Animal Models of Colorectal Cancer: From Spontaneous to Genetically Engineered Models and Their Applications

Colorectal cancer is one of the most common gastrointestinal malignancies in humans, affecting approximately 1.8 million people worldwide. This disease has a major social impact and high treatment costs. Animal models allow us to understand and follow the colon cancer progression; thus, in vivo studies are essential to improve and discover new ways of prevention and treatment. Dietary natural products have been under investigation for better and natural prevention, envisioning to show their potential. This manuscript intends to provide the readers a review of rodent colorectal cancer models available in the literature, highlighting their advantages and disadvantages, as well as their potential in the evaluation of several drugs and natural compounds’ effects on colorectal cancer.


Introduction
Worldwide, colorectal cancer is the third most common cancer in men and second in women [1]. Many risk factors have been considered for the development of colorectal cancer, such as the ingestion of processed meat, alcoholic drinks, body fatness, low intake of vegetables and fruits, smoking, and other concomitant diseases, such as inflammatory bowel disease (IBD), Crohn's disease, and ulcerative colitis [2,3].
Colorectal cancer is characterized by the invasion of neoplastic epithelial cells below the muscularis mucosae of the colorectal wall [4]. Its evolution is slow and characterized by different stages. Progressive changes in the amount or activity of proteins that regulate cell proliferation, differentiation, and cell survival occur, leading to a disorder in cell replication that contributes to the development of proliferative lesions, such as adenoma [5]. Subsequently, the intestinal epithelium undergoes a malignant transformation to invasive carcinoma [4,5]. Besides adenomas, hyperplastic polyps, serrated adenomas, flat adenomas, and dysplastic lesions are also observed in the colon as other types of preneoplastic lesions [5]. In humans, colorectal cancer is histologically classified as an adenocarcinoma [4,6]. In Figure 1 we can observe the progression from normal intestinal epithelium to carcinoma. About 97% of colorectal cancers are spontaneous, and the remaining are due to one of two autosomal dominant inherited diseases: hereditary non-polyposis colorectal cancer (HNPCC) and familial adenomatous polyposis (FAP) [4,5]. The genetic mechanisms of spontaneous CRC are present in the adenoma-carcinoma sequence. Carcinogenesis is initiated with inactivating mutations in the tumor suppressor adenomatous polyposis coli (APC) gene, followed by an accumulation of mutations in the genes K-RAS, PI3K, DCC, SMAD2, SMAD4, and lastly the mutation in the tumor suppressor gene TP53 that determines the progression from the non-invasive to the invasive CRC [7].
Laboratory rodents are commonly used as animal models in experimental research because they are easy and cheap to maintain, their physiology and genetics are well studied, and they are mammals like humans [8]. They allow us to understand and follow the progression of diseases, enable the discovery and development of new preventive strategies, which can be later used in clinical trials. An ideal animal model of human disease should be simple, not expensive, and mimic the disease in terms of morphology, biochemical alterations, and biological behavior [4,9]. Several works have reviewed the use of animal models of CRC [10]. However, this manuscript not only intends to augment the information on rodent models of CRC, highlighting their advantages and disadvantages, but also to review their applications and how they can be used to evaluate natural compounds, nutrition habits, and drugs.

Rat and Mouse Colon and Rectum: Anatomy and Histology
The rat and mouse intestine are similar to that of humans concerning development, structure, and functions [9]. The large intestine comprises the cecum, the colon, the rectum, and anus, and it is responsible for the absorption of water and salt from feces [11] (Figure 2).
The cecum is a curved blind sac responsible for bacterial fermentation and empties into the proximal/right colon. Even though the rodents' colon and rectum represent a percentage of the total size of the large intestine similar to the humans, the cecum is much bigger in rats, which may be attributed to the high fiber content of their diet [12]. The colon continues toward the pyloric region of the stomach and has the same histological structure of the gastrointestinal tract: mucosa, submucosa, inner circular and outer longitudinal tunica muscularis, and serosa [1]. Despite histological similarities, rats and mice do not have adipose tissue in the submucosa, unlike humans who have it in abundance. The colon can be divided into ascending (it leads cranially to the thoracic cavity), transverse colon (from the left to the right side), and descending colon (on the right side of the abdominal cavity). The rodents' middle and distal colon corresponds to the human left colon [13]. The rectum is relatively short and indistinct from the distal colon. The anorectal junction has no stratified columnar epithelium, and the anal canal is lined by keratinized stratified squamous epithelium [11].

Rodents as Models of Colorectal Cancer
Although there is no ideal animal model that replicates all human disease aspects, the rodents are accepted as good models to study colorectal carcinogenesis because of their physiological similarity with humans, reproducible tumor induction, and the possibility to study the disease biopathology and test strategies for cancer prevention and treatment [4]. An ideal rodent model of colorectal cancer should develop carcinomas in the colon and rectum, with a high incidence in a short period, allow non-invasive monitoring of disease progression, and follow the histological and molecular characteristics of human colorectal cancer [8,13]. The models available to study colorectal cancer include spontaneous, induced, genetically engineered, xenograft, and syngeneic models ( Figure 3).

Spontaneous Models
Spontaneous development of colorectal cancer in rats and mice is rare, although some cases were reported in the literature. In 1969 it was reported that C57BL mice developed adenomas in the colon [9], and in 1975 Miyamoto and colleagues showed that 30-40% of animals from the Wistar-Furth/Osaka strain developed adenocarcinomas [14]. More recently, in 2009, Newark and colleagues showed that C57BL/6J developed cancer in the large intestine with an incidence of 1% [15]. These models are not very used due to unpredictability and low reproducibility [4]. In 40% of the spontaneous rat models, the period of latency is approximately eight months [16,17].

Induced Models
Colorectal tumors can be induced in rodents through the administration of chemical carcinogens alone or in combination [5]. There are two types of chemical carcinogenic agents: direct and indirect. Direct carcinogens do not need to be metabolized to induce cancer, while the indirect agents are administered in their inactive form and only acquire carcinogenic activity when biotransformed and converted into their active form in the liver [5,6].

Chemically Induced Models
In 1941, Lorenz and Stewart were the first to induce intestinal mouse tumors by feeding them with dibenzanthracene or methylcholanthrene [4], leading to the development of adenocarcinoma of the small intestine [4]. Later, in 1947, Lisco and colleagues induced carcinomas in the rat colon through feeding with radioactive yttrium [17]. Some years later, in 1963, Laquer and colleagues stated hydrazines are colonic carcinogens. Rats developed adenocarcinomas after feeding with a large quantity of cycad flour, which have hydrazine called cycasin, a form of methylazoxymethanol (MAM) [18,19].
These carcinogens can be indirect-acting agents (DMAB, DMH, AOM, and PhiP), which need an enzymatic reaction to be converted into an active form, or direct-acting carcinogens (MNU and MNNG) that do not need biological catalysis [5].

3,2 -Dimethyl-4-Aminobiphenyl (DMAB)
Walpole and colleagues, in 1952, described the first induction of intestine cancer in male rats by subcutaneous administration of DMAB dissolved in arachis oil, at a mean total dose of 2.8 g/kg, for 141 days [21]. After that, other researchers evaluated the carcinogenic potential of this compound [22]. Of these works, the one made by Reddy et al. [23] stands out; they showed that 30% of F344 rats fed with a low-fat diet and 75% of animals fed with a high-fat diet developed colon cancer after being injected with DMAB (50 mg/kg), once a week, for 20 weeks. DMAB forms carcinogenic DNA adducts through the N-hydroxylation by cytochrome P450, followed by O-acetylation and hydrolysis, reacting with DNA [23]. Nevertheless, this model has some disadvantages because multiple DMAB administrations are needed [23][24][25] and it has low specificity, leading to the development of tumors in various other tissues, such as salivary glands, mammary glands, urinary bladder, ear, and skin [6]. The studies performed using this model may be consulted in Table 1.  Since 1967, after discovering that oral administration of alkylnitrosamide induced adenocarcinomas in the glandular stomach in rats, other works were performed envisioning to address the carcinogenic potential of MNNG and MNU [4,13]. MNU and MNNG are direct DNA alkylating agents; they transfer a methyl group to nucleobases leading to the accumulation of genetic mutations [13]. Intra-rectal instillation of MNNG during 20 weeks at a dose of 1-3 mg/rat/week induced colon cancer in 100% of F344 rats [13,18,20]. Female ICR/Ha Swiss mice instilled with 0.3 mg of MNU intrarectally, three times a week for 10 weeks, developed tumors in the distal colon, rectum, and anus with an incidence of 78% [29]. The intrarectal administration allows a more selective induction in the distal colon and rectum, which is a huge advantage of this model. However, a precise technique is needed, and the quantification of drug volume is difficult [18,20]. In addition, the animals need to be kept in an inverted position for one minute after administration to prevent the return of the compound to the anus [19,20].
This model can be used to evaluate the therapeutic effects of several compounds on colorectal cancer development. More details about colorectal cancer studies using the MNU model to evaluate the influence of diet, drugs, and natural compounds can be consulted in Table 2.

1,2-Dimethylhydrazine (DMH)
DMH is an alkylating agent that needs liver metabolic activation to become a carcinogen. Therefore, DMH is oxidized in the liver into azoxymethane and is then hydroxylated to form methylazoxymethanol (MAM). MAM is converted to formalin and methyldiazonium ion that are responsible for DNA, RNA, and protein alkylation [4,33].
The induction of colon cancer in rats with this compound was described for the first time in 1967 by Druckrey and colleagues, through its subcutaneous administration, at a dose of 21 mg/kg [20,34]. They showed that DMH cancer induction in the distal portion of the colon is histopathologically similar to humans [16,33]. These results were later confirmed by other authors [6,13,19]. DMH can be administered through different routes, including subcutaneous, intraperitoneal, oral, and intrarectal [4]. For example, a subcutaneous injection of 20 mg/kg DMH, once a week, for 20 weeks induces colonic adenomas in about 60% of male F344 rats [21]. Oral administration of 20 mg/kg showed a lower tumor incidence in male Wistar rats, depending on the nature of the diet [35]. Intrarectal administration of 250 mg/kg of DMH in Sprague-Dawley rats induced multiple colorectal adenocarcinomas with a latency period of 34 weeks [36]. Of the routes of administration presented, subcutaneous seems to be the one that leads to high incidence and consequently the most used in chemopreventive studies [4]. More information about other studies with this model may be consulted in Table 3.
Although DMH-induced colon tumors in rodents are similar to human colon tumors [21], this model has disadvantages, e.g., multiple injections of DMH are necessary to induce tumors, it is characterized by at least six months of the latency period, and no hepatic metastases were observed until now ( Figure 4C) [20].

Azoxymethane (AOM)
In 1970, Druckrey and colleagues showed for the first time the ability of azoxymethane (AOM) to induce intestinal tumors. Other works were then published confirming that AOM is a potent inducer of carcinomas of the large intestine in various strains of rats, such as F344, and mice, such as C57BL/6J and SWR/J, among others [21,37,38].
AOM is a metabolite of DMH that has been more frequently used in the induction of colon tumors than DMH, given some of its advantages over the original compound, such as its increased efficacy and greater chemical stability [37]. Like DMH, AOM is also an indirect carcinogenic compound, and it is activated in the liver by N-oxidation through cytochrome P450 2E1, producing metabolites such as methylazoxymethanol and methyl-diazoxide, which induce inflammation [39]. AOM seems to be a more effective carcinogen than DMH because it requires fewer reactions to be activated [6].
AOM induces rodent colon carcinogenesis when administered over 6-8 weeks via subcutaneous or intraperitoneal injection, with a latency period ranging from 20 to 30 weeks [13]. The distribution of tumors developed in the small intestine and colon (predominantly in the distal colon) is similar to that observed in the human colon [21,40]. Histological and histochemical properties of AOM-induced tumors are similar to those described in humans, being classified as adenomas and adenocarcinomas. Using this induction model, it was possible to identify metastases in lymph nodes and the liver similar to those described in humans [6,38].
Details concerning studies using AOM induction model to understand the influence of diet, drugs, or natural compounds in colorectal cancer can be consulted in Table 4.  Reduced tumor incidence when aspirin was administered 1 wk before or after DMH but no effect when administered 4 wks after [51] s.c. 30

Azoxymethane (AOM) and Dextran-Sodium Sulfate Model (DSS)
Because colon cancer is associated with long-standing IBD, such as ulcerative colitis and Crohn's disease, the risk of colorectal cancer development increases with the extent and duration of disease [19]. Chronic and repeated mucosal inflammation may result in tumors through several mechanisms, such as induction of genetic mutations, increased cryptal cell proliferation, changes in crypt cell metabolism, changes in bile acid, and alterations in the bacterial flora [5]. In 2003, Tanaka and colleagues developed a colitis-related mouse model of colorectal cancer initiated with AOM and promoted by dextran-sodium sulfate (DSS) [88]. DSS is an inflammatory compound that causes damages to the epithelial lining of the colon and induces colitis. Using this model, male Crj: Cd-1 (IRC) mice were intraperitoneally injected with AOM (10 mg/kg of body weight) and, one week later, received 2% of DSS in drinking water for seven days. Twenty weeks later, 88% of animals had colonic dysplasia, and the incidence of adenoma and adenocarcinoma was 100% [88]. This model allows the reduction in the number of AOM administrations, avoiding prolonged exposure to this compound, and still allows a reduction in the latency period. After this finding, other researchers have associated DSS with other compounds (PhIP and DMH), with tumor induction in a shorter period than the AOM/DSS model [19,89,90]. Concerning the AOM/DSS-induced model, different mice strains present diverse sensitivity; for example, the incidence and multiplicity of adenocarcinomas appear to be higher in the BALB/c mouse strain [91].
The AOM/DSS model mimetics human colorectal cancer pathogenesis, with a similar location (distal colon) and initiation by a polypoid growth. However, this model has a very low tendency to metastasize, which constitutes a limitation [92]. This model has been used in several chemopreventive studies of colitis-related colon carcinogenesis [88] (Table 5).

2-Amino-1-Methyl-6-Phenylimidazo (4,5-b) Pyridine (PhIP)
The PhIP is a heterocyclic amine isolated from cooked fish and meat, which can be used to induce tumors in the colon, prostate, and mammary gland [93]. After administration, it is rapidly absorbed by the gastrointestinal tract and widely distributed through the body [19]. Then it is bio-transformed by the liver cytochrome P450 s, being converted in the amino group to a hydroxyamino group, which is then activated by forming esters with acetic acid, sulfuric acid, and proline. These esters are responsible to induce carcinogenic DNA adducts and genetic alterations leading to colorectal cancer [94]. The work performed by Ito and colleagues was the first to induce colon tumors in rats with this compound [95]. F344 rats from both sexes were fed with 400 ppm PhIP for up to 52 weeks and presented a high incidence of colon carcinomas [95]. PhIP did not induce colon cancer in mice, it just induced the formation of colonic aberrant crypt foci and lymphomas [4,9]. The mechanisms responsible for the non-induction of cancer in mice are not well understood [94]. This model has been used to evaluate the therapeutic effects of several compounds on colorectal cancer. Detailed information concerning mouse and rat models' studies using PhIP induction can be consulted in Table 6.

Genetically Engineered Models
Genetically engineered models allow the study of genetic predisposition to colorectal cancer development, and its interaction with environmental and modifying factors. These models mimic the genetic alterations of spontaneous and hereditary forms of colorectal cancer [9]. Through the study of hereditary colorectal syndromes, such as FAP and HNPCC, it was possible to discover the mutations under colonic carcinogenesis and replicate genetic lesions in mice and rats by developing genetically engineered models [9].

Adenomatous Polyposis Mouse Models (APC)
It was demonstrated that human colorectal cancer is a multi-step genetic process and that the mutation of the APC gene occurs at the beginning of the carcinogenesis process. The APC gene is responsible for the regulation of β-catenin, cytoskeleton organization, cell cycle regulation, apoptosis, and cell adhesion. When mutated in the germline, this gene is associated with FAP [5,121].
Moser and colleagues discovered that C57BL/6 mice treated with ethylnitrosourea developed a mutation that predisposes to spontaneous development of intestinal cancer, naming the model as ApcMin mouse [122]. The Min mouse model is the only animal model of cancer that contains a single genetic alteration capable of producing a fully penetrating, consistent, and organ-specific tumor phenotype. The adenomas developed rapidly, with lesions identified within 60 days, and high tumor multiplicity. This model allows the study of multiple pathways impacting tumorigenesis and enables numerous entry points for basic or applied studies [10]. Over the years, this model has been tested, improved, characterized, and used to understand the role of the APC gene in colorectal cancer and also in chemopreventive studies [14] (Table 7).

F344-Pirc Rat Model
In 2007, Landgraf and colleagues developed a rat model carrying a knockout allele in the APC gene on an inbred F344/NTac genetic background rat [145]. To distinguish it from the APC mouse, they called this strain Pirc form (polyposis in the rat colon). The Pirc rats developed adenomas similar to those found in humans, showed the same progression to invasive carcinomas, and dependence on gender was observed, with males more prone to develop tumors in the intestinal tract than female rats [145]. In comparison to APC mouse models, this model takes some advantages due to the rat's size, the facility of diagnostic imaging, colonic predisposition, and longevity [10].

Hereditary Nonpolyposis Colon Cancer Mouse Models (HNPCC)
HNPCC is an inherited disease characterized by inactivated DNA mismatch repair genes, such as MLH1, MSH2, MSH6, and PMS2, that leads to the development of a variety of cancers, including colorectal cancer [15]. Mice with a targeted inactivating mutation in the mismatch repair genes are used to study these genes and evaluate their contribution to carcinogenesis [14,146]. Developed tumors are not specific to the colon-rectum; they occur in other organs such as the skin, lung, lymphatic system, stomach, and small intestine [15].

Xenograft and Syngeneic Models
Colorectal cancer cells or grafts of tissue can be implanted into animals to evaluate tumor development and to analyze the effects of several chemicals and natural compounds ( Figure 5). In xenograft models, human tumor cells or tumor fragments are implanted into immunocompromised animals. Syngeneic tumor models are characterized by the use of animal tumor cell lines obtained from chemically induced rodent colorectal cancer and are implanted into animals with the same genetic background as the cell line [13]. These models can be used, among others, to study the effects of treatment on colorectal cancer metastases [4,147,148].
The cells may be implanted subcutaneously, intrasplenically into the renal capsule (heterotopic models), or directly in the colon or rectum (orthotopic models) [13,14]. The subcutaneous inoculation (heterotopic model) is one of the most used methods due to the simplicity, easy access, and high tumor growth [4]; however, the tumor microenvironment is different from the colon [13], and the metastases do not develop [149]. Comparing with the subcutaneous heterotopic model, the intrasplenically and renal capsule heterotopic models and the orthotopic model are technically more advanced and more difficult to work with, requiring the animals' anesthesia and use of imaging modalities (e.g., ultrasonography) to implant the cells specifically in the spleen, the renal capsule or colon, respectively [150].
Orthotopic implantation refers to cells or tumor fragments implanted in the tumor site of origin, i.e., colon or rectum [13]. These models allow replication of tumor invasion, vascular spread, mimic the progression to advanced colorectal cancer in humans, and metastasize to distal organs [12]. For example, MCA-38 cells were intramurally injected into the cecum of C5BL/6J mice, and 40-65% of them developed metastases eight weeks later [151]. In another study, CT26 cells were transanal rectal injected in Balb/c mice with uptake rates of 65%, but only 3.3% developed metastases [152].
These models can be used to evaluate some therapeutic drugs. For example, Tao and colleagues used a commercial human colon cancer cell line, HCT-116, to evaluate the anticolorectal cancer activity of Weichang'an, a Chinese herbal medicine, with 5-fluorouracil. The cells were injected subcutaneously in male BALB/C mice axilla, and after tumor growth, they were transplanted into the cecum. The group concluded that the compounds evaluated inhibited both colon tumor growth and hepatic metastases [153].

Conclusions
Experimental data concerning dietary, drugs, and natural compounds' effects on colon cancer models were reviewed in this work. Although several animal models are available to study colon cancer, there is no perfect model; all constitute an important tool to study human and animal colon carcinogenesis and to evaluate the potential effects of preventive and therapeutic strategies.
Whereas the AOM/DSS model mimetics the pathogenesis observed in human colorectal cancer, others like genetically engineered models allow studying genetic predisposition to the development of this type of cancer. The model selection should consider the studies' goals, the costs, and the advantages and disadvantages of each model, animal, strain, and gender.
Considering dietary patterns and natural products used as chemoprevention or chemotherapy, some like soy isoflavones, β-carotene, dried plums, fuji flavone, and Chinese cabbage showed an inhibitory effect on colorectal carcinogenesis, and adlay bran ethanol extract, grape seed extract, and pomegranate peel extract decreased the development of colonic premalignant lesions. However, groups that studied the effects of wheat bran and heme groups (in form of chicken, beef, black pudding) in the mice diet have concluded the opposite, observing a higher incidence of colorectal carcinogenesis.
In some cases, natural compounds, several drugs, and dietary patterns results are inconsistent and depend on multiple factors, and the best way to obtain better results is to select the most appropriate model and try to reduce most of the external factors. To achieve this goal, more research with controlled parameters is warranted. Moreover, ideally, the studies to evaluate the effects of natural compounds in CRC should not only evaluate the whole compound, but also each active substance in an isolated way. However, these studies imply the use of a higher number of animals, and consequently, higher costs for researchers, which may constitute a limitation.

Conflicts of Interest:
The authors declare no conflict of interest.