Genome-Wide Screening for Radiation Response Factors in Head and Neck Cancer


  • Presented at the Meeting of the Eastern Section of the American Laryngological, Rhinological and Otological Society, Inc., Pittsburgh, Pennsylvania, January 30, 2000.

    Supported in part by grants TA32 CA 09685 (b.s.) from the United States Public Health Service and the Young Investigator Award (b.s.) from the American Society of Clinical Oncology.


Introduction Radiation therapy is an integral part of the treatment of head and neck cancer. Factors predicting radiation response are ill defined. The aim of this study was to identify genetic aberrations associated with radiation response in cell lines derived from head and neck squamous cell carcinomas (HNSCC) using comparative genomic hybridization (CGH) for genome-wide screening.

Methods Five cell lines derived from HNSCC were subjected to a single course of radiation (400 cGy) in parallel with a similarly handled, untreated control. Cellular response to radiation was determined on posttreatment days 1, 2, 3, 4, and 5 using a cell viability assay (MTT assay). Radiation response was defined as 35% or greater decrease in cell survival relative to control. Tumor doubling time was determined by cell counts performed at day 0 and 1 for each cell line. All experiments were done in quadruplicate. CGH analysis was performed by differentially labeling DNA from tumor and normal tissue with fluorescent agents. The labeled DNAs were simultaneously hybridized to normal metaphase chromosomes. Image analysis for fluorescence intensity along the entire length of each metaphase chromosome allowed generation of a color ratio, which was used to detect copy number changes.

Results Radioresistance was identified in two of five cell lines. The tumor doubling time was not a predictor of radiation response. CGH identified a complex pattern of aberrations, with gain of 3q common to all cell lines. The number of genetic aberration was higher in radiation-sensitive cell lines than in radiation-resistant ones. No recurrent aberrations were unique to the radiation-resistant cell lines. Recurrent gains at 7p and 17q and losses at 5q, 7q, and 18q were unique to the radiation-sensitive cell lines.

Conclusions The number of aberrations identified by CGH analysis may be a predictor of radiation response. A large study of primary tumors is warranted to confirm this association and identify specific genetic aberrations associated with radiation response.


Radiation therapy is an important treatment modality in the management of patients with head and neck cancer, either as a single agent or in combination with surgery and/or chemotherapy. 1,2 Radioresistance occurs in a significant number of patients with head and neck cancer and often leads to attempts at surgical salvage with resultant organ sacrifice and diminished survival. The identification of radiation response factors may allow more precise treatment selection and improvements in outcome. However, the detection of clinical and laboratory factors associated with radiation response has been inadequate. 2–5 The most promising of the radiation response studies have used genome-wide analytic methods to identify chromosomal loci potentially associated with radiation response, mainly utilizing traditional cytogenetic techniques. 6 One such study, using karyotyping analysis, showed that a clustering of breakpoints at 1p22, 3p21, and 8p11.2 was associated with radiation response and duplications at 11q13 and distal 14q with radiation sensitivity. 6 However, even though cytogenetic analysis has been a significant adjunct in the identification of important genetic aberrations in human cancers, its usefulness in solid tumors is limited by difficulties in culturing and the presence of polyclonal cell populations. 7,8 Moreover, authors have shown that simple manipulations in culturing conditions can alter the karyotypic findings in solid tumors including head and neck cancer, limiting the confidence in observed results. 9

The advent of modern molecular-cytogenetic techniques has significantly enhanced our ability to screen the genetic composition of all cancers. Comparative genomic hybridization (CGH), originally described by Kallioniemi et al. 10 in 1992, has been a particularly important adjunct to the genetic analysis of solid tumors. This technique allows the simultaneous detection of gains, losses, and amplifications of genetic material, making it an appealing screening tool. 11 In preparation for a larger clinical study of acquired clinical samples, we undertook this preliminary study assessing factors associated with radiation response in head and neck cancer cell lines. The aim of this study was to substantiate the role of genome-wide screening with CGH in the identification of genetic aberrations associated with radiation responsiveness in head and neck cancer.


Cell Culture

Five human cell lines derived from HNSCC were used in this study. Three (HN886, MSKQLL2, and SCC 1483) were isolated and characterized by investigators at Memorial Sloan-Kettering Cancer Center and two (SCC 15 and SCC 25) were purchased from the American Type Culture Collection (ATCC, Manassas, VA). For in vitro studies, cells were grown in minimal essential media (MEM) supplemented with 10% fetal calf serum (FCS) at 37°C in 5% CO2 and 100% humidified atmosphere and subcultured once or twice a week.

Cell Growth and Viability Assay

Cells were grown under controlled conditions, as described above. Cell counts at time zero and at 24 hours were used to determine tumor growth and extrapolated to determine the doubling time. The MTT (3-(4,5-dimethyl thiazole-2-yl)-2,5-diphenyl tetrazolium bromide) cytotoxicity assay was used to determine cellular viability. This was performed by harvesting cells from the exponential phase of maintenance cell cultures. A single-cell suspension with a cell density of 5 × 103/200 mL was made and plated in 96-well culture plate (3075, Falcon Plastics, Brookings, SD) in 200-mL volumes (5 × 103 cells/well). After a 20-hour incubation, cells were irradiated using the 137Cs source irradiator to a total dose of 400 cGy (78 cGy/min). At each study time point (days 1, 2, 3, 4, and 5) the culture medium was aspirated from individual wells and cells washed with PBS, and 200 mL of culture medium containing 50 g of MTT was added. After incubation for 4 hours, MTT culture medium supernatant was removed by slow aspiration and replaced with 200 mL DMSO. The absorbance of each well was measured using a microculture plate reader at 540 nM (EL312e, Bio-Tek, Winooski, VT). All assays were performed in quadruplicate. Cell survival was determined relative to a similarly handled, untreated control. A greater than 35% decrease in cell survival relative to the control was used as the cutoff for determining radiation response.

Comparative Genomic Hybridization

The cells from each cell line were harvested and subjected to DNA extraction. In brief, this was performed by washing 1 × 107 of cells in ice-cold PBS buffer followed by suspension in digestion buffer (10 mmol/L Tris-HCL, 10 mmol/L EDTA [pH 8], 0.5% SDS, and 0.1 mg/mL proteinase K). The suspension was incubated for 12 hours at 50°C, and extraction performed once with phenol, twice with 25:24:1 phenol/chloroform/isoamyl alcohol, and once with pure chloroform. The DNA was precipitated and collected using a glass hook. After washing in 70% alcohol, the DNA was re-suspended in 100 mL of TE buffer (pH 8). Quantification of the DNA was performed using electrophoresis in a 0.8% agarose gel, using Eagle Eye II analytic software (Stratagene, La Jolla, CA).

Comparative genomic hybridization analysis (Fig. 1) was performed using 2 μg of DNA from cell lines and placental DNA confirmed to have a normal karyotype. The DNA from each cell line was labeled using nick translation (Gibco-BRL) with Fluorescein-12-dUTP and from normal placenta with Texas red-5-uUTP (NEN-DuPont). Nick translation was performed at 15°C. The reaction was terminated when the product size was 500 to 2000 bp in size, 200 ng of DNA from each cell line was coprecipitated with 200 ng of normal DNA and 15 μg of unlabeled cot-1 DNA. A Sephadex G-50 column was then used to separate the labeled DNAs and 200 ng of normal and cell line DNA was coprecipitated and suspended in a hybridization mix (ONCOR). The suspension was hybridized for 3 days at 37°C to normal metaphase chromosome spreads prepared from phytohemagglutinin (PHA)-stimulated lymphocytes from normal individuals. On completion of hybridization, the chromosomes were counterstained with 4,6-diamino-2 phenylindole (DAPI) to allow their identification. Then, 7 to 10 separate metaphases were captured for each case using a cooled-charge coupled devices (CCD) camera attached to a Nikon Microphot-SA microscope. The metaphases were processed using the Quantitative Image Processing System (QUIPS, Vysis, Downer's Grove, IL). Chromosomes were identified by DAPI banding analysis, segmented, local background subtracted, and the median axis was defined. Red, green, and blue florescence intensities were analyzed for all metaphase spreads, normalized to a standard length, and statistically combined to show the red:green signal ratio and 95% confidence intervals for the entire chromosome. Copy number changes were detected based on the variance of the red:green ratio profile from the standard of 1. Ratio values of 1.2 and 2.0 were defined as thresholds for gains and amplifications, respectively, and losses were defined as ratios of 0.8 or lower (Fig. 1).

Figure Fig. 1..

Comparative genomic hybridization analysis of cell line HN886. A. Metaphase chromosomes after simultaneous hybridization with differentially labeled tumor and normal DNAs. B. Chromosomes from A identified and positioned for image analysis. C. Combined results of the analysis from 10 individual metaphases. The graphic images represent the mean fluorescence intensity ratio profiles with 95% confidence intervals for each chromosome. The findings from the analysis are summarized by bars to the left of each chromosome for losses of genetic information and to the right for gains of genetic information in the tumor relative to the control.


Response to a single dose of radiation for each cell line is shown in Figure 2. Three cell lines were identified as radiation responsive (SCC 1483, HN886, and SCC15) and two as radioresistant (QLL2 and SCC25). The tumor doubling time was not a predictor of radiation sensitivity (Fig. 3). The genetic composition of the cell lines in the study, identified by CGH, is given in Figure 4. Gain at 3q was present in all cell lines. Losses at chromosome 3p, 4, 8p, and 9p, and gains at 5p and 14q occurred in four of the five cell lines. Increasing genetic complexity, defined by the number of genetic aberrations detected by CGH, was associated with radiation responsiveness (Fig. 5). Recurrent genetic aberrations unique to radiation-responsive cell lines were losses at 5q, 7q, and 18q and gains at 7p and 17q. No recurrent loci were unique to the resistant cell lines, but gains at 10q and 22p were uniquely present in one of the radioresistant cell lines.

Figure Fig. 2..

Radiation response in five head and neck cell lines after single course of 400-cGy radiation exposure. Decrease in cell viability measured relative to similarly handled control. The dashed line represents the cutoff for defining radiation response. Note that SCC1483, HN886, and SCC15 are radiosensitive and SCC25 and QLL2 are radioresistant.

Figure Fig. 3..

Correlation of tumor doubling time in head and neck cancer cell lines with radiation response. Hatched boxes represent radiosensitive cell lines and the solid boxes radioresistant ones. Note absence of a relationship between tumor growth rates and radiation response.

Figure Fig. 4..

Copy number changes in five head and neck cancer cell lines defined by comparative genomic hybridization. Bars to the left of the chromosome represent losses of genetic information in the tumor. Thin bars to the right represent gains and thick bars are regions of high-level amplification.

Figure Fig. 5..

Correlation of genetic composition of head and neck cancer cell lines with radiation response. Numbers of aberrations represent a sum of individual gains, losses, and amplifications in each cell line determined by comparative genomic hybridization. Hatched boxes represent radiosensitive cell lines and the solid boxes radioresistant ones. Note a higher number of aberrations in radiosensitive cell lines.


Many authors have proposed a positive correlation between tumor aggressiveness and radiation response. However, this association has been difficult to establish in laboratory studies. Analyses looking at the relationship between tumor growth rates, as a hallmark of aggressive behavior, and radiation response have been ambiguous in their findings. 2–4,12–14 Similarly, histological and ploidy analyses have also failed to yield reproducible results. 2,4,12–14 Our data suggest that inherent genetic complexity, which can be postulated to represent aggressiveness in HNSCC, may be a predictor of radiation response. The precise mechanism responsible for this association remains unclear. It may be that cell lines exhibiting an increased number of CGH-detectable aberrations have abnormalities in DNA repair mechanisms, which also results in a difficulty in coping with sublethal radiation-induced chromosomal damage. This hypothesis is supported by studies reporting that cell lines accumulating and sustaining increased numbers of radiation-induced chromosomal aberrations are more radioresponsive. 14–18 In this regard, several human genes have been implicated in DNA repair and genomic instability. 19 The role of DNA damage repair genes in radiation response has been shown in cell lines, where increasing expression of XRCC1, XRCC3, and RAD51 gene expression has been correlated with radiation resistance. 19–23

The role of specific genetic aberrations in the determination of radioresponsiveness is more perplexing. Alterations can be radioprotective, induce radiation resistance or response, or simply serve as response markers without functional significance. A comprehensive identification of genetic radioresponse phenomena requires the assessment of a larger series of cases, in the in vivo setting. Accepting the limitation of sample size and in vitro assessment, our analysis failed to show recurrent aberrations unique to radioresistant cell lines. There were two nonrecurrent aberrations that were unique to radioresistant cell lines, namely gains at 10q and 22q, the significance of which needs to be determined in a larger series. Several candidate genes are present at each of these loci, such as EWSR1 on 22q, an oncogene that can transform NIH 3T3 cells, which can potentially impact the cellular response to radiation. 24

The presence of recurrent, unique gains in 7p and 17q and losses in 5q, 7q, and 18q was identified in radiation-sensitive tumors. All of these sites contain candidate genes that have been shown to be markers of poor prognosis. APC and MCC aberrations at 5q occur in 25% to 54% of head and neck cancers and have been associated with a poor prognosis. 25,26 On 7q, our G-banding analysis shows the presence of a breakpoint at 7q31, where FRA7G (7q31.2) has been mapped (Singh B, et al., unpublished data). FRA7G has not been studied in head and neck cancer, but has been found to be abnormal in ovarian, prostate, and breast cancers. 27,28 Of all of these sites, the presence of 7p gain in radiosensitive cell lines is most intriguing. Our work using spectral karyotyping has shown that a clustering of breakpoints occurs at 7p11-13 in 13 SCC cell lines, including in SCC1483 (Singh B, et al., unpublished data). This locus includes the gene encoding for epidermal growth factor receptor (EGFR), which has been related to radioresponsiveness in HNSCC. Interestingly, studies have shown that the application of epidermal growth factor enhances the radiation sensitivity of human HNSCC. 29,30 The growth inhibitory effects of epidermal growth factor on squamous cell carcinoma are enhanced with elevated expression of EGFR. 31 However, studies of EGFR expression show a negative correlation with radiation response in many different cancer models, including HNSCC. 32 Given the high rates of expression and the proven role in radiation responsiveness, the interrelationship between EGF and EGRF merits further study.

When interpreting the findings of this study, is important to stress its limitations. 1) As this is a preliminary study, the sample size is small, limiting the power of the analysis. 2) The in vitro model utilized for radiation response assessment is imperfect, because it does not account for the intricate in vivo milieu of host-tumor interactions, such as oxygen tension and immune response. 3) CGH does not identify all possible alterations, missing all balanced aberrations and those less than 5 to 10 Mb in size. Accordingly, balanced translocations and point mutations are missed by this method. Accepting these limitations, this work defined the genetic aberrations and extrapolated the radiation response of five head and neck cancer cell lines, which can serve as a model for future studies of radiation response. The data also suggest that increasing genetic complexity correlates with radiation sensitivity. This association and the role of specific genetic phenomena in radioresponse require further study.


The authors would like to thank Raju S. K. Chaganti, PhD for his invaluable guidance and mentorship, without which this work would not be possible. We would also like to thank Peter Sachs, PhD, for allowing us to use his HNSCC cell lines.