Gut microbes define liver cancer risk in mice exposed to chemical and viral transgenic hepatocarcinogens.

BACKGROUND AND AIMS: Hepatocellular carcinoma (HCC) frequently results from synergism between chemical and infectious liver carcinogens. Worldwide, the highest incidence of HCC is in regions endemic for the foodborne contaminant aflatoxin B1 (AFB1) and hepatitis B virus (HBV) infection. Recently, gut microbes have been implicated in multisystemic diseases including obesity and diabetes. Here, the hypothesis that specific intestinal bacteria promote liver cancer was tested in chemical and viral transgenic mouse models. METHODS: Helicobacter-free C3H/HeN mice were inoculated with AFB1 and/or Helicobacter hepaticus. The incidence, multiplicity and surface area of liver tumours were quantitated at 40 weeks. Molecular pathways involved in tumourigenesis were analysed by microarray, quantitative real-time PCR, liquid chromatography/mass spectrometry, ELISA, western blot and immunohistochemistry. In a separate experiment, C57BL/6 FL-N/35 mice harbouring a full-length hepatitis C virus (HCV) transgene were crossed with C3H/HeN mice and cancer rates compared between offspring with and without H hepaticus. RESULTS: Intestinal colonisation by H hepaticus was sufficient to promote aflatoxin- and HCV transgene-induced HCC. Neither bacterial translocation to the liver nor induction of hepatitis was necessary. From its preferred niche in the intestinal mucus layer, H hepaticus activated nuclear factor-kappaB (NF-kappaB)-regulated networks associated with innate and T helper 1 (Th1)-type adaptive immunity both in the lower bowel and liver. Biomarkers indicative of tumour progression included hepatocyte turnover, Wnt/beta-catenin activation and oxidative injury with decreased phagocytic clearance of damaged cells. CONCLUSIONS: Enteric microbiota define HCC risk in mice exposed to carcinogenic chemicals or hepatitis virus transgenes. These results have implications for human liver cancer risk assessment and prevention.

Characterization of a mammalian homolog of the Escherichia coli MutY mismatch repair protein.

A protein homologous to the Escherichia coli MutY protein, referred to as MYH, has been identified in nuclear extracts of calf thymus and human HeLa cells. Western blot (immunoblot) analysis using polyclonal antibodies to the E. coli MutY protein detected a protein of 65 kDa in both extracts. Partial purification of MYH from calf thymus cells revealed a 65-kDa protein as well as a functional but apparently degraded form of 36 kDa, as determined by glycerol gradient centrifugation and immunoblotting with anti-MutY antibodies. Calf MYH is a DNA glycosylase that specifically removes mispaired adenines from A/G, A/7,8-dihydro-8-oxodeoxyguanine (8-oxoG or GO), and A/C mismatches (mismatches indicated by slashes). A nicking activity that is either associated with or copurified with MYH was also detected. Nicking was observed at the first phosphodiester bond 3' to the apurinic or apyrimidinic (AP) site generated by the glycosylase activity. The nicking activity on A/C mismatches was 30-fold lower and the activity on A/GO mismatches was twofold lower than that on A/G mismatches. No nicking activity was detected on substrates containing other selected mismatches or homoduplexes. Nicking activity on DNA containing A/G mismatches was inhibited in the presence of anti-MutY antibodies or upon treatment with potassium ferricyanide, which oxidizes iron-sulfur clusters. Gel shift analysis showed specific binding complex formation with A/G and A/GO substrates, but not with A/A, C.GO, and C.G substrates. Binding is sevenfold greater on A/GO substrates than on A/G substrates. The eukaryotic MYH may be involved in the major repair of both replication errors and oxidative damage to DNA, the same functions as those of the E. coli MutY protein.

Possible role for thymine glycol in the selective inhibition of DNA synthesis on oxidized DNA templates.

Single-stranded DNA of coliphage M13mp8 was treated with the oxidizing agent, KMnO4, under conditions that selectively form cis-5,6-dihydro-5,6-dihydroxythymine (thymine glycol). Treatment of DNA with 0.7 and 1.4 mM KMnO4 introduced approximately 200 and 400 thymine glycol residues, respectively, per genome. When these DNAs were used to transform Escherichia coli, it was observed that phage survival was reduced in a dose-dependent manner. In studies designed to investigate the effect of DNA oxidation products on replication in vitro, a complementary 15-mer oligodeoxynucleotide was annealed to the oxidized template and extended with the Klenow fragment of DNA polymerase I from E. coli. It was observed that lesions in oxidized DNA strongly inhibited DNA elongation and that DNA synthesis was stopped opposite thymine residues. This is taken as suggestive evidence that the thymine glycol is inhibitory to DNA replication.

Excretion of an aflatoxin-guanine adduct in the urine of aflatoxin B1-treated rats.

Administration of aflatoxin B1 (AFB1) to rats resulted in the urinary excretion of 2,3-dihydro-2-(N7-guanyl)-3-hydroxyaflatoxin B1. This is the major product formed by the interaction in vivo of AFB1 with rat liver nucleic acids. The adduct was isolated from urine by the combined use of preparative and analytical high-pressure liquid chromatography and was quantitated by measurement of absorbance at 365 nm. The method allowed reproducible quantitation of adduct in urine samples from rats treated with AFB1 by i.p. injection at levels as low as 0.125 mg/kg. Application of the method to urine samples from rats given injections of AFB1 (1 mg/kg) revealed the presence of a compound chromatographically identical to authentic 2,3-dihydro-2-(N7-guanyl)-3-hydroxyaflatoxin B1. Spectral and chemical analysis of microgram quantities of this compound provided strong evidence that this compound is identical to authentic adduct. Measurement of this adduct in the urine of rats given injections of different doses of AFB1 showed that excretion occurs in a dose-dependent manner. Comparison of the dose-response curve for adduct excretion with that previously observed for adduct formation in rat liver DNA in vivo revealed a high degree of qualitative similarity, with the levels of adduct excreted in urine representing 30 to 40% of the levels seen initially in liver DNA.

O6-alkyldeoxyguanosine detection by 32P-postlabeling and nucleotide chromatographic analysis.

The 32P-postlabeling procedure, developed originally by Randerath and coworkers, has been modified for the detection and analytical quantitation of O6-alkyl-2'-deoxyguanosine residues in DNA. Chromatographic techniques were developed to resolve individually the normal deoxyribonucleotide-3'-monophosphates and the O6-alkyldeoxyguanosine-3'-monophosphates by high-pressure liquid chromatography. Selective deoxyribonucleotide-3'-monophosphates (e.g., O6-alkyldeoxyguanosine-3'-monophosphates) were then converted to labeled deoxyribonucleotide-[5'-32P]monophosphates by 32P-postlabeling and nuclease P1 treatment and separated by two-dimensional thin layer chromatography. The O6-methyl- and O6-ethyl-2'-deoxyguanosine-3'-monophosphate nucleotides, and the respective 5'-monophosphates, were chemically synthesized for standardization of these quantitative procedures. The quantitation of O6-methl- and O6-ethyl-2'-deoxyguanosine was observed to be analytically accurate between one O6-alkyl-2'-deoxyguanosine residue per 10(4) and 10(7) 2'-deoxyguanosines. The limit of detection was less than one O6-alkyl-2'-deoxyguanosine in 10(7) 2'-deoxyguanosine residues in a sample size of 100 micrograms of DNA, i.e., approximately 10 pg of adduct. The quantitation of O6-methyl-2'-deoxyguanosine in the liver DNAs of rats treated with [14C-Me]N-nitrosodimethylamine compared well with values obtained by both 14C and high-pressure liquid chromatography coupled with fluorescence detection. Thus, these 32P-postlabeling and nucleotide chromatographic procedures should be useful in monitoring human exposure to methylating and ethylating carcinogens.

Metabolism of aflatoxin B1 and identification of the major aflatoxin B1-DNA adducts formed in cultured human bronchus and colon.

Aflatoxin B1 and benzo(a)pyrene were activated by both cultured human bronchus and human colon as measured by binding to cellular DNA and protein. The binding of aflatoxin B1 to DNA was dose dependent, and the level of binding was higher in cultured human bronchus than it was in the colon. When compared to aflatoxin B1, the binding level of benzo(a)pyrene to both bronchial and colonic DNA was generally higher. The major adducts formed in both tissues by the interaction of aflatoxin B1 and DNA were chromatographically identical to 2,3-dihydro-2-(N7-guanyl)-3-hydroxyaflatoxin B1 (Structure 1) with the guanyl group and hydroxy group in trans-position and an adduct which has been tentatively identified by other investigators as 2,3-dihydro-2-(N5-formyl-2',5',6'-triamino-4'-oxo-N5-pyrimidyl)-3-hydroxyaflatoxin B1 (Structure 11). Seventy % of the radioactivity associated with bronchial DNA was found in these two peaks, and the ratio of radioactivity between the peaks was nearly 1. In colonic DNA, the ratio between Structures 1 and 11 was approximately 2. These observations add aflatoxin B1 to the list of chemical procarcinogens metabolized by cultured human tissues and in which the carcinogen-DNA adducts are similar to the adducts formed in animal tissue susceptible to the carcinogenic action of aflatoxin B1.

Effects of DNA adduct structure and distribution on the mutagenicity and genotoxicity of two platinum anticancer drugs.

cis-Diamminedichloroplatinum(II) (cis-DDP) and cis,trans,cis-ammine(cyclohexylamine)-dibutyratodichloroplatinu m(IV) (ACDDP) are anticancer drugs that bind to DNA, forming replication blocking adducts. ACDDP, probably manifests its cytotoxicity through the metabolite cis-ammine(cyclohexylamine)dichloroplatinum(II) (ACDP). The biological effects of ACDP and cis-DDP were compared by studying polymerase inhibition in vitro and mutagenesis and genotoxicity in vivo in the duplex genome of bacteriophage M13mp18 replicated in Escherichia coli. cis-DDP and ACDP adducts were equally genotoxic within the statistical error limits of the analysis. Survival of genomes platinated by either drug, increased by threefold in cells pretreated with u.v. irradiation to induce the SOS functions of the host. In the M13mp18 lacZ' gene fragment, the mutagenicity of ACDP was lower than that of cis-DDP; the difference was approximately twofold at a dose of two adducts per 370 base-pair mutational target. Mutagenesis by both compounds was SOS-dependent. The structural basis for lower mutagenicity of ACDP is proposed to be its reduced reactivity at d(ApG) sites. This effect is attributed to an orientational isomerism that precludes the formation of one of two possible DNA adducts at d(ApG) residues. The types of mutations induced for both drugs were similar, but they occurred with different distributions. Both compounds induced primarily G-->T transversions at d(GpG) sites whereas G-->A transitions and A-->T transversions, many at d(ApG), d(GpNpG), and d(GpG) sites, were also well represented. The mapping of DNA adducts by DNA synthesis inhibition revealed excellent correlation between the location of DNA lesions and the sites of mutations. Analysis of the distribution of mutations and the distribution of potential platination sites revealed no sequence-dependent mutation hotspots; i.e. mutagenesis was random throughout the lacZ' region of the M13mp18 bacteriophage genome. These results offer insights into the molecular mechanism of mutagenicity of platinum anticancer drugs.

Cisplatin.

Cisplatin is a widely used anti-cancer drug that is exceptionally effective against testicular cancer. trans-DDP, the geometric isomer of cisplatin, is ineffective as a chemotherapeutic agent. The anti-tumour activity of cisplatin is generally attributed to its formation of DNA adducts, both intrastrand and interstrand crosslinks, which induce structural distortions in DNA. The DNA adducts of cisplatin are thought to mediate its cytotoxic effects by inhibiting DNA replication and transcription and, ultimately, by inducing programmed cell death, or apoptosis. The adducts of both cis- and trans-DDP are removed from DNA by the nucleotide-excision-repair pathway. Cellular proteins possessing certain DNA-binding motifs, including the HMG domain, bind selectively to DNA modified by cisplatin, but not to DNA adducts of trans-DDP; evidence suggests a possible role for these proteins in modulating cisplatin cytotoxicity. Both intrinsic and drug-induced resistance often limit the success of cisplatin; several specific mechanisms of cisplatin resistance have been identified.

Context-dependent mutagenesis by DNA lesions.

BACKGROUND: Detailed analyses of mutational hotspots following DNA damage provide an understanding of oncogene activation and tumor suppressor gene inactivation, and hence provide an insight into the earliest steps in the induction of cancer. A mutational hotspot might be created by preferential lesion formation, decreased lesion repair, or increased misinsertion past the lesion during DNA replication. The respective contribution of these factors might be influenced by the DNA sequence context of the hotspot. RESULTS: As a prelude to addressing the contribution of all possible nearest-neighbor contexts on the replication past O6-methylguanine (m6G) and repair of m6G in vivo, we have devised a mutation frequency (MF) detection strategy on the basis of the properties of type IIs restriction enzymes. We also report a method for constructing site-specific single-stranded viral DNA genomes that should yield identical ligation efficiencies regardless of the lesion or its surrounding sequence context. Using repair-deficient Escherichia coli, we discovered that m6G in three sequence contexts was nearly 100% mutagenic in vivo, showing that the DNA polymerase holoenzyme almost always placed a thymine base opposite m6G during replication. In partially repair-proficient cells, the Ada O6-methylguanine-DNA methyltransferase repair protein was twice as efficient on m6G when a guanine base rather than an adenine base was 5' to the lesion. CONCLUSIONS: The system allows the mutagenic potential of, theoretically, any DNA lesion that exhibits point mutations, in any varied local sequence context, to be rapidly determined. The assay demonstrates low background, high throughput, and does not require phenotypic selection, making it possible to discern the effects of sequence context on the processing of m6G.

The mismatch-repair protein hMSH2 binds selectively to DNA adducts of the anticancer drug cisplatin.

BACKGROUND: The antitumor drug cis-diamminedichloroplatinum(II) (cis-DDP or cisplatin) exerts its cytotoxic effects through the formation of covalent DNA adducts. A family of proteins possessing a common HMG box motif that binds specifically to cisplatin DNA adducts has been previously suggested to be important in the clinical efficacy of the drug. RESULTS: We have shown that the human mismatch-repair protein, hMSH2, also binds specifically to DNA containing cisplatin adducts and displays selectivity for the DNA adducts of therapeutically active platinum complexes. Moreover, hMSH2 is overexpressed in testicular and ovarian tissue; tumors in these tissues are most effectively treated by cisplatin. CONCLUSIONS: Our results suggest a role for hMSH2 in mediating cisplatin toxicity. Supporting this view, previous studies in Escherichia coli dam- strains demonstrate that mutations in mismatch-repair proteins confer resistance to cisplatin toxicity. Mismatch-repair deficiency is also correlated with tolerance to O6-methylguanine, a cytotoxic DNA lesion formed by methylating agents. A current model ascribes O6-methylguanine toxicity to unsuccessful attempts at repair of this lesion by mismatch-repair proteins, resulting in a futile cycle of incision and synthesis, leading ultimately to lethal DNA-strand breaks. We propose that mismatch repair may contribute to cisplatin toxicity by a similar mechanism. Alternatively, hMSH2 may shield cisplatin adducts from repair, allowing adducts to persist, thus enhancing lethality.

Multiple pathways of recombination define cellular responses to cisplatin.

BACKGROUND: Cisplatin is a DNA-damaging drug used for treatment of testicular tumors. The toxicity of cisplatin probably results from its ability to form DNA adducts that inhibit polymerases. Blocked replication represents a particular challenge for tumor cells, which are committed to unremitting division. Recombination provides a mechanism by which replication can proceed despite the presence of lesions and therefore could be significant for managing cisplatin toxicity. RESULTS: Recombination-deficient Escherichia coli mutants were strikingly sensitive to cisplatin when compared with the parental strain. Our data identified both daughter-strand gap and double-strand break recombination pathways as critical for survival following treatment with cisplatin. Although it is established that nucleotide excision repair (NER) significantly protects against cisplatin toxicity, most recombination-deficient strains were as sensitive to the drug as the NER-deficient uvrA mutant. Recombination/NER deficient double mutants were more sensitive to cisplatin than the corresponding single mutants, suggesting that recombination and NER pathways play independent roles in countering cisplatin toxicity. Cisplatin was a potent recombinogen in comparison with the trans isomer and canonical alkylating agents. Mitomycin C, which like cisplatin, forms DNA cross-links, was also recombinogenic at minimally toxic doses. CONCLUSIONS: We have demonstrated that all of the major recombination pathways are critical for E. coli survival following treatment with cisplatin. Moreover, recombination pathways act independently of NER and are of equal importance to NER as genoprotective systems against cisplatin toxicity. Taken together, these results shed new light on how cells survive and succumb to this widely used anticancer drug.

Mass spectral identification and positional mapping of aflatoxin B1-guanine adducts in oligonucleotides.

The biological consequences of a carcinogen-DNA adduct are defined by the structure of the lesion and its position within the genome. Electrospray ionization ion trap mass spectrometry (ESI-ITMS) is shown here to be a sensitive and rapid approach capable of defining both of these parameters. Three isomeric oligonucleotides of the sequence 5'-CCGGAGGCC modified by the potent human carcinogen aflatoxin B1 (AFB1) at different guanines were analyzed by ESI-ITMS. All three samples possessed the same molecular ion confirming the presence of an intact aflatoxin moiety in each oligonucleotide. In addition, each sample displayed a characteristic fragmentation pattern that permitted unambiguous identification of the site of modification within the sequence. Furthermore, an AFB1-modified oligonucleotide was converted under alkaline conditions to its more stable formamidopyrimidine (FAPY) derivative. Analysis of this sample revealed the presence of a molecular ion corresponding to the presence of the FAPY adduct and a distinctive fragmentation pattern that paralleled the known chemical stability of the FAPY metabolite. This approach should be of general use in the determination of not only the nature and site of covalent modifications, but also the chemical stability of DNA adducts.

Extrachromosomal probes for mutagenesis by carcinogens: studies on the mutagenic activity of O6-methylguanine built into a unique site in a viral genome.

This work examines the mutagenic activity of O6-methylguanine (O6MeGua), a DNA adduct formed by certain carcinogenic alkylating agents. A tetranucleotide, 5'-HOTpm6GpCpA-3', was synthesized and ligated into a four-base gap in the unique Pst I site of the duplex genome of the E. coli virus, M13mp8. The double-stranded ligation product was converted to single-stranded form and used to transform E. coli to produce progeny phage. The mutation frequency of O6MeGua was defined as the percentage of progeny phage with mutations in their Pst I site, and this value was determined to be 0.4%. To determine the impact of DNA repair on mutagenesis, cellular levels of O6MeGua-DNA methyltransferase (an O6MeGua-repair protein) were depleted by treatment of host cells for virus replication with N-methyl-N'-nitro-N-nitrosoguanidine (MNNG) prior to viral DNA uptake. In these host cells, the mutation frequency due to O6MeGua increased markedly with increasing MNNG dose (the highest mutation frequency observed was 20%). DNA sequence analysis of mutant genomes revealed that in both MNNG treated and untreated cells, O6MeGua induced exclusively G to A transitions.

The relationship between the chemical structures and mutagenic specificities of the DNA lesions formed by chemical and physical mutagens.

The mutagenic activities of radiation, alkylating agents, and aromatic amines and amides are examined from the perspective of relating the observed patterns of mutagenesis with the structures of the premutagenic DNA lesions. The general approach taken to establish such relationships has involved the construction of viral or plasmid genomes containing specific DNA adducts. The in vivo replication of such site-specifically modified genomes results in the induction of mutations, which are characterized by DNA sequencing. It is found that the mutation types that arise in these simple systems often match those occurring in tumors produced in mammals exposed intentionally or unintentionally to carcinogenic agents.

Recognition of cisplatin adducts by cellular proteins.

Cisplatin is a widely used chemotherapeutic agent. It reacts with nucleophilic bases in DNA and forms 1,2-d(ApG), 1,2-d(GpG) and 1,3-d(GpTpG) intrastrand crosslinks, interstrand crosslinks and monofunctional adducts. The presence of these adducts in DNA is through to be responsible for the therapeutic efficacy of cisplatin. The exact signal transduction pathway that leads to cell cycle arrest and cell death following treatment with the drug is not known but cell death is believed to be mediated by the recognition of the adducts by cellular proteins. Here we describe the structural information available for cisplatin and related platinum adducts, the interactions of the adducts with cellular proteins and the implications of these interactions for cell survival.

Mechanisms of resistance to cisplatin.

The use of cisplatin in cancer chemotherapy is limited by acquired or intrinsic resistance of cells to the drug. Cisplatin enters the cells and its chloride ligands are replaced by water, forming aquated species that react with nucleophilic sites in cellular macromolecules. The presence of the cisplatin adducts in DNA is thought to trigger cell cycle arrest and apoptosis. Knowledge of the mechanism of action of cisplatin has improved our understanding of resistance. Decreased intracellular concentration due to decreased drug uptake, increased reflux or increased inactivation by sulfhydryl molecules such as glutathione can cause resistance to cisplatin. Increased excision of the adducts from DNA by repair pathways or increased lesion bypass can also result in resistance. Finally, altered expression of regulatory proteins involved in signal transduction pathways that control the apoptotic pathway can also affect sensitivity to the drug. An improved understanding of the mechanisms of resistance operative in vivo has identified targets for intervention and may increase the utility of cisplatin for the treatment of cancer.

Construction of a shuttle vector containing a single O6-methylguanine: a probe for mutagenesis in mammalian cells.

A shuttle vector, pKE15, was constructed for investigating the mechanisms by which single carcinogen-DNA adducts induce mutations in mammalian cells. pKE15 contains the SV40 origin of replication, the neomycin resistance gene, SV40 polyadenylation sequences and the pML2 origin of replication. Transfection of pKE15 into CHO cells established the G418-resistant phenotype; the frequency of G418-resistant clones was approximately 10(-4), a value that is similar to those obtained with other SV40-based vectors expressing the neomycin resistance gene. A tetranucleotide containing O6-methylguanine, a DNA adduct formed by carcinogenic alkylating agents, was incorporated into a 4-base gap positioned in the center of a PstI site. The tetranucleotide containing the adduct was physically mapped to a 14-base-pair region of the shuttle vector that included the ligation target, the PstI site. It was incorporated approximately equally into either of the complementary strands of the shuttle vector. The ligation efficiency of the tetranucleotide into the gapped genome was approximately 100% and was independent of the concentration of tetranucleotide used at concentrations ranging over one order of magnitude. The potential applications of the site-specifically modified genome for establishing the mutagenic fate of O6-methylguanine in repair-proficient and -deficient CHO cells are discussed.

Mutagenicity and repair of oxidative DNA damage: insights from studies using defined lesions.

Oxidative DNA damage has been implicated in mutagenesis, carcinogenesis and aging. Endogenous cellular processes such as aerobic metabolism generate reactive oxygen species (ROS) that interact with DNA to form dozens of DNA lesions. If unrepaired, these lesions can exert a number of deleterious effects including the induction of mutations. In an effort to understand the genetic consequences of cellular oxidative damage, many laboratories have determined the patterns of mutations generated by the interaction of ROS with DNA. Compilation of these mutational spectra has revealed that GC-->AT transitions and GC-->TA transversions are the most commonly observed mutations resulting from oxidative damage to DNA. Since mutational spectra convey only the end result of a complex cascade of events, which includes formation of multiple adducts, repair processing, and polymerase errors, it is difficult if not impossible to assess the mutational specificity of individual DNA lesions directly from these spectra. This problem is especially complicated in the case of oxidative DNA damage owing to the multiplicity of lesions formed by a single damaging agent. The task of assigning specific features of mutational spectra to individual DNA lesions has been made possible with the advent of a technology to analyze the mutational properties of single defined adducts, in vitro and in vivo. At the same time, parallel progress in the discovery and cloning of repair enzymes has advanced understanding of the biochemical mechanisms by which cells excise DNA damage. This combination of tools has brought our understanding of DNA lesions to a new level of sophistication. In this review, we summarize the known properties of individual oxidative lesions in terms of their structure, mutagenicity and repairability.

Mutagenic specificity of alkylated and oxidized DNA bases as determined by site-specific mutagenesis.

This work demonstrates the use of the tools of site-specific mutagenesis to study the mutagenic activity of two DNA adducts, O6-methylguanine and cis-thymine glycol. The former adduct is one of the methylated bases formed by carcinogenic and mutagenic alkylating agents. It was built into the single-stranded genome of bacteriophage M13 and replicated in Escherichia coli (E. coli). The mutation frequency of O6-methylguanine was 0.4% in physiologically normal cells. In cells in which the repair systems for O6-methylguanine were compromised by challenge with an alkylating agent, the mutation frequency rose to approximately 20%. DNA sequencing revealed that O6-methylguanine induced exclusively G----A transitions, which was most consistent with it pairing with thymine during DNA synthesis. The mutagenic effects also were investigated of cis-thymine glycol isomers, which are major, stable products of ionizing radiation and oxidative damage to DNA. By techniques similar to those employed for the study of O6-methylguanine mutagenesis, a single thymine glycol was situated in an M13 phage genome. The genome was replicated in E. coli that were physiologically normal, induced for SOS functions, or deficient in the nth gene product and, in all cases, the mutagenic processing of thymine glycol in vivo yielded mutant progeny phage at a frequency of 0.3-0.4%. All mutations occurred at the site that originally contained thymine glycol, and all were demonstrated by DNA sequencing to have resulted from targeted T----C transitions. These data suggest that thymine glycol pairs with guanine during replication.