Classifying the variation in response to ionizing radiation in human cells by gene expression profiling

K Ishikawa^a, K Saegusa PhD^a, Y Ohtsuka^a, A Ishikawa^a, K Yasuda^a, A Furuno^a, S Kawai^a, T Kimura^a, Y Michikawa PhD^a, Y Harada PhD^a, M Iwakawa MD PhD^a, M Suzuki PhD^b, T Imai PhD^a

^aRadGenomics Project, National Institute of Radiological Sciences 4-9-1 Anagawa, Inage, Chiba, ^bInternational Space Radiation Laboratory, National Institute of Radiological Sciences 4-9-1 Anagawa, Inage, Chiba, Japan

AIM: To predict an unusual response to clinical radiations, identification of molecular markers and understanding the variation in radiosensitivity at molecular level are required. In this study, microarray assays were performed on the various cell lines to select the candidate genes as biomakers for radiosensitivity. METHODS: Fourteen cell lines including two normal fibroblast cell strains and 12 cancer cell lines were cultured in Eagle's minimum essential medium (MEM) supplemented with 10% FBS. From survival curve of each cell line, D10 value, i.e. one of the parameters, was calculated. D10 value is the dose (Gy) required to reduce the surviving fraction to 0.1, and was defined as the phenotype of each cell line radiosusceptibility. Total RNA was purified from each cell line before irradiation, one or three hours after irradiation. Fluorescent complementary RNA was prepared by reverse transcription and "in vitro" RNA synthesis with Cy3- or Cy5- labeled nucleotides. Hybridization was performed with the custom-made oligonucleotide microarray containing 22,500 probes (Agilent Technologies, CA). After hybridization, slides were washed and scanned using a confocal laser scammer (Agilent Technologies, CA). Gene expression analysis was performed using RESOLVER (Rosetta Inpharmatics, WA) to search for genes whose levels of expression relate to the value of D10. RESULTS: Several gene sets were constructed as follows: genes induced in the radioresistant cells but not in the sensitive cells after irradiation or genes repressed in the radioresistant cells but not in the sensitive cells after irradiation. Expression ratio and measurement accuracy in the microarray experiments were considered for selection of the gene sets. When some gene sets were subjected to the two-dimensional cluster analysis, the cell lines whose D10 value was more than 6 Gy were clearly separated from the cell lines whose D10 values was relatively low, less than 4 Gy. The cell lines whose D10 value ranged between 4 and 6 Gy were also roughly ordered by the gene sets with the Classification tool of the Resolver. CONCLUSIONS: Detection of the expressed genes selected by the microarray analysis is useful to predict radiosensitive or resistant cells that were clarified by D10 value.

Paper presented at the International Symposium on Predictive Oncology and Intervention Strategies; Nice, France; February 7 - 10, 2004; in poster session 892 (Susceptibility genes).