Radiation therapy has played an important role in controlling tumor growth in many patients with cancer. In patients with oral squamous cell carcinoma (OSCC), radiation therapy is currently the standard adjuvant treatment. However, radiation therapy is sometimes ineffective, because cancer cells can be refractory to radiation therapy. A relationship between radioresistance and expression of several genes, i.e., RAS,1RAF12 and BCL2,3 has been reported. It also has been proposed that the Ras/PI3K/Act pathway is associated with radioresistance both in vivo and in vitro in several human cancers, i.e., those of the larynx,4 uterine cervix,4 head and neck,5 bladder,4, 6, 7, 8 colon,4, 7, 8 breast4, 9 and fibrosarcoma.8 Moreover, recent studies have identified genes related to the radioresistance and radiosensitivity of human cancers10, 11, 12, 13, 14, 15 in comprehensive gene expression profiles by microarray analysis.
In oral malignancies including OSCCs, COX-2,16 glycerol,17 14-3-3 sigma protein,18 DNA-PK complex protein,19 DNA contact mutation of p5320 and NF- kappaB21 may be related to radioresistance. Although these findings have achieved a partial understanding of the molecular mechanisms responsible for cellular radiosensitivity, the entire process remains to be clarified. Because the complex mechanism of radioresistance cannot be explained by a small number of genes, it is necessary to analyze simultaneous expression levels of thousands of genes. In this context, the emerging microarray technology provides the ability to comparatively analyze mRNA expression of thousands of genes in parallel. However, little is known about comprehensive gene expression profiles related to the radioresistance of OSCC using microarray analysis.22
Our previous study showed that the radiosensitivity of OSCC cell lines differs greatly in their response to X-ray radiation as assessed by clonogenic survival assay.22 In the current study, we performed microarray analysis using high-density Affymetrix Human Genome-U133 plus 2.0 GeneChip arrays containing 54,675 probe sets (Affymetrix, Santa Clara, CA) to compare gene expression patterns among the cell lines with or without radioresistance after X-ray irradiation. Furthermore, the genes identified were analyzed for network and gene ontology by Ingenuity Pathway Analysis (IPA) software (Ingenuity Systems, Mountain View, CA) to identify networks of interacting genes and other functional groups.23 In addition, further selected genes confirmed the microarray data by real-time quantitative reverse transcriptase–polymerase chain reaction (qRT–PCR).
A calls, absent calls; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; IPA, ingenuity pathway analysis; OSCC, oral squamous cell carcinoma; P calls, present calls; qRT-PCR, quantitative reverse transcriptase–polymerase chain reaction; SD, standard deviation.
Material and methods
Cell lines and culture conditions
The human OSCC-derived cell lines, HSC2 and HSC3 (Human Science Research Resources Bank, Osaka, Japan), were prepared for this study. The cells were maintained in Dulbecco's modified Eagle's medium F-12 HAM (Sigma Chemical, St. Louis, MO) supplemented with 10% heat-inactivated fetal bovine serum and 50 units/ml penicillin and streptomycin. All cultures were grown at 37°C in a humidified atmosphere of 5% carbon dioxide for routine growth.
Irradiation using X-rays
The cells were plated in 75-mm2 tissue culture dishes containing 10 ml of medium, allowed to grow to 70–80% confluence, and then irradiated with 3 single radiation doses (2, 4 and 8 Gy) using X-ray irradiation equipment (MBR-1520R-3; Hitachi, Tokyo, Japan) from a distance of 55 cm.
Isolation of RNA
Total RNA was extracted with TRIzor reagent (Invitrogen Life Technologies, Carlsbad, CA) from X-ray irradiated and nonirradiated cells 1, 2, 4, 8 and 12 hr after X-ray irradiation, according to the manufacturer's instructions. The quality of the total RNA was determined by Bioanalyzer (Agilent Technologies, Palo Alto, CA).
Hybridization of RNAs to oligonucleotide arrays and data analysis
For microarray analysis, 4 hr after X-ray irradiation was selected as the time point to monitor the early response of OSCC cells for X-ray irradiation and to identify differentially expressed early genes that mediate cellular events, such as proliferation and apoptosis. We used Human Genome U133 plus 2.0 GeneChip oligonucleotide arrays. This GeneChip containing 54,675 probe sets analyzes the expression level of over 47,000 transcripts and variants, including 38,500 well-characterized human genes. Microarray analysis compared the most radioresistant cells and the most radiosensitive cells. For hybridization, 20 μg of total RNA per sample was prepared according to the manufacturer's protocols. Fragmented cRNA (15 μg of each) was hybridized to the Human Genome oligonucleotide arrays. The arrays were stained with phycoerythrin–streptavidin and the single intensity was amplified by treatment with a biotin-conjugated antistreptavidin antibody, followed by a second staining using phycoerythrin–streptavidin. The arrays stained a second time were scanned using the GeneChip Scanner 3000 (Affymetrix). Expression data were analyzed using GeneChip Operating Software 1.1 (Affymetrix) and GeneSpring 6.1 (Silicon Genetics, Redwood City, CA). The arrays were normalized by Operating Software 1.1 (Affymetrix) to determine the probe set intensities and “Present” (P) or “Absent” (A) calls. The Expression data of P calls indicated to have reliability. On the other hand, the data of A calls were unreliable.
Network, gene ontology and canonical pathway analysis
Under the criteria combining P calls and fold changes, which suggested a mean enhance in expression level at least 5-fold or more by X-ray irradiation at all doses (0, 2, 4 and 8 Gy) in the radioresistant cell lines, HSC2, compared with the radiosensitive cell lines, HSC3 in GeneChip analysis by GeneSpring 6.1 data mining software (Silicon Genetics, Redwood City, CA), the significantly altered 167 genes were selected and used for the network generation and pathway analysis. Gene accession numbers and mRNA expression values were imported into the IPA software. The genes were categorized based on molecular functions using the software. The identified genes also were mapped to genetic networks in the IPA database and ranked by score. The score reflects the probability that a collection of genes equal to or greater than the number in a network could be achieved by chance alone. A score of 3 indicates that there is a 1/1,000 chance that the focus genes in a network are there by random chance. Therefore, scores of 3 or higher have a 99.9% confidence level of not having been generated by random chance alone. This score was used as the cut-off for identifying gene networks. Moreover, relationships between the network generated in IPA and the known pathways which were associated with metabolism and signaling were investigated by Canonical pathway analysis.
Preparation of cDNA
Total RNA was extracted using TRIzor reagent from X-ray irradiated and nonirradiated HSC2 and HSC3 cells. Total RNA samples were extracted at 1, 2, 4, 8 and 12 hr after 2 Gy of X-ray irradiation to determine the time-dependent effects; the physical doses used in clinical radiotherapy and samples were extracted at 4 hr after 2, 4 and 8 Gy of X-ray irradiation to determine the dose-dependent effects, according to the manufacturer's instructions. Five micrograms of total RNA of each sample was reversed transcribed to cDNA using Ready-To-Go You-Prime First-Strand Beads (Amersham Biosciences, Little Chalfort, Buckinghamshire, UK) and oligo (dT) primer (Sigma Genosys, Ishikari, Japan), according to the manufacturer's protocol.
Analysis of mRNA expression by real-time quantitative reverse transcriptase–polymerase chain reaction (qRT–PCR)
Real-time qRT–PCR was performed to verify the microarray data with a single method using a LightCycler FastStart DNA Master SYBR Green I kit (Roche Diagnostics GmbH, Mannheim, Germany), according to the procedure provided by the manufacturer. Oligonucleotides used as primers and the predicted sizes of amplified PCR products are listed in Table I. Using LightCycler apparatus, we carried out PCR reactions in a final volume of 20 μl of a reaction mixture consisting of 2 μl of FirstStart DNA Master SYBR Green I mix, 3 mM MgCl2 and 1 μl of the primers according to the manufacturer's instructions. Subsequently, the reaction mixture was loaded into glass capillary tubes and subjected to initial denaturation at 95°C for 10 min, followed by 35–45 rounds of amplification at 95°C (10 sec) for denaturation, 58–65°C (10 sec) for annealing, and 72°C for extension, with a temperature slope of 20°C/sec, performed in the LightCycler. The transcript amount for the genes differentially expressed in the microarray analysis was estimated from the respective standard curves and normalized to the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) transcript amount determined in corresponding samples.
Table I. List of the Primers Used for the Quantitative RT-PCR
bp, base pair.
The Mann-Whitney U test was used to determine the statistical significance of the associations between the mRNA expression levels of the most radioresistant cell cells and the most radiosensitive cells. The criterion for statistical significance was p < 0.05.
Microarray analysis of radiosensitive and radioresistant cell lines
To evaluate cell survival after irradiation, a clonogenic survival assay was used as previously described. Previous study showed that the most radioresistant cell line was HSC2 and the most radiosensitive cell line was HSC3 in 6 OSCC cell lines.22
The gene expression profiles of OSCC cells exposed to X-ray irradiation were analyzed using the high-throughput microarray, which contains 54,675 oligonucleotide-based probe sets. The results of microarray analysis showed that the expression levels of 167 genes (220 probes) were elevated at least 5-fold or more by X-ray irradiation at all doses (0, 2, 4 and 8 Gy) in the radioresistant cell lines, HSC2, compared with radiosensitive cell lines, HSC3.
Network and gene ontology analysis
We carried out genetic network analysis of the 167 genes (220 probes) with elevated expression at least 5-fold or more by X-ray irradiation at all doses (0, 2, 4 and 8 Gy) in radioresistant cell compared with radiosensitive cell in GeneChip results. Three networks (Table II) were identified by network analysis using the IPA tool. These networks indicated functional relationships between gene products based on known interactions in the literature. The IPA tool then associated these networks with known biologic pathways. Three networks were highly significant in that they had more of the identified genes present than would be expected by chance. These networks were associated with the cancer, cellular movement, cell cycle, proliferation, cell death and tissue development (Table II). Each network was characterized by different functions. They were merged via overlapping genes (Fig. 1). Of the 167 genes, 40 were mapped to genetic networks identified by the IPA tool. GeneChip results of 40 focus genes in genetic networks (Fig. 1) were showed in Table III. These results were directly correlated with the intensity of the node color (Red), which indicated the degree of up-regulation of focus genes in radioresistant cell (HSC2) compared with radiosensitive cell (HSC3) by microarray analysis in Figure 1. Gene ontology analysis was also performed using the IPA tool. A total of 46 functions were identified as high level functions (Table IV). The cancer-related function had the highest p value (p = 7.54e-5–4.63e-2). Furthermore, to investigate the network of 30 cancer-related genes, we performed network analysis. Consequently, we identified 1 network (Fig. 2) that included 25 cancer-related genes (Table V). These genes, which were part of the network of the 40 genes related to the radioresistance of OSCC, were categorized by function into growth and proliferation, apoptosis and adhesion.
Validation of microarray data by real-time qRT–PCR analysis
To verify the microarray data, we analyzed the levels of mRNAs of the 25 genes by real-time qRT-PCR. The results of the quantitative assessment of expression of the 25 genes after X-ray irradiation are shown in Figure 3. The results were in good agreement with those from the microarray data in a dose-dependent manner. The expression levels of these 25 genes tended to be elevated in radioresistant cells compared with radiosensitive cells not only in a dose-dependent manner but also in a time-dependent manner. Of the 25 genes, 13 had functions of proliferation and antiapoptosis. Another 10 genes had functions of antiproliferation and apoptosis. The remaining 2 genes were unrelated to proliferation and apoptosis (Table V).
Furthermore, among the 25 genes, 11 genes (PEG10, ROBO1, ICAM2, TIMP3, DAB2, MMP13, PLAGL1, ID1, PVRL3, ID3 and FGFR3) had a significant (p < 0.05) elevation of radioresistant cells compared with the radiosensitive cells. The data were expressed as the mean ± standard deviation (SD) of 2 independent experiments with samples in triplicate.
Canonical pathway analysis
Several canonical pathways associated with the cancer-related network (Fig. 2) were found. These were Parkinson's signaling, notch signaling, IL-2 signaling, phospholipid degeneration, FGF signaling, prostaglandin metabolism, and B cell receptor signaling. In these canonical pathways, only FGF signaling pathway (Fig. 4) included the gene (FGFR3) which was confirmed the significant enhancement of gene expression in radioresistant cells (HSC2) compared with the radiosensitive cells (HSC3) by real-time qRT-PCR. FGF signaling pathway showed that the FGFRs (included FGFR3) elevations contributed to cell growth and angiogenesis.
Radiotherapy, an inevitable component of modern cancer management, is a major treatment modality that can potentially provide a cure for patients with OSCC.24 The success or failure of radiotherapy can be affected by the radiosensitivity of the tumor target and the limits imposed on treatment by the radiosensitivity of normal tissues. Recently, several studies have successfully used microarrays to identify and classify a set of human genes that are radiosensitive to X-ray irradiation.10, 11, 12, 13, 14, 15
In the current study, we identified 167 genes with significantly elevated expression in radioresistant cells at all doses (0, 2, 4, 8 Gy) using the high-throughput microarray that contains 54,675 oligonucleotide-based probe sets. When we analyzed these genes regarding functional network using IPA software, we detected 3 genetic networks (Fig. 1) that included 40 genes with higher expression levels in radioresistant cells than in radiosensitive cells. In addition, gene ontology analysis identified 1 network (Fig. 2) that included 25 cancer-related genes that had the highest p value (p = 7.54e-5–4.63e-2). To further evaluate the validity of the 25 genes selected as radioresistant genes, we carried out qRT-PCR analysis. These genes were higher in radioresistant cells than radiosensitive cells not only in a dose-dependent manner but also in a time-dependent manner, and the genes were categorized by function into proliferation, apoptosis, and adhesion. The functions of cell proliferation, apoptosis, DNA repair, and cell cycle have been reported as the radioresistant-related functions.4, 11 We considered that proliferation and apoptosis were noteworthy functions in the 25 cancer-related genes. Ionizing radiation has been proposed to activate both proliferative and antiproliferative signal transduction pathways, the balance of which determines cell fate,25 suggesting that X-ray irradiation may activate functions of apoptosis and antiapoptosis. Thus, it was reasonable to suppose that the functions of proliferation and antiapoptosis were important for the radioresistance of cancer. Among the 25 genes identified, the 13 genes associated with proliferation and antiapoptosis were PTHLH,26, 27ID1,28, 29ID3,30, 31FGFR3,32VAV3,33PTGS2,34PLD1,35, 36JAG1,37FGFBP1,38, 39PEG10,40, 41, 42S100P,43, 44ICAM245 and MMP13.46, 47, 48 In particular, PTGS2 (known as COX-2) was reported to be linked to radioresistance of human glioblastoma,49 esophageal cancer,13 laryngeal cancer,50 OSCC,16 and lung cancer.51 The remaining 12 genes identified have not been reported to be correlated with radioresistance. PTGS2 overexpression leads to increased PGE2 production, resulting in increased cellular proliferation,52 and tends to be resistance to apoptosis by inducing Bcl-2 expression.53 Those studies reported that PTGS2 expression may play a role in radioresistance via proliferative and antiapoptotic pathways. Therefore, we postulated that the other 12 genes are connected to the radioresistance of cancer as well as the mechanism of radioresistance in PTGS2. Of them, 6 genes (ID1, ID3, FGFR3, PEG10, ICAM2 and MMP13) had a significant (p < 0.05) elevation in radioresistant cells compared with radiosensitive cells in a dose-dependent manner and in a time-dependent manner in real-time qRT-PCR analysis (Fig. 3). Therefore, these 6 genes will be useful to guide the choice of appropriate and effective cancer therapy. Especially, FGFR3 was emphasized the relation with the radioresistance by the results of canonical pathway analysis.
In conclusion, this comprehensive gene expression profiling-assisted pathway analysis provided an appealing approach for effectively identifying candidate genes and pathways involved in cellular radioresistance. We suggest that the pathway that includes the 25 genes is related to the radioresistance of OSCC owing to the balance of proliferation, antiproliferation, apoptosis, and antiapoptosis. Moreover, these 25 genes may contribute to a basic understanding of the molecular mechanism underlying the tumor radioresistance to X-ray irradiation in OSCC. The highlight of our study was the detection of a few genes related to cell proliferation and antiapoptosis, i.e., ID1, ID3, FGFR3, PEG10, ICAM2 and MMP13, the expression levels of which were substantially increased in radioresistant cells compared with radiosensitive cells. These genes may help to disclose the molecular mechanisms of the radioresistance of OSCC and could be radiotherapeutic molecular markers for choosing the appropriate radiotherapy in this disease.
We thank Lynda C. Charters for editing this manuscript.