MURINE INTESTINAL CANCER

Murine Intestinal Cancer Models

Mouse models of intestinal cancer can be grouped into five broad categories based on the underlying germline mutation or the means of tumor induction:Apc and Apc-related GEM, GEM with mutations in MMR genes, GEM with altered TGFb signaling, immune-deficient mice, and carcinogen-treated rodents. The following is a summary (adapted from 58) of the characteristics of intestinal tumors in GEM and carcinogen-treated mice.

Apc and related GEM: The Apc^Min/+ (Multiple Intestinal Neoplasia) mouse is the first mouse model of intestinal tumorigenesis induced by a germline mutation (59). Homozygous inheritance of the Apc^Min allele, a nonsense mutation induced by ethylnitrosourea, leads to embryonic lethality (60). As in the human hereditary condition FAP, heterozygous mice develop multiple adenomas exhibiting somatic mutation or loss of the normal Apc allele. Adenomas in Apc^Min/+ mice occur predominantly in the small intestine (59), while in persons with FAP, adenomas are located primarily in the colon. By four months of age, mice with the Apc^Min allele expressed on a C57BL/6 background develop an average of 24 adenomatous polyps in the small intestine, and 5 in the colon. The lesions are generally pedunculated adenomas that arise in the mucosa without inflammation and protrude into the lumen of the gut. Histologically, these tumors are similar to human colonic adenomas with one notable exception: in humans, dysplastic cells are observed on the mucosal surface, while adenomas in Apc^Min/+ mice are covered by a surface layer of normal epithelium. Progression to invasive adenocarcinoma (a common occurrence in persons with FAP) does not occur; likewise, metastasis has not been observed in these mice (61,62).

Gastrointestinal intraepithelial neoplasm from an AKR ApcMin/+ mouse.	Papillary adenoma from a Blm+/- x ApcMin/+ mouse. /td>

Tumors arising in mice carrying different Apc mutations are similar to each other in histological appearance, but vary with respect to age of onset, number, and location. Two other GEM models carrying targeted insertion mutations at other sites in the Apc gene, Apc^Δ716/+ and Apc^1638N/+, also develop polyps located predominantly in the small intestine. Loss of the normal allele is observed in the polyps in both mouse lines (63,64). The onset of tumor development is relatively early in Apc^716/+ mice (3 weeks of age) and there is a three-fold increase in the number of polyps per mouse as compared to Apc^Min/+. Apc^1638N/+ mice exhibit fewer polyps than either Apc^Min/+ or Apc^Δ716/+ mice. Similar to tumors occurring in Apc^Min/+ and Apc^Δ716/+, tumors from Apc1638N/+ mice demonstrate LOH for markers on the entire chromosome 18; however, they do not carry mutations in K-Ras and p53 (64).

Two other Apc GEM, Apc^1638T/+ and Apc^1309/+, have been described. Apc^1638T/+ mice do not develop intestinal tumors. These mice survive into adulthood, but are smaller than wild type animals (65). Apc^1309/+ mice develop an average of 34 adenomas by 14 weeks of age, a slightly higher incidence of polyp formation than the Apc^Min/+ mouse.

The APC protein exerts its effects as a tumor suppressor, in part, through its regulatory role in the Wnt-signal-transduction pathway. Nuclear localization of β-catenin, an indication of activated WNT signaling, occurs in human CRC and in tumors from Apc^Min/+ mice. Three GEM have been generated in which overexpression of mutated β-catenin acts as a dominant activator of WNT signaling. In one of these GEM, adenoma development occurrs by 3-4 weeks of age; at the same time the mice begin to die from polycystic kidney disease that also accompanies expression of the transgene (66). In contrast, no tumors were observed in transgenic mice overexpressing an NH2-terminally deleted β-catenin in the small intestine (67). A third transgenic mouse overexpressing an activated β-catenin developed tumors that were primarily restricted to the duodenum and jejunum. These tumors appeared around 3 weeks of age and were similar histopathologically to those of Apc^Δ716 heterozygous mice, but occurred in much greater numbers (68). The reasons for this variability in phenotype are not well understood, but may be due to differences in the promoter and/or coding sequences used to construct the transgenes, or differences in genetic background. In fact, specific modifier genes that contribute to differences in tumor susceptibility among mouse strains have been identified (69,70).

GEM with alterations in mismatch repair genes: MMR genes encode proteins that correct DNA base-pair mismatches resulting from replication errors, genetic recombination, and chemical modification. In humans, germline mutations in most of these genes, MSH2, MSH6, MLH1, and PMS2, but not MSH3, cause HNPCC. GEM have been generated with mutations in each of these genes. Unlike HNPCC patients, there is no increase in tumor formation in mice heterozygous for mutations in MMR genes. Homozygous mice do develop tumors at an increased rate and are, thus, used as models for HNPCC (42). Most of these mice have an attenuated life span due to the development of lymphomas, but those that survive are prone to intestinal epithelial tumors similar to those observed in HNPCC. Four MMR GEM-Msh2^-/-, Msh3^-/-, Mlh1^-/-, and Msh6^-/--develop intestinal lesions, and all four exhibit a higher incidence of neoplasia than any of the Apc GEM.

Msh2^-/- mice in a mixed background (C57BL/6 and 129/Ola) exhibited a high incidence of skin neoplasms as well as lymphomas that lead to the death of approximately 50% of mice between the ages of 2 and 6 months. Mice that survived 6 months or longer developed intestinal adenomas at a high frequency (15/22), with most located in the duodenum and jejunum. Adenomas were typically plaque lesions, unlike the pedunculated tumors seen in Apc^Min/+ mice and other MMR GEM. Intestinal neoplasms in Msh2^-/- mice were associated with a marked inflammatory response unlike other MMR GEM models (71).

Msh3^-/- mice on a mixed background of C57BL/6 (60%), 129/Sv (37.5%), and SJL/J (2.5%) strains did not have an attenuated lifespan. Overall, tumors appeared in these mice at an advanced age at a rate similar to the wild-type animals examined; however, GI tumors were seen only in Msh3-/- mice. Tumors occurred in 24 of the 39 Msh3^-/- mice examined, and 4 of these had a total of 5 GI tumors: 3 adenomas, and 2 adenocarcinomas. (72). Intestinal lesions occured in 10% of mice and include low-grade polypoid adenomas, with deep cystically dilated crypts lined with goblet cells. Rectal prolapse with mucosal hyperplasia associated with inflammation also occurs in this model. In the double mutant model, Apc^1638/+ Msh3^-/-, mice display adenomas with numerous apoptotic bodies, and adenocarcinomas consisting of infiltrative glands in a desmoplastic stroma. Rectal prolapse also occurs in these double mutants, manifested morphologically by mucosal hyperplasia at the anorectal junction, with surface erosion and early displacement of deep crypts classified as non-neoplastic herniation.

Msh6^-/- mice developed on the same mixed background as the Msh3^-/- mice have an attenuated lifespan of only 11 months, on average. Intestinal tumors, with a high ratio of carcinomas to adenomas, were observed in 38% of mice (72). On a C57BL/6 background, Msh6-deficiency led to intestinal adenomas (primarily in the duodenum and jejunum) with densely packed crypts with high-grade nuclear features (73). Extra-intestinal lesions, including lymphomas and benign skin and hepatic neoplasms, were also common in these mice (73). Intestinal tumors were rare in Msh6^-/- mice developed on a different mixed background (129/OLA and FVB/n), a finding that underscores the importance of genetic background in mouse tumorigenesis (74).

Msh3 and Msh6 double-mutant mice were generated to evaluate the overlap in function of the MMR proteins. The incidence of intestinal tumors doubled to two tumors per mouse compared to Msh6^-/- mice, while the frequency of mice with tumors increased to 75%. The ratio of carcinomas to adenomas did not change. The lifespan of the mice decreased from 50% survival at 22 months in Msh3^-/- mice, 11 months for Msh6^-/- mice and 6 to7 months for the double-mutant mice (72). These results suggest that Msh3 can slightly ameliorate the effects of Msh6-deficiency.

Among MMR mice developed on mixed C57BL/6 and 129/Sv backgrounds (Mlh1^-/-, Pms1^-/- and Pms2^-/-) intestinal tumors were observed only in Mlh1^-/- mice. In these animals, most lesions were in the jejunum and ileum; however, a few occurred in the colon. Multiple neoplasms, including carcinomas, were observed in Mlh1^-/- mice between 4 and 12 months of age. These mice developed lymphomas, skin tumors, and sarcomas like other MMR mice (75). Although no spontaneous intestinal lesions were detected in the Pms2 mutant mice, heterozygotes were more likely than wild-type mice to develop intestinal adenomas and carcinomas after exposure to a mutagen (76).

GEM with alterations in the TGFb signaling pathway: Homozygous targeted inactivation of Tgfβ1 in GEM results in death before one month of age due to autoimmune disease. Thus, the Tgfβ1^-/- mouse is most easily maintained on an immunodeficient background (Rag2^-/-). To enable the investigation of the role of Tgfβ1 in the development and progression of intestinal cancer, Tgfβ1(^+/+, ^+/-, or ^-/-) x Rag2^-/- mice were generated in a mixed 129S6/CF-1 background. Cecal and colonic neoplasms arose spontaneously in all resulting mouse lines regardless of the status of Tgfβ1-expression; however, tumor incidence and severity was markedly increased in Tgfβ1^-/- x Rag2^-/- mice (77). In these animals, intestinal lesions occurred with 100% penetrance; sessile adenomas appeared by two months of age while carcinomas were detectable at 3 to 6 months. Carcinomas often contained areas with mucin-filled cysts. Immunohistochemical analysis showed no evidence of changes in β-catenin localization, suggesting that the tumor-suppressive effects of TGFβ1 are not mediated through an Apc-dependent pathway (77).

Mucinous Adenocarcinoma of the Murine Intestinal Tract

The SMAD family of proteins includes downstream effectors of the TGFβ signaling pathway. Two Smad-deficient mice (Smad3^-/- and Smad4^+/-) developed intestinal lesions. Smad3^-/- mice were developed on a 129/Sv and a 129/Sv X C57BL/6 background (83). On the 129/Sv background, 100% of Smad3^-/- mice developed colonic tumors by 6 months of age. These lesions included mucosal hyperplasia, sessile adenomas, adenocarcinomas, and metastatic carcinomas with metastasis to the liver and mesenteric lymph nodes. Tumors were also characterized by transmural inflammation with slight mucosal hyperplasia of the lamina propria in some areas of the colon. Occasionally, more advanced colitis was observed. As in Tgfβ1^-/- x Rag2^-/- mice, there was no indication of activation Wnt signaling in tumors. On the hybrid background of 129/Sv and C57BL/6, Smad3^-/- mice developed similar lesions; however, the tumors occurred in only 30% of mice and their onset was delayed to 10 months of age (83).

Homozygous Smad4-deficient mice die during embryonic development (84). Heterozygous Smad4 mice developed polyps in the duodenum and stomach after 1 year of age. Most of the gastric polyps in this model are hamartomatous, although dysplastic glands and signet ring cell carcinoma were also reported. A mild inflammatory response was associated with the tumors, and there were no lesions in the distal intestines of the Smad4 mice. The Smad4 GEM may be a useful model for juvenile polyposis, a human syndrome in which germ-line mutations of SMAD4 have been identified (85).

Combinations of Apc mutants with other GEM: Several double mutants have been generated by crossing Apc mutants with other GEM that develop intestinal tumors. Tumors of the small intestine in Smad4^+/-x Apc^Δ716/+ mice were larger, but not more numerous than in either parental mouse line. There was also a significant increase in the severity of the lesions, indicated by increased desmoplasia, invasion, and the presence of signet ring cells, suggesting that mutation of Smad4 plays a role in tumor progression (84).

Tumor progression was not enhanced when MMR GEM were crossed with mice carrying Apc mutations; however, most of these double mutants demonstrated increased tumor numbers. For example, Pms2^-/- x Apc^Min/+ mice had a 3-fold increase in small intestinal adenomas, a 4-fold increase in colonic adenomas, but no increased incidence of carcinoma (86). Similar increases in tumor number were observed in Msh2^-/-x Apc^Min/+ and Mlh1^-/-x Apc^1638N/+ mice (87,88). Likewise, Mlh1^-/-x Apc^Min/+ mice exhibited a 3-fold increase in average tumor number, but no differences in tumor size or invasiveness (89).

Adenocarcinoma of the Murine Intestinal Tract

Immune-deficient mouse models with intestinal inflammation: Several GEM with mutations that interfere with the regulation of immune function (i.e. Il-10, IL-2, Gα_i2, and Tcrα knockouts) exhibit spontaneous intestinal inflammation. These GEM are generally characterized by inflammation of the large bowel with proliferative lesions that occasionally progress to adenocarcinomas (90-93). Inflammation does not develop when these mice are re-derived in germ-free (bacteria- and virus-free) environments. In specific-pathogen free environments (where normal flora are still present in the gastrointestinal tract), Il-10- and Il-2-deficient mice exhibit delayed development of lesions that are smaller and less numerous (91,94).

Severe hyperplasia and inflammation in the colon of an IL-10-/- mouse.

The intestinal flora plays an important pathogenic role in the immune-deficient IBD models. Helicobacter sp. have been associated with lesion development in these GEM models of inflammatory bowel disease (IBD), but there are conflicting data concerning a causative role for these bacteria in the disease process. For example, some studies show that Helicobacter hepaticus infection is not required for the development of IBD in Il-10^-/- mice (95,96). However, in another study, Il-10^-/- mice failed to develop IBD when cleared of Helicobacter infections, while pathogen-free Il-10^-/- mice developed colitis after infection with Helicobacter hepaticus (91). This study is not conclusive, however, since it is possible that bacteria other than Helicobacter sp, were eliminated by antibiotic treatment, and infection with H. hepaticus may have exacerbated an already existing condition rather than causing the colitis.

Immune-deficient GEM models of IBD provide strong evidence that intestinal inflammation is a significant precursor of adenocarcinoma in mice. Colonic lesions arise in 60% of Il-10^-/- mice (C57BL/6 X 129) by 6 months of age (97), and 32% of Il-2,β₂-microglobulin^-/- mice (C57BL/6 X 129) develop colonic adenocarcinomas between 6 and 12 months of age (98). These tumors are plaque-like lesions located in regions of hyperplastic colitic mucosa or rectal prolapse. This association with inflammation complicates the histological classification of lesions in these animal models of colitis. Neither altered β-catenin expression nor loss of heterozygosity at the Apc locus was detected in adenocarcinomas from Il-2^-/- or Il-10^-/- mice.

Rodent models of carcinogen-induced intestinal neoplasia AOM is a potent procarcinogen that specifically induces colorectal tumors (99,100). In AOM-treated rodents, most tumors arise in the colon and form grossly visible exophytic polypoid or plaque-like growths. Low-grade lesions are similar in microscopic appearance to human colonic adenomas. There is evidence that AOM-treated mice may be useful as a model of metastasis in colorectal cancer (101). The molecular abnormalities identified in AOM-induced tumors include alterations in β-catenin expression (102) and p53 mutations (103). Gastrointestinal tumors can be induced in rodents by a variety of carcinogens including N-methyl-N'-nitro-N-nitrosoguanidine (104), N-ethyl-N'-nitro-N-nitrosoguanidine, 1,2-dimethylhydrazine (105), 2-amino-3,4-dimethylimidazo[4,5-f]quinoline (106), and N-methyl-N-nitrosourea (76). Other carcinogens that induce intestinal tumors have been identified in chronic-exposure, long-term bioassays (reviewed in 107). These include capsaicin, Captafol, Captan, hydrogen peroxide, and N-(Trichloromethylthio)phthalimide.

Multiple polyps in the colon of an AOM-treated mouse.	Carcinoma with severe desmoplasia in the colon of an AOM-treated mouse.

Other GEM that develop intestinal tumors The caudal-like homeobox gene, Cdx2, encodes a homeodomain transcription factor that is restricted to the intestinal epithelium in adult mice. Cdx2 plays a broad role in the development of caudal body structures during early development and in later development promotes differentiation of the intestine by activating the transcription of intestine-specific genes (108). In a Cdx2 knockout mouse, the homozygous mutation proved to be lethal in early embryonic stages. In heterozygous (Cdx2^+/-) mice, disruption of mucosal differentiation contributed to the development of colonic lesions that appeared very early in life. These mice developed heterotopic villiform structures in the cecum and proximal colon, and some hamartomatous polyps in the colon. The duplication of the mucosa and formation of hamartomas was apparent at embryonic day 11.5. The lesions persisted into adulthood, but grew slowly and did not progress to invasive adenocarcinomas (109).

N-cadherin is a mediator of calcium-dependent cell-cell adhesion. GEM that overexpress a dominant negative N-cadherin along the entire crypt-villus axis exhibit inflammation and occasional adenomas in the small intestine. Inflammation in these mice involves lymphocyte infiltration of the lamina propria, IgG- and IgA-secreting plasma cells, and histiocytes. Neutrophilic infiltration, Paneth cell hyperplasia, perturbed crypt-villus architecture, and ulceration were observed in inflammatory lesions in early stages. The incidence of neoplasia in this model has not been described (110).

The intestinal epithelium is protected and lubricated by a layer of mucus that is comprised mainly of highly glycosylated proteins called mucins. Muc2 is the most abundant gastrointestinal mucin. Mice with an inactivated Muc2 gene exhibit intestinal epithelial tumor formation without apparent inflammation. Spontaneous progression to carcinoma is also observed in Muc2^-/- mice. Tumor cells in Muc2^-/- mice exhibit increased proliferation, migration, and c-Myc expression; and decreased apoptosis. Tumors in mice younger than 6 months were classified as adenomas, while the majority of tumors in older mice were adenocarcinomas. A total of 13 tumors were found in 19 one-year-old Muc2^-/- mice; 9 of these were in the small intestine, while 4 were in the large intestine. Rectal tumors, observed in 3 of the 19 Muc2^-/-mice, are a unique feature of this GEM model. It is not yet known whether tumor formation and cellular alterations are a secondary response to insufficient protection and lubrication, or if they are a primary response to altered Muc2 signaling (111).

LKB1 (STK11) is a serine/threonine kinase that is mutated in most persons with Peutz-Jeghers syndrome (PJS). PJS is a hereditary disorder that involves gastrointestinal hamartomatous polyposis associated with mucocutaneous pigmentation. Three groups independently generated Lkb1 knockout mice (112-114). Hamartomatous polyps throughout the gastrointestinal tract-from the stomach to the colon-were observed in all the resulting GEM models. No groups observed progression of the polyps to carcinoma, and only one (114) detected loss of heterozygosity of the wild-type Lkb1 allele.