Lung Cancer Models
Erica L. Jackson and Tyler Jacks
Howard Hughes Medical Institute
Massachusetts Institute of Technology
Welcome to the MMHCC Lung Cancer Site. As you enter the site you will find an introductory section providing background information on lung cancer and a general overview of current methods for diagnosis and treatment, molecular alterations occuring in lung cancer, and existing murine models of lung cancer. The introductory section will be followed by several sections providing more in depth discussion of the topics listed below.
Introduction
Lung cancer is the leading cause of cancer deaths worldwide, with 169,500 new cases and 157,400 deaths predicted for 2001 in the United States alone (32). The majority of lung cancer cases are related to tobacco use with approximately 10% of lung tumors arising in non-smokers (7). However, the incidence of lung cancer deaths that are not associated with smoking or other environmental factors is increasing at a higher rate than in any other group (54).
Lung cancer patients suffer a high case:fatality ratio with a 5 year survival rate of ~14%. For treatment purposes lung cancer is divided into two histopathologic classes, small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC), which differ in their responses to therapy. Approximately 80% of lung cancer cases are classified as NSCLC, while SCLC accounts for ~18% (80). The two classes of lung cancer are characterized by distinct patterns of oncogene activation, tumor suppressor gene mutation and chromosomal alterations, which may explain their different biologies. Little progress has been made in the treatment of lung cancer over the past 30 years. Since 1970 to the present, the 5 year survival rate has only increased from 7 to 14%.
The development of murine models of lung cancer may aid in our understanding of lung tumor biology and facilitate the development and testing of novel therapeutic approaches and methods for early diagnosis. To this end, mouse models should mimic the genetic alterations found in human lung tumors, the histological characteristics of human tumors or both. To date, several approaches have been taken in the development of murine lung cancer including chemically induced tumors, transgenic strains expressing relevant oncogenes, and strains in which important tumor suppressor genes have been knocked out. More recently, mice carrying latent and conditional alleles of oncogenes and tumor suppressor genes known to be mutated in human lung cancer have been developed. These models more accurately mimic the human situation in which genetic mutation occurs in a subset of cells within adult somatic tissues.
Diagnosis and Treatment
Chest x-ray and sputum cytology are regularly used to detect lung cancer in symptomatic patients. Recent data suggest that low radiation dose spiral CT is capable of detecting early abnormalities in lungs of high risk individuals, but it remains to be seen whether this screening method will result in a reduction in lung cancer mortality (56). The ultimate diagnosis of both SCLC and NSCLC is based on fiber-optic bronchoscopy and histologic analysis of sputum or biopsy samples.
SCLCs are neuroendocrine (NE) tumors of highly aggressive nature. Often the cancer has metastasized to distant organs by the time of diagnosis. The size of the primary lesion and extent of metastasis dictate the treatment regimen. Due to its propensity for metastasis, SCLC is rarely treated by surgical resection (30). Combination chemotherapy regimens that include a platinum agent are the standard of care for most SCLCs, with the PE regimen (cisplatin and etoposide) being the most commonly used in the US. Although most SCLCs are initially highly responsive to therapy, typically the primary tumor or metastasis becomes resistant to chemotherapy and >90% of patients succumb to the disease. For more information on treatment options for SCLC please see: Small Cell Lung Cancer (PDQ?: Treatment (www.cancer.gov)
NSCLCs can be further subclassified into squamous cell carcinoma, adenocarcinoma and large cell carcinoma, with adenocarcinoma being the most common (histology for each subclass will be shown). However, for treatment purposes NSCLC is considered a uniform group of aggressive cancers, and again treatment options are based on the stage of disease at the time of diagnosis. Surgery and radiotherapy are used to treat early-stage disease. For patients with unresectable metastatic disease, platinum based combination chemotherapy is again the treatment of choice. Recent randomized trials demonstrate that as long as the therapy contains a platinum compound (cisplatin or carboplatin) and a modern agent active against NSCLC (paclitaxel, docetaxel, gemcitabine or vinorelbine) the survival benefits are the same (16). The addition of concurrent radiotherapy to the PE regimen may also provide some additional survival benefits. However, most patients become resistant to therapy, relapse and die from the disease. For more information regarding treatment options for NSCLC see: Small Cell Lung Cancer (PDQ?: Treatment Option Overview (www.cancer.gov)
For both SCLC and NSCLC patients the performance status of the individual is considered when choosing a course of treatment. Platinum compounds have poor toxicity profiles and may not be suitable for all patients. Due to the the high rate of relapse and toxic side affects under the standard therapies, many new therapeutic strategies are being developed. Advances in the understanding of the molecular events underlying the development of lung cancer have enabled researchers to develop rationally targeted therapies. These biologic agents specifically target proteins used by cancer cells to promote inappropriate growth and survival. Such agents include selective protein kinase inhibitors, a variety of antisense oligonucletides, and antibodies all used to inhibit the expression or function of growth factor receptors, signal transduction proteins or anti-apoptotic mediators. The efficacy of such agents is still being evaluated but targeted therapeutics may provide treatment options with increased efficacy and decreased toxicity. Genetically modified murine lung cancer models may provide a useful reagent for pre-clinical testing of therapeutics directed against the specific molecular lesions driving tumorigenesis in these mice.
Molecular Characterization of Human Lung Cancer Overview
(For a more detailed description see the Human Lung Cancer Molecular Alterations section )
Lung cancer arises as the result of numerous genetic lesions often caused by exposure to cigarette smoke or other environmental carcinogens. It is thought that 10 or more genetic or epigenetic abnormalities must occur before a lung tumor becomes clinically evident (63). Genetic alterations can occur at the chromosomal level including large gains and deletions, at the nucleotide level, or through epigenetic changes such as DNA methylation. The changes that occur result in the activation of oncogenes and other growth promoting genes, and in the inactivation of tumor suppressor genes. NSCLC and SCLC exhibit distinct but overlapping patterns of genetic alterations.
Oncogene/Growth promoters
The protein products of oncogenes are involved in processes that stimulate cellular proliferation or survival. During tumorigenesis oncogene activation can occur through point mutations resulting in constitutively active proteins, through gene amplification and through over expression. In addition, the acquisition of growth promoting autocrine loops, in which individual tumor cells express both growth factors and their cognate receptor. Several oncogenes and growth promoting factors are known to be altered in NSCLC including K-ras, Erb-B1 (EGFR), Erb-B2 (HER-2/neu), myc, raf, bcl-1, bcl-2 and cyclin D1 (reviewed in (95), (97)). SCLCs are characterized by activation or overexpression of myc, raf, myb, Erb-B1, Bcl-2, fms, rlf and by Kit/SCF and GRP/GRP receptor co-expression (reviewed in (95), (81)).
Tumor Suppressor Genes (TSGs)
The protein products of tumor suppressor genes are involved in the inhibition of cell growth or survival. Several TSGs have been found to be deleted or mutationally inactivated in lung cancer. In addition, several chromosomal regions are specifically deleted in tumors, suggesting the presence of unidentified TSGs in these locations. LOH of p53 is frequent in both SCLC and NSCLC. RB mutations are common in SCLC; they are less frequent in NSCLC. However, mutations of p16INK4, a regulator of RB function, are found in many NSCLCs (81). In addition, deletions of several regions of chromosome 3p, 4q, 8p, 9q, Xp, Xq are found in both SCLC and NSCLC. Several regions, including 6q, 9p and 19p are frequently lost in NSCLC, while deletions of 5q, 10q and 22q are unique to SCLC (28). Many additional regions of allelic loss are found at lower frequencies in lung tumors. Candidate TSGs have been proposed for some regions, but in others the underlying TSGs remain to be discovered.
Overview of Existing Mouse Models
(For a more in depth discussion see the Murine Models of Lung Cancer section)
Murine lung cancer models provide opportunities to characterize the serial stages of tumor progression and to investigate the molecular alterations associated with them. In addition mouse models allow for the testing of novel chemopreventives, therapeutics and screening methods. The original models include spontaneous and carcinogen induced tumors in sensitive mouse strains such as A/J and SWR (JAX MTB Tumor Frequency Grid). A broad range of chemicals can induce the development of adenomas/adenocarcinomas in susceptible strains including tobacco smoke (106), urethane, metals and individual constituents of tobacco smoke such as polyaromatic hydrocarbons and nitrosamines (Reviewed in Stoner 1988 (86) and Tuveson 1999 (97)). These murine adenocarcinomas contain certain molecular alterations observed in human lung carcinomas including K-ras mutations (~90%) and LOH of chromosomal regions containing the murine p16Ink4a gene. Of note, the only exisiting murine models of squamous cell carcinoma were created either by direct application of carcinogen through intratracheal instillation or by rigorous topical application of carcinogen. Based on an understanding of the molecular alterations that occur in both murine and human lung tumors, several transgenic mouse NSCLC models have been created.
More recently mice carrying latent or conditional alleles of oncogenes and tumor suppressor genes have been created in order to more closely model the human situation in which tumorigenesis is initiated through somatic mutations occuring in adult tissues. Alleles of tumor suppressor genes flanked by loxP sites (floxed alleles) are expressed normally in the germline configuration, but after expression of Cre recombinase in the cell, all or a critical portion of the gene is deleted, leading to its inactivation. Furthermore, several inducible mouse lung adenocarcinoma models have been created using the doxycycline regulatable tet-operator system, which provides a unique opportunity to study the genes required for tumor maintenance and thus may help to identify potential therapeutic targets.
Molecular Alterations
(reviewed in 22, 78, 81, 91)
Oncogenes and Signal Transduction
The Ras family of proto-oncogenes encode a family of small GTPase proteins that transduce proliferation and survival signals from RTKs at the cell membrane. Activating mutations in the K-ras proto-oncogene are found in about 20-50% of NSCLCs, especially adenocarcinomas, and are associated with smoking (76). Point mutations at codon 12 are the most frequent, followed by mutations at codons 13 and 61, and result in a decreased intrinsic GTPase activity and inappropriate constitutive signaling for cell proliferation.
The Myc proto-oncogene family encodes three basic-helix-loop-helix transcription factors, C-Myc, N-Myc and L-Myc. The MYC proteins regulate the expression of key cell cycle regulators and genes involved in DNA synthesis and RNA metabolism. Activation of MYC occurs through gene amplification or transcriptional dysregulation, both resulting in overexpression of the MYC protein. Results of numerous studies of Myc gene amplification have shown that one Myc family member is amplified in 18-38% of SCLCs and 8-20% of NSCLCs with the lower end of the range representing findings in primary tumors and the upper end in cell lines (75). Furthermore, Myc mRNA expression has been noted in 33-67% of NSCLCs (41). Myc family DNA amplification has been associated with the highly malignant variant class of SCLC (V-SCLC) (50). It is also seen more often in patients that have been previously treated than in untreated patients, and is associated with reduced survival (41).
Autocrine/Paracrine Loops
Many growth factors or neuropeptides and their cognate receptors are expressed by individual cancer cells or the adjacent stroma. This results in several autocrine or paracrine loops that provide a driving force for tumor cell proliferation. Autocrine loops involving co-expression of neuropeptides and their specific G-protein coupled receptor (GPCR) are especially common in SCLC. Activated GPCRs have been shown to produce proliferative signals and to elicit a mitogenic response in a variety of cell types (35). Bombesin/gastrin releasing peptide (GRP), bradykinin, cholecystokinin (CCK), gastrin, neurotensin and vasopressin are all thought to be involved in driving SCLC growth (78).
One well characterized autocrine system involves gastrin-releasing peptide or other bombesin-like peptides (GRP /BN) and their receptors. Expression of GRP was demonstrated in 20-60% of SCLC by immunohistochemical analysis. Neutralizing antibodies against GRP/BN and bombesin antagonists inhibit both "in vitro" and "in vivo" growth of SCLC cell lines, and monoclonal antibodies show anti-tumor activity against SCLC in clinical trials (81). Thus GRP /BN autocrine signaling appears to play an important role in stimulating growth of SCLCs. Interestingly, the aberrant expression of these genes does not seem to be linked to gene amplification or rearrangement. GRP /BN is known to be involved in embryonic lung development suggesting that perhaps the cells of these tumors have de-differentiatiated to a more primitive state or have reactivated developmentally important signaling pathways. In addition the high percentage (57%) of SCLCs expressing both gastrin and its receptor CCK-B (74) further support a prominent role of neuropeptide autocrine signaling as a driving force for SCLC proliferation.
Peptide growth factor autocrine loops are more commonly found in NSCLC than SCLC. These growth factors bind to and activate receptor tyrosine kinases (RTKs) which then initiate intracellular signaling cascades. Expression of the Neuregulin receptor ERBB2 (also known as HER2/neu) has been noted in ~30% of NSCLCs and has been associated drug resistance and metastatic potential. The ERBB1 receptor (also known as the Epidermal Growth Factor Receptor) is also commonly overexpressed in NSCLC along with its ligands TGF-a, amphiregulin and EGF (78),(81). Several additional growth factor/RTK autocrine loops may also play a role in SCLC lung cancer proliferation including KIT and its ligand stem cell factor (SCF) as well as insulin like growth factors (IGFs) and their receptor IGF-R which are expressed in both SCLC and NSCLC.
Anti-apoptotic Genes
Bcl-2 was first identified as a proto-oncogene located at a translocation breakpoint in many B cell lymphomas. Bcl-2 is an anti-apoptotic protein that functions at the mitochondrial membrane. It is thought to promote cell survival by inhibiting linker proteins necessary for the activation of caspases. More than 90% of SCLCs express the bcl-2 protein. Most Bcl-2 positive tumors express the protein in a high percentage of the tumor cells (40, 109). A smaller subset of NSCLCs are bcl-2 positive. The bcl-2 protein is expressed in ~25% of squamous cell carcinomas and ~12% of adenocarcinomas (68). Interestingly, bcl-2 expression is thought to correlate with good prognosis in NSCLCs, but does not seem to correlate with prognosis in SCLC.
Tumor Suppressor Genes and Cell Cycle Regulation
Mutations or deletions of p53 are very common in both NSCLC (50%) and SCLC (80%). P53 normally acts to induce cell cycle arrest or apoptosis in response to cellular stresses such as DNA damage. P53 functions as a sequence specific transcription factor activating genes responsible for G1 arrest such as p21 Waf1/Cip1, to allow cells to repair damaged DNA before replication. Alternatively p53 can activate the transcription of genes involved in apoptosis such as BAX and PERP others. Mutations in the p53 gene are found in ~70% of SCLC and ~45% of NSCLC (9,31,88).
Alterations in the Rb pathway are also important in both SCLC and NSCLC. The RB protein acts as a growth suppressor by inactivating proteins that promote transcription of genes required for DNA replication, thus blocking the G1/S transition. RB mutations are found in 90% of SCLCs and 15-30% of NSCLCs most of which result in a truncated RB protein (6, 14, 36, 73). Furthermore, Cyclin D1 is overexpressed in up to 47% of NSCLCs (4, 60). Cyclin D1 acts to inhibit RB function by inducing it phosphorylation by Cdk4. Mutations in p16INK4A, an inhibitor of Cdk4 kinase activity, are also common in NSCLC (~60%)(67).
A second protein p14ARF is encoded by the p16INK4A locus. p14ARF is transcribed from an alternate reading frame that largely overlaps that of p16INK4A, but results in a totally unrelated protein. p14ARF prevents p53 degradation by MDM2 , resulting in p53 activation. p14ARF mutations are found in 19-37% of NCLCs (25, 66). The p14ARF protein is frequently lost in SCLC (65%), although the mRNA transcript is still present suggesting a post-transcriptional mechanism of inactivation. Interestingly, loss of p14ARF often occurs in the presence of p53 mutations in both NSCLC and SCLC, suggesting an alternative tumor suppressor function for p14ARF, distinct from that of p16INK4A (25).
Common regions of chromosomal loss and LOH suggest the existence of other tumor suppressor genes involved in lung tumorigenesis. Deletions of 3p are observed in 50% of NSCLCs and 90% of SCLCs. Such deletions often include FHIT, a candidate TSG , and abnormal FHIT mRNAs have been found in 40-80% of lung cancer (85). However many of these tumors also express the wild type FHIT transcript as well, raising the question of whether FHIT acts as a classical tumor suppressor. Additional sites of chromosomal loss for which the candidate TSGs remain unknown include 4p, 4q, 5p, 5q, 10p, 10q, 13q34 for SCLC; 1p, 6p, 13q11, 18q, 19p and Xq22.1 for NSCLC; and 8p, 9q, Xp for both SCLC and NSCLC (28).
Tumor Vasculature
Small tumors (1-3 mm) can obtain nutrients and oxygen by passive diffusion from their surrounding tissues. However, neovasculariztion is needed to support tumor growth, progression and metastasis. For tumors to induce angiogenesis, tumor cells and stromal cells must secrete factors that induce endothelial migration and proliferation. Vascular endothelial growth factor (VEGF ) stimulates neovasculariztion in a paracrine fashion. It is expressed by >50% of NSCLCs, and is associated with an increase in intratumoral microvascular density (IMD) and poor prognosis (59, 100). Platelet derived endothelial growth factor (PD-ECGF ) was initially identified as a novel angiogenic factor in platelets (98). PD-ECGF is expressed by ~32% of squamous cell carcinomas, 42% of adenocarcinomas and 33% of adenosquamous carcinomas (26). IL-8 is a member of the CXC chemokine family and has been reported to be a potent angiogenic factor (46). IL-8 is expressed in approximately 45% of NSCLCs and is associated with increased IMD (59). However, neither IL-8 nor PD-ECGF is expressed at significant levels in SCLC (110, 108). Finally, 49-70% of pulmonary adenocarcinomas express bFGF and expression correlates with poor prognosis (89, 90).
Murine Lung Cancer Models
Several approaches have been taken for creating murine lung cancer models. Specific inbred strains of mice are susceptible to the development of spontaneous lung tumors (JAX MTB Tumor Frequency Grid). The most sensitive strains include A/J and SWR while others range from intermediate sensitivity (BALB/c and O20), somewhat resistant (CBA and C3H) to nearly fully resistant (DBA and C57BL/6). The susceptible strains are also sensitive to chemically-induced lung tumors, and this sensitivity has been employed as a carcinogenicity bioassay (84). A polymorphism in the second intron of K-ras, that may affect gene expression levels, is one major modifier of sensitivity to lung tumorigenesis (111).
These strain differences in tumor susceptibility have been exploited for the mapping of additional loci that confer sensitivity to lung cancer. Analysis of progeny from crosses between recombinant inbred (RI) strains derived from the sensitive A/J strain and the resistant C57BL/6J strain, suggested the existence of three pulmonary adenoma susceptiblitiy (Pas) loci (55). Pas-1 was later identified by analysis of F2 progeny from a cross between strain A/J and the C3H/He resistant strain and was mapped to the distal region of chromosome 6 (24). Linkage analysis has demonstrated K-ras to be tightly linked to the Pas-1 locus, suggesting K-ras as a candidate for Pas-1 (47). Additional Pas loci have been mapped to chromosomes 9, 17 and 19 (13, 18). Furthermore, several numerous susceptibility to lung cancer (Sluc) loci have been identified by using a multilocus mapping method to analyze F2 mice generated from recombinant congenic strains (RCS). The Sluc loci are involved in complex genetic interactions that control susceptibility to the development of lung cancer (19, 20).
A wide variety of chemical carcinogens can induce pulmonary adenoma and adenocarcinoma formation in mice although they vary in their potencies (for a review on spontaneous and chemically induced mouse lung tumors see (53, 84, 86)). Some well characterized tumorigenic agents include urethane, metals, aflatoxin, tobacco smoke and tobacco smoke constituents including polyaromatic hydrocarbons and nitrosamines. Of note, the only murine model of squamous cell carcinoma existing to date is a carcinogen induced model resulting from the topical administration of Nitrosobris-(2-chloroethyl) urea (NTCU) twice a week for 35-40 weeks (71). The study of chemically induced lung tumors has provided insights into the histiogenesis of murine lung tumors suggesting that murine pulmonary adenocarcinomas are derived from cells of the alveolar epithelium of the type II cell lineage, or from the bronchiolar epithelium of the Clara cell lineage (34, 70, 72, 92).