QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Meuwissen, R. Articles by Berns, A. Search for Related Content PubMed PubMed Citation Articles by Meuwissen, R. Articles by Berns, A. Related Content Cancer and Disease Models Social Bookmarking                   What's this?

REVIEW

Mouse models for human lung cancer

Division of Molecular Genetics and Center of Biomedical Genetics, The Netherlands Cancer Institute, 1066 CX Amsterdam, The Netherlands

	Abstract

Top Abstract Prominent molecular... Activation of oncogenes in... Tumor-suppressor genes... Deregulating growth factors via... Tumor vascularization and... Mouse models for spontaneous... Transgenic murine lung cancer... Histogenesis and ontogeny of... Stem cells in lung... Murine lung tumor models... Future directions Acknowledgments References

 
In recent years several new mouse models for lung cancer have been described. These include models for both non-small-cell lung cancer (NSCLC) and small-cell lung cancer (SCLC). Tumorigenesis in these conditional mouse tumor models can be initiated in adult mice through Cre-recombinase-induced activation of oncogenic mutations in a subset of the cells. They present a marked improvement over mouse models that depend on carcinogen induction of tumors. These models permit us to study the consecutive steps involved in initiation and progression and allow us to address questions like the cell of origin, and the role of cancer stem cells in the maintenance of these tumors. They now need to be validated as suitable preclinical models for intervention studies in which questions with respect to therapy response and resistance can be addressed.

[Keywords: Human; lung cancer; mouse models; NSCLC; SCLC; tumorigenesis]

Lung cancer can be divided into two major histopathological groups: non-small-cell lung cancer (NSCLC) (Van Zandwijk et al. 1995) and small-cell lung cancer (SCLC) (Schiller 2001). About 80% of lung cancers are NSCLC, and they are subdivided into adenocarcinomas, squamous cell, bronchioalveolar, and large-cell carcinomas (Travis 2002). Squamous cell carcinomas and adenocarcinomas are the most prominent. The remaining 20% of lung cancers show properties of neuroendocrine cells. These neuroendocrine lung tumors can be divided into four subgroups based upon their morphological characteristics (Wistuba et al. 2001). SCLC, which accounts for close to 18% of all lung tumors, and large-cell neuroendocrine carcinomas both have a very high proliferative and metastatic potential. The remaining neuroendocrine tumors consist of low- and intermediategrade typical and atypical carcinoids, respectively. The high-grade tumors have a significantly worse prognosis compared to the relative benign carcinoids.

SCLC and NSCLC show major differences in histopathologic characteristics that can be explained by the distinct patterns of genetic lesions found in both tumor classes (Zochbauer-Muller et al. 2002). Responsiveness to treatment with chemotherapy and/or radiation also differs significantly between NSCLC and SCLC and has a dramatic effect on clinical treatment outcome. The overall 5-yr survival rate for lung cancer is 14% (Travis et al. 1995); for SCLC alone it is even worse, 5% (Worden and Kalemkerian 2000).

Spontaneous lung tumors in mice are similar in morphology, histopathology, and molecular characteristics to human adenocarcinomas. Mouse models for lung cancer can thus serve as a valuable tool not only for understanding the basic lung tumor biology but also for the development and validation of new tumor intervention strategies as well as for the identification of markers for early diagnosis.

To meet these goals, the various mouse lung tumor models should each resemble the different human lung cancer types with respect to both critical genetic alterations and tumor cell characteristics. Obviously, it is important to compare the genetic lesions found in lung tumors of man and mouse (Festing et al. 1998; Sargent et al. 2002). Similar genotype-phenotype correlations in murine versus human lung cancer would emphasize the general relevance of these genetic alterations in lung cancer.

A range of mouse models for lung cancer have been described. These include spontaneous models of murine lung cancer as observed in susceptible strains, models in which tumors are induced by carcinogens, and transgenic and knockout models in which lung tumors arise as a result of distinct introduced genetic lesions. This latter approach has recently been improved substantially by the generation of mouse strains carrying conditional oncogenes and tumor-suppressor genes allowing somatic induction of these mutations in a locotemporal fashion, thereby closely mimicking the sporadic character of human lung cancer. A detailed knowledge of the recurrent genetic lesions in human lung cancer is thus a prerequisite for the proper design of mouse models for lung cancer. Therefore, we first describe molecular abnormalities in human lung cancer before we focus on the various mouse models for lung cancer and discuss their use for basic research purposes as well as preclinical tumor intervention studies.

	Prominent molecular abnormalities in lung cancer

Top Abstract Prominent molecular... Activation of oncogenes in... Tumor-suppressor genes... Deregulating growth factors via... Tumor vascularization and... Mouse models for spontaneous... Transgenic murine lung cancer... Histogenesis and ontogeny of... Stem cells in lung... Murine lung tumor models... Future directions Acknowledgments References

 
Many genetic abnormalities have been identified in human lung cancer (Girard et al. 2000). Genetic aberrations can already be found in clonal lesions of preneoplastic bronchial epithelium damaged by smoking (Park et al. 1999; Girard et al. 2000). However, some lesions can also be found within normal lung tissue from smokers (Hussain et al. 2001). In NSCLC, in particular, in squamous cell carcinomas (Wistuba and Gazdar 2003), tumor onset and progression proceed via morphologically distinct lesions: hyperplasia, metaplasia, dysplasia, carcinoma in situ, and fully invasive tumors. Microdissections of these well-defined lesions from the same patient showed a sequential loss of heterozygosity (LOH), first in chromosome 3p, followed by 9p, 8p, 17p (including Tp53 mutations), 5q, and finally RAS mutations (Wistuba et al. 2002). Other precursor lesions such as atypical adenomatous hyperplasia progressing to adenocarcinomas and neuroendocrine hyperplasia leading to carcinoids show similar lesions (Hung et al. 1995; Kishimoto et al. 1995). However, contrary to squamous cell carcinoma, RAS mutations are not late events in adenocarcinoma development since they can already be found in atypical adenomatous hyperplasia (Westra et al. 1993; Kitamura et al. 1999). Early events such as LOH on chromosome 3p can already be found in multiple small clonal and subclonal patches of morphological normal or slightly abnormal bronchial epithelium of smokers (Smith et al. 1996; Park et al. 1999). This chromosome 3p LOH is one of the most prominent lesions in lung cancer as it is observed in 50% of adenocarcinomas and even in >90% of the SCLC and squamous cell lung cancers (Wistuba et al. 2001).

However, genetic alterations do not only occur at the chromosomal level as large deletions or amplifications but also through nucleotide mutations and epigenetic changes. Some of the polymorphisms in alleles at loci of carcinogen-activating and -detoxifying enzymes such as cytochrome P450, glutathione S-transferase, p53, and DNA repair proteins have been found in human populations (Bouchardy et al. 2001; Kiyohara et al. 2002). These genetic polymorphisms are believed to increase susceptibility to lung cancer after tobacco exposure (Husgafvel-Pursiainen 2004). Mounting evidence shows that familial predisposition might play a role in inherited susceptibility to lung cancer (Jonsson et al. 2004). A recent mapping of a human lung cancer susceptibility locus at chromosome 6q23 confirms this (Bailey-Wilson et al. 2004).

In this respect, lung tumorigenesis conforms to the multistep model of tumorigenesis (Hanahan and Weinberg 2000). Activation of oncogenes and inactivation of tumor-suppressor genes are the events underlying this process, and the pattern of genetic alterations found in NSCLC versus SCLC shows both substantial overlap as well as differences (Zochbauer-Muller et al. 2002). An overview of prominent genetic aberrations found in both NSCLC as well as SCLC is presented in Table 1.

	Activation of oncogenes in human lung cancer

Top Abstract Prominent molecular... Activation of oncogenes in... Tumor-suppressor genes... Deregulating growth factors via... Tumor vascularization and... Mouse models for spontaneous... Transgenic murine lung cancer... Histogenesis and ontogeny of... Stem cells in lung... Murine lung tumor models... Future directions Acknowledgments References

 
The RAS genes (HRAS, KRAS, and NRAS) encode GTPase proteins that play a role in transducing growth-promoting and survival signals from membrane-bound RTKs. Hydrolysis of bound GTP to GDP abrogates RAS signaling, but oncogenic mutations in RAS impair GTP hydrolysis, causing persistent signaling. The RAS oncogenes acquire their transforming capacity by point mutations that are detected in 20%-30% of lung adenocarcinomas and 20% of all NSCLCs (Slebos et al. 1990; Rodenhuis and Slebos 1992). Most point mutations are G-T transversions and are correlated with smoking (Slebos et al. 1990). The mutations are found most frequently in codon 12, followed by mutations in codons 13 and 61 (Rodenhuis et al. 1988). Ninety percent of the mutations are found in KRAS in lung adenocarcinomas, whereas no RAS mutations have been detected in SCLC. The presence of KRAS mutations marks a poor prognosis for both early- and late-stage NSCLC (Van Zandwijk et al. 1995; Graziano et al. 1999).

The MYC proto-oncogenes, MYCL, MYCN, and CMYC, encode basic helix-loop-helix transcription factors that regulate the expression of genes involved in DNA synthesis, RNA metabolism, and cell cycle regulation (Oster et al. 2002). Activation of MYC genes occurs by amplification or loss of transcriptional control, which results in MYC protein overexpression. In SCLC MYCN, MYCL or CMYC are often amplified and aberrantly expressed, whereas in NSCLC exclusively CMYC is found affected and only in a fraction of the tumors. MYC amplification occurs in 15%-30% SCLCs and 5%-10% NSCLCs (Richardson and Johnson 1993). Increased MYC mRNA levels were detected in 36% of SCLC cell lines derived from resected metastatic tumors from patients with a relapse after chemotherapy. MYC amplification could therefore be indicative for poor prognosis (Johnson et al. 1996).

A member of the NOTCH gene family, NOTCH-3, was found to be overexpressed in NSCLCs after chromosome 19p translocation (Dang et al. 2000). Notch-3 is involved in differentiation and neoplasia (Campese et al. 2003) and likely influences differentiation of lung cancer cells (Dang et al. 2003).

Finally, overexpression of the proto-oncogene BCL-2 is often found in lung cancer (Pezzella et al. 1993; Kaiser et al. 1996). BCL-2 is an antiapoptotic protein and is expressed in 75%-95% of SCLCs (Jiang et al. 1995), whereas it is expressed in 25%-30% of the squamous cell carcinomas and 10% of adenocarcinomas (Pezzella et al. 1993). BCL-2 counteracts BAX, a proapoptotic protein and a downstream target of p53. High BCL-2 and low BAX expression are frequently found in SCLCs that are p53-deficient (Brambilla et al. 1996). Interestingly, SCLCs with high BCL-2 expression levels are mostly very sensitive to chemotherapy. Moreover, BCL-2 expression in NSCLC is believed to be a favorable prognostic factor, while BCL-2 expression does not influence survival in SCLCs (Maitra et al. 1999; Martin et al. 2003).

	Tumor-suppressor genes frequently inactivated in lung cancer

Top Abstract Prominent molecular... Activation of oncogenes in... Tumor-suppressor genes... Deregulating growth factors via... Tumor vascularization and... Mouse models for spontaneous... Transgenic murine lung cancer... Histogenesis and ontogeny of... Stem cells in lung... Murine lung tumor models... Future directions Acknowledgments References

 
One of the most commonly found aberrations is mutation or deletion of Tp53. p53 is critical for maintaining genomic integrity after DNA damage inflicted by and UV irradiation or carcinogen exposure (Khanna and Jackson 2001; Hanawalt et al. 2003). Cellular stress such as DNA damage or hypoxia causes up-regulation of p53, which then acts as a sequence-specific transcription factor driving expression of a range of genes such as p21, controlling G1/S cell cycle transition, or GADD45, involved in the G2/M DNA damage checkpoint. Apoptosis can be induced through p53 by activating BAX, PERP (Ihrie et al. 2003), and other genes (Mori et al. 2004). Loss of p53 function in lung cancer often occurs through mis-sense mutations and rarely by deletions and is found in 75% and 50% of SCLCs and NSCLCs, respectively (Toyooka et al. 2003a). The Tp53 mutations in lung tumors, mostly G-T transversions, carry the hallmark of smoking (Lewis and Parry 2004).

Normally, expression levels of p53 are kept low through an autoregulatory feedback loop with MDM2, which itself is transcriptionally controlled by p53. MDM2-p53 binding enhances the proteasome-dependent degradation and therefore keeps p53 levels in check. Overexpression of MDM2 is found in 25% of NSCLCs (Higashiyama et al. 1997).

The p16^INK4A-cyclin D1-CDK4-RB pathway is critical in controlling the G1/S cell cycle transition, and one of its components is invariably mutated or functionally altered in lung cancer. Allelic loss, mutations, or promoter hypermethylation of p16^INK4A occur frequently in NSCLC but rarely in SCLC (Fong et al. 2003). Up to 30%-50% of primary NSCLC does not express p16^INK4A. p16^INK4A functions by binding to cyclin-dependent protein kinase 4 (CDK4), which inhibits the ability of CDK4 to interact with cyclin D1. The cyclin D1-associated CDK4 phosphorylates RB, thereby releasing the cell from RB-mediated cell cycle arrest (Malumbres et al. 2003). CDK4 as well as cyclin D1 overexpression have been found in NSCLCs (Borczuk et al. 2003; Ratschiller et al. 2003) and is correlated with a poor prognosis (Caputi et al. 1997). The key component of this pathway, the RB gene, can be inactivated by point mutations, alternative splicing, or deletions. Abnormalities in the RB protein have been found in >90% of SCLCs and 15%-30% of NSCLCs (Reissmann et al. 1993; Dosaka-Akita et al. 1997). As would be expected from proteins acting in the same pathway, mutations of both RB and p16^INK4A are rarely found in the same lung tumor. Interestingly, in spite of the mutual exclusiveness of mutations in cyclin D1, CDK4/6, p16^INK4A, and RB, alterations in p16^INK4A-cyclin D1-CDK4 are most commonly seen in NSCLC, whereas RB gene inactivation is a typical feature for SCLCs (Zochbauer-Muller et al. 2002).

Of the other members of the RB family, p107 and pRB2/p130, only pRB2/p130 is found mutated or lowly expressed in both NSCLCs (Claudio et al. 2000) and SCLCs (Helin et al. 1997).

The alternative reading frame product p14^ARF encoded by the p16^INK4A locus, was inactivated in 65% of SCLCs (Gazzeri et al. 1998), whereas p14^ARF mutations were found in 20% of NSCLCs (Nicholson et al. 2001). Since p14^ARF interacts with MDM2 and thus prevents p53 degradation, it is an integral member of the p53-MDM2-p14^ARF pathway (Fong et al. 2003). There remains, however, a possibility that p14^ARF acts also through a yet unknown pathway (Weber et al. 2000) in lung cancer since loss of p14^ARF can be found independent from p16^INK4A and concurrent with p53 mutations in both NSCLC and SCLC (Gazzeri et al. 1998; Nicholson et al. 2001). Also other tumor-suppressor genes are of importance as is evident from the recurrent chromosomal losses. Several candidate tumor suppressors have been identified in the chromosome 3p region that so frequently shows LOH in lung cancer (Wistuba et al. 2001). One candidate is FHIT in region 3p14.2, showing aberrant transcripts in 80% of SCLC and 40% of NSCLCs, while no FHIT protein is seen in 50% of all lung cancers (Sozzi et al. 1996; Zochbauer-Muller et al. 2001). Other candidate tumor-suppressor genes from the 3p region include RASSF1 (Dammann et al. 2000; Burbee et al. 2001), SEMA3B (Sekido et al. 1996), FUS1 (Kondo et al. 2001), and RAR (Virmani et al. 2000).

An alternative way for inactivating tumor-suppressor genes in lung cancer is hypermethylation of promoter regions resulting in transcriptional inactivation of one allele while the remaining allele is lost via LOH. This epigenetic inactivation is often found in both NSCLC and SCLC (Zochbauer-Muller et al. 2002) but can also be detected in early preneoplastic lesions of smokers. Methylated promoter regions of the individual genes TIMP-3, p16^INK4A, p14^ARF, CDH13 (H-Cadherin), CDH1 (E-Cadherin), DAPK, GSTP1 (for review, see Zochbauer-Muller et al. 2002), and the genes of the chromosome 3p region (RASSF1, SEMA3B, RAR, and FHIT) have been reported. Several regional hypermethylation spots at chromosomal regions 4q, 10q, and 17p are present in both NSCLC and SCLC, but so far no adjacent candidate tumor-suppressor genes have been identified in these regions (Fong et al. 2003).

	Deregulating growth factors via autocrine/paracrine loops

Top Abstract Prominent molecular... Activation of oncogenes in... Tumor-suppressor genes... Deregulating growth factors via... Tumor vascularization and... Mouse models for spontaneous... Transgenic murine lung cancer... Histogenesis and ontogeny of... Stem cells in lung... Murine lung tumor models... Future directions Acknowledgments References

 
Multiple growth-promoting and inhibitory growth factors and their receptors are expressed by lung cancer cells and their neighboring normal cells (Maulik et al. 2003). The resulting autocrine and paracrine regulatory loops can both stimulate or impair tumor growth. Overexpression of one or more growth factor receptors in tumor cells is often associated with poor prognosis. One very common autocrine loop in especially SCLC is the gastrin-releasing or other bombesin-like peptides (GRP/BN) and their G-protein-coupled receptors (GPCR) (Fathi et al. 1996). These activated GPCRs modulate proliferative signals leading to mitogenic responses in various cell types (Rozengurt 1998). Along with GRP, there are other neuropeptides such as cholecystokin (Reubi et al. 1997), gastrin, neurotensin, bradykinin, and vasopressin that confer proliferative signaling in SCLC (Rozengurt 1998). Immunohistochemical analysis showed that a substantial fraction of SCLC and NSCLC express GRP/BN receptors and GRP/BN peptides, although different studies vary with respect to the extent in which these components are expressed in SCLC and NSCLC (Johnson and Kelley 1993; Fathi et al. 1996; Reubi et al. 1997). No mutations or amplifications of the GRP/BN or the GPCR genes have been found, and the mechanisms underlying activation of this growth-stimulatory pathway remain unknown (Forgacs et al. 2001). Blocking this autocrine loop with monoclonal antibodies against GRP/BN or bombesin antagonists caused an in vitro and in vivo growth arrest of SCLC cells (Halmos and Schally 1997).

Another autocrine loop encompasses signaling through receptor tyrosine kinases (RTKs) such as the neuregelin receptor Erbb-2, which is aberrantly expressed in 30% of NSCLC (mainly adenocarcinomas) (Rachwal et al. 1995; Zochbauer-Muller et al. 2002) but not expressed in SCLC. The use of monoclonal antibodies against ERBB-2 resulted in in vitro growth inhibition of ERBB-2-expressing NSCLC cell lines (Kern et al. 1993). ERBB-1 or epidermal growth factor receptor (EGFR) is overexpressed together with its ligands EGF or TGF- in 13% of NSCLCs (Reissmann et al. 1999). In addition to neutralizing antibodies against ERBB-1, specifically designed tyrosine kinase-inhibiting drugs such as gefitinib (Ciardiello et al. 2000) also inhibit tumor growth of ERBB1-overexpressing lung cancer cell lines.

The hepatocyte growth factor (HGF) does induce proliferation and morphological differentiation of lung epithelial cells. HGF overexpression is found in NSCLCs but not in SCLCs (Harvey et al. 1996; Olivero et al. 1996). The HGF receptor, c-Met, however, is expressed in both NSCLC and SCLC.

Other autocrine RTK loops found in both NSCLC and SCLC are the insulin-like growth factors IGF-1 and IGF-2 with their receptors (Quinn et al. 1996; LeRoith and Roberts 2003).

Also, the c-KIT receptor and its ligand are highly expressed and provide autocrine growth in many SCLCs (Krystal et al. 1996). C-KIT overexpression serves as an important negative prognostic factor (Potti et al. 2003) but is seen far less often in NSCLCs (Pietsch et al. 1998).

	Tumor vascularization and metastasis

Top Abstract Prominent molecular... Activation of oncogenes in... Tumor-suppressor genes... Deregulating growth factors via... Tumor vascularization and... Mouse models for spontaneous... Transgenic murine lung cancer... Histogenesis and ontogeny of... Stem cells in lung... Murine lung tumor models... Future directions Acknowledgments References

 
Angiogenesis is an important step in both tumor growth and metastasis (Onn et al. 2003). Tumor cells and their surrounding stromal cells can induce neovascularization by the secretion of factors such as vascular endothelial growth factor, VEGF. VEGF regulates both angiogenesis and vascular permeability (Yano et al. 2003). VEGF is expressed in >50% of NSCLCs and is correlated with an increase in intratumoral microvascular density and poor prognosis (Masuya et al. 2001). Expression of VEGF by lung cancer cells is correlated with loss of p53 function (Niklinska et al. 2001). IL-8 is a strong angiogenic factor belonging to the CXC chemokine family, and its expression is seen in 50% of NSCLCs and is, like VEGF, correlated with intratumoral microvessel density (Masuya et al. 2001). Other angiogenic factors such as the platelet-derived endothelial growth factor PD-ECGF are expressed in 30%-40% of adenocarcinomas, adeno squamous carcinomas, and squamous cell carcinomas (Giatromanolaki et al. 1997). Contrary to VEGF, PD-ECGF and IL-8 are expressed at very low levels in SCLC (Yatsunami et al. 1997; Yamashita et al. 1999). Fifty percent to 70% of NSCLC express bFGF, which is linked with poor prognosis (Takanami et al. 1996; Junker 2001). Finally, the family of metalloproteinases (MMP) and their inhibitors (MMPIs) play an important role in metastasis and the promotion of tumor-related angiogenesis (Chambers and Matrisian 1997; Nelson et al. 2000). However, MMP expression in NSCLC as well as SCLC is not well documented as there are conflicting reports on the prognostic significance of MMPs in lung cancer (Bonomi 2002).

An important factor for maintaining normal tissue architecture is the E-cadherin-catenin complex (Bremnes et al. 2002). Loss of E-cadherin expression is observed in local lung cancer invasion as well as in regional metastasis and is associated with poor prognosis (Hirata et al. 2001; Kalogeraki et al. 2003). The reduced expression of laminins and integrins is often associated with disrupted interaction of lung tumor cells with extracellular matrix, which could lead to fragmentation of the basement membrane and subsequent invasion of the surrounding stroma. Impaired expression of both protein types is, indeed, observed in lung cancer (Akashi et al. 2001) and serves as a poor prognostic factor (Moriya et al. 2001; Vitolo et al. 2001).

We now proceed to describe the various mouse models for human lung cancer. As we shall see, many of the above-mentioned molecular aberrations were also found in the mouse models, causing an often striking pathobiological resemblance between murine and human lung cancer.

	Mouse models for spontaneous or chemically induced lung tumors

Top Abstract Prominent molecular... Activation of oncogenes in... Tumor-suppressor genes... Deregulating growth factors via... Tumor vascularization and... Mouse models for spontaneous... Transgenic murine lung cancer... Histogenesis and ontogeny of... Stem cells in lung... Murine lung tumor models... Future directions Acknowledgments References

 
The susceptibility to and incidence of spontaneous lung tumors vary largely between mouse-inbred strains. Strains that have a high spontaneous lung tumor incidence are also very responsive to chemical induction of lung tumors, for example, exposure to cigarette smoke, tar, or chemically pure carcinogens (Shimkin and Stoner 1975). A/J and SWR mice are among the most sensitive strains, while others range from intermediate sensitive (O20 and BALB/c) to more resistant (CBA and C3H) to almost complete resistant (C57BL/6 and DBA). A polymorphism in intron 2 of K-Ras found in most susceptible strains (You et al. 1993; Chen et al. 1994) is believed to influence K-Ras expression (Malkinson and You 1994), thereby influencing the sensitivity to lung cancer (You et al. 1992). A candidate for the susceptibility locus Papg1 on chromosome 4 is Cdkn2a, which encodes the tumor suppressor p16^ink4a (Manenti et al. 1997; Zhang et al. 2002). A Cdkn2a polymorphism was found between the intermediate resistant BALB/cJ and susceptible A/J strains (Herzog et al. 1999).

Three pulmonary adenoma susceptibility (PAS) loci have been mapped in recombinant inbred strain crosses derived from susceptible A/J and resistant C57BL/6 strains (Malkinson et al. 1985; Malkinson 1999). One of the candidate genes, Pas-1, was assigned to the distal end of chromosome 6 (Gariboldi et al. 1993; Fijneman et al. 1996). Subsequent genetic mapping analysis did link K-Ras to the Pas-1 locus (Lin et al. 1998). However, K-Ras is not the only candidate for Pas-1. Two other nearby lung tumor susceptibility loci (M. Wang et al. 2003) on chromosome 6 might also contribute to lung tumor susceptibility. At least 12 other Pas loci have been mapped within the mouse genome (Obata et al. 1996; Devereux and Kaplan 1998; Festing et al. 1998; Malkinson 1999). Using recombinant congenic strains for multilocus fine mapping of F2 mice, 30 different loci conferring susceptibility to lung cancer (Sluc) were identified (Fijneman et al. 1998; Tripodis et al. 2001). Most of these Sluc loci are involved in complex genetic interactions influencing lung tumorigenesis (Tripodis et al. 2001; Demant 2003). However, to date none of the Sluc genes has been identified (Demant 2003). The other way around, namely, introduction of human polymorphic susceptibility alleles in a defined mouse model background might directly reveal their relevance for lung cancer susceptibility.

Induction of lung tumors with chemical carcinogens is very reproducible and almost invariably results in pulmonary adenoma and adenocarcinomas (Shimkin and Stoner 1975; Malkinson 1989). Very potent carcinogens are polycyclic aromatic hydrocarbons and nitrosamines derived from tobacco and ethyl carbamate (urethane). However, only two studies have so far reported on mouse models for pulmonary squamous cell carcinoma. In the first, intratracheal intubation of methyl carbamate (MC) was applied (Nettesheim and Hammons 1971), while in the second prolonged topical application of N-nitrosomethyl-bis-chloroethylurea (NMBCU) or N-nitroso-trischloroethylurea (NTCU) was used (Rehm et al. 1991).

After carcinogen treatment there is a transient decrease in the number of proliferating Clara and alveolar type 2 cells. The cell numbers recover and then soon surpass the number of cells of control mice (Shimkin et al. 1969). In an early phase multiple hyperplastic foci in bronchioles and alveoli can be detected (Foley et al. 1991). Many of these foci, but not all, then develop further into adenomas and finally, after several months, into adenocarcinomas with in situ invasiveness. Latency and tumor number depend on susceptibility of the strain and can be increased by transplacental carcinogenesis (Miller et al. 2000). However, benign papillary and solid adenomas are also found that do not further develop into malignant adenocarcinomas. The malignant adenocarcinomas rarely metastasize. Histologic analysis of adenocarcinomas showed an equal distribution between papillary and solid subtypes (Malkinson 2001; Nikitin et al. 2004). Further immunohistochemical examination with antibodies against SP-C and/or SP-B and CC10 to identify alveolar type II cells (SP) and Clara cells (CC10) showed only staining for alveolar type II cells. It is not known, however, if murine papillary adenocarcinomas or adenomas arise from alveolar type II cells or expression of CC10 is down-regulated during tumor progression (Malkinson 1991) as is the case in human adenocarcinomas (Szabo et al. 1998).

Molecular characterization of spontaneous and carcinogen-induced tumors revealed various genetic alterations of which activating mutations of K-Ras (Chen et al. 1993; Li et al. 1994) is a prominent early event already detectable in hyperplastic lesions (Horio et al. 1996). Besides frequent overexpression of c-MYC (Re et al. 1992), well-known tumor-suppressor genes like Trp53, APC, Rb, Mcc, Cdkn2a, and Fhit are often inactivated (Malkinson 2001). Interestingly, methylation of CpG islands in the promoter region of Cdkn2a was found in early hyperplasias, while deletions were only found in adenomas and adenocarcinomas (Belinsky et al. 1996). Mutations in Trp53 were never found in hyperplasias but were frequently detected in adenocarcinomas (Horio et al. 1996). This indicates that tumor-suppressor gene inactivation often occurs late in chemically induced mouse lung tumorigenesis (Malkinson 2001).

An even better recapitulation of human lung cancer without the need of applying random mutagenesis can be achieved by the design and use of transgenic mouse models. Furthermore, crossing these models with suitable resistant strains permits a systematic search for modifiers of tumor susceptibility or tumor phenotype.

	Transgenic murine lung cancer models

Top Abstract Prominent molecular... Activation of oncogenes in... Tumor-suppressor genes... Deregulating growth factors via... Tumor vascularization and... Mouse models for spontaneous... Transgenic murine lung cancer... Histogenesis and ontogeny of... Stem cells in lung... Murine lung tumor models... Future directions Acknowledgments References

 
Introduction of genetic lesions found in human lung cancer into the mouse germline or pulmonary tissue has resulted in murine lung tumors that resemble in many aspects lung tumors found in man. Various transgenic models have been developed in which oncogene expression is targeted to a specific subset of lung epithelial cells, thus allowing us to examine the role of these oncogenes in lung tumor initiation and progression (Table 2). Papillomavirus uses two viral proteins, E6 and E7, to sequester the host cell's p53 and Rb, respectively. The SV40 virus, on the other hand, uses a single dual-purpose protein called large T antigen, for the same purpose. The E6 protein binding leads to ubiquitylation of its p53 partner, inducing p53 proteolysis. Mice expressing SV40 large T antigen (Tag) from a Clara cell specific CC10 (or CCSP) promoter or alveolar type II cell specific promoter develop early multifocal bronchioloalveolar hyperplasias followed by mixed solid and papillary adenocarcinomas from which the mice succumb by 4-5 mo of age (DeMayo et al. 1991; Wikenheiser et al. 1992; Sandmoller et al. 1994). Ectopic expression of a HPV-16 E6/E7 transgene under the control of the keratin-5 promoter resulted in lung adenocarcinomas after roughly 6 mo (Carraresi et al. 2001). Mice in which c-Myc expression is driven from an SP-C promoter also developed multifocal bronchio-alveolar adenomas and finally adenocarcinomas (Ehrhardt et al. 2001). Tumor penetrance is, however, incomplete, and no metastases were observed. The same study showed that SP-C-driven expression of the secretable form of EGF, IgEGF, gives rise to alveolar hyperplasias, again with incomplete penetrance. However, bitransgenic mice expressing both c-Myc and IgEGF developed bronchioloalveolar adenocarcinomas with a reduced latency period, suggesting cooperation between c-Myc and EGF in tumor progression (Ehrhardt et al. 2001). When c-Myc was placed under control of the CC10 promoter, only bronchioloalveolar hyperplasias were seen that originated from Clara cells (Geick et al. 2001).

Germline deletion of prominent tumor-suppressor genes involved in lung tumorigenesis often resulted in embryonal or perinatal lethality. When mice showed a longer life span, the tumor spectrum was often broad with only a small fraction developing lung tumors. Mice homozygous null for p19^ARF (Kamijo et al. 1999), p16^INK4A (Serrano 2000), and Trp53 (Donehower et al. 1992) rarely developed adenocarcinomas. However, one has to bear in mind that complete null alleles for Trp53 are seldom found in human lung cancer. Genetic engineering into endogenous p53 of two point mutations that are commonly found in Li-Fraumeni patients imposed to these p53^R270H/+ and p53^R172H/+ (p53^M/+) mice a very different tumor spectrum as compared to conventional p53^+/- mice (Olive et al. 2004). The relative frequency of epithelial carcinomas and adenomas increased significantly in a p53^M/+ and p53^M/- background, although the mean survival times for p53^M/+ and p53^M/- were similar to those of p53^+/- and p53^-/- mice. Among others, p53^R270H/+ developed lung adenocarcinomas (7/36) with more malignant features such as nuclear atypia, desmoplasia, and even metastasis that are never found in p53^-/- mice. This lung adenocarcinoma phenotype is very reminiscent of human lung adenocarcinoma. Up to 18% of p53^M/- mice developed carcinomas as compared to none in the p53^-/- mice. The spectrum of carcinomas in p53^M/- mice is somewhat more mixed as compared with p53^M/+ mice. These data strongly suggest that Tp53 point mutations found in human cancer show enhanced oncogenicity as compared to conventional Tp53 loss (Lang et al. 2004; Olive et al. 2004). Use of carcinogens can augment lung tumor multiplicity in knockout mice with an otherwise normal pulmonary phenotype. To gain insight into the role of transforming growth factor-1 (TGF-1) and the TGF- type II receptor (TGF- RII) as tumor-suppressor genes in lung tumorigenesis, two different models have been explored. Mice hemizygous for TGF-1 showed an increased incidence of adenocarcinomas compared to their wild-type littermates (Kang et al. 2000; McKenna et al. 2001), and transgenic mice with dominant-negative TGF- RII transgene under control of a MMTV promoter showed an increase in the number of adenocarcinomas after carcinogen treatment (Bottinger et al. 1997).

However, classical transgenic and knockout mice do not recapitulate the events underlying the development of sporadic cancer. Widespread expression of oncogenes or absence of tumor-suppressor genes likely creates a microenvironment that substantially deviates from that in cancer development in which only a small number of mutated cells are surrounded by normal cells (Meuwissen et al. 2001a; Jonkers and Berns 2002). The use of conditional alleles of tumor suppressor and oncogenes has enabled the development of murine lung cancer models, which more closely mimic this sporadic tumorigenic process. In conditional lung cancer models, only a subset of cells acquire mutations in an adult mouse in which lung development has been completed. Cre/loxP technology (Gu et al. 1993; Kuhn et al. 1995) has been used to develop multiple conditional alleles of tumor-suppressor genes as well as oncogenes. In the case of tumor-suppressor genes, loxP sites can flank essential coding exons of the tumor-suppressor gene. Then tissue-specific Cre recombinase expression results in the deletion of the floxed gene element with concomitant inactivation of the tumor-suppressor function. Oncogene activation can be achieved by Cre/LoxP-mediated removal of a stop element preventing expression of an oncogene or by the use of inducible oncogenes, taking advantage of hormone receptor fusions or tetracycline-inducible promoters. So far three strains of mice carrying conditional alleles of oncogenic K-Ras^G12D or K-Ras^G12V containing a floxed transcriptional stop element have been generated (Jackson et al. 2001; Meuwissen et al. 2001b; Guerra et al. 2003). Infection of the lungs with Adeno-Cre virus, a recombinant adenovirus expressing Cre-recombinase, showed 4 wk post-infection onset of adenomatous alveolar hyperplasia, which further developed into adenocarcinomas at 9-12 wk post-infection (Jackson et al. 2001; Meuwissen et al. 2001b). No metastases could be detected, possibly due to fast local tumor progression and concomitant short life span (Meuwissen et al. 2001b). Lung tumor multiplicity could be controlled by the dose of the Adeno-Cre virus. The possibility to define precisely the initiation of tumorigenesis by Adeno-Cre virus administration facilitates the analysis of tumor progression. Moreover, these models mimic sporadic tumor development, as activated K-Ras is present in tumor cells that are surrounded by normal cells.

A different murine lung tumor model based on sporadic K-Ras activation was generated using "hit and run" targeting (Johnson et al. 2001). These mice have a latent allele of oncogenic K-Ras^G12D, which is only expressed after a spontaneous recombination event. In addition to pulmonary adenocarcinomas, these mice also developed skin papillomas and aberrant intestinal crypt foci, which indicates the sensitivity of particular tissues to K-Ras mutations. Systemic activation of a conditional K-Ras^G12D allele gave similar results (Guerra et al. 2003), although with a longer latency (8 mo). Neither of these models showed any evidence for metastatic spread of the tumors. Interestingly, in mice with systemic activation of conditional K-Ras^G12D, only a subset of bronchio-alveolar cells form hyperplasias, and of those only a fraction progress to adenocarcinomas, indicating that only a small subset of cells (e.g., progenitor cells of Clara and alveolar type II cells) respond to a K-Ras mutation and, furthermore, that progression to higher malignancy likely requires additional mutations. In addition, it is very possible that the microenvironment in which the cells reside also plays a decisive role in permitting tumor initiation and progression. The role of K-Ras mutation and its effect on downstream effector pathways is not well understood in murine lung cancer. Recent evidence showed that K-Ras-induced lung tumorigenesis requires Rac1 (J.L. Kissil, M.J. Walmsley, K.M. Haigis, C.F. Bender Kim, A. Sweet-Cordero, M.S. Eckman, D.A. Tuveson, V.L.J. Tybulewicz, and T. Jacks, in prep). Rac1, as a member of the Rho protein family, plays a role in Ras-induced transformation (Sahai and Marshall 2002). Cre-mediated deletion of a conditional Rac1 allele in combination with activating a conditional K-Ras^G12D showed a marked decrease in tumor progression and number of adenocarcinomas in these mice compared to single K-Ras^G12D controls (J.L. Kissil, M.J. Walmsley, K.M. Haigis, C.F. Bender Kim, A. Sweet-Cordero, M.S. Eckman, D.A. Tuveson, V.L.J. Tybulewicz, and T. Jacks, in prep). Another intriguing observation suggested that wild-type K-Ras can function as a tumor suppressor in lung cancer (Zhang et al. 2001). Chemical induction of lung tumors in K-Ras^+/- showed a higher susceptibility as compared to K-Ras wild-type mice. The wild-type K-Ras alleles were efficiently mutated in both mice groups. Moreover, K-Ras^+/+ mice had only one allele mutated but gave a very high percentage of allelic loss of the remaining wild-type K-Ras allele. If, indeed, K-Ras has a dual function as mutated oncogene and loss-of-function tumor-suppressor gene, then this should be confirmed in the somatic K-Ras lung tumor models. If so, exploring the molecular pathways on which the inhibitory effect of wild-type K-Ras works will offer a new exciting way for tumor intervention strategies.

So far two binary transgenic systems have been used as conditional expression systems for generating murine lung cancer models. The first is based on a fusion protein of a mutated ligand-binding domain (LBD) of the human progesterone receptor with the DNA-binding domain of the yeast Gal4 transcription factor and an activation domain of the nuclear factor B p65 protein (p65-AD). The resulting chimeric receptor, GLp65, cannot bind endogenous progesterone, but only progesterone antagonists such as mifepristone. GLp65 binds and activates a minimal UAS_Gal4 promoter only in the presence of mifepristone. By using an SP-C promoter, GLp65 expression was targeted to alveolar type II cells. Combining this SPC-GLp65 transgene with a target fibroblast growth factor FGF-3 transgene under control of the minimal UAS_Gal4 promoter (UAS_Gal4-FGF3) ensured controlled FGF-3 expression in the presence of mifepristone. High induction levels of FGF3 caused diffuse alveolar type II hyperproliferation, whereas low FGF3 levels caused macrophage infiltration. Both phenotypes were completely reversible after mifepristone withdrawal (Zhao et al. 2001).

The other bitransgenic system uses the tetracycline-responsive regulatory expression elements. In this system, a tetracycline-controlled reverse transactivator (rtTA) consisting of a chimeric tetR (from Escherichia coli Tn10) and mammalian transcription factor VP16 transactivating domain serves as an effector. One transgene consists of a tissue-specific promoter controlling rtTA effector transcription. This rtTA binds particularly efficient to the seven tandemly repeated tetO sequences (tetO₇) placed in front of a minimal CMV promoter that drives a target gene of choice. However, rtTA binding to a tetO₇ promoter only occurs in the presence of doxycycline, a tetracycline-like antibiotic. Thus, specific gene expression in these bitransgenic mice can be switched on or off by administration and withdrawal of doxycycline (Gossen and Bujard 1992).

So far both CCSP (CC10)-rtTA and SP-C-rtTA transgenic mice have been generated (Perl et al. 2002), directing doxycycline or "tet"-responsive expression to Clara or alveolar type II cells, respectively.

This enabled the generation of bitransgenic CCSP-rtTA;(tetO₇)CMV-FGF-7 mice that, after post-natal doxycycline administration, developed epithelial cell hyperplasia, adenomatous hyperplasia, and pulmonary infiltration with mononuclear cells. Epithelial cell hyperplasia caused by FGF-7 was largely resolved after removal of doxycycline (Tichelaar et al. 2000).

When a CCSP-rtTA;(tetO₇)CMV-K-Ras4^G12D bitransgenic mouse was generated (Fisher et al. 2001), the CCSP-rtTA transgene showed only ectopic rtTA expression in alveolar type II cells, but not as expected in Clara cells. CCSP-rtTA;(tetO₇)CMV-K-Ras4^G12D mice developed normally into adulthood and gave rise to focal epithelial hyperplasia already after 1 wk of doxycycline application. Prolonged tetracycline exposure (2 mo) caused multiple adenomas and adenocarcinomas. However, upon doxycycline withdrawal, K-Ras^G12D transcription was reduced to background levels with concomitant apoptotic regression of both early proliferative and late tumor lesions. No tumors could be detected 1 mo after doxycycline withdrawal (Fisher et al. 2001). Similar experiments performed in CCSP-rtTA;(tetO₇)CMV-K-Ras4^G12D mice in a Trp53 or p16^Ink4a/p19^ARF-deficient background resulted in an even faster tumor onset. In Trp53 and p16^Ink4a/p19^ARF-null backgrounds, adenocarcinomas already developed after 1 mo and showed a more malignant phenotype. Interestingly, these adenocarcinomas also regressed after doxycycline withdrawal, illustrating the importance of mutant K-Ras expression for not only tumor initiation but also tumor maintenance (Fisher et al. 2001).

All the above-described murine lung tumor models developed pulmonary adenocarcinomas with limited malignancy albeit with a striking histopathologic similarity to human adenocarcinomas. No metastases have so far been observed in these models.

Which genetic or epigenetic alterations cause NSCLC to metastasize is presently unclear. In order to generate models that show metastatic spread of NSCLC, one might add additional conditional lesions known to contribute to the invasive and metastatic potential of the tumor cells. Alternatively, one might allow the system to progress to metastatic spread to gain access to the lesions that can give rise to this metastatic phenotype in an unbiased fashion. To enable tumors to progress further, it is necessary to reduce tumor multiplicity, for example, by inducing lesions in a highly sporadic fashion, and by promoting the occurrence of additional lesions, for example, by introducing genetic instability or by applying insertional mutagenesis. As a starting point, one might use mice that combine a conditional K-Ras or Myc oncogene with conditional tumor-suppressor genes such as Trp53 or p16^Ink4a/p19^ARF (Vooijs et al. 1998; Jonkers et al. 2001; Krimpenfort et al. 2001; Jonkers and Berns 2002) that are frequently mutated in lung tumors.

Inducing sporadic lung tumors in one or more of these compound mice will likely lead to a further improvement of murine lung tumor models for NSCLC. Currently we do not know if there is a fundamental difference in metastasis propensity between human and murine lung cancer. A more close analysis of expression profiles of murine NSCLC is needed to know if they can be compared with late or maybe only early-stage human NSCLC. This will, of course, have a major impact on the utility of murine NSCLC for tumor intervention studies. However, promising more malignant lung cancer phenotypes were recently found for the abovementioned Trp53^M/+ mice (Olive et al. 2004). It remains very interesting to combine one mutated Trp53 with a conditional Trp53 allele. This combination, with or without an additional K-Ras^G12D, would enable onset of NSCLC in a somatic fashion and holds great promise for malignant and maybe even metastazing murine NSCLC. So far there are no somatic models for squamous cell carcinoma (SCC). Although epithelial cells of the lower tracheal and proximal bronchial tract are infected by Adeno-Cre, no SCCs have been reported in the conditional K-Ras mice. The rapid onset of lung adenocarcinomas might preclude the development of SCC in these mice. Therefore, a more direct targeting of cell-type-specific Cre recombinase expression into the putative cells of origin of SCC would be a first choice.

Contrary to the many different transgenic models resembling human NSCLC, there are only a few models for neuroendocrine lung tumors and so far a single model for SCLC.

Mice carrying a mutated H-Ras under control of a neuroendocrine (NE) specific, calcitonin gene-related protein (CGRP) promoter, developed both NE and non-NE hyperplasias. Subsequently, primarily non-NE tumors resembling adenocarcinomas were found (Sunday et al. 1999).

Two models have been presented so far for murine NE carcinomas. In one model, proneural transcription factor human achaete-scute homolog-1 (hASH-1), which is highly expressed in SCLC (Ball et al. 1993; Bhattacharjee et al. 2001), was constitutively expressed under the CC10 promoter (Linnoila et al. 2000a), and these mice exhibited a rapid and progressive bronchial hyperplasia obliterating the bronchioloalveolar junction with only minimal changes further away in the alveolar septum. This bronchiolization via hyperplastic ciliated and non-ciliated cells did not exhibit NE differentiation but remained positive for CC10 (Linnoila et al. 2000a). However, when crossed with a transgene CC10-SV40 large T antigen (CC10-Tag), the bitransgenic CC10-hASH1;CC10-Tag mice developed already after 3 mo progressive NE dysplasias and aggressive lung adenocarcinomas with both focal NE differentiation and CC10 expression (Linnoila et al. 2000a,c). These adenocarcinomas did not resemble human SCLC but were very similar to human non-SCLCs with NE differentiation, NSCLC-NE (Linnoila et al. 1994).

The second model makes use of Cre/lox-based somatic deletion of both conditional alleles for Rb (Vooijs et al. 1998) and Trp53 (Jonkers et al. 2001), respectively. Based on the extremely high degree of Rb and Trp53 inactivation in human SCLC, mice carrying conditional alleles for both tumor suppressors were subjected to intrabronchial Adeno-Cre infection (Meuwissen et al. 2001b) to induce murine SCLC. At 3 mo post-Adeno-Cre infection, multiple foci of neuroendocrine hyperplasia developed throughout the bronchi, and after a median latency of 6 mo, major lung tumor lesions with typical histologic features of SCLC could be detected. Immunohistochemical characterization of these tumors showed that they, indeed, had neuroendocrine features with a striking similarity to human SCLCs. All the major neuroendocrine differentiation markers like synaptophysin, calcitonin gene-related protein, neuron-specific enolase, neuron cell adhesion molecule (NCAM), and mASH-1 were found to be expressed as is the case for human SCLC. Moreover, these murine SCLCs or MSCLCs did readily metastasize to sites that are commonly affected in human SCLC patients: thoracic lymph nodes, liver, brain, adrenal gland, ovary, and bone marrow (Meuwissen et al. 2003). All primary tumors and their metastases had both Rb and Trp53 alleles deleted. No tumor lesions with complete Rb deletion but with a wild-type Trp53 allele were detected in any of the tumors that arose. On the other hand, lesions were found retaining one Rb wild-type allele in a Trp53-null background. These were invariably adenocarcinomas with NSCLC features (Meuwissen et al. 2003).

So development of MSCLC requires complete Rb and Trp53 inactivation, and in the conditional Rb;Trp53 model both SCLC and NSCLC can coexist. Loss of Rb by itself is therefore not sufficient to induce NE neoplasia but might need concomitant suppression of p53-mediated apoptosis for further tumor progression. Interestingly, although an identical technique of Adeno-Cre infection was used, K-Ras activation leads to NSCLC-like tumors. This might indicate that the same target cells develop different tumors along specific pathways depending on the gene modifications that are introduced, for example, K-Ras activation versus Rb/Trp53 inactivation. Alternatively, non-identical target cells might respond differently to specific oncogenic mutations. Using recombinant Adeno-Cre viruses with cell-type specific promoters that drive Cre-expression to NE, Clara, and/or alveolar cells, respectively, might address this.

Another intriguing observation is that early lesions in MSCLC do not automatically progress to full-blown tumors. Mice succumbing from tumors still harbor the small lesions first seen 3 mo after Adeno-Cre induction, suggesting that tumor progression requires additional genetic or epigenetic events that occur at a relatively low frequency. Alternatively, these early lesions might not be the precursors of MSCLC.

Both murine lung tumor models for NSCLC and SCLC share unmistakable characteristics with their cognate counterparts in man. This offers unique opportunities to unravel additional molecular events involved in tumor progression for both NSCLC and SCLC as the different stages of tumor development can now be sampled in a well-defined model system.

	Histogenesis and ontogeny of murine lung cancer

Top Abstract Prominent molecular... Activation of oncogenes in... Tumor-suppressor genes... Deregulating growth factors via... Tumor vascularization and... Mouse models for spontaneous... Transgenic murine lung cancer... Histogenesis and ontogeny of... Stem cells in lung... Murine lung tumor models... Future directions Acknowledgments <