Oncogene (2006) 25, 622–632 & 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc ORIGINAL ARTICLE Transcriptional responses to ionizing radiation reveal that p53R2 protects against radiation-induced mutagenesis in human lymphoblastoid cells

M-H Tsai1, X Chen1, GVR Chandramouli2, Y Chen3, H Yan1, S Zhao1, P Keng4, HL Liber5, CN Coleman1, JB Mitchell1 and EY Chuang1,6

1Radiation Biology and Oncology Branches, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA; 2Advanced Technology Center, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA; 3Cancer Genetics Branch, National Research Institute, National Institutes of Health, Bethesda, MD, USA; 4Radiation Oncology, School of Medicine and Dentistry, University of Rochester, Rochester, New York, USA; 5Department of Environmental and Radiological Health Sciences, Colorado State University, Fort Collins, CO, USA and 6Department of Electrical Engineering, National Taiwan University, Taipei, Taiwan

The has been implicated in multiple cellular induced by p53 protein and is involved in protecting responses related to DNA damage. Alterations in any of against radiation-induced mutagenesis. these cellular responses could be related to increased Oncogene (2006) 25, 622–632. doi:10.1038/sj.onc.1209082; genomic instability. Our previous study has shown that published online 10 October 2005 mutations in p53 lead to hypermutability to ionizing radiation. To investigate further how p53 is involved in Keywords: microarray; radiation; p53R2; TK6 regulating mutational processes, we used 8K cDNA microarrays to compare the patterns of expression among three closely related human cell lines with different p53 status including TK6 (wild-type p53), NH32 (p53- Introduction null), and WTK1 (mutant p53). Total RNA samples were collected at 1, 3, 6, 9, and 24 h after 10 Gy c-irradiation. Ionizing radiation induces damage to DNA, and this is Template-based clustering analysis of the followed by a variety of cellular responses such as cell- over the time course showed that 464 are either up cycle arrest, transformation, and cell death (Lakin and or downregulated by at least twofold following radiation Jackson, 1999; Gong and Almasan, 2000; Backlund treatment. In addition, cluster analyses of gene expression et al., 2001). The p53 protein has been implicated in profiles among these three cell lines revealed distinct multiple cellular responses related to radiation-induced patterns. In TK6, 165 genes were upregulated, while 36 DNA damage including apoptosis, cell cycle control, as genes were downregulated. In contrast, in WTK1 75 genes well as DNA replication and repair. Our recent work were upregulated and 12 genes were downregulated. In focused on radiation-induced gene mutations with NH32, only 54 genes were upregulated. Furthermore, we respect to different p53 status. The human B-lympho- found several genes associated with DNA repair namely blast cell lines WTK1, NH32, and TK6 are derived from p53R2, DDB2, XPC, PCNA, BTG2, and MSH2 that the same progenitor, WIL2; all are heterozygous at the were highly induced in TK6 compared to WTK1 and autosomal thymidine kinase (TK) (Skopek et al., NH32. p53R2, which is regulated by the tumor suppressor 1978; Benjamin et al., 1991; Chuang et al., 1999), but p53, is a small subunit of ribonucleotide reductase. To differ in their p53 status. WTK1 (Little et al., 1995; Xia determine whether it is involved in radiation-induced et al., 1995) overexpresses a mutant form of p53 mutagenesis, p53R2 protein was inhibited by siRNA in (methionine to isoleucine substitution at codon 237), TK6 cells and followed by 2 Gy radiation. The back- whereas TK6 is wildtype for p53. NH32 is a double p53 ground mutation frequencies at the TK locus of siRNA- knockout cell line derived from TK6 by using a transfected TK6 cells were about three times higher than promoterless gene-targeting approach (Chuang et al., those seen in TK6 cells. The mutation frequencies of 1999). These lines respond quite differently to ionizing siRNA-transfected TK6 cells after 2 Gy radiation were radiation. After irradiation, NH32 and WTK1 cells significantly higher than the irradiated TK6 cells without showed a delayed and reduced X-ray-induced apoptosis p53R2 knock down. These results indicate that p53R2 was compared with TK6 cells (Yu and Little, 1998; Coelho et al., 2002). In addition, the background and radiation- induced mutation frequencies of NH32 cells were Correspondence: Dr EY Chuang, Radiation Biology Branch, Center approximately equal (Chuang et al., 1999) or somewhat for Cancer Research, National Cancer Institute, Building 10, Room higher (Peng et al., 2002) as TK6 cells at the TK locus, B3-B69, Bethesda, MD 20892, USA. whereas WTK1 was much more sensitive to sponta- E-mail: [email protected] Received 26 May 2005; revised 27 July 2005; accepted 31 July 2005; neously arising and radiation-induced mutation published online 10 October 2005 (Chuang et al., 1999). Thus, these cell lines provide a p53R2 protects against radiation-induced mutagenesis M-H Tsai et al 623 powerful system to study the effect of radiation-induced WTK1 downregulated subcluster; in addition, GAPD p53-dependent or independent pathways. was chosen as a control gene. A very good agreement With the recent development of DNA microarray between microarray data and the real-time RT–PCR technology it is possible to analyse gene expression analysis was achieved (Figure 2). patterns for large numbers of genes in a single experiment (Schena et al., 1995). In this study, we used Patterns of gene expression of different function the cDNA microarray to identify TK6, NH32, and categories WTK1 gene expression patterns after 10 Gy g-ray To better exhibit the different expression profiles caused radiation at different time points. After clustering by irradiation among the three cell lines, the responses analysis, there were many p53-dependent or DNA of selected genes over the time course were extracted repair-related genes that were upregulated in TK6 cells. from the hierarchical map in Figure 1B. For example, Among these genes, p53R2, which has been identified by some DNA repair-associated genes including p53R2, recent studies (Nakano et al., 2000; Tanaka et al., 2000), DDB2, XPC, PCNA, BTG2, and MSH2 from showed the highest expression level in TK6 cells. Figure 1B were selected and replotted in Figure 3a. To determine whether p53R2 is involved in radiation- They showed a distinct upregulation pattern in TK6 induced mutagenesis, p53R2 protein levels were reduced cells. Specifically, p53R2 showed the most increase of by siRNA in TK6 cells and followed by 2 Gy g-ray expression ratio and this was sustained throughout the radiation. The background mutation frequencies of time course, reaching maximal levels at 9 h after siRNA-transfected TK6 cells were three times higher irradiation. Some p53-related genes, such as BTG2, than untransfected TK6 cells at the TK locus. However, PPM1D, and TP53TG1, were also selected. They the p53R2 knock-down cells were much more sensitive showed upregulation in TK6 cells, but not in NH32 or to radiation-induced mutation. Furthermore, knock- WTK1 (Figure 3b). Genes that were distinctively down of p53R2 did not alter the sensitivity of the cells to upregulated in WTK1 and NH32 cells were also radiation-induced cell killing, as surviving fractions were selected. In Figure 3c, genes displayed were found unchanged. Our results suggest that p53R2 exhibits a significantly upregulated in WTK1, and all of them p53-dependent response and plays a pivotal role in relate to cell signal transduction. In Figure 3d, some preventing radiation-induced mutations. filament-related genes, such as KRT19, COL6A1, and COL6A2, were significantly upregulated in NH32 cells. Results Irradiation-mediated gene expression Gene expression clustering analysis We screened through annotations of the 464 genes To identify genes normally induced or repressed by found in this study and identified 66 genes that can be ionizing radiation, we applied a template-based cluster- grouped into nine subcategories of biological signifi- ing algorithm, a method developed in our laboratory cance (Table 1). Included in the table are the ratios which helps to reduce systematic errors that are between the maximal/minimal induction and the time at detectable by multiple time point measurements which either the maximal or minimal induction was (Chuang et al., 2002). A total of 464 genes showed achieved. Close inspection of these genes revealed temporal changes that were significant (>twofold) and distinct patterns between TK6 and the other two cell lines. For example, some of these subcategories showed consistent (correlation coefficient of fitting rk >0.85). The hierarchical clustering map is shown in Figure 1B. elevated expression levels in TK6 cells but not in other In order to assure that the changes were induced by two types of cells. This included genes related to p53, radiation and not a result of genetic differences among TGF-signaling pathway, cell cycle, DNA damage/ the three cell lines, preirradiation expression data of the repair, and signal transduction. Among genes involved same 464 genes were collected and clustered. The in apoptosis, more were observed in TK6 than the other differences of gene expression patterns among three cell two cell lines. This was confirmed by the result of lines before irradiation were in fact ignorable apoptosis assay conducted on all three cell lines. TK6 (Figure 1A). Given that, we conducted gene expression displayed rapid induction of apoptotic cells after 24h analysis based on Figure 1B and identified the following compared to the other two cell lines (data not shown). trends. The expression of the selected 464 twofold Some expression patterns that were new in connection changes genes showed distinct patterns among the three with irradiation were also observed. For example, cell lines. Several subclusters can be identified. In all, 175 collagen and extracellular matrix have been known to genes in TK6 were upregulated and 36 were down- proliferate excessively in response to irradiation. How- regulated. In WTK1, 75 were upregulated and 12 were ever, significant upregulation of filament-related genes downregulated. In NH32, only 54were upregulated and was only found in WTK1 and NH32 cells but not in no genes were prominently downregulated. TK6 cells. To verify the above results, the real-time RT–PCR analysis was performed for 10 selected genes at the 24h Effect of p53R2 protein knock-down in radiation-induced time point for each cell line. These included seven genes toxicity and mutagenicity from the TK6 upregulated subcluster, one gene from the The observed upregulation of p53R2 in the TK6 cell line NH32 upregulated subcluster and one gene from the and its nascent yet unclear role in radiation-induced

Oncogene p53R2 protects against radiation-induced mutagenesis M-H Tsai et al 624

Figure 1 (A) Hierarchical clustering of RNA expression observed for untreated TK6, NH32, and WTK1 cells of 464 twofold change genes using the Pearson correlation and complete linkage. (B) Hierarchical clustering of RNA expression observed for TK6, NH32, and WTK1 cells following by 10 Gy radiation using Pearson correlation and complete linkage. Average expressions of the three replicates of 464 twofold changes in genes are shown in logarithmic scale. All expressions shown are relative to untreated cells. For each cell line, expressions shown are for 0, 1, 3, 6, 9, and 24h after irradiation (left to right). Selected subclusters (a–e) are shown on the right. Panels a and d show the clusters of TK6 up- or downregulated gene, panel b and e are the clusters of WTK1 up- and downregulated genes, panel c is the cluster of NH32 upregulated genes.

mutagenesis made it a target for in-depth analysis in this p53R2 protein knock-down showed a small but study. A siRNA targeted to p53R2 in normal TK6 cells nonsignificant decrease compared to normal TK6 cells was utilized in a transfection and the Western blot (Figure 4b). The toxicity induced by 2 Gy g-ray analysis is shown in Figure 4a. The level of knock-down radiation in TK6 cells was also not influenced by of p53R2 protein was around 80% 1 day after siRNA p53R2 knock-down (survival of 7.571.3% and transfection, but this decreased to about 40% after 9.471.6%, respectively). However, knock-down of 2 days and 10% after 3 days. No knock-down effect of p53R2 did have significant effects on mutagenesis at p53R2 protein was observed in transfected controls. the TK locus in TK6 cells (Figure 4c). Without Therefore, knock-down was successful. irradiation, the background TKÀ mutation frequencies To further investigate whether p53R2 affected radia- of TK6 cells were elevated after knock-down, from tion sensitivity, g-ray-induced toxicity and mutagenicity 573 Â 10À6 in controls to 1475 Â 10À6 in knockdowns. were compared in TK6 cells 1 day after knock-down The mutation frequencies induced by 2 Gy of radiation with p53R2. The plating efficiency of TK6 cells after at the TK locus were significantly higher in p53R2

Oncogene p53R2 protects against radiation-induced mutagenesis M-H Tsai et al 625 differences, we clustered the expression data of the 464 genes obtained before irradiation (Figure 1A). All three cell lines showed similar patterns, and expression levels of the 464 genes were close. This suggested that the one known genetic difference among the three cell lines, namely p53 status, has little effect on gene expression among the selected 464 genes in unirradiated cells. The gene influence was observed in the whole gene set level. TK6 and NH32 were found to have similar expression but differed from WTK1. This was expected since NH32 is directly derived from TK6 (two cloning steps), but WTK1 is separated by at least six cloning steps from TK6. This difference did not affect the expression profile of our selected genes, as indicated in Figure 1. In contrast, after irradiation gene expression patterns of the three cell lines differed dramatically, and expression levels of these genes changed more than twofold from their unirradiated state (Figure 1B). This observation suggests that p53 does not exert its effect on the expression of these genes until after irradiation. We therefore hypothesize that the changes in expression level of the 464 genes in our experiment might be linked to p53 status. Many of these genes, found to be related directly to p53 in the TK6 cell line, could yield interesting results with further experimentation. The identification and analysis of p53R2 demonstrates the potential of this approach. We are working on other genes in this pool with the hope of identifying as yet unreported p53-dependent genes. In addition to investigating individual genes that may be related to the p53-dependent response to irradiation, we tried to identify pathways with such a dependency. We anticipated that the acute dose of irradiation could trigger responses from genes in certain signaling path- ways, and activation of these signaling pathways could also cause changes in the expression of their member genes. Therefore, if the number of genes within one pathway showed significant changes in expression level, this pathway may very likely play a role in p53- dependent response to irradiation. The use of TK6, Figure 2 Validation of cDNA microarray data using the real-time RT–PCR analysis. The expression ratios between 24h after NH32, and WTK1, with their difference in p53 status, irradiation and 0 h (unirradiated control) for TK6 (a), WTK1 (b), helped to reveal the p53 dependency of the pathways. or NH32 (c) are compared for nine selected genes and GAPD (used Out of 161 upregulated genes in TK6, 19 of them as a control gene). Ratios >1 means upregulation, whereas ratio were reported as p53-related. Among these genes, o1 means downregulation. The RNAs used for the real-time RT– MSH2 was reported as a essential repair components PCR analysis were taken from one set of the three replicated experiments, and the data shown represent the average of three for DNA homologous recombination repair (Evans independent real-time RT–PCR reactions. et al., 2000; Elliott and Jasin, 2001), and can be co- immunoprecipitated with p53 (Yang et al., 2004). These data suggested that MSH2 formed a complex with p53 in response to the damaged DNA during double- protein knock-down cells, at 3927107 Â 10À6, compared stranded break repair. Moreover, DDB2 and XPC were to irradiated normal TK6 cells at 84740 Â 10À6. This reported in p53-dependent global genomic repair difference was statistically significant (Po0.05). (Sugasawa et al., 2001; Tan and Chu, 2002; Fitch et al., 2003) and also showed a linear dose response relationship between 0.2 and 2 Gy at 24and 48h after ionizing radiation in human peripheral blood lympho- Discussion cytes (Amundson et al., 2000). This high ratio suggests that irradiation effectively influenced the expression of To address the question of whether the different p53-related genes. It is therefore reasonable to hypothe- expression levels observed among the three cell lines size the discovery of novel p53-related genes among the are caused by irradiation, or just reflect genetic rest of the upregulated genes.

Oncogene p53R2 protects against radiation-induced mutagenesis M-H Tsai et al 626

Figure 3 (a) Selected DNA repair-related genes from TK6 upregulated subcluster, (b) selected p53-related genes from TK6 upregulated subcluster, (c) selected genes from WTK1 upregulated subcluster, and (d) the genes from NH32-upregulated subcluster.

Among the 161 significantly upregulated genes in the and NH32 with null p53. However, the genes involved TK6 cell line, eight are components of an apoptosis (TNFRSF8, NFKB1, NFKB2, TRAF1, and TNFAIP3) pathway (GG2-1, Fas, TANK, CASP1, TNFSF10, are different than the set observed in TK6. This STAT1, NFKB1E, and CFLAR), and all are tumor observation demonstrated a connection between acute necrosis factor related genes. As shown in the time irradiation and increased apoptotic gene activities, and course data in Figure 1B, acute irradiation activated an yet the connection involves different genes depending on apoptosis pathway in p53 wild-type cells at around 3 h, the p53 status of their cell lines. The signaling pathway and gene activities remained high through 24h. Such of NFkB has been shown to have roles in both concerted activation of multiple genes in the apoptosis protective and apoptotic responses after genotoxic stress pathway was also observed in WTK1 with mutant p53 (Holbrook et al., 1996; Smith and Fornace, 1996). For

Oncogene p53R2 protects against radiation-induced mutagenesis M-H Tsai et al 627 Table 1 Gene expression for TK6, NH32, and WTK1 lines according to functional categories (only representative genes are shown here, the complete gene list can be found at http://home.ncifcrf.gov/ROSP-Microarray-Lab/p53r2/Table1_complete_gene_list.htm) Name CloneID Cell lines

TK6 NH32 WTK1 p53-related genes Protein phosphatase Wip1 (PPM1D) IMAGE:243882 2.59 (6) 1.74(6) 1.24(3) Pig10 ¼ p53-inducible gene ¼ Similar (ENC1) IMAGE:510856 3.8 (24) 1.28 (9) 1.85 (24) TP53 target gene 1 (TP53TG1) IMAGE:506646 2.05 (9) 1.3 (24) 1.11 (6) Pig3 ¼ p53-inducible gene (PIG3) IMAGE:859359 5.09 (9) 1.44 (3) 1.21 (6) RB ¼ retinoblastoma susceptibility (RB1) IMAGE:487777 2.21 (24) 1.17 (6) 2.35 (24) Small-inducible cytokine A3 (SCYA3) IMAGE:460398 6.23 (24) 4.2 (24) 2.57 (9) Cyclin-dependent kinase inhibitor 1A (CDKN1A) IMAGE:2549557 4.36 (9) 1.38 (24) 1.5 (24) p53-inducible-ribonucleotide reductase (p53R2) IMAGE:768466 4.96 (9) 1.22 (9) 1.12 (9)

Apoptosis TNF-induced protein (GG2-1) IMAGE:627401 2.32 (6) 1.41 (3) 1.4 (24) CD95 ¼ Fas (TNFRSF6) IMAGE:714213 3.1 (9) 1.41 (9) 1.89 (9) TRAF family member-associated NF (TANK) IMAGE:502486 2.63 (24) 1.33 (1) 1.28 (24) Caspase 1 (CASP1) IMAGE:120106 2.48 (24) 1 (0) 1.68 (9) TRAIL ¼ Apo-2 ligand (TNFSF10) IMAGE:203132 2.43 (24) 1.38 (3) 1.29 (3) STAT1 ¼ IFNa/b-responsive (STAT1) IMAGE:840691 3.67 (24) 1.35 (24) 1.62 (24) Nuclear factor of k light (NFKBIE) IMAGE:1573311 2.07 (24) 1.27 (24) 1.92 (6) CASP8 and FADD-like apoptosis regulator (CFLAR) IMAGE:8137142.18 (24)1.02 (6) 1.26 (6) CD30 ¼ Ki-1 antigen ¼ TNFR family (TNFRSF8) IMAGE:505538 1.57 (3) 2.83 (24) 2.34 (6) NFkB1 ¼ NF-kB p105 ¼ p50 (NFKB1) IMAGE:789357 1.57 (6) 1.55 (24) 2.08 (9) NFkB2 ¼ NF-kB p100 ¼ p49 ¼ p50B (NFKB2) IMAGE:682529 1.28 (3) 3.21 (24) 4.05 (9) TRAF1 ¼ TTNF -associated factor 1 (TRAF1) IMAGE:155583 1.61 (24) 3.91 (24) 3.55 (6) Tumor necrosis factor (TNFAIP3) IMAGE:770670 1.31 (6) 2.13 (24) 1.72 (9)

TGF-b signaling pathway TGF-b receptor type III (TGFBR3) IMAGE:209655 2.08 (9) 1.28 (9) 1.21 (24) Homo sapiens cDNA: FLJ23037 fis, clone IMAGE:345935 2.87 (24) 1.14 (9) 1.21 (9) E1A binding protein p300 (EP300) IMAGE:309591 2.07 (9) 1.09 (6) 1.34(9) P8 protein (P8) IMAGE:80484 2.89 (24) 1.9 (24) 2.3 (9) Prostate differentiation factor (PLAB) IMAGE:788832 3.28 (24) 2.07 (24) 1.57 (3)

Cell cycle regulated genes Cyclin I (CCNI) IMAGE:248295 2.77 (24) 1.2 (6) 1.15 (24) Cyclin K ¼ CPR4 (CCNK) IMAGE:855949 4.2 (9) 1.24 (9) 1.17 (24) Cyclin E2 (CCNE2) IMAGE:826273 3.18 (6) 2 (24) 2.35 (9) Cyclin G1 (CCNG1) IMAGE:547058 2.14 (24) 1.24 (24) 1.12 (6) Cyclin G2 (CCNG2) IMAGE:823691 2.1 (24) 1.34 (1) 1.89 (6) Cyclin F (CCNF) IMAGE:455128 4.04 (6) 1.23 (24) 1.34 (24) Chaperonin containing TCP1, subunit 3 (CCT3) IMAGE:2577143 2.53 (24) 1.33 (24) 1.36 (24) Chaperonin containing TCP1, subunit 4 (CCT4) IMAGE:897880 2.01 (9) 1.64 (24) 1.57 (24) Chaperonin containing TCP1, subunit 7 (CCT7) IMAGE:882484 2.04 (24) 1.18 (9) 1.28 (24) DP-2 ¼ DP-2 (TFDP2) IMAGE:814101 2.3 (24) 1.38 (24) 1.55 (24) Cyclin-dependent kinase 6 (CDK6) IMAGE:214572 1.74 (6) 1.6 (24) 2.19 (9) Cyclin-dependent kinase inhibitor 1C (CDKN1C) IMAGE:2413955 1.27 (1) 2.16 (6) 1.71 (3)

Filament-related genes Collagen, type IV, a2 (COL4A2) IMAGE:769959 1.15 (1) 1.66 (9) 2.19 (9) Collagen, type VI, a1 (COL6A1) IMAGE:263716 1.57 (1) 2.92 (24) 2.25 (9) Collagen, type VI, a2 (COL6A2) IMAGE:2409635 1.22 (3) 2.07 (9) 2.06 (9) Procollagen-lysine, 2-oxoglutar (PLOD3) IMAGE:810928 1.8 (3) 1.89 (24) 2.04 (9) Ficolin (FCN3) IMAGE:858877 1.21 (3) 4.18 (6) 1.6 (9) Keratin 8 (KRT8) IMAGE:897781 1.62 (24) 1.91 (9) 2.09 (9) Keratin 14 (KRT14) IMAGE:183602 1.41 (6) 2.33 (24) 2.43 (3) Keratin 19 (KRT19) IMAGE:810131 1.43 (9) 1.77 (6) 2.01 (9) Integrin, a2 (ITGA2) IMAGE:811740 1.24 (6) 4.11 (6) 1.5 (9) Microtubule-associated protein 7 (MAP7) IMAGE:144834 1.3 (3) 1.36 (1) 2.01 (9) Plasminogen activator, urokinase (PLAU) IMAGE:714106 1.53 (1) 1.56 (24) 2.2 (6)

DNA damage/repair genes PCNA ¼ proliferating cell nuclear (PCNA) IMAGE:789182 2.18 (6) 1.67 (6) 1.35 (6) MSH2 ¼ DNA mismatch repair mutS homolog 2 (MSH2) IMAGE:630013 2.29 (6) 2.02 (24) 1.31 (24) Damage-specific DNA binding protein (DDB2) IMAGE:753447 2.7 (9) 1.67 (24) 1.18 (24) GADD153 IMAGE:361456 2.09 (9) 1.61 (9) 1.7 (9)

Signal transduction genes Interferon-g receptor 1 (IFNGR1) IMAGE:47900 2 (24) 1.37 (24) 2.5 (1)

Oncogene p53R2 protects against radiation-induced mutagenesis M-H Tsai et al 628 Table 1 (continued ) Name CloneID Cell lines

TK6 NH32 WTK1

Signal-induced proliferation-associated gene 1 (SIPA1) IMAGE:593114 2.19 (9) 2.05 (24) 1.74 (24) coactivator 3 (NCOA3) IMAGE:502333 2.16 (24) 1.35 (24) 1.38 (6) Protein tyrosine phosphatase, (PTPN11) IMAGE:1604158 1.86 (9) 2.89 (9) 1.44 (9)

Trascription factor-regulated genes Transcription factor 21 (TCF21) IMAGE:461351 2.54(3) 3.14(24) 2.17 (9) Basic transcriptional factor (BATF) IMAGE:1558151 1.98 (6) 2.42 (1) 2.31 (6) Nuclear factor (NFE2L1) IMAGE:755821 1.22 (9) 1.31 (6) 2.06 (9) Regulatory factor X, 5 (RFX5) IMAGE:767753 1.55 (9) 2.14(24) 1.23 (9)

Stress/oxidative related genes sgk ¼ putative serine/threonine protein kinase (SGK) IMAGE:840776 2.2 (24) 1.43 (6) 1.47 (1) Cytochrome c oxidase subunit VI (COX7B) IMAGE:258120 1.44 (24) 1.59 (24) 2.47 (1) Carnitine acetyltransferase (CRAT) IMAGE:744417 1.33 (24) 1.22 (9) 2.03 (1) NADH dehydrogenase ubiquinone b subcomplex (NDUFB5) IMAGE:307933 2.18 (24) 1.16 (6) 1.24 (6) UQCRC2—ubiquinol-cytochrome c reductase core protein 2 (HK) IMAGE:255842.21 (24)2.37 (1) 2.9 (6)

Data in table are maximum/minimum ratios. Numbers in parentheses are the time point at which the maximum or minimum ratio occurred.

example, activation of NFkB by ionizing radiation has analysis indicated that CCT is essential for CCNE1 been associated with enhanced survival in colon cells maturation and accumulation (Won et al., 1998); it (Russo et al., 2001) and lymphoma cells (Kawai et al., therefore will be interesting to find evidence of possible 1999). It was also found to enhance transcription of coordination between cyclins and CCT and their antiapototic genes, such as TRAF1, TRAF2, IAP1, relation to radiation. IAP2 (Wang et al., 1998), and A 20 (Krikos et al., 1992). Components of the transforming growth factor-b Therefore, the observation of NFkB activation showing (TGF-b) signaling pathway were also found to be a universal response to radiation stress in our experi- significantly upregulated in TK6 (TGFBR3, EP300, ment came as no surprise. The interesting fact, however, NFKB1, P8, and PLAB). TGF-b is a multifunctional is that the activation of apoptotic genes are seemingly cytokine known to modulate several tissue develop- correlated to their p53 status. No clear relationship ments and repair processes (Wakefield and Roberts, between NFkB and p53 has been defined in the 2002). After ionizing radiation, TGF-B1 can be rapidly literature. Webster and Perkins (1999) suggested that activated in mouse mammary gland (Barcellos et al., activity of NFkB may directly oppose the proapoptotic 1994). A recent study reported that p53 status can affect function of p53 activation through competition for the responses to TGF-B1 and vice versa (Teramoto et al., p300/CREB-binding protein transcriptional coactivator 1998). Both p53 and TGF-B1 are induced by a variety of complexes. However, evidence of cooperation between cytotoxic agents, specifically ionizing radiation, and p53 and NFkB also exists (Wadgaonkar et al., 1999), both can induce apoptosis and cell cycle arrest. The (Yang et al., 2000). Our data provides evidence for the early loss of TGF-B1 could cause reduced action of p53 existence of a relationship between p53 and the NFkB (Ewan et al., 2002). The reverse could also be true, that pathway. To elucidate their hypothetical opposing/ is, loss activity of p53 could reduce action of TGF-B1. cooperating relationship, further experiments are This coincided with our data in which wild-type p53 cell needed and the three cell lines and microarray analysis line displayed high TGF-b gene expression while mutant used in this experiment can be of great use. and null showed none. Six cyclin genes (CCNI, CCNK, CCNE2, CCNF, Our survey of radiation-influenced genes in NH32 CCNG1, and CCNG2) were found to be significantly and WTK1 showed that a large number of filament- upregulated in the TK6 cell line. Given the fact that related genes are upregulated. There are five collagen- cyclins are regulated by NFkB, it is not surprising to see related genes (COL6A1, COL6A2, COL4A2, PLOD3, their accumulation in the cell line in which genes of FCN3), three keratin genes (KRT8, KRT14, KRT19), NFkB pathway are upregulated. When relating cyclin one integrin (ITGA2) and one microtubule-associated expression to p53 status, we found upregulation of protein (MAP7). Upregulation of lesser degrees was cyclins in TK6 p53 wild-type but not in WTK1 mutant observed in TK6. It is known that one common late and NH32 null cell lines. It is premature to conclude any complication of radiation therapy is tissue fibrosis, connection based on this observation. In fact, a recent characterized by accumulation of collagen and extra- study indicated that cyclin B1 and cyclin D may play a cellular matrix, and excessive proliferation of fibroblast. role in p53-independent radiation resistance (Chen et al., Specific alterations in gene expression have been 2002). We also found that three chaperonin containing associated with the development of fibrosis following TCP1 complex (CCT3, CCT4, and CCT7) are signifi- radiation injury and include upregulation of tenascin-C cantly downregulated in the TK6 cell line. Mutational (Geffrotin et al., 1998) and plasminogen activator-C

Oncogene p53R2 protects against radiation-induced mutagenesis M-H Tsai et al 629 provides precursors for DNA synthesis during the S-phase and functions in DNA repair. The amino-acid sequence of p53R2 showed three characteristic domains: two putative nuclear localization signal sequences and one RR small subunit signature (Tanaka et al., 2000). This gene contains a p53-binding site in intron 1 and responds to several stress signals that activate the p53 response (Tanaka et al., 2000). UV light, g-irradiation, or adriamycin treatment can induce p53R2 protein expression in a p53-dependent manner (Lozano and Elledge, 2000; Tanaka et al., 2000; Guittet et al., 2001; Yamaguchi et al., 2001; Lin et al., 2004). Tanaka et al. (2000) reported that the discovery of p53R2 suggests a relationship between RR activity and repair of dam- aged DNA and that p53R2 supports DNA repair by increasing the dNTP pools needed for repair. Xue et al. (2003) showed that hRRM1, hRRM2, and p53R2, three human RR subunits, translocate from the cytoplasm to the nucleus in response to UV irradiation. Inhibition of endogenous p53R2 expression in cells that have a p53-dependent DNA damage checkpoint led to reduc- tion in RR activity, DNA repair, and cell survival after exposure to various genotoxins (Tanaka et al., 2000). Based on these findings, it has been proposed that p53R2 mutations result in cancer susceptibility and this underscores the importance of understanding the role of p53R2 in cancer. Although p53R2 has been clearly defined in the DNA repair pathway, however, its role in radiation mutagenesis remains to be elucidated. In the present study, we have shown that suppression of p53R2 gene expression resulted in a significant increase in radiation-induced mutability. In a previous study by Tanaka et al. (2000), inhibition of p53R2 with antisense DNA caused a significant decrease in in- corporation of dNTPs into DNA, following DNA damage induced with UV light or adriamycin. Kimura et al. (2003) showed that dNTP pools were severely Figure 4 (a) The effect of p53R2 siRNA on protein level in TK6 attenuated in p53R2À/À mice under , and cells. Protein samples were collected from normal TK6 cells (c)or p53R2 deficiency caused higher rates of spontaneous transfected with control buffer (TC) or at 1, 2, and 3 days after À/À siRNA transfection. Aliquots of 20 mg protein from each sample mutation in the kidney of p53R2 mice. Taken were subjected to 12% SDS–PAGE. p53R2 and cytochrome c together, it appears that p53R2 plays an important role were detected by enhanced chemiluminescence system in repairing DNA damage. A disruption of p53R2 gene using polyclonal antibodies. (b, c) The effect of p53R2 knock-down expression would be expected to hinder the production on survival and mutation frequency at the TK locus in TK6 cells of RNR, which in turn reduces dNTP pools. Such following 2 Gy radiation. TK6 cells were transfected with or without p53R2 siRNA for 1 day, then irradiated with 0 or 2 Gy of damage might cause misincorporation of dNTP into irradiation. The TKÀ mutation frequencies of TK6 cells with or DNA and lead to accumulation of spontaneous or without p53R2 protein knock-down were 1475 Â 10À6 and induced mutations. 573 Â 10À6, respectively. The TKÀ mutation frequencies of 2 Gy The basal level of all 464 twofold changes genes in irradiated TK6 cells with or without p53R2 protein knock-down were 3927107 Â 10À6 and 84740 Â 10À6, respectively. Each point is three cell lines were clustered and showed in Figure 1A. the mean of three independent experiments. The differences of gene expression patterns among three cell lines before irradiation were in fact ignorable. For p53R2, the ratios of unirradiated RNA against common reference RNA were 1.18, 0.83, and 0.71 in TK6, NH32, (Zhao et al., 2001). Upregulation of plastimogen and WTK1, respectively. After 80% knock down of activator (PLAU) was also observed in our experiment. p53R2 protein, the level of p53R2 in TK6 cells should be It will be interesting to investigate the cause(s) of higher lower than NH32. In addition, the TK mutation degrees of accumulation of collagen in p53 null and frequency after p53R2 knock-down was higher than mutant than in wild type. the NH32 spontaneous rate (Chuang et al., 1999). p53R2 was recently discovered and encodes a 351- We observed that siRNA knock-down of p53R2 amino-acid peptide similar to the human ribonucleotide only slightly decreased survival of g-irradiated cells. This reductase (RR) small subunit 2 (R2), an enzyme that indicated that, although cells were damaged by irradia-

Oncogene p53R2 protects against radiation-induced mutagenesis M-H Tsai et al 630 tion, and presumably were deficient in the DNA repair Apoptosis analysis mechanism involving p53R2, their survival was - SubG1 peak was used to detect apoptotic cells. Cells were tively unaffected. A previous study showed that RNR collected 1, 2, 3, and 4days after treatment, washed in ice-cold can be formed via two different pathways, one PBS, fixed with 75% ice-cold ethanol and stored at 41C for at composed of p53R2 and R1 and controlled by p53, least 24h. The fixed cells were resuspended in 1 ml RNase (1 mg/ml) for 30 min, centrifuged, and resuspended in 0.5 ml the other composed of R2 and R1, and independent of propidium iodide (10 mg/ml). All samples were analysed on an p53 (Tanaka et al., 2000). Another study also showed Epics Elite ESP (Coulter Electronics, Hialeah, FL, USA) set to that DNA damage caused elevated levels of the R2 collect 10 000 events. Cell cycle and subG1 peak were analysed protein and dNTPs, and consequently enhanced the to determine the percentage of apoptosis. survival of p53(À/À) cells (Lin et al., 2004). This suggests that R2 can be employed to supply dNTPs for DNA damage repair, independent of p53R2, there- Probe labeling and microarray hybridization fore independent of p53. Altogether, we propose an Methods for the probe labeling reaction and microarray hybridization were as described previously (Chuang et al., hypothesis where lack of p53R2 results in a different, 2002). Briefly, 20 mg of sample RNA or 40 mg universal human but error-prone repair to account for the lack of a reference (Stratagene, La Jolla, CA, USA) RNA was labeled difference in survival but a big increase in mutation with Cy3 and Cy5, respectively, by using Superscript II frequency. Reverse Transcriptase (Invitrogen, Carlsbad, CA, USA). The In summary, we have shown that p53 does not exert samples were incubated at 421C for 60 min. The reaction was its effect until after irradiation, a novel observation that stopped with 5 ml of 0.5 M EDTA and 1 M NaOH was added to serves as the basis for our effort to identify novel p53- hydrolyse the residual RNA; the samples were then incubated dependent genes from the different expression profiles at 651C for 1 h. Samples were purified in Bio-6 Chromatograph among the three cell lines. The adoption of microarray columns (Bio-Rad, Hercules, CA, USA) and concentrated analysis in this study helps to identify genes and with Microcon 30 membranes (Millipore Corp., Bedford, MA, USA) to 17 ml. The cDNA samples were mixed with 1 ml pathways that react concertedly to irradiation in a COT1-DNA (10 mg/ml) (Invitrogen, Carlsbad, CA, USA), p53-depenedent manner and is shown to be an effective polyA (8–10 mg/ml) (Amersham Pharmacia Biotech, Piscat- tool in such pursuit. We have also shown that blocking away, NJ, USA), and yeast tRNA (4 mg/ml) (Ambion, Austin, p53R2 gene expression in p53 wild-type cells signifi- TX, USA). The mixed samples were denatured at 1001C for cantly increased mutagenicity but did not affect cell 1 min and cooled on ice prior to hybridization onto microarray survival. These results further support recent discoveries slides. that p53R2 might play a pivotal role in DNA repair. The microarray slides used for the experiments contained 7680 human cDNA clones and the detail of this array and the methods for microarray fabrication were described previously (Chuang et al., 2002; Kimura et al., 2003). The microarray Materials and methods slides were incubated in prehybridization buffer (5 Â SSC, 0.1% SDS, 1% BSA) at 421C for 1 h. Then, the prehybridized Cell culture slides were washed in deionized water and dipped in 2- Three human lymphoblastoid cells, TK6, WTK1, and NH32, propanol before they were allowed to air dry. In all, 20 mlof were maintained as exponentially growing cultures in RPMI 2 Â hybridization buffer (50% formamide, 10 Â SSC, 0.2% 1640 supplemented with 10% horse serum, 100 U/ml penicillin, SDS) was added and the entire sample was loaded onto the and 100 mg/ml streptomycin. Cells (1 Â 108) were grown slide. A cover glass was placed on the array and the unit was exponentially in stationary culture in loosely capped tissue placed in a sealed, humidified hybridization chamber in a 421C 5 culture flasks at 371Cin5%CO2 at densities of 4–10 Â 10 water bath for 16–20 h. Following hybridization, slides were cells/ml. Cells were treated with 10 Gy g-ray radiation and washed in 2 Â SSC/ 0.1% SDS, 1 Â SSC/ 0.1% SDS, and 0.2 Â total RNA and protein samples were extracted from untreated SSC for 4min each followed by a 1 min wash in 0.05 Â SSC. or treated cells at 1, 3, 6, 9, and 24h after irradiation for Slides were then placed in 2-propanol and spin dried. cDNA microarrays and Western blots. Three independent cultures were used for cDNA microarray experiment. Microarray image and data analysis RNA preparation Hybridized arrays were scanned at 10 mm resolution on a Briefly, for each collection point, cells were washed twice with GenePix 4000A scanner (Axon Instruments Inc., Union City, PBS (41C) and followed by centrifugation (1000 r.p.m. at 41C, CA, USA) at variable PMT voltage settings to obtain maximal 5 min). Trizol reagent (1 ml; Invitrogen, Carlsbad, CA, USA) signal intensities with o0.1% probe saturation. The resulting was added to cell pellets, which were vortexed and incubated tiff images were analysed by GenePix Pro 3.0 software (Axon for 5 min at 41C. The mixtures were transferred into 1.5 ml Instruments Inc., Union City, CA, USA). The ratios of the centrifuge tubes and 200 ml of chloroform per ml of Trizol was sample intensity to the reference red/green intensity for all added to partition the phases. The mixtures were iced for 5 min targets were determined, and ratio normalizations were and then centrifuged (10 000 r.p.m., 41C) for 10 min. The performed to normalize the center of ratio distribution to 1.0. aqueous layers were then transferred into new 1.5 ml centrifuge A template-based clustering algorithm was used to study the tubes, and a 1/2 volume of 2-propanol added. Samples were temporal changes of gene expression after radiation treatment pelleted by centrifugation (14000 r.p.m.) for 30 min at 4 1C and (Chuang et al., 2002). Selected genes satisfied the following washed twice with 1 ml of 70% ethanol. Then, 100 mgof criteria in at least one of the cell lines: (a) at least 60% of data extracted RNA was further purified using Qiagen Rneasy Mini points (at least three time points) in each profile had a Kit (Valencia, CA, USA) according to the manufacturer’s maximum intensity >500 and a minimum intensity >200 instructions. (gray-levels); (b) correlation coefficient of fitting rk>0.85;

Oncogene p53R2 protects against radiation-induced mutagenesis M-H Tsai et al 631 (c) maximum fold-change >2; and (d) total P of three Valencia, CA, USA). This double-stranded siRNA is as replicates o0.001. Among 7680 genes, 464 genes were selected. follows.

Real-time RT–PCR analysis AGAGUUCUCGCCGGUUUGUdTdT After initial expression analysis, 10 genes were selected for dTdTUCUCAAGAGCGGCCAAACA real-time RT–PCR analysis. Quantitative RT–PCR was performed using 1 mg of total RNA, which was reverse A BLAST search of the human genome database was carried transcribed using Superscript II enzyme (Invitrogen, Carlsbad, out to ensure that the sequence would not target other gene CA, USA). The reaction mixture was incubated at 421C for 1 h transcripts. The concentration of siRNAs was 0.15 mM during and then at 701C for 10 min to inactivate Superscript II. After transfections, which were facilitated by Oligofectamine reverse transcription all samples were diluted 1:9 with sterile (Qiagen, Valencia, CA, USA), also according to the protocols water and 2 ml was used for each SYBR Green PCR assay. of Elbashir et al. (2001) and the siRNA supplier. Briefly, Real-time PCR was performed using MJ research Opticon2 1 Â 105 cells were plated and grown in six-well plates to 90% detection system (Waltham, MA, USA) with reagents pur- confluence and immediately before transfection washed with chased from Applied Biosystems (Foster City, CA, USA). serum-free medium, and 900 ml of serum-free medium was Primer sequences were designed using Primer Express software added per well. For each well, 200 nM siRNA was mixed with (Applied Biosystems, Foster City, CA, USA). 5 ml of Oligofectamine (Invitrogen, Carlsbad, CA, USA) in PCR mixes contained all reagents (1 Â SYBR Green PCR 100 ml of serum-free medium. The mixtures were incubated for master mixture, 100 pM each forward and reverse primers in a 20 min at room temperature and then added to cells. Serum final volume of 10 ml) and templates were added into 96-well was added 4h later to a final concentration of 10%. After 24h reaction plate. Thermal cycling was performed with a hot-start transfection, total protein was collected for Western analysis at 951C for 10 min and then 40 cycles of reactions (15 s at 951C or the transfected cells were treated with 0 or 2 Gy g-ray and 1 min at 601C). All SYBR Green PCR data were analysed radiation for survival and mutation analyses. using the opticon monitor 2 software (MJ Research, Waltham, MA, USA). Determination of toxicity and mutagenicity of ionizing radiation Following treatment, lymphoblastoid cells were seeded into Western blot analysis 96-well microtiter plates at densities of 1–10 cells/well, Protein samples were extracted as described previously (Yu depending on the treatment; after 12 days, colonies were et al., 1997). Briefly, cells were pelleted and washed twice with counted, and the Poisson distribution was used to calculate the cold PBS, and lysed on ice for 20 min in 50 mM Tris-HCl (pH plating efficiency. The surviving fraction was determined by 8.0), 150 mM NaCl, 0.02% sodium azide, 100 mg/ml phenyl- dividing the plating efficiency of a treated culture by the methylsulfonyl fluoride, 1 mg/ml aprotinin, and 1% NP40. The plating efficiency of the untreated control (Furth et al., 1981). protein concentration of each sample was quantified, and For mutagenicity analysis, cultures were grown in non- 50 mg of protein from each sample was loaded onto a 12% selective medium for 3 days after treatment, then cells were SDS polyacrylamide gel. After electrophoresis, proteins were plated in 96-well plates in the presence of TFT (2.0 mg/ml) to transferred onto a nitrocellulose membrane. Filters were select TKÀ mutants. Cells from each culture were also plated at probed with different antibodies. The signals were detected 1 cell/well in the absence of TFT to determine plating by the enhanced chemiluminescence system (PerkinElmer, efficiency. All plates were incubated for 11 days prior to Boston, MA, USA). scoring colonies. Mutation plates were refed with fresh TFT medium and incubated for an additional 7 days to observe the SiRNA transfection appearance of any late-appearing mutants. The mutant The siRNA sequences used for targeted silencing of p53R2 fractions were calculated with the Poisson distribution (Furth were chosen as recommended by the siRNA supplier (Qiagen, et al., 1981). Three indenpent experiments were analysed.

References

Amundson SA, Do KT, Shahab S, Bittner M, Meltzer P, Trent Evans E, Sugawara N, Haber JE, Alani E. (2000). Mol Cell 5: J et al. (2000). Radiat Res 154: 342–346. 789–799. Backlund MG, Trasti SL, Backlund DC, Cressman VL, Ewan KB, Henshall-Powell RL, Ravani SA, Pajares MJ, Godfrey V, Koller BH. (2001). Cancer Res 61: 6577–6582. Arteaga C, Warters R et al. (2002). Cancer Res 62: Barcellos CK, Schetinger MR, Battastini AM, Silva LB, 5627–5631. Dias RD, Sarkis JJ. (1994). Braz J Med Biol Res 27: Fitch ME, Cross IV, Turner SJ, Adimoolam S, Lin CX, 1111–1115. Williams KG et al. (2003). DNA Repair (Amsterdam) 2: Benjamin MB, Potter H, Yandell DW, Little JB. (1991). Proc 819–826. Natl Acad Sci USA 88: 6652–6656. Furth EE, Thilly WG, Penman BW, Liber HL, Rand WM. Chen X, Shen B, Xia L, Khaletzkiy A, Chu D, Wong JY et al. (1981). Anal Biochem 110: 1–8. (2002). Cancer Res 62: 1213–1221. Geffrotin C, Tricaud Y, Crechet F, Castelli M, Lefaix JL, Chuang YY, Chen Q, Brown JP, Sedivy JM, Liber HL. (1999). Vaiman M. (1998). Radiat Res 149: 472–481. Cancer Res 59: 3073–3076. Gong B, Almasan A. (2000). Cancer Res 60: 5754–5760. Chuang YY, Chen Y, Chandramouli GVR, Cook JA, Coffin Guittet O, Hakansson P, Voevodskaya N, Fridd S, Graslund D et al. (2002). Cancer Res 62: 6246–6254. A, Arakawa H et al. (2001). J Biol Chem 276: 40647–40651. Coelho D, Fischer B, Holl V, Jung GM, Dufour P, Bergerat JP Holbrook NJ, Liu Y, Fornace Jr AJ. (1996). EXS 77: 273–288. et al. (2002). Radiat Res 157: 446–452. Kawai H, Yamada Y, Tatsuka M, Niwa O, Yamamoto K, Elbashir SM, Harborth J, Lendeckel W, Yalcin A, Weber K, Suzuki F. (1999). Cancer Res 59: 6038–6041. Tuschl T. (2001). Nature 411: 494–498. Kimura T, Takeda S, Sagiya Y, Gotoh M, Nakamura Y, Elliott B, Jasin M. (2001). Mol Cell Biol 21: 2671–2682. Arakawa H. (2003). Nat Genet 34: 440–445.

Oncogene p53R2 protects against radiation-induced mutagenesis M-H Tsai et al 632 Krikos A, Laherty CD, Dixit VM. (1992). J Biol Chem 267: Wadgaonkar R, Phelps KM, Haque Z, Williams AJ, 17971–17976. Silverman ES, Collins T. (1999). J Biol Chem 274: Lakin ND, Jackson SP. (1999). Oncogene 18: 7644–7655. 1879–1882. Lin ZP, Belcourt MF, Cory JG, Sartorelli AC. (2004). J Biol Wakefield LM, Roberts AB. (2002). Curr Opin Genet Dev 12: Chem 279: 27030–27038. 22–29. Little JB, Nagasawa H, Keng PC, Yu Y, Li CY. (1995). J Biol Wang CY, Mayo MW, Korneluk RG, Goeddel DV, Baldwin Chem 270: 11033–11036. Jr AS. (1998). Science 281: 1680–1683. Lozano G, Elledge SJ. (2000). Nature 404: 24–25. Webster GA, Perkins ND. (1999). Mol Cell Biol 19: Nakano K, Balint E, Ashcroft M, Vousden KH. (2000). 3485–3495. Oncogene 19: 4283–4289. Won KA, Schumacher RJ, Farr GW, Horwich AL, Reed SI. Peng Y, Zhang Q, Nagasawa H, Okayasu R, Liber HL, (1998). Mol Cell Biol 18: 7584–7589. Bedford JS. (2002). Cancer Res 62: 6400–6404. Xia F, Wang X, Wang YH, Tsang NM, Yandell DW, Kelsey Russo SM, Tepper JE, Baldwin Jr AS, Liu R, Adams J, Elliott KT et al. (1995). Cancer Res 55: 12–15. P et al. (2001). Int J Radiat Oncol Biol Phys 50: 183–193. Xue L, Zhou B, Liu X, Qiu W, Jin Z, Yen Y et al. (2003). Schena M, Shalon D, Davis RW, Brown PO. (1995). Science Cancer Res 63: 980–986. 270: 467–470. Yamaguchi T, Matsuda K, Sagiya Y, Iwadate M, Fujino MA, Skopek TR, Liber HL, Penman BW, Thilly WG. (1978). Nakamura Y et al. (2001). Cancer Res 61: 8256–8262. Biochem Biophys Res Commun 84: 411–416. Yang CR, Wilson-Van Patten C, Planchon SM, Wuerzberger- Smith ML, Fornace Jr AJ. (1996). Mutat Res 340: 109–124. Davis SM, Davis TW, Cuthill S et al. (2000). FASEB J 14: Sugasawa K, Okamoto T, Shimizu Y, Masutani C, Iwai S, 379–390. Hanaoka F. (2001). Genes Dev 15: 507–521. Yang Q, Zhang R, Wang XW, Linke SP, Sengupta S, Hickson Tan T, Chu G. (2002). Mol Cell Biol 22: 3247–3254. ID et al. (2004). Oncogene 23: 3749–3756. Tanaka H, Arakawa H, Yamaguchi T, Shiraishi K, Fukuda S, Yu Y, Li CY, Little JB. (1997). Oncogene 14: 1661–1667. Matsui K et al. (2000). Nature 404: 42–49. Yu Y, Little JB. (1998). Cancer Res 58: 4277–4281. Teramoto T, Kiss A, Thorgeirsson SS. (1998). Biochem Zhao W, Spitz DR, Oberley LW, Robbins ME. (2001). Cancer Biophys Res Commun 251: 56–60. Res 61: 5537–5543.

Oncogene