer

Apc Mutant Mice 

The first mouse mutant in the Apc gene (the mouse homologue of human APC on mouse chromosome 18) was identified in a colony of mice following random mutagenesis (Table 1).6 This mutant, Min (Multiple intestinal neoplasia), carries a truncation mutation at codon 850 of the Apc gene, hence ApcMin.7 Heterozygote ApcMin mice on a C57BL/6 background typically develop ∼30 polyps in the small intestine per animal. Using gene knockout technology in embryonic stem cells, several Apc mutations have been constructed. ApcΔ716 contains a truncating mutation at codon 716, whereas Apc1638N contains a truncating mutation at codon 1638.8, 9 Similar to ApcMin mice, both knockout mutants develop polyps mainly in the small intestine. Histologically, all these Apc mutants form polyp adenomas indistinguishable from each other (Figure 2A and B). Interestingly, however, the polyp numbers are very different, even in the same C57BL/6J background. Namely, ApcΔ716 develops usually ∼300 polyps, whereas Apc1638N forms only ∼3 polyps. Despite the numerous polyps developing in the small intestine of the ApcΔ716 (as well as in ApcMin) mice, only a few polyps are formed in the colon, although genetic penetrance of the latter phenotype is complete. With an additional mutation in Cdx2, a compound mutant strain can develop numerous polyps in the distal colon (see following text).10 Using ApcΔ716 mice, it was shown that polyp formation is initiated by loss of heterozygosity (LOH) at the Apc locus in the proliferative zone cells, followed by formation of an outpocket in the intestinal crypt.5, 9 These results suggest that APC is required for proliferative zone cells to migrate along the crypt-villus axis, which is supported by additional findings. For example, APC interacts with microtubules and accumulates at their plus ends in membrane protrusions,11, 12, 13, 14, 15 as well as the Rac-specific guanine nucleotide exchange factor Asef. APC stimulates Asef activity, regulating the actin cytoskeletal network and cell morphology.16 Moreover, Asef is required for migration of colon cancer cells and contributes to their aberrant migratory properties in the presence of truncated APC.17 Additional proteins have been shown to interact with APC, such as IQGAP1 and Asef2, which influence cell polarization and migration.18, 19

Table 1. Apc Mutant Mice and Their Phenotypes

	Mutation (gene symbol)	Tumor no./mousea		Tumor size and tumor histopathology	Reference
		Small intestine	Colon
	Simple Apc+/− mutations
	ApcMin (or ApcΔ850)	∼30	∼3	Benign adenomas	6
	Apc1638N	∼3	∼0	Benign adenomas	8
	ApcΔ716	∼300	∼3	Benign adenomas	9
	ApcΔ14	∼40	∼4	Benign adenomas	20
	Mutations that affect intestinal adenomas of Apc+/− mice
	Genes in the arachidonic acid pathway (gene product)
	Ptgs2 (COX-2)	(+/−) 34% and (−/−) 14%		Smaller adenomas	28
	Ptgs1 (COX-1)	(+/−) 57% and (−/−) 23%		Smaller adenomas	27
	Ptgerep2 (EP2)	(−/−)58%		Smaller adenomas	34
	Pla2g4a (cPLA2)	(+/−) 100% and (−/−) 85%		Smaller adenomas	179
	Ptges (mPGES-1)	(−/−)34%		Smaller adenomas	180
	Ptgds2 (H-PGDS)	(−/−)150%		Same as in Apc+/− mice	181
	Other genes (gene product)
	IL6 (IL6) (inflammation)	(−/−)70%			24
	Trp53 (p53 tumor suppressor)	Unchanged			182
	Blm (Bloom syndrome DNA helicase)	(−/−)4X,(+/−)2X			183
	Terc (telomerase RNA)	Increase or decrease depending on generations			184
	Cdkn1a (p21)	(−/−)2X			185
	Cdkn1b (p27)	(−/−)∼5X			186, 188
	Mmp7 (MMP-7)	(−/−)40%			187
	Nos2 (iNOS)	SI <50%, colon 10%			189
	Dnmt1 (DNA methylase, maintenance)	< 2%			190
	Mbd2 (methyl CpG binding repressor)	∼10%			191
	Mbd4 (methyl-CpG binding domain)	(−/−)2–3X			192, 193
	Abcb1b (Mdr1; multidrug resistance 1)	∼50%			194
	Mutations that affect small intestinal vs colon polyp adenoma distribution of Apc+/− mice
	Cdx2 (homeobox gene)	SI, 1/6; Colon 6X		Same as in Apc+/− mice	10
	Bub1b (BubR1; yeast homologue)	Colon 10X		Same as in Apc+/− mice	42
	ApcloxP (or Apc580S) × Adeno V-cre	Tumors in infected colon		Benign adenomas	196
	ApcloxP × Cdx2P-NLS-cre	Tumors predominantly in colon		Light invasion in old mice	112
	ApcloxP × Villin-cre	Tumors predominantly in SI			112
	Dietary and lifestyle factors that affect intestinal adenoma numbers of Apc+/− mice
	Docosahexaenoic acid	(Females) 31% (No effects on males)			43
	High-fat diet	SI, 44%; colon 36%			44
	Western-style diet	2.9 × of minimal diet			45
	Caloric intake	40%			46
	Exercise	60%–70%			47

SI, small intestine.

a Number in the C57BL/6 genetic background. Note that polyp number is affected by age, diet, and genetic background.

View Large Image Download to PowerPoint

Figure 2. Histopathology of intestinal tumors in mice with Apc heterozygous mutations and those with Apc/Smad4 compound mutations. (A and B) A relatively early small intestinal adenoma in an ApcΔ716 mouse stained with H&E and silver, respectively, in adjoining serial sections. Note the well-preserved basement membrane stained with silver in B. (C) A representative small intestinal adenocarcinoma in an Apc/Smad4 compound mutant mouse. Green arrowheads show the smooth muscle that forms a triangular shape (trabeculation), into which adenocarcinoma glands are invading (asterisks). (D and E) An invasion front of a colonic poly in an Apc/Smad4 compound mutant mouse (E shows a higher magnification of the squared area in D, arrowhead in E indicates “cap cells”). (F) Immunohistochemical staining of CD34 (green) and CD31 (red) in a section adjoining that in E. Closed arrowheads indicates immature myeloid cells expressing CD34, whereas open arrowheads show normal blood vessels that express both CD34 and CD31 (yellow). The red staining on the left upper corner is muscularis mucosae. Nuclei are stained in blue (D–F reprinted with permission from Kitamura et al57).

Another Apc knockout strain that has been reported is the ApcΔ14 mouse line.20 The histopathology of mice of this strain appears essentially the same as in ApcMin or ApcΔ716, with differences in only polyp multiplicity.

While intestinal polyposis is the most visible phenotype in Apc heterozygous mutant mice, they display a wide variety of additional phenotypes. For example, ApcMin mice show progressive loss of immature and mature thymocytes from ∼80 days of age, with complete regression of the thymus by 120 days.21 They also deplete splenic natural killer cells, immature B cells, and B progenitor cells in the bone marrow due to complete loss of interleukin (IL)-7–dependent B-cell progenitors. Transplantation experiments suggest that an altered bone marrow microenvironment is responsible for the selective lymphocyte depletion in ApcMin mice.21

Along this line, it has been reported that CD4+CD25+ regulatory lymphocytes induce regression of intestinal tumors in ApcMin mice.22 In cell transfer experiments, the recipients of regulatory cells show increased apoptosis and down-regulation of COX-2 within the tumors. Functions of regulatory cells may be compromised during thymic atrophy.

Because of the heavy tumor load in the small intestine, most Apc mutant mice die young (4–5 months) due to anemia and cachexia, and some of them die of intestinal intussusception. It has been reported that ApcMin mice have perturbations in ammonia metabolism in the liver that appears to be responsible for the mortality.23 On the other hand, ApcMin mice have a 10-fold increase in the level of circulating IL-6, which can cause severe cachexia as exemplified by loss of muscle weight and fat tissues.24 The role of IL-6 in the cachexia has been verified in compound mutant ApcMin/+ IL-6−/− where the symptoms disappear with mild (∼30%) decrease in the polyp multiplicity.

Mutations That Affect Intestinal Adenomas of Apc+/− Mice 

Introduction of an inactivating cyclooxygenase (COX)-2 gene mutation dramatically decreases the polyp number in Apc mutant mice (Table 1).25, 26 Likewise, COX-1 mutation also reduces polyp multiplicity.27 Expression of COX-2 protein is found in intestinal polyps of various sizes, and COX-2 is expressed from a very early stage of polyp formation on.28 Introduction of a COX-2 gene (Ptgs2) knockout mutation into the ApcΔ716 mice causes dramatic reductions in both the number and size of polyps in the compound mutant mice in a mutant gene dosage-dependent manner.28, 29 These results provided the rationale to treat human patients with FAP with COX-2 inhibitors such as celecoxib or rofecoxib, and clinical trials confirmed the results of the animal experiments.30 Following these experiments, a number of reports have been published in which the Apc mutant mice were dosed with various drugs or drug candidates (for a review, see Steinbach et al30). The constitutive isozyme COX-1 also has a significant role in polyposis. Introduction of COX-1 mutation into ApcMin mice reduces the number and size of intestinal polyps by ∼80%, a similar effect to that caused by COX-2 mutation.27 In fact, COX-1 and COX-2 cooperate in polyp formation by supplying prostaglandin (PG) E2, which stimulates polyp angiogenesis.31 Unfortunately, COX-2 inhibitors caused cardiovascular side effects in some patients who participated in sporadic polyposis prevention trials.32, 33 Accordingly, COX-2 inhibitors, as well as regular nonsteroidal anti-inflammatory drugs, must be prescribed with caution to patients who are at high risk for cardiovascular accidents or to younger patients with FAP who are carefully monitored.

To further assess the role of the arachidonic acid/COX-2/PGE2 pathway (Figure 3) in intestinal polyposis, several additional compound mutants of ApcΔ716 mice have been constructed with other genes in the pathway (Table 1). Homozygous null mutations in the genes that are involved in the metabolic pathway caused similar reductions in the polyp number, although milder than the effects by COX-2 or COX-1 depletion.

View Large Image Download to PowerPoint

Figure 3. Arachidonic acid metabolism and intestinal polyposis. Metabolites of arachidonic acid are shown with the enzymes that catalyze the conversion steps of the metabolites. The receptors for PGE2 are also shown. TX, thromboxane.

PGE2 is known to bind to some of the four G protein–coupled cell surface receptors: EP1, EP2, EP3, and EP4. To determine which particular EP receptor is involved in the polyposis phenotype, knockout mutations for the respective receptors were introduced into ApcΔ716 mice, and the polyp number and size were scored. The results clearly show a significant suppression of the polyposis phenotype only in the compound mutant mice with EP2 mutation.34 Moreover, induction of COX-2 itself, as well as other phenotypes associated with COX-2 induction such as angiogenesis and basement membrane changes, appears to be mediated by PGE2 and EP2 receptor. These phenotypes suggest that a positive feedback signal, mediated through EP2 and intracellular cyclic adenosine monophosphate, regulates COX-2 expression.34 Although the role of the EP4 receptor could not be assessed by genetic experiments because homozygous EP4 deletion is lethal, cell culture experiments showed that increased proliferation and motility in response to PGE2 treatment was mediated by EP4,35 and inhibitors for EP4 have been reported to correct such cell behaviors.36, 37

COX-2 is expressed not only by adenomatous polyps, but also by hamartomatous polyps and malignant tumors. For example, COX-2 expression has been reported in Peutz–Jeghers gastrointestinal hamartomas38 as well as in hamartomas that developed in Smad4, Cdx2, and Lkb1 mutant mice.39 COX-2 induction is therefore a common feature of gastrointestinal tumors.

Many antitumor mechanisms have been proposed for nonsteroidal anti-inflammatory drugs40, 41; genetic studies should help to elucidate the pathways involved, that is, crossing mice with disruptions in putative target genes with Apc mutant mice.

Mutations of many genes have been introduced into Apc mutant mice; some increase the number of polyps, and many others reduce them. Such effects are summarized in Table 1.

As described previously, all Apc mutant mice develop adenomatous polyps in the small intestine rather than in the colon. However, an additional mutation in the Cdx2 gene in ApcΔ716 mice reverses the polyp localization, shifting most polyps to the colon as in human FAP.10 Interestingly, the dramatic increase in the colon polyp number is caused by the increased frequency of Apc LOH due to chromosomal instability. The latter appears to result from the activation of the mTOR pathway and an acceleration of the G1 to S phase transition in the cell cycle.10 These results present a new mechanism for chromosomal instability and suggest a possibility for treatment and prevention of colon cancer with chromosomal instability. Consistently, introduction of a BubR1+/− mutation into ApcMin mice causes 10 times more colonic tumors, and MEFs derived from compound mice show a higher rate of genomic instability.42

Another advantage of studying colon cancer in genetic mouse models is that the effects of various diets and food additives on tumor formation and progression can be investigated. For example, docosahexaenoic acid reduces intestinal polyp development, although the effect is moderate and found only in female mice,43 whereas feeding ApcΔ716 mice with a high-fat diet increases polyp numbers significantly.44 Likewise, a Western-style diet (high fat and low calcium) accelerates tumor formation in Apc1638N mice.45 Restriction of caloric intake by 40% reduces intestinal polyp numbers in ApcMin mice by ∼60%, suggesting that dietary interventions can partially offset genetic susceptibility to intestinal carcinogenesis.46 Likewise, exercise can reduce the polyp multiplicity in ApcMin mice by 30% to 40%.47 A variety of chemopreventive agents have been evaluated using ApcMin mice, but their effects appear to be affected by the tumor stages of the cells derived from these mice.48

β-Catenin Mutant Mice 

Because APC forms a complex with other proteins that mediate the Wnt signaling pathway, mouse models have been used to determine whether mutations in other components of the complex can also induce polyp formation. Stabilizing mutations in serine/threonine residues of β-catenin have been identified in a subpopulation of human colon tumors that do not carry APC mutations; therefore, conditional mutations that stabilize β-catenin have been specifically expressed in the intestines of mice. When stabilized β-catenin was expressed from the calbindin promoter, mice developed only a few polyps in the small intestine.49 By contrast, Cre recombinase expression under the control of cytokeratin 19 (K19) or fatty acid binding protein gene promoters allows expression of a stabilizing β-catenin mutant from a floxed allele and causes the formation of 700–3000 polyps in the small intestine.50 These findings confirm the role of Wnt signaling activation in polyp formation and suggest that polyps are initiated from the transient amplifying cells in the proliferative zone where K19 and fatty acid binding protein are expressed at significant levels but calbindin is only scarcely. This conditional system has been used to conditionally express β-catenin in several other organ systems and has provided evidence for a role of Wnt signaling in prostate tumorigenesis51 and embryonic and immune system development.52, 53, 54

Modifiers of Apc That Affect Tumor Progression 

While all Apc mutant mice develop adenomatous polyps, they do not progress into invasive or metastatic adenocarcinomas at any significant frequencies. Mutations in other genes can modify the tumor progression phenotypes of Apc mutant mice. For example, introduction of Smad4 mutation into the ApcΔ716 polyposis mice results in locally invasive malignant adenocarcinomas.55 Although the human homologues SMAD4 and APC are on separate chromosomes, the mouse genes are both found on mouse chromosome 18, about 30 cM apart. Because polyps are initiated by Apc LOH in ApcΔ716 intestines, and because this LOH is caused by loss of the entire chromosome 18 due to recombination at the ribosomal DNA locus near the centromere, LOH of Apc also results in LOH at Smad4. Taking advantage of this observation, mice were constructed that carried ApcΔ716 and Smad4 mutations on the same chromatid in the cis-configuration (cis-Apc/Smad4 mice). In the intestinal polyps, both Apc and Smad4 loci are lost, resulting in homozygous mutant cells for both loci. Importantly, the intestinal polyps in these mice progress rapidly into very invasive adenocarcinomas (Figure 2C).55 Interestingly, however, these adenocarcinomas do not metastasize during the short life span of these mice. The histopathology is somewhat similar to that of human right-sided colon cancer, which is associated with mutations in the type II receptor of transforming growth factor (TGF)-β. This model verifies tumor progression by sequential mutations in multiple genes. Moreover, some of the cis-Apc/Smad4 mice also developed adenocarcinomas at the duodenal papilla of Vater, which is one of the complications in human FAP after colectomy is performed.55 Consistently, homozygous disruption of the type II receptor of TGF-β gene (Tgfbr2) in Apc1638N mice caused malignant transformation of the intestinal adenomas caused by the Apc mutation.56

Recently, a molecular mechanism of tumor invasion has been proposed for the cis-Apc/Smad4 mice.57 In this model, a novel type of immature myeloid cells is recruited from the bone marrow to the tumor invasion front (Figure 2D and E). These CD34+ immature myeloid cells (Figure 2F) express metalloproteinases MMP9/2 and CC-chemokine receptor 1 (CCR1) and migrate toward the CCR1 ligand CCL9. In adenocarcinomas, expression of CCL9 is increased in the tumor epithelium. Deletion of the Ccr1 gene in cis-Apc/Smad4 mice prevents accumulation of CD34+ immature myeloid cells at the invasion front and suppresses tumor invasion. These results suggest that preventing the recruitment of matrix metalloproteinase–expressing cells with chemokine receptor antagonists is a therapeutic strategy for patients with advanced colon cancer (Figure 4),58 especially because matrix metalloproteinase inhibitors were not effective in cancer clinical trials.59 In models of intestinal polyposis caused by stabilization of β-catenin50 or Apc mutation, CD34+ mast cells have been reported to be an essential component for polyp development.60 The possible relationship between mast cells and immature myeloid cells remains to be investigated.

View Large Image Download to PowerPoint

Figure 4. Molecular mechanism of colon tumor invasion in the cis-Apc/Smad4 mutant mouse. SMAD4-deficient tumor cells produce the chemokine CCL9 and recruit receptor-expressing cells that promote tumor invasion. The inactivation of the TGF-β family signaling within the tumor epithelium causes increased production of chemokine CCL9, because it is suppressed by TGF-β, activin-A, and BMPs (1). Increased levels of CCL9 recruit immature myeloid cells that carry the CCL9 receptor CCR1 from the bloodstream to the tumor invasion front (2). These immature myeloid cells produce MMP9 and 2 (3), which allow the tumor to invade the stroma (4). Modified with permission from Kitamura and Taketo.58

Persistent activation of the phosphatidylinositol 3-kinase/Akt pathway has been implicated in colorectal cancer as well as in other neoplasms. It has been reported that ApcMin mice with heterozygous disruption of Pten develop invasive carcinomas that are larger in size.61 However, the histopathology of these mice is not so severe as in the cis-ApcSmad4 mice.

Likewise, mutation of Prox1 in ApcMin mice has been reported to induce progression of colon adenomas to cancer.62 Although the pathologic changes in the tumor epithelium are of a higher-degree dysplasia than seen in the ApcMin adenomas, little signs of invasion are observed, and tumors appear to remain as carcinoma in situ. Whether these lesions can really progress to invasive cancer remains to be investigated.

The Mammalian DNA MMR System 

The MMR system has several functions that are important for maintaining the integrity of mammalian genomes, including the postreplicative repair of base substitution mutations as well as small insertion/deletion mutations (IDLs), the signaling of cell cycle arrest and apoptosis in response to exposure to DNA damaging agents, and the suppression of recombination between homologous sequences. The characteristic repair steps of MMR are conserved between prokaryotes and eukaryotes and include the recognition of mismatched bases by MutS proteins and the recruitment of MutL proteins to initiate the subsequent repair steps resulting in the removal and resynthesis of the DNA strand carrying the mutated base(s).72 The MMR system in eukaryotes is complex, and several MutS and MutL homologues have been identified. In mammalian cells, the initiation of the repair of mispaired bases is facilitated by 3 different homologues of the bacterial MutS protein.73 The MSH2-MSH6 complex (MutSα) initiates the repair of base-base mispairs as well as single-base IDLs, whereas the MSH2-MSH3 complex (MutSβ) initiates the repair of larger IDLs of 2 to 4 bases (Figure 5).74, 75, 76, 77 Both MutSα and MutSβ complexes are redundant in the repair of some single-base IDL mutations. Two additional MutS homologues are known in mammals, termed MSH4 and MSH5, both of which were shown to play a specialized role in the control of meiotic recombination in mice.78, 79, 80

View Large Image Download to PowerPoint

Figure 5. Model for mammalian MMR. MMR is initiated by the recognition of mispaired bases by MutSα and MutSβ complexes that act as sliding clamps. The activation of downstream repair events requires the interaction of MutSα and MutSβ with MutLα. In addition, MutLγ participates in the repair of single-base mismatches and 1–base pair IDLs. MutSα bound to a mismatch recruits EXO1, which initiates mismatch excision from 5′ nicks in the template strand. Although the mechanisms that generate 5′ and 3′ single-stranded nicks is not clear, MutLα contains an endonuclease activity (encoded by PMS2) that can introduce 5′ nicks into the template strand carrying the mismatch (not shown). Mismatch excision proceeds past the site of the mismatch, and the resulting gap is stabilized by RPA. In mammalian cells, other exonucleases possibly participate in the removal of mispaired bases whose identity remain unknown. Later repair steps require the interaction of MutS and MutL complexes with DNA replication proteins, including proliferating cell nuclear antigen, replication factor C, RPA, and polymerase δ, to coordinate the transfer of information between mismatch recognition and excision/resynthesis (not shown). Little is known about the nature of these interactions and the precise composition of the late MMR complexes.

Following initial mismatch binding, the MutSα complexes interact with heterodimeric complexes of eukaryotic MutL homologues to activate subsequent repair events, including a complex between MLH1-PMS2 (MutLα), a complex between MLH1-PMS1 (MutLβ), and a complex between MLH1-MLH3 (MutLγ). The primary MutL activity for mitotic MMR is provided by the MutLα complex and its interaction with MutSα complexes is essential for the activation of the later MMR steps, which include the excision of the DNA strand carrying the mismatched base(s) and its resynthesis (Figure 5).72, 73, 81 It is not entirely clear how the MutS and MutL proteins interact at mismatched bases to initiate the excision steps, but different models emphasize the importance of adenosine triphosphate (ATP) binding and hydrolysis by MutS and MutL proteins for the activation of repair excision.82, 83, 84, 85

In human cells, mismatch excision can be initiated from a single-strand break located either 5′ or 3′ to a mispair.86, 87 Biochemical studies suggest that for the 5′ to 3′ directed excision, MutSα facilitates the loading of EXO1, a 5′ to 3′ exonuclease, at single-strand DNA breaks in a mismatch-dependent manner. EXO1 proceeds to excise the mismatch-carrying strand, creating a gap that is filled by RPA (Figure 4).88 MutLα enhances the mismatch dependence of the reaction89 and promotes the inactivation of EXO1.90 Recently it was found that PMS2 encodes an endonuclease activity that can introduce 5′ and 3′ nicks into substrate DNA to facilitate the 5′ to 3′ degradation of the mismatch-encoding strand by EXO1.91 Replication factor C and proliferating cell nuclear antigen support the bidirectional excision of mismatched bases in vitro.72 The completion of the MMR reaction also requires repair synthesis by DNA polymerase δ.88 Although EXO1 is currently the only known MMR exonuclease, the analysis of Exo1 knockout mice indicates the presence of additional not yet identified exonucleases that participate in the excision reaction.92 Similarly, many other proteins that are required for completion of MMR such as helicases and topoisomerases remain unknown.

In addition to correcting replication errors, MMR also participates in the cytotoxic response to DNA damaging agents including SN1 DNA methylators such as temozolomide and N-methyl-N′-nitro-N-nitrosoguanidine (MNNG), cross-linking agents such as cisplatin, and base analogues such as 6-thioguanine.72, 93, 94, 95 Whereas the MMR-dependent cellular response to DNA damage initiates G2 arrest and apoptosis, MMR-deficient tumor cells display an increased resistance to DNA-damaging agents, including a variety of chemotherapeutic drugs. The MMR-dependent activation of apoptosis by these agents was proposed to involve the recognition of DNA lesions by MutSα, recruitment of MutLα, and the subsequent creation of DNA double-strand breaks that result from futile attempts of the MMR pathway to repair damaged sites in the template strand (“futile cycle model”).96 An alternative “direct signaling model” proposes that MutSα and MutLα complexes function as sensors of DNA damage93, 97 that directly recruit cell cycle effectors such as ATM and ATR/ATRIP to damaged sites to activate CHK1 and CHK2, respectively.98, 99, 100, 101, 102 Both models are not mutually exclusive, and a combination of both mechanisms may be responsible for the MMR-dependent DNA damage response.

MMR Knockout Mouse Lines 

Mice have been generated with disruptions in all known mammalian MutS and MutL homologues as well as in Exo1 (Table 2). Analysis of their phenotypes reveals that, in addition to their functions in genome maintenance, some of the MMR proteins have essential functions in mammalian meiosis, as well as in immunoglobulin class switch recombination and somatic hypermutation during B-cell maturation. Some MMR proteins promote triplex repeat expansion associated with neurologic disorders (reviewed by Peled et al,103 Kolas and Cohen,104 and McMurray105).

Table 2. MMR Mutant Mouse Lines

	Mouse line	Tumor incidence	Tumor spectruma				Repair defect (MSI)		DNA damage responseb	Reference
			Gastrointestinal	Lymphoma	Skin	Other	Mononucleotide	Dinucleotide
	Knockout mouse lines
	MutS homologues
	Msh2−/−	High	√	√	√	√	High	High	Defective	108, 109, 110
	Msh3−/−	Low	√	—	—	√	Moderate	High	Normal	74, 115
	Msh6−/−	High	√	√	√	√	None	Low	Defective	113
	Msh3−/−Msh6−/−	High	√	√	√	√	High	High	Defective	74, 115
	Msh4−/−	None	—	—	—	—	N/A	N/A	N/A	80
	Msh5−/−	None	—	—	—	—	N/A	N/A	N/A	78, 79
	Msh2loxP/loxP; Vill-cre	High	√	—	—	—	High	High	Defective	Kucherlapati et al, unpublished data
	MutL homologues
	Mlh1−/−	High	√	√	√	√	High	High	Defective	117, 118, 119, 120, 132
	Pms1−/−	None	—	—	—	—	Low	Low	N/A	119
	Pms2−/−	High	—	√	√	√	High	High	Defective	119, 121
	Mlh3−/−	High	√	√	√	√	Moderate	N/A	Defective	124
	Pms2−/−Mlh3−/−	High	√	√	—	—	High	N/A	Defective	124
	Exonuclease
	Exo1−/−	Moderate	—	√	—	—	High	Low	Defective	92, 195
	Knock-in mouse lines
	Msh2G674A/G674A	High	√	√	√	√	High	High	Normal	139
	Msh6T1217D/T1217D	High	√	√	√	√	High	High	Normal	142
	Mlh1G67R/G67R	High	√	√	√	√	High	High	Normal	144

N/A, data not available. GI tumors: adenoma, adenocarcinoma, and flat adenoma. Skin tumors: squamous cell carcinoma. Others include tumors of the uterus, brain, lung, liver, and mammary gland.

a MSI was analyzed in tumor samples, normal tissue, or cultured cells. Tumor spectrum includes mainly the following. lymphoma: B and T cell lymphoma (NHL, non-Hodgkin's lymphoma). b DNA damage response was analyzed in MEF lines, ES cell lines, and various tissues of MMR mutant mouse lines. Agents include MNU, MNNG, temozolomide, 6-thioguanine, cisplatin, and 1,2-dimethylhydrazine, among others. Details can be found in the cited references for each MMR knockout mouse line.

The cancer susceptibility phenotypes of MMR knockout mice correlate with the associated DNA repair defects and the frequency of MMR mutations found in patients with HNPCC. However, phenotypic differences between MMR knockout mice and patients with HNPCC exist. For example, unlike patients with HNPCC with mutations in MSH2, MSH6, and MLH1, heterozygous mice carrying the corresponding knockout mutations do not develop early-onset tumors. Homozygous Msh2, Msh6, and Mlh1 knockout mice are cancer prone with a tumor spectrum that includes gastrointestinal cancers, but unlike patients with HNPCC, most mice die prematurely due to aggressive lymphomas (Figure 6).106 The differences between humans and MMR knockout mice are likely due to the shorter life span of mice, making somatic loss of the wild-type allele and subsequent tumorigenesis unlikely. The recent identification of patients with biallelic mutations in MSH2, MSH6, MLH1, and PMS2 revealed that MMR deficiency in humans results in a severely reduced life span due to hematologic and other malignancies similar to those of MMR knockout mice.107 A comparison of the phenotypes of human and mice indicates that the basic mechanisms of DNA repair and tumor suppression have been conserved; MMR mutant mice have been highly useful in determining how the loss of MMR function results in tumorigenesis.

View Large Image Download to PowerPoint

Figure 6. Histopathology of tumors in MMR mutant mice. (A) Lymphoma in Msh6 mutant mouse. (B) Higher magnification showing large anaplastic cells admixed with small lymphocytes. (C) Adenoma in Msh2 mutant mouse. (D) Higher magnification showing columnar to cuboidal neoplastic intestinal epithelial cells. (E) Invasive carcinoma in Msh2 mutant mouse with the formation of mucin lakes (asterisks) and desmoplasia (arrows). (F) Higher magnification showing invasive cells at all stages of intestinal epithelial cell maturation (goblet [triangles] and Paneth [arrows] cells).

MutS Homologue Knockout Mouse Lines 

Several Msh2 knockout mice have been generated.108, 109, 110 Msh2 is the central component of both MutSα and MutSβ complexes, and its inactivation results in complete MMR deficiency. As a consequence, Msh2−/− mouse cells are unable to repair single-base mismatches and 1 to 4 base IDLs. The loss of MMR in Msh2-deficient mice causes a severe reduction in survival and a strong cancer predisposition phenotype. The majority of animals develops T-cell lymphomas and succumbs to these tumors by 6 to 8 months of age. Msh2−/− mice that survive to 6 to 12 months of age develop small intestinal adenomas and invasive adenocarcinomas (Figure 6C–F), and a subset of these older animals also develop sebaceous gland tumors similar to patients with Muir–Torre syndrome with MSH2 mutations. The tumors in these mice display an MSI-high phenotype similar to patients with HNPCC109, 111; in contrast to Msh2−/− mutant mice, Msh2+/− heterozygous mice display overall normal survival and only a small number of these mice develop similar tumors later in life.

Recently, a mouse line for the conditional inactivation of Msh2 has been generated (M. Kucherlapati and W. Edelmann, 2009, unpublished data). This mouse line carries an Msh2loxP allele that allows the somatic deletion of Msh2 in the intestinal epithelium by expression of the Villin-Cre recombinase transgene. Tumorigenesis in Msh2loxP × Villin-cre mice is limited to the small intestinal epithelium, and the animals develop 1 to 2 adenoma and/or adenocarcinoma within their first year of life. The tumors that develop in these mice display an MSI-high phenotype. Recently, a Cdx2-NLS-cre recombinase mouse line has been established that directs tumorigenesis to the large intestine in mice carrying a floxed Apc knockout allele.112 The cancer phenotype of Msh2loxP × Villin-cre mice, and possibly Msh2loxP × Cdx2-NLS-cre mice, more closely resembles that of patients with Lynch I syndrome, which will make these mice useful preclinical models.

Msh6−/− mice display a strong cancer phenotype with a tumor spectrum similar to that of Msh2−/− mice.113 However, Msh6-deficient mice survive longer (up to 18 months) and the onset of tumor development is delayed compared with Msh2-deficient mice. The longer survival of Msh6−/− mice is the result of the partial repair defect caused by Msh6 inactivation. In Msh6-deficient cells, the repair of base substitution mutations and single-base IDLs is impaired; however, in contrast to Msh2-deficient cells, the repair of 2 to 4 base IDLs by MutSβ is not affected. As a consequence, the mutator phenotype in Msh6−/− mice is characterized predominantly by an accumulation of base substitution mutations rather than frameshift mutations that occur frequently in Msh2−/− mice. Because the repair of 2 base IDLs is not impaired, the tumors that develop in Msh6−/− mice do not display the MSI phenotype that is characteristic of HNPCC tumors. These observations in Msh6-deficient mice suggest that mutations in MSH6 can result in late-onset cancer phenotypes in patients whose tumors could be characterized by low rates of MSI. Indeed, MSH6 mutations are frequently associated with atypical HNPCC cases that are characterized by cancer onset at age 60 years and older and variable MSI phenotypes.114 In addition, Msh6−/− mice develop endometrial cancers that are also seen in a significant number of patients with MSH6 mutations.115, 116

In contrast to Msh2−/− and Msh6−/− mice, the disruption of Msh3 in mice does not result in a strong cancer phenotype. Msh3-deficient mice display normal survival and develop gastrointestinal tumors only very late in life and at low incidence.74 The absence of a significant tumor phenotype is caused by the moderate repair defects conferred by the loss of Msh3 function. Msh3-deficient cell extracts are defective in the repair of 1 to 4 base IDLs but can still efficiently repair single-base substitution mutations due to the presence of functional MutSα complexes. The combined inactivation of Msh6 and Msh3 in mice results in complete MMR deficiency and a strong cancer phenotype similar to Msh2−/− mice.74, 115 Therefore, in humans, inactivating mutations in MSH3 are likely not implicated in HNPCC due to the functional redundancy between the MutSα and MutSβ complexes. However, it is possible that MSH3 missense mutations are associated with increased cancer susceptibility, especially in older patients.

MutL Homologue Knockout Mouse Lines 

Similar to Msh2, the common component of both MutSα and MutSβ complexes, Mlh1 is important for the formation of all 3 MutL complexes. For example, loss of Mlh1 results in complete MMR deficiency and a strong mutator phenotype in the tissues of Mlh1−/− mice. Mlh1-deficient mice have a shortened life span (up to 12 months) and display a strong cancer predisposition phenotype that is very similar to that of Msh2-deficient mice. The tumor spectrum of Mlh1−/− mice includes T-cell lymphomas, intestinal adenomas and adenocarcinomas, and skin tumors. The tumors in these mice also display an MSI-high phenotype.117, 118, 119, 120

Pms2 is the other component of the MutLα complex; Pms2−/− mice also display a strong cancer predisposition phenotype. However, in contrast to Mlh1-, Msh2- or Msh6-deficient mice, Pms2-deficient mice develop lymphomas and sarcomas but not intestinal tumors and these tumors develop later in life.119, 121 The delayed cancer onset and the different tumor spectrum in Pms2-deficient mice are most likely related to a milder mutator phenotype that is caused by Pms2 deficiency. Although the inactivation of Pms2 increased the mutation frequencies at mononucleotide repeat tracts, the increase in mutation frequency was almost 3-fold lower as compared with Mlh1-deficient mice. This difference in the strength of the mutator phenotype indicates that Pms2 inactivation, similar to Msh6 inactivation, causes only a partial repair defect in the tissues of mice. Consistent with this notion, PMS2 mutations in human patients are associated with late-onset Lynch syndrome cancers and the overall cancer risk is lower as compared with patients with MSH2 or MLH1 mutations.122

Recently, an interesting Pms2cre knock-in mouse line has been generated.123 In this mouse line, Cre recombinase is activated by frameshift reversion at a dinucleotide repeat within the Cre coding sequence. This mouse line allows the stochastic inactivation of loxP-flanked gene loci in single cells surrounded by normal tissue and should be useful for modeling spontaneous cancer-causing mutations.

Similar to Pms2−/− mice, the inactivation of Mlh3 results also in a late-onset cancer phenotype.124 However, in contrast to Pms2−/− mice, Mlh3−/− mice develop adenomas and adenocarcinomas in their small intestines. In addition, Mlh3-deficient mice develop extra-gastrointestinal tumors, including lymphomas, basal cell carcinoma of the skin, and other tumors. Mlh3−/− mice display increased MSI at mononucleotide repeat sequences in their genomic DNA but to a lesser extent than that seen in Pms2−/− mice. The combined inactivation of both Mlh3 and Pms2 in mice increases the levels of MSI to levels that are similar to those seen in Mlh1-deficient mice. In addition, Mlh3-Pms2 double-deficient mice display severely reduced survival, increased cancer susceptibility, and a tumor spectrum that is comparable to Mlh1-deficient mice. These results indicate that Pms2 and Mlh3 as components of the MutLα and MutLγ complexes, respectively, share overlapping in vivo repair and tumor suppressor functions. They also provide an explanation as to why PMS2 and MLH3 mutations are less frequently found in patients with colorectal cancer than MLH1 mutations and also suggest that PMS2 and MLH3 mutations might be more common in patients with late-onset Lynch syndrome–associated cancers.

The inactivation of Pms1 in mice does not result in a detectable repair defect or increased cancer susceptibility.119 It is possible that the MutLβ complex shares redundancy in its repair functions with the 2 other MutL complexes in mice. Consistent with this notion, mutations in PMS1 seem not to play a role in HNPCC.

Exo1 Knockout Mice 

EXO1 is the only known human exonuclease that facilitates, in conjunction with MutSα and MutLβ, the excision of mismatched bases. The inactivation of Exo1 in mice results in increased predisposition to lymphoma, although these tumors develop only late in life.92 The loss of Exo1 function causes defects in the repair of single-base mismatches and 1-base IDL but does not affect the repair of 2-base IDLs. As a result, the tissues and tumors in Exo1−/− mice display MSI at mononucleotide repeat markers but not dinucleotide repeat makers. The comparison with the repair defects in Msh2- or Mlh1-deficient mice indicates that Exo1 deficiency results only in a partial repair defect and indicates the existence of redundant exonucleases in mammalian MMR, whose identity remains unknown.

The role of EXO1 germline mutations in HNPCC is currently uncertain. Several EXO1 mutations were identified in patients with HNPCC-like colorectal cancers that displayed MSI.125 The lack of intestinal tumors and the absence of a significant MSI phenotype in Exo1-deficient mice suggest that EXO1 mutations might only rarely be involved in HNPCC.92 Similarly, the incidence of intestinal tumors in Exo1+/− mice that carried an Apc1638N allele was only slightly increased.126 However, based on the studies with Exo1-deficient mice, it is possible that EXO1 mutations might be associated with late-onset cancers, cancer phenotypes with reduced penetrance, and/or different tumor etiology. Interestingly, a recent analysis of single nucleotide polymorphisms (SNPs) in EXO1 showed that 2 coding SNPs, T439M and P757L, were associated with a 5-fold increase in colorectal cancer risk. In addition, the tumors from patients who carried the risk genotypes for these SNPs tended to have increased MSI.

Intestinal Tumorigenesis in MMR/Apc Mutant Mice 

MMR deficiency increases the mutator phenotype, and it is thought that the accumulation of somatic mutations in key tumor suppressor genes is responsible for the accelerated progression of MSI-positive tumors. A significant number of MSI-positive human colorectal cancers carry somatic mutations in the APC tumor suppressor gene, indicating that loss of APC function is critical for tumor initiation and/or progression in MMR-deficient tumors. To test the impact of MMR deficiency on Apc-driven intestinal tumorigenesis, mouse lines with homozygous mutations in Msh2, Msh6, Msh6, Mlh1, and Pms2 that also carry heterozygous Apc germline mutations have been constructed. The combination of MMR deficiency with the predisposing Apc mutations in mice limits the tumor development almost exclusively to the intestinal tract and the tumor incidence correlates with the severity of the MMR defects in the different MMR knockout mice. Whereas the loss of Msh2 or Mlh1 function in Apc mutant mice results in a dramatic increase in the number of intestinal tumors, the loss of Msh6 or Pms2 causes a more moderate increase in intestinal tumor numbers and the loss of Msh3 does not increase the tumor load.120, 127, 128, 129 The increase in tumor number is caused by the accumulation of somatic mutations within the wild-type Apc allele.110, 129 Most of the Apc mutations are truncation mutations and affect β-catenin binding or down-regulation, which are essential for tumor suppression. MMR-defective mice have also been crossed with mice that carry mutations in other tumor suppressor genes considered important for gastrointestinal tumorigenesis. For example, the combination of Msh2 or Msh6 deficiency with p53 deficiency in mice results in accelerated tumorigenesis.130, 131 However, the Msh2−/−/p53−/− and Msh6−/−/p53−/− compound mutant mice mainly succumb to T-cell lymphoma. Therefore, it will be necessary to determine the effects of p53 mutation on intestinal tumorigenesis in conditional knockout MMR mice.

MMR Knock-in Mouse Lines 

The generation of MMR knockout mouse lines allowed researchers to systematically assess the biological functions of each MMR protein and to evaluate their importance for tumor suppression. However, the phenotypes seen in knockout mice reflect the combined loss of all MMR genome maintenance functions. For example, Msh2 and Msh6 deficiency in mice results not only in DNA repair deficiency but also defects in DNA damage response and increased recombination between homologous sequences.108, 115 Although it is reasonable to assume that the loss of all of these functions contributes to the cancer phenotype in these mice, the individual importance of each MMR function for tumor suppression is difficult to assess. Recent studies tested the effect of DNA-damaging agents on MMR-deficient mice and showed that exposure to MNU led to increased mutation frequencies and lymphomagenesis in Mlh1- and Pms2-deficient mice.132, 133 Similarly, exposure of Msh2-deficient mice to 1,2-dimethylhydrazine accelerated lymphomagenesis and also induced colorectal carcinogenesis.134 Interestingly, the intestinal crypt cells in Msh2-deficient mice display a reduced apoptotic response to 1,2-dimethylhydrazine and also MNNG, temozolomide, and cisplatin.134, 135 Collectively, these results suggest that the failure to effectively eliminate DNA damage–bearing cells increases the mutator phenotype in MMR-defective cells, which may also provide a selective advantage to tumor cells, especially in the early stages of tumorigenesis.93

To test this assumption and better define the importance of the MMR-dependent DNA repair and damage response functions for tumor suppression, knock-in mouse lines with separation-of-function mutations in the Msh2, Mlh1, and Msh6 genes have been generated (Table 2). The first MMR knock-in mouse line carried a mutation that creates a glycine-to-alanine change at codon 674 within the conserved adenosine triphosphatase domain of the Msh2 protein (Msh2G674A, termed Msh2GA). ATP processing is critical for the initiation of the repair process by MutS homologues,136, 137 and a significant number of MSH2 mutations in patients with HNPCC affect this function.138 The analysis of Msh2GA mice shows that Msh2 missense mutations can have distinct effects on the DNA repair and damage response functions. Although the mutant MutSαMsh2GA complex retains normal mismatch recognition, it is defective in ATP-mediated mismatch release and lost its ability to signal MMR. The Msh2GA mutation also affects the antirecombination function of MutSα, and Msh2GA/GA embryonic stem cells display increased homologous recombination frequencies (Avdievich and Edelmann, unpublished data). By contrast, the Msh2GA mutation does not affect the DNA damage response function of the MutSαMsh2GA complex and Msh2GA/GA MEF cells display a normal apoptotic response to cisplatin, 6-thioguanine, and MNNG, indicating that the DNA repair and damage response functions can be separated by MutS missense mutations. Msh2GA/GA mutant mice display a strong cancer predisposition phenotype and develop T-cell lymphoma, intestinal adenocarcinoma, and squamous basal cell carcinoma, similar to Msh2−/− mice. However, the 50% survival rate differs between Msh2−/− (6 months) and Msh2GA/GA (9–10 months) mice, indicating that the MMR-dependent DNA damage signaling is important for suppressing the early stages of tumorigenesis. Nevertheless, the strong cancer phenotype of the Msh2GA/GA mice indicates that the increased mutation rate causing the MMR deficiency is sufficient to drive tumorigenesis. Overall, the analysis of Msh2GA mutant mice shows that the combination of increased mutation rates and defective apoptosis accelerates tumorigenesis in MMR-deficient mice and that human tumors carrying MMR missense mutations may remain more responsive to treatment with chemotherapeutic agents.139

Msh6T1217D knock-in mice carry a threonine to aspartic acid mutation at amino acid 1217 in the Msh6 coding region (termed Msh6TD). The mutation is located opposite the Msh2 ATP binding domain in the MutSα complex, and studies of the corresponding Msh6G1067D mutation in yeast indicate that the mutant MutSα complex can bind to mispaired bases but is defective for ATP-induced sliding clamp formation. In addition, it prevents complex formation MutLα and occludes mispaired bases from other MMR pathways.140, 141 In mice, the mutant MutSαMsh6TD complex also retains mismatch binding capacity and is resistant to ATP-induced mismatch release.142 The Msh6TD mutation causes a dominant repair defect that not only impairs the repair of single-base mutations by MutSα but also interferes with the MutSβ-mediated repair of 2-base IDLs. By contrast, the loss of Msh6 in Msh6−/− mice only impairs the repair of base substitution mutations. As a consequence, genomic DNA in the tissues of Msh6TD/TD mice displays increased MSI at both mononucleotide and dinucleotide repeat markers, whereas no MSI is seen in Msh6−/− mice. Similar to the Msh2GA mutation, the Msh6TD mutation does not affect the DNA damage response function of MutSα. Msh6TD/TD mice display a strong cancer predisposition phenotype and develop B-cell lymphoma, intestinal adenocarcinomas, and skin tumors. It is important to note that unlike Msh6+/− mice, survival of Msh6TD/+ mice is reduced due to late-onset lymphoma that displays an MSI phenotype. Interestingly, a patient with HNPCC with early-onset colorectal cancers that displayed MSI was recently identified that carried a threonine to isoleucine germline mutation at the corresponding amino acid position T1219 in MSH6.143 The analysis of the corresponding mutation in yeast and mice provides a mechanistic explanation for the observed cancer and MSI phenotype. In addition, the studies with Msh6TD mutant mice show for the first time that MMR missense mutations can exert dominant effects on the DNA repair functions of mammalian MMR complexes in vivo and increase cancer susceptibility in heterozygous carriers.142

The Mlh1G67R mouse line is the first mouse model of a recurrent HNPCC mutation found in patients of different ethnic origin and from different countries (termed Mlh1GR).138 The Mlh1GR mutation affects one of the Mlh1 adenosine triphosphatase domains and also has distinct effects on MMR-dependent cellular DNA repair and damage response functions. Similar to the Msh2GA and Msh6TD missense mutations, these functions are clearly distinguishable by the Mlh1GR mutation. Mlh1GR/GR mice display increased MSI in the genomes of intestinal epithelial cells and T lymphocytes, indicating the loss of MMR activity similar to Mlh1−/− mice. However, Mlh1GR/GR mice retain a robust apoptotic response to cisplatin exposure in these tissues, whereas this response is impaired in Mlh1−/− mice. Mlh1GR/GR mice succumb to T-cell lymphomas and intestinal adenomas and adenocarcinomas. However, the incidence of intestinal tumors is significantly reduced in Mlh1GR/GR mutant mice compared with Mlh1−/− mice. These results show that the DNA damage response function of Mlh1 affects tumorigenesis in a tissue-specific manner. In addition, Mlh1GR/GR mice are sterile due to the inability of the mutant Mlh1GR protein to interact with meiotic chromosomes at pachynema, showing that the adenosine triphosphatase activity of Mlh1 is essential for fertility in mammals.144

The generation of MMR knock-in mice provides the opportunity to model MMR gene variants found in humans. A significant proportion of patients with HNPCC carry MSH2, MLH1, and MSH6 missense mutations138; however, it is often difficult to assess the impact of these mutations on tumorigenesis. Similarly, MMR genes in humans carry a significant number of SNPs of unknown significance. The generation of knock-in mouse lines provides the opportunity to determine the extent to which certain HNPCC missense and SNPs impact the DNA repair and damage response functions of MMR proteins and to assess their pathogenicity in vivo.

Mouse Models for Colon Cancer Metastasis 

Although many useful genetic mouse models of benign adenoma have been developed, there have been few models of colon cancer except for the locally invasive models described previously. Essentially, no genetic mouse models are available of spontaneous tumors that arise in the intestines, become invasive, and metastasize to organs such as liver, lungs, and lymph nodes, as they do in humans. Because metastasis is responsible for most colon cancer mortality, it is important to develop genetic mouse models of colon cancer metastasis. In the meantime, it is also necessary for metastasis researchers to use the transplantation (xenograft) models most suitable for their own purposes. These models have been important tools for evaluating a variety of therapeutic compounds and biologics, especially before genetic mouse models were available (for reviews, see Kobaek–Larsen et al,170 Heijstek et al,171 Alencar et al,172 and Céspedes et al173). In these studies, human colon cancer cell lines or tissue pieces were transplanted into immunocompromised mice such as nude mice or NOD-SCID mice.

Some human cancer cell lines have been tested for lymph node metastasis after intrarectal implantation of the cells suspended in Matrigel.174 These cell lines, including the colon cancer line HT-29, produce locally aggressive rectal tumors that subsequently metastasize to the para-aortic lymph node, although they do not produce distant metastases in organs such as liver or lungs.

Serial surgical orthotopic implantation of a human colon cancer–derived liver metastasis fragment in nude mice led to the establishment of cells that reproducibly metastasized to the liver, peritoneum, lymph nodes, and spleen but not to the lungs.175 The metastatic abilities of these cells might result from the mixture of the epithelial and stromal cells within the original tumor sample; these findings have been reproduced.176 The orthotopic injection of human colon cancer cell lines has recently been improved using a specially designed micropipette to deliver the tumor cells into the cecum of nude mice.173 The injected cells metastasize to the lymph nodes, liver, lungs, and peritoneum at significantly higher rates than reported previously.

Although not many cell lines have been derived from mouse colon cancers, some of them are characterized and are being extensively studied. For example, the Colon 26 (CT26) line was established from a chemical carcinogen-treated C3H mouse, whereas the CT38 cell line came from a C57Bl/6 mouse.177 Another cell line, CMT93, shows typical epithelial morphology and was derived by chemical carcinogenesis from a C57Bl/6 mouse.178