Tobacco Documents Online

, u Sclencs On-Une: Mikhalf F. Denlssenko, et al., Science 274(5208):430 1Oq to 5CiENC°. On-Line with fewergrn hg icsl SCIENCE MAGAZiNE ~ E QUESTSIX1>K. - 19~EDBACK CU11f~11~5 :~ = 'ih~ isscte • llr!lde.lsnta :°- Sut~aFpHon Reports Preferential Formation of Benzo[a ]pyrene Adducts at Lung Cancer Mutational Hotspots in P53 Mikhail F. Denissenko, Annie Pao, Moon-shong Tang, * Gerd P. Pfeifer * Friday, October 18, 1998 Cigarette smoke carcinogens such as benzo[a Jpyrene are implicated in the development of lung cancer. The distribution of benzo[a Jpyrene diol epoxide (BPDE) adducts along exons of the P53 gene in BPDE-treated HeLa cells and bronchial epithelial cells was mapped at nucleotide resolution. Strong and selective adduct formation occurred at guanine positions in codons 157, 248, and 273. These same positions are the major mutational hotspots in human lung cancers. Thus, targeted adduct formation rather than phenotypic selection appears to shape the P53 mutational spectrum in lung cancer. These results provide a direct etiological link between a defined chemical carcinogen and human cancer. . M. F. Denissenko and G. P. Pfeifer, Department of Biology, Beckman Research Institute of the City of Hope, Duarte, CA 91010, USA. A. Pao and M.-s. Tang, M. D.'Anderson Cancer Center, University of Texas, Science Park, Smithville, TX 78957, USA. * To whom correspondence should be addressed. Lung cancer is currently the leading cause of cancer death in the United States and is also the most common type of tumor worldwide. Tobacco smoking is the single most important risk factor for lung cancer. Among the many components of cigarette smoke, polycyclic aromatic hydrocarbons are strongly implicated as causative agents in the development of these cancers (1). Benzo[a ]pyrene, which occurs in amounts of 20 to 40 ng per cigarette, is by far the best studied of these compounds and is one of the most potent mutagens and carcinogens known. The compound requires metabolic activation to become the ultimate carcinogenic metabolite, BPDE [(±)-anti-V,8Ct.-dihydroxy-9tx,10m-epoxy-7,8,9,10-tetrahydrobenzo[a]pyreneJ, which binds to DNA and forms predominantly covalent (+) trans adducts at the N2 position of guanine (2_). About 60% of human lung cancers contain mutations in the P53 tumor suppressor gene (3). The P53 mutation database (4) includes more than 500 entries of sequenced P53 mutations for lung cancer. There is a large percentage of G to T transversion mutations in these tumors. Such mutations are hallmarks of mutagenesis involving certain types of polycyclic aromatic hydrocarbons, including BPDE (5-), but they can also be induced by other agents, including oxidative DNA damage (6). The distribution of mutations along the P53 gene in lung cancer is nonrandom but rather is characterized by several mutational hotspots, in particular, at codons 157, 248, and 273 (Fig. 1), which correspond to amino acids within the DNA binding domain of p53. Codon 157 is a mutational hotspot specific for lung cancer and does not occur as a hotspot in any other cancer, by,t the other two codons are affected in many different tumor types (3, 2). The majority of lung cancer mutations at these three codon positions are G to T transversions (4). aY t ( i ~ t, ;( ` J ,~1'si'.tRf 112! ,iJ,~tLt?,Idf i x4R8eagpsA, ~ ; . c-1~ SP Q C '~ccs. Vo-J C -t-zU B~O(^ ccrt dE c_~4 ~ .: ~"~ . :_.,.,,,~ dt!(t rSPz2 G= L-W . 4w l(a .! ----. ~.-.---=---------- ' h ttp:l/www.sciencemag. orghdence/scrtpts/ dlsplay/tu II/274/5288/430. htmf B REAK'THR~~?~J~ QF'THE YEAR ~ ~ Fig. 1. Frequency of P53 mutations in lung cancer by codon position. Numbers were obtained from the P53 database (4). Radon-associated lung cancers and cancers from nonsmokers were excluded. The sequences surrounding the mutational hotspot codons 157, 248, and 273 are indicated. The asterisks mark the mutated Gs within these codons. iViecv Larager Version of this Imaee (22K GIF file)l a /'2ur7 ~ ~C fl , w .6 /°."%c` - ~/ac_ page: i

Sclence On-7Jne: Mlkhak F. Denbsenko, et al., Sclence 274(5288):430 Friday, October f8, f998 To investigate the relation between BPDE adduct formation and P53 mutations, we mapped the distribution of BPDE adducts along the P53 gene using a modification of the ligation-mediated polymerase chain reaction (LMPCR) (8). HeLa cells were treated with various concentrations of BPDE (2), and DNA was isolated and cleaved at the sites of modified bases with the UvrABC nuclease complex from Escherichia coli (0). UvrABC makes a dual incision 5° and 3r to the adducted base, and the 31 incision occurs ~ specifically at the fourth nucleotide position 31 to a BPDE adduct ((1). These break positions can then be visualized by LMPCR in which P53 -specific oligonucleotide primers were used (12, 13). Figure 2A shows an analysis of the upper (nontranscribed) DNA strand of exon 5. One of the strongest BPDE-derived signals along the exon is seen at codon 157, which is one of the major mutational hotspots in lung cancer. In exon 7, the two guanine positions within the frequently mutated codon 248 are the preferred targets for BPDE adduct formation (Fig. 2B). The same is true for exon 8, where the strongest signal corresponds to a BPDE adduct at the guanine within the mutational hotspot codon 273 (Fig. 2_C). Fig. 2. Distribution of BPDE adducts along P53 exons in HeLa cells. Cells were treated with various concentrations of BPDE, and the distribution of adducts in P53 was determined after cleavage with UvrABC nuclease and LMPCR (t 1, L2). Adduct-specific bands migrate four nucleotide positions faster than the corresponding bands in the Maxam-Gilbert sequencing ladders (left three lanes). Some bands in the sequencing lanes are absent because 5-methylcytosines are not cleaved C13). (A) Exon 5, nontranscribed strand. (B) Exon 7, nontranscribed strand. (C) Exon 8, nontranscribed strand. Brackets indicate the'positions of selected P53 codons. Asterisks mark the strongly modified G positions within codons 157, 248, and 273. LView Latger Version of this lmaLye (65K GIF file)1 To analyze a cell type that is more representative of the target cell population during lung tissue transformation, we performed similar experiments with normal human bronchial epithelial cells U. The BPDE adduct pattems were generally similar between HeLa cells and normal bronchial cells. Most important, the adduct hotspots were the sanre in the bronchial epithelial cells (Fig. 2). Fig. 3. Distribution of BPDE adducts along P53 exons in bronchial epithelial cells. Cells were treated with 4 µM of BPDE for 30 min, and the distribution of adducts in P53 was determined after cleavage with UvrABC nuclease and LMPCR. (A) Exon 5, nontranscribed strand. (B) Exon 7, nontranscribed strand. (C) Exon 8, nontranscribed strand. Asterisks mark the strongly modified G positions within codons 157, 248, and 273. j_View Larger Version of this Imar?e (90K GIF file)1 To test whether the sequence specificity is related to chromatin structure, we compared the adduct pattern in BPDE-treated HeLa cells with the pattern in BPDE-treated isolated genomic DNA. The two patterns were almost identical (15), which excludes chromatin tructure as a major modulating factor for the cell types analyzed. It should be noted tlat the histogenesis of the different types of lung cancer is incompletely understood. Therefore, it is important that a similar adduct pattem was seen in three different cell types: HeLa cells (Fig. 2), bronchial cells (Fig. 2), and normal human fibroblasts U5). This pattern does not appear to be greatly modified by cell type-specific chromatin structure, which suggests that the same adduct pattern is likely to be present in the unidentified target cells for lung tissue transformation. Strong selectivity of BPDE binding at guanine positions in codons 157, 248, and 273 was not observed in previous experiments in which a DNA polymerase fingerprint assay was used to detect adducts formed in carcinogen-treated plasmid DNA L). The apparent discrepancies between our findings and those of this previous study could be due to different methylation patterns in E. coli versus human DNAs; however, the discrepancies may also arise from differences in specificity and sensitivity of the detection method. The BPDE adduct hotspots are on the nontranscribed DNA strand, which is expected to be repaired relatively inefficiently, according to the concept of transcription-coupled repair (17, ~. A strand bias in repair is consistent with the majority (>90%) of G to T mutations in lung cancer attributable to guanines on the nontranscribed strand (3, 4). Codon 179, which is also frequently mutated in lung tumors, is not a strong target for BPDE adduct formation (Fig. 2_A). However, this codon does not contain a guanine on the nontranscribed strand, and the majority of mutations are A to G transitions at the second codon base (_4). BPDE binds to guanine 20 times more efficiently than to adenine; thus, it is likely that these mutations are caused by another mutagenic component of cigarette smoke. Pronounced adduct formation was observed at codon 267 (Figs. 2C and .3_C), for http://www.sciencemag.arg/sdence/scrtpts/ Paa}e: 2 d(splay/tu111274/5288/430l h tmt

a . Science On-Une: MddieH F. Denbsenlro, et al., Friday, October 18, 1998 Science 274(5286):430 which there is only one mutation entry in the P53 database. Here, the most strongly adducted base corresponds to the third codon position (CGG), and a mutation would not produce an amino acid change. It has been generally assumed that P53 cancer mutations occur frequently at specific codons because they are selected for in the cell transformation process. One possibility is that mutational hotspot codons are sites of preferred gain of function mutations or sites that are most important for tumor suppressor function of the protein. The presence of mutational hotspots has been correlated with crystal structure data obtained from a p53 protein-DNA complex (J2). The most frequently mutated amino acids (residues 248 and 273) contact DNA directly, whereas some of the other commonly mutated amino acids contribute to stabilization of the protein (12). In lung cancers, mutations in the P53 gene are found at more than 100 different sequence positions (Fig..L), and it is likely that all of these mutations can provide a growth advantage. These results, together with our current finding that P53 mutation hotspots 157, 248, and 273 act as selective BPDE binding sites, suggest that P53 mutation hotspots are preferential targets for DNA-damaging agents and that selection may not necessarily play a major role in the occurrence of mutations at these sites. It is also of interest that two of the adduct hotspots (at codons 248 and 273) are at positions that are common mutational sites not only in lung cancer but also in many other cancers. Almost all of the adduct hotspots were at CpG dinucleotides, although not all CpG sites were strong binding sites for BPDE. Because the CpG sites in the P53 gene are methylated in every human tissue or cell type examined (112Q), the preferentially adducted sequence in vivo is 5-methyl-CpG. Whether selective DNA damage also plays a role in the frequent occurrence of transition mutations at specific CpG codons (codons 175, 245, 248, 273, and 282) remains to be determined. The coincidence of mutational hotspots and adduct hotspots suggests that benzo[a ]pyrene metabolites or structurally related compounds are involved in transformation of human lung tissue. Our study thus provides a direct link between a defined cigarette smoke carcinogen and human cancer mutations. REFERENCES AND NOTES .4 1. S. S. Hecht, S. G. Carmeila, S. E. Murphy, P. G. Foiles, F.-L. Chung, J. Cell, Biochem. Suppl. 17F, 27 (1993) . 2. B. Singer and D. Grunberger, Molecular Biology of Mutagens and Carcinogens (Plenum, New York, 1983). 3. M. Hollstein, D. Sidransky, B. Vogelstein, C. C. Harris, Science 253, 49 (1991) ; M. S. Greenblatt, W. P. Bennett, M. Holistein, C. C. Harris, Cancer Res. 54, 4855 (1994) . 4. M. Hollstein et al., Nucleic Acids Res. 24, 141 (1996) . 5. E. Eisenstadt, A. J. Warren, J. Porter, D. Atkins, J. H. Miller, Proc. Natl. Acad. Sci. U.S.A. 79, 1945 (1982) ; M. Mazur and B. Giickman, Somatic Cell Mol. Genet. 14, 393 (1988) ; R.-H. Chen, V. M. Maher, J. J. McCormick, Proc. Narl. Acad. Sci. U.S.A. 87, 8680 (1990) ; B. Ruggeri et al., ibid. 90, 1013 (1993) . 6. T. Lindahl, Nature 362, 709 (1993) . 7. A. J. Levine et al., Ann. N.Y. Acad. Sct. 768, 111 (1995) . 8. G. P. Pfeifer, R. Drouin, A. D. Riggs, G. P. Holmquist, Proc. Natl. Acad. Sci. U.S.A. 88, 1374 (1991) . 9. HeLa S3 cells (American Type Culture Collection, Rockville, MD) were treated with medium containing 1, 2, or 4 µM of freshly prepared BPDE (obtained from the National Cancer Institute repository, Midwest Research Institute, Kansas City, MO) for 30 min at 37°C in the dark [ S. Venkatachalam, M. Denissenko, A. A. Wani, Carcinogenesis 16, 2029 (1995) ]. Controls were treated with solvent only (95% ethanol). 10. A. Sancar and M.-s. Tang, Photochem. Photobiol. 57, 905 (1993); B. Van Houten and A. Snowden, Bioessays 15, 51 (1993). 11. The purified DNA was treated with UvrABC (a 10-fold molar excess of protein over 104 nucleotides of DNA) under standard reaction conditions as described [M.-s. Tang, in Technologies for Detection of DNA Damage and Mutations, G. P. Pfeifer, Ed. (Plenum, New York, 1996), pp. 139-153]. After 90 min of incubation at 37°C, the proteins were removed by phenol extractions followed by diethyl ether extraction, and the qNA was purified further by repeated ethanol precipitations. UvrABC nuclease incises six to seven bases 51 and four bases 3f to a BPDE-modified purine, and under these reaction conditions, the cleavage at BPDE-DNA adducts by UvrABC nucleases is quantitative [M.-s. Tang, J. R. Pierce, R. P. Doisy, M. E. Nazimiec, M. C. MacLeod, Biochemistry 31, 8429 (1992)]. These results validate the UvrABC incision method for analysis of the sequence selectivity of BPDE binding. Because the UvrABC incision at the 3' side of BPDE-DNA adducts is very specific (four bases 3'to the adduct), LMPCR can be used to determine the BPDE adduct distribution at nucleotide resolution. 12. UvrABC-induced strand breaks in the P53 gene were detected by use of LMPCR with P53 -specific primers [ S. Tornaletti, D. Rozek, G. P. Pfeifer, Oncogene 8, 2051 (1993) ; S. Tornaletti and G. P. Pfeifer, Science 263, 1436 (1994) 1. 13. S. Tornaletti and G. P. Pfeifer, Oncogene 10, 1493 (1995) . 14. Normal human bronchial epithelial cells (Clonetics, San Diego, CA) were cultured in growth medium recommended by the supplier. The cells were treated with 4 µM of BPDE as described (2). 15. M. F. Denissenko, A. Pao, M.-s. Tang, G. P. Pfeifer, unpublished observations. 16. A. Puisieux, S. Lim, J. Groopman, M. Ozturk, Cancer Res. 51, 6185 (1991) . 17. I. Mellon, G. Spivak, P. C. Hanawalt, Cell 51, 241 (1987) . 18. R.-H. Chen, V. M. Maher, J. Brouwer, P. van de Putte, J. J. McCormick, Proc. Natl. Acad. Sci. U.S.A. 89, 5413 (1992) 19. Y. Cho, S. Gorina, P. D. Jeffrey, N. P. Pavletich, Science 265, 346 (1994) . http://www.solencecaq.orq/actence/scripts/ C+ el7i dispiay/iull/27,4/5286/430.html (,Ar,~,~` ,~ r .v~ a . s c.'fii. t ~ f 1 - _.u r c, )5,, 10 Y r'v Paqe:3

Sdence On-Line: Mpchail F. Denlseenko, et ai., ~ Solence 274(5266):430 20. W. M. Rideout III, G. A. Coetzee, A. F. Olumi, P. A. Jones, ibid. 249, 1288 (1990). Friday, Odober 15, . , 21. We thank S. Bates for cell culture work. Supported by NIH grants CA65652 to G.P.P. and ES03124 to M.-s. T. 29 May 1996; accepted 19 August 1996 r Volume 274, Number 5286, Issue of 18 October 1996, pp. 430-432 ®1996 by The American Association for the Advancement of Science. ®I996 by The Amsricw Asc:_iation for t_t±e Advancement of Science. a http://www.eciencemeg.org/acience/ecrlpta/ Page: 4 diepley/f uli/274/5266/430. html