|Home | About | Journals | Submit | Contact Us | Français|
Dietary exposure to aflatoxin B1 (AFB1), in addition to other known factors, increases risk for human hepatocellular carcinoma (HCC). HCCs from AFB1-exposed individuals frequently have distinct TP53 mutations, such as G to T transversions in the 2nd guanine of codon 249 (AGG to AGT), and a characteristic mutational spectrum predominated by G:C to T:A mutations.
To recapitulate the distinctive features of TP53 mutations in AFB1-associated HCC, we investigated AFB1-induced DNA adduction in relation to mutagenesis in transgenic mouse fibroblasts exposed to AFB1 in vitro.
Immuno-dot-blot determination of DNA adducts in the overall genome of AFB1-exposed cells revealed the dose-dependant formation of persistent imidazole ring-opened AFB1-DNA adducts. DNA footprinting analysis of the cII transgene in AFB1-exposed cells verified the dose-dependent and sequence-specific formation of DNA adducts. Preferential formation of AFB1-induced DNA adducts along the cII transgene was almost exclusively localized to guanine-containing sequences encompassing CpG dinucleotides. Mutation analysis of the cII transgene in AFB1-exposed cells revealed a dose-dependent induction of cII mutant frequency (P < 0.001) and a unique induced mutational spectrum characterized by predominant induction of G:C to T:A transversions that occurred within CpG sequence contexts. Notably, codons 42 and 45 of the cII transgene, which have identical sequence contexts to that of codon 249 of human TP53, constituted 2 frequently mutated sites in AFB1-exposed cells that contain the G to T transversion signature mutation at their 3rd base positions.
In this model system, AFB1-induced DNA adduction and mutagenesis recapitulate the unique mutational features of TP53 in AFB1-associated human HCC.
Human hepatocellular carcinoma (HCC) is the 5th most common malignancy and the 3rd most frequent cause of cancer-related death, worldwide 1–3. Cancer epidemiology investigations have implicated a multifaceted etiology for human HCC 4, 5. In addition to chronic infections with hepatitis B and hepatitis C viruses, and lifestyle choices, e.g., alcohol abuse, dietary exposure to aflatoxin B1 (AFB1) is a major risk factor for development of human HCC 1–5. The etiologic involvement of AFB1 in human HCC can be inferred from the high incidence of this disease in certain geographical regions, e.g., Southeast Asia and sub-Saharan Africa, where warm and humid climate promotes mold growth and contamination of food supplies by this naturally occurring mycotoxin 4, 6. A substantial number of HCC tumor samples harbors a unique mutation in the TP53 tumor suppressor gene, i.e., G to T transversion in the 2nd guanine of codon 249 (AGG to AGT), and a characteristic TP53 mutational spectrum predominated by G:C to T:A mutations (see, Fig. 1a, b) 7. Strikingly, this signature of TP53 mutations is highly augmented in HCC tumor samples from individuals residing in areas where contamination of foodstuffs by AFB1 is prevalent (see, Fig. 1c, d) 8–11.
The carcinogenicity of AFB1 is partly ascribed to its ability to induce DNA damage (adduct) and mutagenesis 6, 8, 9. Upon absorption, AFB1 undergoes metabolic activation primarily by cytochrome P450 monooxygenases 12–14, rendering an exceedingly reactive metabolite, i.e., AFB1-exo-8,9-epoxide 15–18. The exo-epoxide interacts with DNA predominantly through binding to guanine residues at N7 position forming trans-8,9-dihydro-8-(N7-guanyl)-9-hydoxyaflatoxin B1 (AFB1-N7-Gua) adducts, although small quantities of the corresponding adenine adducts can also be formed 15, 16, 19–21. Alternatively, the highly reactive AFB1-exo-8,9-epoxide can be detoxified through conjugation with glutathione catalyzed by glutathione S-transferase, or conversion to AFB1-dihydrodiol catalyzed by microsomal epoxide hydrolase (reviewed in 8, 9, 18). The predominant AFB1-N7-Gua is a cationic labile adduct, which can either depurinate to yield abasic (AP) sites, or undergo a base-catalyzed rearrangement through which the imidazole ring of dG opens and a stable 8,9-dihydro-8-(2,6-diamino-4-oxo-3,4-dihydropyrimid-5-yl-formamido)-9-hydroxyaflatoxin B1 (AFB1-FAPY) adduct is formed 8, 9, 16, 22.
The unique features of TP53 mutations in AFB1-associated HCC provide an attractive context to investigate the underlying mechanism of carcinogenesis through correlation studies of AFB1-DNA adduction and mutagenesis. Attempts have been made to recapitulate the characteristic TP53 mutagenesis in human HCC by investigating the formation and kinetics of repair of AFB1-induced DNA adducts or by establishing the AFB1-induced mutagenesis in various in vitro and in vivo model systems (reviewed in 8, 9, 11, 18). To this date, however, the correlation between base-specific AFB1-induced DNA adducts and mutagenesis has not been investigated in a single mammalian model system. To simultaneously investigate AFB1-induced DNA adduction and mutagenesis, we have used the transgenic Big Blue® mouse model system 23, which is proven to be an invaluable means for correlation studies of DNA damage and mutagenesis at both genomic and single nucleotide level 24, 25. Here, we have determined the induction of AFB1-DNA adducts in the genome overall and specifically in the cII transgene of Big Blue mouse embryonic fibroblasts treated in vitro with AFB1. We have also investigated the mutagenic fate of the induced AFB1-DNA adducts by establishing the mutant frequency and mutation spectrum of the cII transgene in the AFB1-treated cells. Subsequently, we have elicited correlations between AFB1-DNA adduction and mutagenesis both at the genomic and single nucleotide levels. Methodologically, we have used an immuno-dot-blot assay with an AFB1-DNA adduct-specific antibody 26, and a lesion-specific cleavage assay with a specialized DNA repair enzyme 27 to determine AFB1-induced DNA adduction in the genome. We have utilized the terminal transferase-dependent polymerase chain reaction (TD-PCR) assay 28 for footprinting of AFB1-induced DNA adducts, at the nucleotide resolution level, in the cII transgene. We have employed the λ Select-cII assay and DNA sequence analysis to establish AFB1-induced cII mutant frequency and mutational spectrum, respectively.
Early passage Big Blue mouse embryonic fibroblasts were grown as monolayers at ~20–25% confluence in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum. Prior to chemical treatment, the culture media were removed, and the cells were washed thoroughly with phosphate buffered saline. The culture dishes were filled with serum free DMEM, and subsequently AFB1 (Sigma-Aldrich Inc. Saint Louis, MO), dissolved in dimethylsulfoxide (DMSO), was added to the media at increasing concentrations of 0.04–100 μM, in the presence and absence of standard S9-activation system 29, 30. The S9-activation system comprised of Aroclor 1254-induced rat liver preparations and cofactor reagents (Moltox, Inc. Boone, NC). The freshly made S9-activation system, prepared according to the manufacturer’s instructions (Moltox, Inc.), was added to the media at a final concentration of 1%, and incubation was performed at 37°C for 6 hours in the dark. Immediately after treatment, the cells were harvested by trypsinization for evaluation of cytotoxicity and DNA damage formation. Alternatively, the treated cells were cultured in complete growth medium for an additional 4 days, and afterward were subjected to mutation analysis of the cII transgene. The 4-day growing period is essential for the fixation of all mutations into the genome 24. At the time of harvesting, all cultured cells had undergone 3–4 population doublings. All experiments were conducted in triplicates.
To assess the formation of AFB1-induced DNA adducts, genomic DNA of differently treated cells vs control were subjected to immuno-dot-blot assay using the mouse monoclonal anti-AFB1 (6A10) antibody (Novus Biologicals, Inc. Littleton, CO), which is raised against the imidazole ring-opened persistent form of the major N7 guanine adduct of AFB1 26. The immuno-dot-blot assay was performed as described earlier 31 with some modifications (see, “Supplementary Material”).
To determine the formation of AFB1-induced AP sites, genomic DNA of differently treated cells vs control were digested with AP endonuclease (APE), a multifunctional enzyme with DNA repair activity that cleaves DNA at AP sites 27 (see, “Supplementary Material”).
To map the formation of AFB1-induced DNA adducts in the cII transgene, we have used the versatile TD-PCR footprinting assay, which enables sensitive and specific detection of DNA adducts at the level of nucleotide resolution 28, 32 (see, “Supplementary Material”).
To establish AFB1-induced cII mutant frequency and mutational spectrum, we have used the λ Select-cII assay (Stratagene, La Jolla, CA) and DNA sequence analysis, respectively (see, “Supplementary Material”).
AFB1 treatment of mouse embryonic fibroblasts, in the absence of S9-activation system, caused concentration-dependent cytotoxicity, most visibly in the micromolar dose range, as determined by the trypan blue dye exclusion assay (Fig. 2). In the presence of S9-activation system, however, AFB1 treatment, in the corresponding dose range, resulted in much intensified cytotoxicity. In the latter case, cell survival was diminished to as low as 67.0 ± 3.5% and 16.5 ± 6.7% (Median ± SD) at doses of 5 and 10 μM AFB1, respectively. In keeping with the objective of our study to investigate AFB1-induced DNA adducts in relation to mutagenesis, we determined a relevant dosing range to establish viable and proliferative cells in which the induced AFB1-DNA adducts could be efficiently translated into mutations. Thus, we set the upper dose limit of AFB1 for all DNA adduct and mutagenesis experiments no higher than 5 μM, which rendered a moderate cytotoxicity, in the presence of S9-activation system, and a marginal cytotoxicity in the absence thereof (see, Fig. 2). For verification purposes, we also performed total cell counts in AFB1–treated cells over defined periods of time post-treatment, and ensured that the proliferative capacity of the cells was not compromised subsequent to various AFB1 treatments (data not shown).
We used immuno-dot-assay and lesion-specific cleavage assay coupled to alkaline agarose gel electrophoresis to assess the formation of DNA adducts in AFB1– treated mouse embryonic fibroblasts. Our immuno-dot-blot assay results verified a dose-dependent formation of persistent imidazole ring-opened (AFB1-FAPY) adducts in the treated cells, in the presence of S9-activation system (see, Fig. 3a). In the absence of S9-activation system, however, there was no appreciable formation of AFB1-FAPY adducts in the treated cells at all doses tested. To determine the presence of parent AFB1-N7-Gua adducts in AFB1-exposed cells, we pre-treated their genomic DNA with 15 mM Na2CO3, 30 mM NaHCO3, pH 9.6 for 4 hours at 37°C. This pre-treatment is known to convert the cationic AFB1-N7-Gua adducts to the imidazole ring-opened form 30. As shown in Figure 3b, there was no considerable difference in the detected levels of AFB1-FAPY adducts between the pre-treated and non-pre-treated samples. This implies that the initially formed AFB1-N7-Gua adducts were mainly converted to the persistent imidazole ring-opened derivatives by the end of AFB1 treatment in cell culture. Moreover, our APE-cleavage assay in combination with gel electrophoresis analysis ruled out the formation of AP sites, at a detectable rate, in the genome of AFB1-treated cells, in the presence or in the absence of S9-activation system (see, Fig. 3c). This can be inferred from the absence of detectable fragmentation of the genomic DNA in AFB1-treated cells digested with APE, since the DNA repair function of this enzyme serves to cleave DNA at AP sites 27.
We used the TD-PCR assay to map the formation of DNA adducts, at the level of nucleotide resolution, in the cII transgene of AFB1– treated mouse embryonic fibroblasts 24, 28. As shown in Figure 4, there was a dose-dependent and sequence-specific formation of AFB1–DNA adducts in the treated cells, in the presence of S9-activation system. In the absence of S9-activation system, however, there was no significant formation of AFB1– DNA adducts in the treated cells at all doses tested (see, Fig. 4). Preferential formation of AFB1– DNA adducts along the cII transgene was at nucleotide positions: 7–11 (CGTG), 19 (C), 22–25 (AACG), 47–48 (CG), 62–64 (CG), 90–95 (GGAAGC), 101–105 (GCGTT), 120–135 (CAGCAGGTGGAAGAGG), 144–146 (TCC), 159–169 (GCTGCTTGCTG), 178–187 (TGGGGGGTCG), 192 (G), 210–215 (GGCGCG), 225–228 (TGCG), and 247–256 (CGCCCGGCG). All but one of these guanine-containing nucleotide positions (144–146 nt) encompass CpGs within their sequence contexts.
We used the λ Select-cII assay to determine the induction of cII mutant frequency in AFB1–treated mouse embryonic fibroblasts. As shown in Figure 5 and Table S1, AFB1 treatment of these cells, in the presence of S9-activation system, resulted in a dose-dependent induction of cII mutant frequency. The induced cII mutant frequency reached a 4.1-fold-increase relative to background at the highest administered dose of AFB1 (P < 0.001). In contrast, AFB1 treatment of these cells, in the absence of S9-activation system, was not mutagenic at none of the tested doses as it only marginally elevated the cII mutant frequency relative to background (see, Fig. 5 and Table S1). The small increases in relative cII mutant frequency in cells treated with rising concentrations of AFB1, in the absence of S9-activation system, were not appreciably different from one another, either.
To establish the spectra of AFB1-induced and control mutations, we sequenced the DNA of cII mutants obtained from cells treated with 4 μM AFB1 or control solvent (DMSO), in the presence of S9-activation system. We randomly selected 95 cII mutant plaques induced by AFB1 treatment in comparison to 100 control plaques. Of the respective number of plaques, 89 and 95 contained a minimum of one mutation in the cII transgene. As shown in Table S2, single mutations, mainly single base substitutions, predominated in both the AFB1-induced and control mutation spectra. Detailed mutation spectra induced by AFB1 treatment as compared with control are illustrated in Figure 6. The overall spectrum of mutations induced by AFB1 treatment was significantly different from that of control (P < 0.0001).
To specify the difference between AFB1-induced and control mutation spectra, we compared the frequency of each type of mutation, e.g., transversions, transitions, etc. in the respective mutation spectra. In the Big Blue system, the cII transgene is a non-transcribed gene 25. Hence, the strand bias of mutagenesis, a phenomenon caused by transcription-coupled DNA repair 33, is unlikely to influence the spectrum of mutations found in this transgene. Conventionally, therefore, the strand mirror counterparts of all transversions (e.g., G to T + C to A) or transitions (e.g., G to A + C to T) are combined when the specific types of mutation between different treatment groups are compared 24. As shown in Table 1, G:C to T:A transversions prevailed in the spectrum of mutations produced by AFB1 treatment, i.e., 65.6% of all induced mutations vs 8.3% of control mutations (P < 0.0001). Approximately, half of these G:C to T:A transversion mutations induced by AFB1 treatment occurred within CpG sequence contexts. For the most part, the CpG-targeted G:C to T:A transversions co-localized with preferential sites of AFB1– DNA adduction along the cII transgene (see, Fig. 4 and Fig. 6). However, not all hypermutable sites in the cII transgene in AFB1-treated cells were necessarily the major hotspots of AFB1-DNA adduction. For example, the frequently mutated codons 9, 30, and 60 in the cII transgene of AFB1-exposed cells were not the strongest DNA-adduct formation sites. Conversely, the main hotspots of DNA adduction in the cII transgene of AFB1-treated cells were not the most highly mutated positions. Of significance, codon 42 and codon 45 of the cII transgene, which have identical sequence contexts to that of human TP53 codon 249, were two frequently mutated codons in AFB1-treated cells, harboring the signature mutation of G to T transversion at their 3rd base positions 7–11 (AGG to AGT) (see, Fig. 6).
Epidemiologic studies have the potential to identify suspect carcinogens, which may etiologically be involved in human cancers 34. This objective is often achieved by the demonstration of a significant association between exposure to specific carcinogens and the incidence of cancer in defined human populations 24. Not only can the specificity of carcinogen exposure determine the type of human cancers, but it may also influence the genetic and/or epigenetic aberrations that are unique for certain malignancies 24. The correlative nature of human cancer development and carcinogen exposure can be used for causality inference at the molecular level if the observed correlation can be recapitulated experimentally 10, 11, 35. Human HCC associated with dietary AFB1 exposure represents an exceptional case for inferring cancer causality 24. The HCC tumor samples from individuals residing in areas with known contamination of foodstuffs by AFB1 commonly harbor TP53 mutations with distinctive mutational features, i.e., G to T transversion at the 3rd base position of codon 249 (AGG to AGT), and a typical mutational spectrum predominated by G:C to T:A transversions 7–11. This mutation is exceedingly rare in cancers other than HCC and in HCC of patients from areas of the world where contamination of foodstuff with AFB1 is low. To recapitulate these unique features of TP53 mutations in AFB1-associated HCC, we have investigated AFB1-induced DNA adduction in relation to mutagenesis in transgenic mouse embryonic fibroblasts treated in vitro with AFB1.
In keeping with the objective of our study to correlate AFB1-induced DNA adducts and mutations at a physiologic level, we initially determined AFB1 cytotoxicity to mouse embryonic fibroblasts. We 29, 30 and others 36, 37 had previously shown that in vitro AFB1 requires an external metabolic activation system to exert its biological effects. We had also verified that the herein used rat liver S9-activation system, similarly to human liver S9 preparations or the classical double-induced S9 mix (Aroclor + phenobarbital or dexamethasone), has the capacity to convert AFB1 to DNA reactive derivatives in vitro 29, 30. The rat liver S9-activation system is proven to contain cytochrome P450 3A4 and 1A2 enzymes 38 that are primarily involved in oxidative bioactivation of AFB1 12–14. Thus, we included the S9-activation system in our cytotoxicity experiments and established a relevant dosing range for AFB1, which served as a guideline for DNA adduct and mutagenesis experiments (see, Fig. 2).
Determination of DNA adducts in the genome overall and specifically in the cII transgene of AFB1-treated mouse embryonic fibroblasts confirmed that S9-activation system is necessary for biotransformation of AFB1 to DNA reactive species in this experimental model system (see, Fig. 3 and Fig. 4). Our immuno-dot-blot assay results revealed that AFB1 treatment of these cells, in the presence of S9-activation system, gave rise to a dose-dependant formation of persistent imidazole ring-opened (AFB1-FAPY) adducts in the genomic DNA (see, Fig. 3a). The parent AFB1-N7-Gua adducts were not, however, detectable in the genome of AFB1-treated cells, which implies an in situ rearrangement of these labile adducts to the stable AFB1-FAPY adducts (see, Fig. 3b). Furthermore, our APE-cleavage assay coupled to gel electrophoresis analysis ruled out the formation of AP sites, at a significant rate, in the genomic DNA of AFB1-treated cells (see, Fig. 3c). In the mammalian genome, AFB1-FAPY adducts have been shown to evade DNA repair and persist for several days to weeks 39, 40. AFB1-FAPY adducts have also displayed higher mutagenic potency and stronger blocking effects on DNA replication relative to their molecular predecessors (AFB1-N7-Gua adducts) 22. Nuclear magnetic resonance studies have demonstrated that AFB1-FAPY adducts have a more subtle effect on DNA architecture and a pronounced effect on the melting temperature of the DNA duplex than AFB1-N7-Gua adducts41, 42. These properties enable the AFB1-FAPY adducts to evade DNA repair and exhibit great mutagenic potentials 40, 43, 44.
In confirmation, our TD-PCR footprinting data verified a dose-dependent and sequence-specific formation of DNA adducts in the cII transgene of AFB1– treated cells, in the presence of S9-activation system only (see, Fig. 4). Preferential formation of AFB1– induced DNA adducts along the cII transgene was almost exclusively localized to guanine-containing sequences encompassing CpG dinucleotides. The high reactivity of AFB1 with guanine residues, and its sequence context selectivity for guanines adjacent to cytosines on the 5′ end, have been demonstrated both in vitro and in vivo 9, 11, 14, 20, 22, 36, 45, 46. Of relevance, we have previously shown that all CpGs in the cII transgene of the Big Blue mouse system are highly methylated 47.
Consistent with the above-mentioned AFB1-DNA adduction data, our mutation analysis of the cII transgene in AFB1– treated mouse embryonic fibroblasts showed a significant and characteristic mutagenic response in the treated cells, in the presence of S9-activation system only. More specifically, there was a dose-dependent induction of cII mutant frequency in AFB1– treated cells, which reached a 4.1-fold-increase relative to background at the highest administered dose of AFB1 (P < 0.001) (see, Fig. 5 and Table S1). The AFB1– induced cII mutation spectrum differed significantly from that of control (P < 0.0001), and was characterized by a predominant induction of G:C to T:A transversions occurring mostly within CpG sequence contexts (see, Fig. 6 and Table 1). Notably, codon 42 and codon 45 of the cII transgene, which have the same sequence contexts as codon 249 of human TP53 gene, constituted two frequently mutated sites in AFB1-treated cells, harboring the signature mutation of G to T transversion at their 3rd base positions (see, Fig. 6). These findings are in good agreement with previously reported results by others showing that sequence-specific G:C to T:A transversions are the hallmark of mutations produced by AFB1 in various in vitro or in vivo model systems 14, 20, 22, 36, 45, 46. Aguilar et al. 48 have shown a preferential induction of G to T transversions at the 3rd base position of codon 249 in AFB1-treated human hepatocarcinoma cells (HepG2), although other types of mutations in neighboring bases were also observed, albeit at much lower frequencies. For example, the absolute frequencies of AFB1-induced G to T transversions at the 2nd and 3rd base positions of codon 249 were 0.8 and 8.4 ×10−7, respectively. Methodologically, the occurrence of the former types of mutations could partially be ascribed to a misincorporation of DNA polymerase during the amplification step of the restriction fragment length polymorphism-PCR used in the Aguilar et al. 48 study. In fact, the relative frequencies of G to T transversions at the 2nd base position of codon 249 in AFB1-treated cells and control were 4.3% and 4.5%, respectively 48.
Moreover, there was a good concordance between our DNA footprinting and mutagenicity data as the CpG-targeted G:C to T:A transversions in the cII transgene accorded well with the preferential formation of AFB1– DNA adducts found therein (see, Fig. 4 and Fig. 6). This implies that AFB1– DNA adduction essentially drives the induction of mutagenesis in our experimental model system. However, not all sites of AFB1– DNA adduction in the cII transgene were hypermutable per se. For example, whereas the CG rich nucleotide positions 247–256 were a major hotspot of AFB1–DNA adduction, no mutation was detectable in these sequences. Of importance, mutations in the last ⅓rd part of the cII gene does not severely impair the function of the cII protein 23. Thus, although AFB1–DNA adduction is a driving force behind AFB1-induced mutagenesis in our experimental system, it cannot, in and of itself, elicit the mutagenic outcome. It appears that the formation of AFB1–DNA adducts in critically important genomic loci is integral for a sequential chain of events, which may act additively or synergistically in relation to one another.
The extent to which AFB1-DNA adduction, in addition to or in combination with other determining factors, e.g., viral infection, drives its mutagenicity and carcinogenicity is an important area of research, which requires further exploration 9, 11. We have previously demonstrated that while AFB1-DNA adducts were substantially formed in the 2nd guanine of codon 249 of the TP53 gene in human liver cells, more extensive formation of these adducts was also detectable in other positions in several additional codons 30. This situation is well reflected in the present study as the hypermutable codon 42 and codon 45 of the cII transgene were two substantial, but by no means the sole or the most significant, AFB1-DNA adduct formation sites. As explained above, the major hotspots of AFB1-DNA adduction in the cII transgene were not necessarily the most frequently mutated positions, either. Thus, these observations suggest that additional biological mechanism may contribute to the predominance of G to T transversions in the 2nd guanine of codon 249 of the TP53 gene in human HCC.
Lastly, we should acknowledge that in the present study, we have used mouse embryonic fibroblasts as a surrogate for human hepatocytes, which are the target cell type for AFB1-mutagenicity. Species- or cell type-dependent influential factors that contribute to mutagenesis, such as metabolic activity, DNA repair capacity, fidelity of polymerases for DNA synthesis, etc. 24 should be taken into account while interpreting the data obtained in model systems, and finding their relevance to humans.
In conclusion, we have shown a DNA adduct-mediated mutagenicity of AFB1 in transgenic mouse embryonic fibroblasts, both at the genomic and single nucleotide levels. Although the correlated aspects of AFB1-DNA adduction and mutagenesis in this experimental system recapitulate the unique mutational features of the TP53 gene in AFB1-associated human HCC 1, 7, they also reiterate the multifactorial nature of the etiology of this disease 4, 5. Notwithstanding the crucial role of AFB1-DNA adducts in inducing mutagenesis, our findings support an interplay among different etiologic factors in shaping the signature of TP53 mutations in dietary AFB1-associated human HCC. Altogether, DNA adduct-derived mutagenicity of AFB1, in addition to or in combination with other determinants, is likely to account for the mutation signature of this tumor type.
We would like to thank Myriam Lereau for assistance with our preliminary studies, and Steven E. Bates for help with cell culture experiments.
Grant Support: This work was supported by a grant from the National Cancer Institute (CA84469 to G.P.P.).
Ahmad Besaratinia: Designed and performed research, analyzed data, and wrote the paper; Sang-in Kim: Performed research; Pierre Hainaut: Supervised research; Gerd P. Pfeifer: Obtained funding and supervised research. DISCLOSURES: All the authors declare that they have no conflict of interest.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.