Nasopharyngeal carcinoma is a special type of squamous cell carcinoma of head and neck associated with EBV infection, environmental factors and genetic aberrance. Clinically, poorly differentiated squamous cell carcinoma may account for 98% NPC patients while well differentiated squamous cell carcinoma and adenocarcinoma are rarely encountered in Southern China where this disease is particularly prevalent. The present study focused on alteration of gene expression in poorly differentiated squamous cell carcinoma of the nasopharynx. Owing to anatomical limitations, NPC and NP specimens collected from the clinic are usually very scarce and, because in most cases biopsies are semi-blindly performed through a forceps rather than a scalpel, it, is hard to accurately exclude normal NP tissue contamination from cancerous ones. Even in NPC biopsies without normal NP cells, the content of NPC cell appears to vary greatly. Therefore, screening the specimens obtained from the clinic for presence of a sufficient number of cancer cells was the critical step in order to assure the uniformity and validity of the microarray results. We found that tumor cell content varied from 5% to 95% in 102 pathologically diagnosed primary NPC samples among which only 32 samples contained more than 70% of cancer cells; the latter were qualified for further analysis.
In the present study, we applied the cDNA microarray technology to study differential gene expression between NPC and NP samples from Southern China. A pooling strategy was used to improve the statistical power and reduce individual heterogeneity. Class comparison analysis revealed 692 deferentially expressed genes with 2-fold or greater difference at a significance p-value of < 0.005. Of these genes, 435 were up-regulated and 257 were down-regulated. Subsequently, all the differentially expressed genes were further screened by MILANO analysis. After excluding the genes without association with limited oncogene and tumor suppressor terms when searching in MILANO, 10 genes were found to be reported consistently as known oncogenes and tumor suppressor genes in various human malignancies in previous studies. 4 genes were up-regulated and 6 genes down-regulated. Interestedly, of these 10 genes, three genes (MIF, BIRC5 and ATM) have been documented in NPC associated articles, and the rest genes (PTTG1, FOXO1A, TGFBR2, PRKAR1A, CCND2, KLF5, and PDCD4) have never reported or poorly understood in NPC before.
It is well known that oncogenes tend to be up-regulated and tumor suppressor genes are down-regulated in malignant condition. Based on CGH, chromosomal abnormalities associated with primary and metastatic NPC have been identified in all human chromosomes except chromosome Y [3, 4, 7, 17–20]. BIRC5 up-regulated in NPC (oncogenes) was located on 17q21.3-25. As down-regulated tumor suppressors in NPC, FOXO1A and KLF5 existed on 13q14.1-34, ATM on 11 q13-23, and TGFBR2 on 3p21.3-24. The expression level of these genes was consistent with corresponding chromosomal gains or losses. However, the gene expression of the rest of the genes did not correspond with their chromosomal status. MIF and PTTG1 up-regulated in NPC (oncogenes) were on 19p13.3 and 5q35.1 respectively and PRKAR1A and PDCD4 down-regulated in NPC (tumor suppressors) were mapped to chromosome 17q22-q24 and 10q24 respectively, but no corresponding high frequency of chromosomal alterations in those region have been reported in association with NPC. Down-regulation of some genes might be involved in other mechanisms including promoter hypermethylation or additional modification, promoter mutation and gene mutation.
Several studies have shown that several differentially expressed genes identified in this study have functions associated with cell proliferation, cell apoptosis, cell cycle, angiogenesis and signal transducer activity, creating an active proliferation and progression status at tumor site. As an up-regulated oncogene, MIF, a macrophage migration inhibitory factor, may function as an autocrine mediator of both growth factor- and integrin-dependent sustained ERK MAPK activation, cyclin D1 expression, and cell cycle progression. Swanthttp://linus.nci.nih.gov/BRB-ArrayTools.html found that MIF could stimulate cyclin D1 by recruiting Rho GTPase and its downstream signaling for MAPK activation. EBV oncogene LMP1 in NPC, PTGS2 in endometrial carcinoma and MYC in breast cancer may induce the BIRC5 expression by different pathways. As a tumor suppressor, KLF5, a transcription factor associated with cellular signaling involved in cell proliferation and oncogenesis, inhibits the expression of BIRC5 by binding p53 in acute lymphoblastic leukemia, suggesting that anti-apoptotic factor BIRC5 may play a key role in gene-regulated network of NPC. Pituitary tumor-transforming gene-1 (PTTG1) is overexpressed in a variety of endocrine-related tumors and nonendocrine-related cancers involving the central nervous, pulmonary, and gastrointestinal systems. It is a potent oncogene because of its ability to combine with p53 and thus to prevent p53 from binding to DNA and inducing cell death. DNA damage and the consequent activation of the DNA damage response (DDR) pathway is one of the occasions for the tumor initiation and progression. Among tumor suppressors, ATM, a major regulator of the cellular response to DNA double-strand breaks, may be a key factor in the DDR pathway. Knocking down ATM with a short hairpin RNA (shRNA) could block p53 induction in response to aberrant STAT5A activation and bypass the senescence response to this oncogene when the Rb pathway was also inactivated. In addition, knocking down ATM could inhibit E2F1-induced senescence and, in combination with Rb inactivation, suppress RasV12-induced senescence. TGF-β exerted its tumor-suppressor effects on many tumors by binding to the transmembrane TGF-β type II receptor (TGFBR2, a tumor suppressor), which caused the recruitment of the TGF-β type I receptor (TGFBR1) with subsequent activation of the receptor complex. Decreased expression of TGFBR2 found in NPC may block this important pathway of tumor-growth inhibition and promote the NPC malignant transformation. Programmed cell death-4 (PDCD4) is a new discovered tumor suppressor protein that inhibits protein synthesis by suppression of translation initiation. A recent study showed that PDCD4 suppressed tumor progression in human colon carcinoma cells by the novel mechanism of down-regulating MAP4K1 transcription, with consequent inhibition of c-Jun activation and AP-1-dependent transcription. Down-regulated expression of PDCD4 in NPC tissue and cells suggested a negatively regulation role in NPC pathogenesis. Tumor suppressor FOXO1A, a FOXO transcription factor, has been implicated in several human cancers. However, it has not been reported in NPC to date. Shore found that Expression of the EBV genes for latent membrane protein 1 and latent membrane protein 2A could decrease Foxo1A expression by phosphatidylinositol 3-kinase-mediated nuclear export. Oncoprotein P3k and Akt can suppress the FoxO1 expression by a common denominator of their pathway. The tumor suppressor gene, PRKAR1A, coding for the Type 1alpha regulatory subunit of protein kinase A, a critical cellular component of a number of cyclic nucleotide-dependent signaling pathways, is mutated in Carney complex, a familial neoplasia syndrome that is associated with thyroid tumors. Its decreased expression in NPC may be caused by similar gene mutation, which is remained to be further elucidated.
Interestedly, several studies really supported the dysregulation of several pathways, such as Wnt signaling, MARK signaling [31, 32], NF-kB-Apoptosis resistance, integrin signaling in NPC although most of pathways involving cell growth and death, cell communication, and immune system, etc. have not been well documented. In our present study, our computational pathway analysis of 692 differentially expressed genes strongly supported that multiple biological pathways were indeed involved in NPC oncogenesis. Particularly, we found that most of immune system associated pathways, such as Natural killer cell mediated cytotoxicity, Hematopoietic cell lineage, B cell receptor signaling, T cell receptor signaling, Fc epsilon RI signaling, Complement and coagulation cascades, and Toll-like receptor signaling, were involved, suggesting that they may play important roles in NPC.
Recently, there has been increasing interest in identify diagnostic tools that could complement standard histopathologic evaluation to determine the presence of cancer cells in tissues, a previous study from our group applied a high throughput microarray technology to investigate a broad range of cancer types and identified 16 proposed universal cancer biomarkers with high prediction accuracy. These biomarkers could be used broadly to improve the sensitivity and specificity of cancer staging and early detection of loco-regional or systemic recurrence . Among the 16 common cancer biomarkers identified as highly predictive by the previous study which did not include NPC samples, 6 genes (CYC1, MIF, LAMB3, TSTA3, TUBB2, and UBE2C) were found to be highly expressed in NPC by the study. The seventh gene, TRAP1, was also included in the first 50 genes. Therefore, CYC1, MIF, LAMB3, TSTA3, TUBB2, UBE2C, and TRAP1 may possibly represent NPC candidate biomarkers adding a molecular dimension to the histo-pathological diagnosis of the disease and these biomarkers may be added to the pathologist's repertoire for the uncovering of NPC pathogenesis when comprehensive histologic evaluation is not sufficient. Obviously further validation will be needed in the future. Moreover, NPC could share the expression of various biomarkers with other cancers of disparate histology suggesting the existence of common pathways of oncogenesis that may be most relevant as therapeutic targets.