© 2003 American Thoracic Society DOI: 10.1165/rcmb.F274
Functional Genomics in Lung Cancer and Biomarker DetectionDepartments of Medicine, Environmental Medicine, and Microbiology, Bellevue Chest Service, Division of Pulmonary and Critical Care Medicine, NYU School of Medicine, New York, New York Address correspondence to: William N. Rom, M.D., M.P.H., Bellevue Chest Service, Division of Pulmonary and Critical Care Medicine, NYU School of Medicine, Bellevue 7N24, 550 First Avenue, New York, NY 10016. E-mail: romw01{at}gcrc.med.nyu.edu
Abbreviations: non–small cell lung cancer, NSCLC • small cell lung cancer, SCLC
Lung cancer is a major health problem in the United States and throughout the developed world. Lung cancer causes Pulmonologists spend a substantial portion of their time evaluating patients who have lung cancer in their differential diagnosis. Many diagnostic bronchoscopies are performed for this disease. Low dose spiral computed tomography screening likely will increase referrals for evaluation. Clinical practices are witnessing a shift to more adenocarcinomas and early stage lung cancer. Despite playing a significant part of pulmonary practice, the basic science of lung cancer finds little space in the ATS journals. It's exciting to see the article by Powell and colleagues (2) that brings functional genomics to the Red Journal. Hopefully more basic science contributions will follow. Lung cancer is also a disease that slips through the cracks at the National Institutes of Health. The Lung Division of NHLBI does not fund cancer research (although it funds tuberculosis and AIDS research), and there are few pulmonologists funded by the National Cancer Institute (NCI). The NCI focuses on basic mechanisms and therapy, and its natural constituency is oncologists rather than pulmonologists. There needs to be a patient-oriented research effort from the combined institutes on the high-risk smoker who has preneoplasias and likely will develop lung cancer. Hahn and Weinberg (3) described the genetic rules for making human tumor cells. Dominant oncogenes such as ras were necessary, and early studies showed that ras and myc were oncogenes that cooperated in transforming animal cells. K-ras is mutated at codon 12 in 50% of lung adenocarcinomas, allowing unfettered growth signals to reach the nucleus (4). Normal human cells need external growth factors to signal cell proliferation, but oncogenic ras can substitute signals for cell growth without external growth stimuli (5). The mitogen-activated protein kinase signaling pathways downstream of ras are also upregulated, e.g., p38, that further enhance growth signals to reach the cell nucleus (6). The K-ras oncogene is important for cancer maintenance; for example, induction of lung-specific expression of oncogenic ras under a tetracycline control switch can produce lung adenocarcinomas within weeks in transgenic mice. Withdrawal of tetracycline shuts off the expression of oncogenic ras and tumors and preneoplastic lesions shrink in size (7). Tumor suppressor genes are especially important to lung carcinogenesis, and sequencing has identified point mutations or major deletions resulting in their inactivation. These genes are frequently checkpoint inhibitors (p16, p27) or control genes (p53) for complex circuits in the cell cycle and apoptosis pathways (8). An important mechanism for their loss is through silencing by the methylation of cytosines in the promoter sequences in CpG islands, resulting in the lack of gene expression. The mechanism underlying CpG methylation in lung carcinogenesis remains to be determined. We do know that the checkpoint inhibitor, p16, is practically always silenced in non–small cell lung cancer (NSCLC) due to methylation of its promoter. Methylation-specific polymerase chain reaction can detect these changes in the tumor itself as well as in sputum and bronchial brushes. p16 methylation is a proposed biomarker because this may be detected before the cancer can be observed on computed tomography scans. However, methylation-specific polymerase chain reaction methodology is not standardized, and there are considerable variations from laboratory to laboratory. The persistence of promoter methylation is not known, nor whether chemopreventive agents may alter this phenomenon. The loss and inactivation of p53 and Rb tumor suppressor functions are two essential contributors to transformation of human cells (Figure 1). In small cell lung cancer (SCLC), Rb is invariably mutated, and in NSCLC, p16 is silenced by methylation. Loss of p16 expression leads to Rb phosphorylation by the cyclin D: cyclin-dependent kinase 4, 6 complex releasing E2F with onset of the S phase of the cell cycle (9). The pathway controlled by p53 is altered in up to 70% of lung cancer. Point mutations occur in exons 5–7 in the DNA binding and activation domain, preventing this transcription factor from activating dozens of genes, especially those involved in DNA repair or apoptosis. These point mutations occur at random in nonsmokers, but in cigarette smokers they occur at so-called "hot spots" such as codons 157, 175, 248, 249, 272, and 283 (10). Structure-function maps identify these codons as key binding and activation sites on p53 DNA. Methylation may occur at CpG sites near these "hot spots" where carcinogen-DNA adducts form.
DePinho has maintained that telomerase is essential for the maintenance of telomere length, and induction of telomerase activity is the final prerequisite for cell immortalization (11). Transfection of the hTERT gene that encodes telomerase, the SV40 early region gene which encodes the large T and small t oncoproteins that bind p53 and Rb, and the ras oncogene were the minimum necessary to transform human fibroblast or epithelial cells to a tumorigenic state (12, 13). As tumors grow beyond 2 mm, they also need to signal for angiogenic factors such as vascular endothelial growth factor or basic fibroblast growth factor to attract endothelial cells and build a tumor circulation system (14). Hypoxia in the center of the growing tumor mass can induce hypoxia-inducible factor-1, which functions as a transcription factor for vascular endothelial growth factor. Thus a small set of intracellular pathways, namely DNA repair and apoptosis, cell cycle signaling, telomere maintenance, and mitogenic signaling, proved to be the minimal essentials for tumorigenesis. These pathways have been summarized as resistance to growth inhibition, evasion of apoptosis, immortalization, independence from mitogenic stimulation, and angiogenesis (15). Microarray experiments allow us to measure increased/decreased gene expression for thousands of genes in a simple experiment; lung cancers and adjacent uninvolved lung are obvious targets to use this technology. Goals and objectives should be the identification of new or known target genes, or gene function patterns along the lines of the pathways enumerated above (16). The Affymetrix U133 A and B GeneChips contain 33,000 human genes. Tumors may contain stroma and uninvolved lung contains vessels, airways, and interstitium with numerous cell types confounding any results. This can be avoided by laser capture microdissection. A general rule is that 30–50% of the genes on a GeneChip should hybridize (i.e., 30–50% genes expressed) for an experiment to be valid. Like many other scientific experiments, they should be done in triplicate with the results averaged. GeneChips are expensive, and one patient with adenocarcinoma and adjacent lung could consume six GeneChips in a single experiment. Garber and coworkers (17) and Bhattacharjee and colleagues (18) analyzed over 250 lung cancers by microarrays. They found separate gene profiles for squamous cell carcinoma, adenocarcinoma, and small cell lung carcinoma with distinct subclasses in adenocarcinoma. The strength of their studies was the large sample size where gene expression analysis could be verified across many samples. SCLC and carcinoid tumors both showed high levels of expression of neuroendocrine genes, and carcinoids also expressed a distinct group of genes diverging from malignant lung tumors. SCLC had the highest level of gene expression for proliferative markers. SCLC express high levels of certain keratin-type genes and the keratinocyte-specific protein, stratifin. Four adenocarcinoma clusters were defined by cell division and proliferation genes, neuroendocrine markers such as dopa decarboxylase and the serine protease kallikrein 11, ornithine decarboxylase 1 and surfactant proteins B, C, and D genes, and Type II pneumocyte genes. Interestingly, adenocarcinomas with upregulated neuroendocrine genes had a less favorable survival outcome (18). Beer and colleagues (19) were able to separate low- and high-risk stage I lung adenocarcinomas based on 50 genes from two equivalent but independent training and testing sets. They confirmed their microarray findings with Northern Blot analysis showing increased IGFBP3 mRNA levels from stage I to stage III adenocarcinomas, and substantiated gene upregulation with immunohistochemistry documenting protein expression in tumors for IGFBP3 and cystatin C. Miura and colleagues (20) expanded on these findings using laser capture microdissection to separate out malignant cells from endogenous stroma in the tumor. Forty-five genes separated smokers from nonsmokers after two rounds of amplification of extracted RNA from microdissected tumor samples. The expression of two genes localized in the commonly-deleted region on 3p21.3, the putative tumor suppressor 101F6 and carnitine/acyl carnitine translocase, was high in nonsmokers and low in smokers. Lerman and Minna (21) described 25 genes localized on the 630-kb homozygous deletion region on chromosome 3p21.3, a well-known region for frequent homozygous deletion in lung cancers. Sixteen of these genes showed lower expression in the adenocarcinomas from smokers as compared with nonsmokers, supporting the hypothesis that there is a high frequency of LOH in certain chromosomal regions, including 3p21.3. In this issue, Powell and colleagues (2) have extended these reports by comparing tissue from adjacent lung and adenocarcinoma obtained from resections of six never-smokers and six smokers. Their central finding was that four times more genes were altered between tumor and lung in nonsmokers as compared with smokers. This is consistent with the concept of "field cancerization" in heavy smokers where genes are altered in wide areas of the epithelium. Increased/decreased expression of genes is only part of the spectrum of genetic changes, e.g., methylation may occur in promoters of tumor suppressor genes leading to silencing; point mutations may result from cigarette smoking–related DNA adducts; and large deletions in certain chromosomal regions may occur. These observations suggested that the accumulation of > 6 molecular events might be necessary for lung cancer to develop. In addition, this may also explain the long latency and why there is a paucity of evidence for familial association.
To identify potential genes commonly involved in lung carcinogenesis that are independent of smoking, we have compared the lists of significant genes from (i) tumor versus nontumor in smokers with that from (ii) tumor versus nontumor in nonsmokers (2). The genes that were commonly regulated in lung cancer include two upregulated genes (tyrosine-protein kinase receptor TIE-2 precursor, and SSR1 signal sequence receptor- Recently, Campa and colleagues (22) reported the use of matrix-assisted laser desorption/ionization time-of-flight mass spectrometry discovery platform to generate protein expression profiles and identified macrophage migration inhibitory factor and cyclophilin A as potential molecular targets for lung cancer diagnosis and therapeutics. Macrophage migration inhibitory factor was also identified as a gene increased in lung tumors as compared with nontumor tissues (2). By comparing the gene lists from Beer and coworkers (19) and Powell and colleagues (2), we identified two genes commonly induced in lung adenocarcinomas as the 63-kD transmembrane protein and 5T4 oncofetal trophoblast glycoprotein. In fact, Kopreski and coworkers (23) have reported the potential for detecting circulating mRNA for 5T4 in the serum as a tumor marker for the identification of patients with cancer and for 5T4-targeted therapies. On the other hand, laminin ß 1, a gene found to be significant for survival (19), was also found to be decreased in tumors as compared with normal tissues (2). Moldvay and colleagues (24) reported that when stratified by tumor staging, independent markers of longer survival (i.e., high histological degree of tumor differentiation, positive bcl-2, and A+B+H blood group antigen expression in tumors) and shorter survival (i.e., O blood groups and p53 tumor staining) could be used as predictive survival markers in patients with surgically resected lung adenocarcinomas. Kaminski and Friedman (25) recently reviewed approaches to microarray experiments; importantly, pulmonary scientists need expertise from bioinformatics and biostatistics. Significance analysis of microarrays is an approach to identify genes significantly increased or decreased in expression. Multiple comparisons are plagued by the false discovery rate, and this should be kept low to increase the validity of the data. Permutation-based neighborhood analysis can be used to select the top 50 genes that distinguish the comparative groups, and identified 16 of 50 top genes that changed in a similar fashion in the transition from noninvolved lung to tumor in both smokers and nonsmokers (2). Many novel genes were identified in functional clusters that are frequently deregulated in lung cancer. Cluster analysis has been the most revealing in identifying subgroups and correlations with clinical outcome or response to chemotherapy. Their more precise role in experimental carcinogenesis can now be pursued.
This work was supported by NIH Grant: UO1 CA86137-04. Received in original form May 1, 2003
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
160,000 deaths in the United States compared with 231,000 deaths from chronic respiratory disease. Lung cancer is the leading cause of death from cancer in men and women, exceeding the combined total of colon, breast, and prostate cancer (
) and seven downregulated genes (ABC3 ATP-binding cassette 3, CDH5 cadherin 5, surfactant protein SP-C1, KIAA0257, RAGE, forkhead protein FREAC-1, and placenta copper monamine oxidase). We also found three genes, mitogen-inducible gene 2, metalloproteinase inhibitor 3 precursor, and ID1, to be reduced in nontumor in smokers versus nontumor in nonsmokers, suggesting that smoking may be involved in the downregulation of these genes. Interestingly, these same genes were also suppressed in tumors of nonsmokers when compared with nontumors from nonsmokers. By comparing gene profiles among these various groups of tissues, it is hopeful that additional insights into the lung carcinogenesis process and its relationship to smoking can be deciphered. 
Top
References
