Gene expression profiling for molecular distinction and characterization of laser captured primary lung cancers

Lung cancer represents a heterogeneous group of diseases in terms of their biology and the clinical course. The diagnosis and classification of lung cancers are primarily based on the histological morphology and immunohistological methods for distinguishing between small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC) [1]. The molecular pathogenesis of lung cancer, as far as it has been deciphered, consists of genetic and epigenetic alterations, including the activation of proto-oncogenes and inactivation of tumor suppressor genes [2–4]. This leads to a malignant phenotype, resulting in changes in cell structure, adhesion and cell proliferation [5].

Oligonucleotide microarray studies are commonly used to extend the knowledge of the differences in the biology of lung tumors and to identify new candidate genes with diagnostic, prognostic and therapeutic value [6–9]. Several gene expression profiling studies in lung cancer have been published, however, it is still difficult to compare these studies due to the differences in methodologies, array platforms, normalization of the data and biostatistical analyses approaches, which may influence the reproducibility and comparability [10–12]. Such differences could have led to divergent results, with limited overlap of described genes.

Another crucial step in the field of oligonucleotide microarray studies is the preparation of the solid tumor sample itself. It contains a variable amount of mesenchymal stroma cells, blood vessels, fibroblasts, tumor-invading lymphocytes and necrotic areas next to the tumor cells themselves. Analyzing the complete tumor sample without efficient separation of the tumor cell confounds the true gene expression profile of the tumor.

In order to overcome these methodological limitations, we followed the guidelines from the Microarray Gene Expression Data Society [13] and the MicroArray Quality Control (MAQC) Consortium [14, 15], the External RNA Controls Consortium (ERCC) [16] as well as the European consensus guidelines for gene expression experiments [17]. The purification of the tumor cells was carried out by laser capture microdissection (LCM), which has been shown to greatly improve the sample preparation for microarray expression analysis [18]. Few reports on LCM and microarray gene expression analysis have been published to date, comparing all distinct lung cancer subtypes to normal lung tissue [19–21].

In this report, we performed a comparison of gene expression profiles, using microarray analysis and LCM, according to the methodological quality consensus guidelines for microarray experiments, with the aim of identifying genes that are differentially expressed in the major histological lung cancer subtypes, as compared to normal lung tissue. In addition, 14 differentially expressed genes in human lung tumors were corroborated by quantitative real-time PCR. Furthermore, using genetic programming, we found a subset of 40 genes, that could be utilized for the classification of different types of lung tumors.

In this study, a comparison of the expression pattern of the 3 major histological lung cancer subtypes, as measured by array analysis, is presented. In comparison to the normal lung tissue, 205, 335 and 404 genes in AC, SCC and SCLC were found to be at least 2-fold differentially expressed. Fourteen genes of different gene families were corroborated using RT-PCR.

In AC, we found an up-regulation of keratin 7, a characteristic finding for pathologists to diagnose this subtype of lung cancer. On the other hand, keratin 5 was downregulated in AC. The differential expression is already described as a separator between AC and SCC, in line with our results [28, 29]. Looking at adhesion molecules in AC, a down-regulation of the desmosomes desmoglein 3 and desmocollin 3 was found. In this context, it was shown that the invasive behavior of cells is inhibited when transfected with desmosomal components [30], suggesting that down-regulation of the desmosomes in adenocarcinomas of the lung plays a role in the loss of cell to cell contact and tumor spreading.

The extracellular cell matrix receptors integrin alpha-3 and integrin beta-2 as well as the collagen binding protein-1 (SERPHINH1) were upregulated in AC. These genes have a high affinity to collagen IV and laminin, both essential components of the basement membrane [31], possibly mediating adhesion and invasion. Additionally, we found intercellular adhesion molecule 1 (ICAM1), a cell-adhesion molecule also binding to integrin beta-2 and promoting metastasis due to tumor cell adhesion to endothelium overexpressed in AC [32, 33].

Looking at the oncogenes in SCC, we found genes of the RAS associated gene family, the myc myelocytomatosis viral oncogene homolog (MYC) and musculoaponeurotic fibrosarcoma oncogene (MAF) upregulated. MAF encodes for nuclear transcriptional regulating proteins with a leucine zipper motif, and was identified in the genome of the acute transforming avian retrovirus AS42, which induces fibrosarcomas and has the ability to transform chicken embryo fibroblasts [34].

It is noteworthy that in SCC 5 members of the collagen family type I, V, VI, and XI were upregulated. An increased collagen synthesis might be associated with carcinogenesis, as in patients with breast cancer the emerging fibrotic focus is regarded as an indicator of tumor angiogenesis and independent predictor of early metastasis [35].

SCLCs show an up-regulation of 3 proto-oncogenes, which have not been described in this context so far. The DEK oncogene encodes for a 375 amino acid chromatin binding protein, which introduces supercoiling in DNA. It has been described to be upregulated in other tumor types, such as bladder cancer, glioblastoma, melanoma and leukemia [36]. The Qin oncogene, originally isolated from avian sarcoma virus, causes oncogenic transformation. Qin is the avian orthologue of mammalian brain factor-1 or forkhead box G1 (FOXG1B), a gene which belongs to the human forkhead-box gene family [37]. Possibly related to the neuroendocrine differentiation of SCLC, forkehead box G1 is essential for the proliferation and survival of cerebro-cortical progenitor cells [38]. Further, we found the Fyn oncogene upregulated in SCLC. Fyn is a member of the src family which is activated in colorectal cancer, and has also been identified in melanoma cells with elevated cell motility and spreading ability [39, 40].

With regard to adhesion molecules, the overexpressed neural cell adhesion molecule 1 is useful for the diagnosis of SCLC [41, 42]. Next to neural cell adhesion molecule 1 we found other genes significantly upregulated such as the Purkinje cell protein 4, secretory granule neuroendocrine protein 1, synaptotagmin 1 and the neuronal cell adhesion molecule (NRCAM) that seems to reflect the neuronal heritage of this particular lung tumor subtype. NRCAM belongs to the L1 family immunoglobulin-like CAMs, which are involved in the guidance, growth and fasciculation of neuronal cells [43]. Neuronal cell adhesion molecule has also been described in 2006 by Taniwaki and colleagues', who performed comprehensive gene expression profiles of pure SCLC cells derived from laser-microdissected tissue samples [44]. In order to confirm the overexpression of the neuronal cell adhesion molecule using a different technique, we corroborated the result of microarray analysis using RT-PCR, showing a 9.3-fold overexpression of the neuronal cell adhesion molecule in SCLC in comparison to lung tissue.

The imbalance of activated oncogenes and lost tumor suppressor genes, found in different types of lung cancer, may be associated with the different tumor growth kinetics. SCLC is the fastest growing lung tumor with a median tumor doubling time of 50 days [45]. This is reflected by our data with regard to the number and strength of upregulated cell cycle genes affecting growth rate. Several cyclines, their associated cyclin-dependent kinases and cell division cycle (CDC) genes controlling cell cycle progression, such as cyclin A2, B2 and E2, and cyclin-dependent kinase 2 and 4, as well as cell division cycle 2, 20 and 25B were upregulated [46]. The activation level of different cell cycle genes may be relevant with regard to new antitumor agents, which selectively target cell cycle proteins. For example, flavopiridol has the ability to induce cell cycle arrest by binding and inhibiting different cyclin-dependent kinase such as 2 and 4 [47, 48]. Both CDKs are significantly upregulated in SCLC. On the other hand, the upregulation of cyclin-dependent kinase 2, that is critical for cell entry and progression through S phase of the cell cycle, is missing in NSCLC. Preclinical data support this finding since most NSCLC cell lines are resistant to flavopiridol-induced apoptosis unless they were treated during S phase. Furthermore, the IC 50 of flavopiridol-treated cells in SCLC cell lines is three times lower compared to NSCLC cell lines [49]. Consequently, this drug might be more promising in patients with SCLC.

We have further shown that genes involved in mismatch repair, such as mutS homolog 2 or 6, were upregulated in SCLCs, which is in line with other reports showing that these gene transcripts and proteins are present [50, 51], in contrast to NSCLCs, where high resolution deletion mapping reveals frequent allelic losses at the DNA mismatch repair loci mutS homolog 3 [52]. Similar to the latter report, we have observed a downregulation of mutS homolog 3 in ACs.

After outlining potentially important molecular differences in different subtypes of lung cancer to normal lung tissue, we were interested in defining how many and which genes are necessary for correct classification of the lung tumor subtype. Using genetic programming (GP), a training set of 34 tissue samples was applied. With an evolutionary algorithm of GP, 40 genes were sufficient for a correct discrimination between all lung tumor tissue types and normal lung tissue. The 40 selected genes, identified using GP, were a subset of the genes, which were previously identified to be differentially expressed using cluster analysis. Following identification of the 40 genes with GP, further 13 tissue samples of previously treated patients NSCLC lung cancers were correctly classified with 100% prediction accuracy. It is important to note that the samples of the training set were from treatment naïve patients, while the test set came from those that were previously treated for their cancer using platinum-based chemotherapy. Nevertheless, the presented genes for distinction seem to maintain their value, independent from whether or not the patient had been treated. However, caution must be applied, since in the test set did not contain additional SCLC samples and larger sample size is needed which includes samples from all lung cancer subtypes in order to confirm the predictor.

Our data show the different gene expression profiles in dependence from the histological type of lung cancer, which reflects the specific biological characteristics of the respective tumor subtype. These data may form the basis for a molecular classification system and allows a further insight into the altered genomic progress of the lung cancer cell, which may help to develop molecularly targeted drugs.