Mouse models of DNA mismatch repair in cancer research

Abstract

Germline mutations in DNA mismatch repair (MMR) genes are the cause of hereditary non-polyposis colorectal cancer/Lynch syndrome (HNPCC/LS) one of the most common cancer predisposition syndromes, and defects in MMR are also prevalent in sporadic colorectal cancers. In the past, the generation and analysis of mouse lines with knockout mutations in all of the known MMR genes has provided insight into how loss of individual MMR genes affects genome stability and contributes to cancer susceptibility. These studies also revealed essential functions for some of the MMR genes in B cell maturation and fertility. In this review, we will provide a brief overview of the cancer predisposition phenotypes of recently developed mouse models with targeted mutations in MutS and MutL homologs (Msh and Mlh, respectively) and their utility as preclinical models. The focus will be on mouse lines with conditional MMR mutations that have allowed more accurate modeling of human cancer syndromes in mice and that together with new technologies in gene targeting, hold great promise for the analysis of MMR-deficient intestinal tumors and other cancers which will drive the development of preventive and therapeutic treatment strategies.

1. Introduction

Since the identification of the mammalian MutS and MutL homolog (MSH and MLH, respectively) gene family and the discovery that mutations in some of these genes underlie Hereditary nonpolyposis colorectal cancer/Lynch syndrome (HNPCC/LS), mouse models with mutations in all known MSH and MLH genes have been generated to study their roles in genome maintenance and tumor suppression. The analysis of these mouse lines has revealed that specific mammalian MSH and MLH family members have different significance for these processes largely consistent with their known roles in the repair of DNA replication errors (RERs) and the cellular DNA damage response (DDR). For example, mouse lines with knockout mutations in Msh2, Msh6, Mlh1 and Pms2 that form the MutSα (Msh2-Msh6) and MutLα (Mlh1-Pms2) complexes and play major roles in the repair of RERs in eukaryotic cells and the MMR-dependent DDR, display strong mutator and cancer predisposition phenotypes [1–8]. In contrast, mouse lines with mutations in Msh3 or Mlh3 that participate in other MMR complexes such as MutSβ (Msh2-Msh3) and MutLγ (Mlh1-Mlh3) display milder cancer phenotypes [9–11]. While mouse lines deficient in Msh4 and Msh5 that form the Msh4-Msh5 complex, together with some of the MutL proteins, play essential roles in meiotic recombination but do not display any obvious cancer phenotypes [12–15] [16] this issue. Overall, the cancer predisposition phenotypes of MMR mutant mouse lines show concordance with the incidence of MutS and MutL gene mutations in HNPCC/LS patients in which MSH2 and MLH1 are the most frequently mutated genes, while mutations in MSH6, PMS2 and MLH3 occur less frequently [17].

The analysis of MutS and MutL mutant mouse models also revealed essential roles for some of the MMR proteins in a number of different biological processes in mammals, including somatic hypermutation and class switch recombination during B cell maturation [18,19] this issue. In addition, the analysis of Msh2 and Msh3 knockout mouse lines revealed that the MutSβ complex facilitates triplet repeat expansion, which is associated with a number of neurological diseases [20] this issue. The roles of MMR proteins in these latter processes are beyond the scope of this article, but are described in more detail in other chapters of this review issue.

In addition to constitutive knockout MMR mouse models, several separation-of-function MutSα and MutLα knock-in mice have been generated (termed Msh2^G674A, Msh2^G674D, Msh6^T1217D, Mlh1^G67R and Pms2^E702K) [21–25]. These knock-in mutations selectively impaired mismatch repair of base substitutions and 1-2 base insertion/deletions but largely left the MMR-dependent DDR intact allowing dissection of the significance of the two MMR functions for tumor suppression. All of the above mentioned knock-in mice displayed increased cancer susceptibility phenotypes highlighting the critical role of mismatch repair of RERs in tumor suppression. Nevertheless, the onset of tumorigenesis was delayed in these mouse models, suggesting that the MMR-dependent-DDR function is important for suppressing the initial stages of tumorigenesis. The analysis of MMR knockout and knock-in mutant mice was described in previous review articles and their phenotypes are summarized in Table 1 [26,27].

Table 1.

MMR knockout and knock-in mouse lines.

Gene		Molecular Defects				Phenotypic Defects		Ref.
Knock-out		MSI	DDR	TRI	SHM/CSR	Cancer	Fertility
MutS	Msh2−/−	+++	+++	+++	+++	+++	-	1,2,3,7
	Msh3−/−	+	-	+++	-	+	-	9,10
	Msh6−/−	+	+++	-	++	++	-	4
	Msh4−/−	-	NA	NA	NA	-	+++	12
	Msh5−/−	-	NA	NA	NA/-	-	+++	13,14
MutL	Mlhl−/−	+++	+++	+++	+++	+++	+++	5,6,8
	Mlh3−/−	++	+++	+++	-	+++	++	11
	Pms1−/−	+	NA	NA	NA	-	++	5
	Pms2−/−	+++	+++	+++	-/+++	+++	+++	5

Knock-in		MSI	DDR	TRI	SHM/CSR	Cancer	Fertility
MutS	Msh2G674A	+++	-	+++	+++/+	+++	-	21
	Msh2G674D	+++	-	NA	NA	+++	-	25
	Msh6T1217D	+++	-	NA	++/−	+++	-	22
MutL	Mlh1G67R	+++	-	NA	−/+++	+++	+++	23
	Pms2E702K	+++	-	NA	−/+++	+++	-	24

Abbreviations: MSI, microsatellite instability; DDR, DNA damage response; TRI, triplet repeat instability; SHM, somatic hypermutation; CSR, class switch recombination.

The loss of MMR in Msh 2 and Mlh1 homozygous mutant mice results in strong cancer predisposition phenotypes that are, however, predominantly characterized by T-cell lymphomas and to a lesser extent by skin tumors and small intestinal adenomas and adenocarcinomas. Therefore, in recent years, a major focus in the development of MMR mutant mouse lines was the establishment of mouse models that more accurately model the cancer phenotype of HNPCC/LS patients and to create preclinical models for the assessment of potential preventative and therapeutic interventions. In this review, we will provide an overview of these efforts with an initial description of the generation of conventional MMR knockout mouse lines that carried mutations in MMR genes and tumor suppressor or oncogenes genes to study the impact on intestinal and other HNPCC/LS-associated tumorigenesis. The main emphasis of this review will be an overview of more recently generated mouse lines with conditional MMR knockout mutations and their use as preclinical models of HNPCC/LS.

2. Mouse lines carrying constitutive MMR and tumor suppressor gene knockout mutations

To develop mouse models that more closely mimic the tumor spectrum seen in HNPCC/LS patients, MMR-deficient mouse lines were intercrossed with mouse lines carrying knockout mutations in some of the tumor suppressor or oncogenes frequently mutated in human colorectal cancers such as APC, TRP53 and KRAS (summarized in Supplemental Table 1). Initially, MMR mutant mice were crossed with mouse lines carrying heterozygous germline mutations in the Apc tumor suppressor gene. The large majority of MSI-positive colorectal cancers in human patients carry somatic mutations in the APC tumor suppressor gene indicating that loss of APC function is critical for the initiation and/or progression of MMR-deficient intestinal tumorigenesis. The combination of homozygous mutations in Msh2, Msh6, Mlh1 and Pms2 with heterozygous Apc mutations limited the tumor development to the intestinal tract, and the tumor incidence correlated with the severity of the MMR defects in the different mice. For example, loss of Msh2 or Mlh1 caused dramatic increases in tumor numbers, while loss of Msh6 or Pms2 caused more moderate increases in intestinal tumor numbers and loss of Msh3 did not increase intestinal tumor numbers [6,28–30]. Similar to human MSI-positive colorectal cancers, MMR-deficiency in these mice was associated with somatic mutations in the remaining wild type Apc allele in the intestinal tumors [3,30]. MMR deficient mice have also been crossed with p53 mutant mice since p53 mutations occur frequently in human MMR-deficient colorectal cancers. Although p53-deficiency resulted in greatly accelerated tumorigenesis, the tumor spectrum in Msh2^−/−; p53^−/− and Msh6^−/−; p53^−/− mice was characterized mainly by T-cell lymphoma [31,32]. Interestingly, loss of Msh3 did not accelerate tumorigenesis in Msh3^−/−; p53^−/− mice; however, it changed the tumor spectrum from lymphoma to sarcoma, indicating that Msh3 is a potent modulator of p53-driven tumorigenesis [33]. The analysis of Msh3^−/− cells further revealed increased chromatid breaks and persistence of γH2AX foci after ionizing radiation indicating a defect in DNA double strand break repair (DSBR). As a consequence, Msh3; p53 mutant tumors displayed increased loss of heterozygosity, elevated genome-wide copy number variation in addition to a moderate MSI phenotype. Overall, these studies indicated that the MutSβ complex is important for tumor suppression due to its roles in both DSBR and MMR, especially in late onset tumors or on predisposing genetic backgrounds with mutations in the p53 gene or other tumor suppressor genes.

Another study tested the effect of oncogenic K-ras mutations on MMR-deficient intestinal tumorigenesis in mice. The conditional expression of a K-ras^V12 transgene in K-ras^V12/Ah-Cre; Msh2^−/− mice showed that oncogenic K-ras cooperates with MMR-deficiency and accelerates tumorigenesis in both the small and large intestine. However, the underlying molecular mechanisms remain unclear [34].

In addition to colorectal cancers HNPCC/LS patients develop endometrial cancers at high frequency. To study the effect of MMR-deficiency on endometrial tumorigenesis Msh2 knockout mice were intercrossed with mice carrying a heterozygous knockout mutation in the E-cadherin gene which is frequently inactivated in human endometrial cancers (Msh2^−/−; Cdh1^+/− mice) [35]. A proportion of Msh2^−/−; Cdh1^+/− mice developed endometroid-like tumors that displayed loss or reduced expression of E-cadherin. Interestingly, the GC-box, a cis-acting regulatory transcription element in the promoter region of the E-cadherin gene was mutated in several tumors in Msh2^−/−; Cdh1^+/− mice, suggesting that a reduction in the promoter activity of the remaining E-cadherin wild type allele accounted for the loss of E-cadherin expression. Similarly, Mlh1^−/− mice that carried a heterozygous Pten^+/− knockout allele displayed accelerated endometrial tumorigenesis compared to Pten^+/− mice that was also associated with the loss of the wild type Pten allele [36]. MMR knockout mouse lines have also been intercrossed with mouse lines carrying mutations in other DNA repair genes to study their interaction in genome maintenance. Although not a topic of this review, these mouse lines are also summarized in Supplemental Table 1.

3. Mouse lines carrying conditional MMR mutations and their use as preclinical models

Most of the MMR mutant mouse lines that were previously developed carried homozygous knockout or knock-in mutations and as a result developed predominantly T and B cell lymphomas. Genetically, these mouse lines are more akin to human patients with constitutional mismatch repair deficiency (CMMR-D) syndrome that carry biallelic germline mutations in the MSH2, MSH6, MLH1 and PMS2 genes [37]. Similar to MMR knockout mouse models, these patients display MMR-deficiency in all tissues and develop early onset cancers mainly including haematological malignancies and/or brain tumors, as well as early-onset colorectal cancers. In addition, although the combination of MMR-deficiency with tumor suppressor or oncogene gene mutations in the compound mutant mice described above led to accelerated tumorigenesis in the target tissues, these mouse models display complete MMR-deficiency in all of their tissues. In contrast, HNPCC/LS patients normally carry heterozygous germline mutations in MSH2 or MLH1 and tumorigenesis is associated with a second mutagenic event in the remaining wildtype MMR allele in intestinal epithelial cells later in life.

In an attempt to create a more faithful mouse model for HNPCC/LS, a conditional knockout mouse model was created in which an Msh2^loxP allele can be tissue-specifically inactivated by Cre-recombinase mediated deletion of a floxed exon 12 [25]. The combination of this Msh2^loxP allele with an intestine specific Villin-Cre recombinase transgene [38] (Villin-Cre; Msh2^loxP/loxP, termed VCMsh2^loxP) led to the loss of Msh2 and MMR deficiency exclusively in the intestinal epithelium. As a consequence, VCMsh2^loxP mice developed 1–2 intestinal adenocarcinomas and/or carcinomas that were pathologically identical to HNPCC/LS tumors at very high penetrance within the first year of life. The VCMsh2^loxP model also provided the possibility to create compound mouse models by combining the Msh2^loxP allele with either a complete loss-of-function Msh2 knockout allele (Msh2⁻) or with an Msh2 separation-of-function allele (Msh2^G674D) to mimic mutations previously found in HNPCC/LS tumors and to study the effects of allelic phasing on intestinal tumorigenesis and drug response in vivo. The combination of these alleles with the Msh2^loxP allele resulted in the development of intestinal tumors in both VCMsh2^loxP/− and VCMsh2^loxP/G674D mice similar to the original VCMsh2^loxP mouse line. However, the response of intestinal tumors to the commonly used anticancer agent FOLFOX (5-Fluorouracil, Leucovorin, Oxaliplatin) differed between the two mouse lines. MRI imaging revealed that in VCMsh2^loxP/− mice, the response of the intestinal tumors to FOLFOX treatment was impaired due to the complete loss of the MMR-dependent DDR function in the intestinal tumor cells. In contrast, in VCMsh2^loxP/G674D mice the presence of the Msh2^G674D allele mediated a significant FOLFOX response that resulted in tumor regression. These results indicate that the MMR-dependent DDR plays an important role in the in vivo response of intestinal tumors to FOLFOX treatment and suggests that the precise evaluation of MMR gene mutations could be helpful in determining more adequate chemotherapeutic treatment strategies for HNPCC/LS patients.

The VCMsh2^loxP mouse line was also employed as an in vivo model to determine the potential of long-term treatment with nonsteroidal anti-inflammatory drugs (NSAIDs), such as acetylsali-cyclic acid (ASA or aspirin) and nitric oxide-donating ASA (NO-ASA), as a chemopreventive approach in HNPCC/LS patients [39]. Previous epidemiological studies and studies in rodents indicated that ASA use may be effective in reducing the incidence of colorectal cancers [40,41] and that the NO moiety in NO-ASA may reduce NSAIDs-associated side effects such as gastropathy [42,43]. Furthermore, studies in cell lines previously showed that both ASA and NO-ASA suppressed the MSI mutator phenotype in MMR-deficient colon cancer cell lines through genetic selection that appears to enhance apoptosis in critically unstable cells [44,45]. The exposure of VCMsh2^loxP mice to ASA delayed but did not prevent intestinal tumorigenesis and moderately extended the life span of the animals. The analysis of ASA-treated VCMsh2^loxP mice further suggested that the positive effect of ASA on survival is caused by suppression of MSI in the intestinal epithelia, although further studies are required to support this notion. In contrast, while exposure to low doses of NO-ASA also extended life span, exposure to high doses of NO-ASA had a negative effect on survival and actually increased the intestinal tumor incidence and MSI in intestinal epithelial cells. These studies indicate that long-term treatment with ASA may provide a chemopreventive option for HNPCC/LS patients and that treatment with low doses of NO-ASA may help reduce the gastropathy associated with long-term ASA exposure (see also [46] this issue).

The potential of NSAIDs in the chemopreventive treatment of HNPCC/LS patients was also shown in another study employing the VCMsh2^loxP model [47]. In this study, the effect of mesalazine (5-aminosalicyclic acid, 5-ASA), a drug that is structurally related to aspirin and used in the treatment of ulcerative colitis, was studied in VCMsh2^loxP mice. Different to ASA treatment, mesalazine reduced both the incidence and multiplicity of intestinal tumors in VCMsh2^loxP mice. Mesalazine also reduced MSI in the intestinal epithelial cells, but had no effect on cell viability or induction of apoptosis at the concentrations used. These results are consistent with the idea that mesalazine improves replication fidelity in MMR-deficient cancer cells and decelerates carcinogenesis. In the same study thymoquinone, a natural compound from Nigella sativa, was studied and had similar effects on replication fidelity and intestinal tumorigenesis in VCMsh2^loxP mice providing another candidate compound for chemoprevention in HNPCC/LS patients.

While in VCMsh2^loxP mice the Villin-Cre recombinase transgene is expressed during development and adulthood resulting in constitutive MMR-deficiency in the intestinal epithelium, the Msh2^loxP allele can also be employed to model MMR-deficient sporadic colon cancer [48]. To achieve this, the Msh2^loxP allele was combined with an Apc^loxP allele to generate Msh2^loxP/loxP; Apc^loxP/loxP mice and tumorigenesis was initiated by viral infusion of Adenoviral-Cre into the colon by laparotomy. This approach led to the development of multiple colonic adenomas and a small number of advanced adenocarcinomas. This model was then used to study the effect of NVP-BEZ235, a dual PI3K/mTOR inhibitor, on small and large intestinal tumorigenesis. Dysregulation of the PI3K/AKT/mTOR pathway is frequently observed in many human tumor types [49], and this mouse model together with advanced imaging techniques offered the possibility of closely monitoring the histologic and genetic changes in MMR-deficient intestinal tumors before and after drug treatment. The tumors that developed in these mice displayed variable dysregulation of a number of genes relevant to intestinal tumorigenesis, including down-regulation of Pten, upregulation of Akt and pMek among others. Treatment of colonic tumors with NVP-BEZ235 alone or in combination with the MEK inhibitor ADZ4266 resulted in partial tumor regression, however, the response was not seen in highly advanced tumors. Overall, these studies suggested that gene directed treatment of MMR-deficient intestinal tumors with NVP-BEZ235 alone or in combination with ADZ4266 may be effective in a subset of MMR-deficient colorectal cancers and that other genes or pathways that are dysregulated in drug resistant tumors such as PDPK1 (phosphoinositide dependent protein kinase 1) may be useful targets in treating these tumors.

In a subsequent study, another Lynch-like mouse model was created. In this model an Msh2^loxP allele was created in which exons 12 and 13 are flanked by two inverted loxP sites. This loxP configuration leads to the inversion of both exons upon Cre-loxP mediated recombination [50]. This allele (Msh2^flox) was combined with a constitutive Msh2 knockout allele (Msh2⁻) and the Lgr5-EGFP-IRES-CreERT2 transgene [51] to generate Lgr5-CreERT2; Msh2^flox/− mice. In this mouse line the Lgr5 promoter drives the expression of a tamoxifen-responsive Cre-recombinase (creERT2) in self-renewing intestinal stem cells that are located at the crypt base making it possible to induce mosaic Msh2 inactivation in a small number of intestinal crypts in adult mice after tamoxifen injection. As expected, a proportion of tamoxifen treated mice developed Msh2-deficient intestinal adenocarcinomas late in life. In addition, this model was used to study the response to temozolomide (TMZ), a methylating chemotherapeutic agent that is being used in the treatment of gliomas and melanomas. Exposure to TMZ causes the formation of O⁶-methylguanine lesions that if left unrepaired can cause the formation of aberrant base pairs upon replication. These aberrant base pairs are subsequently recognized by MMR complexes, which mediate cell cycle arrest and apoptosis [52,53] this issue. In MMR-deficient cells, this genotoxic response is impaired resulting in the survival of MMR-deficient cells that also display increased mutation rates. The exposure to high doses of TMZ caused a significant expansion of Msh2-deficient crypts in Lgr5-CreERT2; Msh2^flox/− mice consistent with the idea that Msh2-deficient crypts acquire a growth advantage over wildtype crypts. In addition, TMZ exposure accelerated intestinal tumor development that most likely was caused by an increased mutational load. Therefore, the authors concluded that HNPCC/LS patients should not be exposed to TMZ to avoid expansion and selection of highly tumorigenic Msh2-deficient cells.

Besides the intestinal epithelium, other tissues have been targeted using conditional MMR mouse models. For example, the Msh2^loxP mouse line [25] was used to investigate the role of Msh2 in the development of Huntington's disease. Huntington's disease gene (HTT) CAG (HdhQ111) knock-in mice were crossed with Msh2^loxP mice and D9-Cre recombinase transgenic mice to study the role of Msh2 in CAG-triplet repeat expansion specifically in medium-spiny GABA-ergic projection neurons (MSNs) in the striatum that are known to be extremely susceptible to the negative effects of the mutant huntingtin protein [54]. This study demonstrated that Msh2 acts within MSNs as a genetic enhancer of somatic HTT CAG expansions and as a modifier of HTT CAG-dependent phenotypes in mice. Based on these findings, it was suggested that intervening in the expansion process in these neurons could have therapeutic benefit.

Besides Msh2^loxP mice, a mouse line for the conditional inactivation of Mlh1 has been generated in which exon 4 can be deleted by Cre-loxP recombination [55]. Deletion of exon 4 mediated by an EIIa-Cre recombinase transgene [56] in EIIa-Cre; Mlh1^loxP/loxP mice led to MMR-deficiency from the zygote stage onward, resulting in a cancer and infertility phenotype indistinguishable from constitutive Mlh1^−/− knockout mice. To investigate the tumor suppressor function of Mlh1 specifically during T-cell development, the Mlh1^loxP mice were mated with Lck-Cre recombinase transgenic mice [57] to generate Lck-Cre; Mlh1^loxP/loxP mice (Mlh1^{T Δ ex4/T Δ ex4}). In Mlh1^{T Δ ex4/T Δ ex4} mice Mlh1 is lost in thymocytes at the first stages of maturation resulting in a reduced T-cell lymphoma incidence when compared to Mlh1^−/− mice. These results indicated a critical role for Mlh1 in suppressing lymphomagenesis at the very early stages of T-cell development and that loss of MMR at later stages causes only moderate increases in T-cell lymphomagenesis.

4. Future directions

Although constitutive knockout and knock-in mouse lines have been invaluable in determining the functions of individual Msh and Mlh genes in diverse biological processes and evaluating their roles in tumor suppression, the availability of Msh2^loxP and Mlh1^loxP conditional knockout mouse lines enables the analysis of MMR in specific cells or tissues. A summary of previously generated conditional mutant MMR mouse lines is shown in Table 2. The availability of different Cre-recombinase transgenic mouse lines, active in different tissues and/or inducible at different times, has greatly expanded new avenues for studying MMR functions in vivo. Moreover, the use of Cre-recombinase alleles targeting MMR in the stem cell compartment in the intestine or other tissues, similar to the Lgr5-EGFP-IRES-CreERT2 allele, will provide novel insights into the role of MMR in tumor suppression. The Lgr5-EGFP-IRES-CreERT2 transgene allows the identification and isolation of MMR-deficient Lgr5⁺ stem cells not only within the normal intestinal mucosa but also in intestinal adenocarcinomas. The analysis of intestinal tumor cells expressing Lgr5 or other stem cell markers will likely be important to determine the roles these cells play in the response of MMR-deficient cancers to anticancer drugs. In addition, these mouse lines will provide important preclinical models to study the interaction of diet or environmental carcinogens with MMR-deficient stem cells in the development of intestinal and other cancers.

Table 2.

Conditional MMR knockout mouse lines.

Gene	Cre-Transgene	Target Tissue	Observed Phenotype	MSI	Ref.
Msh2loxP	Villin-Cre	Small and large intestine	Small intestinal tumors	+++	25
Msh2loxP	Ella-Cre	All tissues	T-cell lymphomas Small intestinal tumors	na	25
Msh2loxP	Lgr5-CreERT2	Lgr5 expressing stem cells	Small intestinal tumors	na	50
Msh2loxP	Adenoviral-Cre	Cre encoding adenovirus	Large intestinal tumors in combination with ApcloxP allele	na	48
Msh2loxP	D9-Cre	Medium-spiny GABA-ergic projection neurons (MSNs)	Reduced HTT-CAG expansions and modified huntingtin phenotypes	na	54
Mlh1loxP	Ella-Cre	All tissues	T-cell lymphomas Gastrointestinal tumors	+++	55
Mlh1loxP	Lck-Cre	Lck expressing early stage thymocytes	T-cell lymphomas	++	55

An important future goal will be to engineer mouse models that develop MSI-positive colonic tumors similar to those that develop in HNPCC/LS patients. However, similar to other mouse models of intestinal cancer, and as mentioned above, all currently available MMR-deficient mouse models develop tumors predominantly in the small intestine. Although the reasons for this are not well understood, it is likely that specific genetic differences between humans and mice are responsible. For example, colonic tumors in HNPCC/LS patients frequently carry mutations in coding repeat sequences of genes thought to be important for tumorigenesis and the vast majority of these tumors with microsatellite instability carry frameshift mutations in a 10 bp polyadenine repeat tract in the coding region of the TGFβRII gene [58,59]. This repeat is disrupted by a single guanine in the mouse TgfβRII gene making it resistant to RERs that result in frameshift mutations. The “humanization” of such repeats in the mouse genome might present a possible solution to this challenge. CRISPR/Cas9 technology is a relatively new and powerful approach for the manipulation of the mouse genome and allows the rapid introduction of changes into relevant genes in new or already existing mouse models [60]. Another possible genetic approach could be to induce MMR-deficiency specifically in large intestinal stem cells by combining the Msh2^loxP or Mlh1^loxP alleles with Lrig1-CreERT2 transgenic mice, that were recently shown to induce colon cancer in an Apc^loxP mouse model [61].

The development of mouse models that develop HNPCC/LS-like colon cancers will also be essential in analyzing differences that exist in the microbiome between the large and small intestine that likely affect tumorigenesis [62]. For example, a recent study revealed a significant role of the gut microbiota in colonic polyp formation in Msh2^−/−; Apc^min mice [63]. The gut microbes did not induce colon tumorigenesis by the production of mutagens or inflammatory responses, but rather by producing carbohydrate-derived metabolites such as butyrate that was proposed to induce hyper-proliferation of Msh2-deficient colon epithelial cells and increased colon polyp formation in these mice [63].

Finally, another approach to establish “humanized” HNPCC/LS mouse models might be the development of patient-derived tumor xenografts (PDTX) mouse models in which patient-derived tumor tissues are directly transplanted into immunodeficient mice. In contrast to cell line-derived xenografts, PDTX models can retain the histopathologic features and molecular characteristics of the original tumor even after extensive passaging in mice [64]. These PDTX models in combination with the genetic models promise to be highly useful for drug development and biomarker discovery and also provide an important tool for personalized medicine for HNPCC/LS tumors.