Recognition and repair of 2-aminofluorene- and 2-(acetylamino)fluorene-DNA adducts by UVRABC nuclease

Recognition of damage induced by N-hydroxy-2-aminofluorene (N-OH-AF) and N-acetoxy-2-(acetylamino)fluorene (NAAAF) in both phi X174 RFI supercoiled DNA and a linear DNA fragment by purified UVRA, UVRB, and UVRC proteins was investigated. We have previously demonstrated that N-OH-AF and NAAAF treatments produce N-(deoxyguanosin-8-yl)-2-aminofluorene (dG-C8-AF) and N-(deoxyguanosin-8-yl)-2-(acetylamino)fluorene (dG-C8-AAF), respectively, in DNA. Using a piperidine cleavage method and DNA sequence analysis, we have found that all guanine residues can be modified by N-OH-AF and NAAAF. These two kinds of adducts have different impacts on the DNA helix structure; while dG-C8-AF maintains the anti configuration, dG-C8-AAF is in the syn form. phi X174 RF DNA-Escherichia coli transfection results indicate that while the uvrA, uvrB, and uvrC gene products are needed to repair dG-C8-AAF, the uvrC, but not the uvrA or uvrB gene products, is needed for repair of dG-C8-AF. However, we have found that in vitro the UVRA, UVRB, and UVRC proteins must work in concert to nick both dG-C8-AF and dG-C8-AAF. In general, the reactions of UVRABC nuclease toward dG-C8-AF are similar to those toward dG-C8-AAF; it incises seven to eight nucleotides from the 5' side and three to four nucleotides from the 3' side of the DNA adduct. Evidence is presented to suggest that hydrolysis on the 3' and 5' sides of the damaged base by UVRABC nuclease is not simultaneous and that at least occasionally hydrolysis occurs only on the 3' side or on the 5' side of the damage site. The possible mechanisms of UVRABC nuclease incision for AF-DNA are discussed.     

The formation of 2-aminofluorene-DNA adducts in vivo: evidence for peroxidase-mediated activation

Formation of DNA adducts in various tissues of dogs fed a single dose of the carcinogen 2-aminofluorene was investigated. Adduct analysis was performed using a technique that allows measurement of both N-(deoxyguanosin-8-yl)-2-amino-2-aminofluorene-DNA adduct formed by reaction of N-hydroxy-2-aminofluorene with DNA, as well as the polar 2-aminofluorene-DNA adducts formed when 2-aminofluorene is activated by prostaglandin H synthase-peroxidase in vitro. Two male beagle (A and B) dogs were examined and a different DNA adduct profile was observed with each dog. For the dog A, N-(deoxyguanosin-8-yl)-2-aminofluorene was the major adduct found in hepatic DNA; no peroxidase-derived adducts were detected in this tissue. In contrast, adducts eluting similarly to peroxidase-derived adducts were found in urinary tract tissues of this dog with the relative abundance of these adducts in the order urothelium greater than renal medulla greater than renal cortex, which correlates with the respective tissues' prostaglandin H synthase activity. N-(Deoxyguanosin-8-yl)-2-aminofluorene was detected in the renal tissues, but not in urothelium. For dog B, only the N-(deoxyguanosin-8-yl)-2-aminofluorene adduct was observed in all tissues examined, including the urothelium. However, total binding to liver, kidney, and bladder were two-, two-, and four-fold lower, respectively, than dog A. These data indicate that both prostaglandin H synthase-mediated activation and N-hydroxylation of 2-aminofluorene occur in vivo and may be subjected to pharmacodynamic considerations. Furthermore, the tissue distribution of the peroxidase-mediated 2-aminofluorene adducts suggests this process may also be of importance in the bladder-specific carcinogenicity of aromatic amines.     

Inhibition of enzymic incision of thymine dimers by covalently bound 2-[N-[(deoxyguanosin-8-yl)acetyl]amino]fluorene in deoxyribonucleic acid

The effects of DNA adducts of the carcinogen 2-[N-(acetoxyacetyl)amino]fluorene on enzymic incision of thymine dimers was investigated. Escherichia coli DNA labeled with [3H]thymidine was reacted with the carcinogen. Thymine dimers were then introduced into the modified DNA by irradiation with monochromatic 254-nm light in the presence of the photosensitizer silver nitrate. This DNA containing both types of damages, mainly 2-[N-[(deoxyguanosin-8-yl)acetyl]fluorene and thymine dimers, was then used as substrate for pyrimidine dimer-DNA glycosylase, purified from E. coli infected by bacteriophage T4. Activity was assayed by measuring release of free labeled thymine after photoreversal of the enzyme-reacted DNA by 254-nm light. The Vmax of the enzyme was decreased when it was reacted with the extensively arylamidated substrate. This inhibition of incision of pyrimidine dimers was increased with the number of carcinogen-DNA adducts, although no enzymic activity against modified guanines was present. Therefore, carcinogen-modified purine moieties can interfere with initiation of excision repair of ultraviolet-induced pyrimidine dimers. This suggests an indirect pathway by which modified DNA bases can be mutagenic.     

Mutagenesis by single site-specific arylamine-DNA adducts. Induction of mutations at multiple sites

Two related carcinogen adducts, N-(deoxyguanosin-8-yl)-2-aminofluorene (AF) or N-(deoxyguanosin-8-yl)-N-acetyl-2-aminofluorene (AAF), were introduced into the lacZ' gene at base position 6253 of the minus strand of M13mp9 viral DNA. The construction of this site-specifically modified DNA was accomplished by first preparing a gapped heteroduplex missing 7 nucleotides at position 6251-6257 followed by ligation with an unmodified heptamer or with a heptamer containing either an AF or AAF adduct. These site-specifically modified templates were transfected into competent wild-type Escherichia coli cells (JM103) and a uvrA strain (SMH12). The mutation spectrum was determined by phenotypic selection of colorless plaques indicating a defective beta-galactosidase marker enzyme and by an in situ hybridization procedure to detect single base pair mismatches in the adduct region. DNA sequencing was used to characterize 179 of the mutants obtained. We found that both adducts were capable of inducing base substitution mutations at the adduct site and in the local region of the adduct. A specific frameshift (+1G) was also observed at a displaced site. All of the frameshift mutations occurred at the ligation site of the modified oligonucleotide. Control experiments with an unmodified oligonucleotide did not show an enhancement of mutations at this site, indicating that the adducts may have been responsible for these frameshifts. The mutations spectra induced by these adducts suggest that mutagenesis depends not only on adduct structure but also the sequence in which the adduct is located and the host cell type used for mutation expression.     

Differential effects of N-acetyl-2-aminofluorene and N-2-aminofluorene adducts on the conformational change in the structure of DNA polymerase I (Klenow fragment)

The carcinogen N-acetyl-2-aminofluorene forms two major DNA adducts: the N-(2'-deoxyguanosin-8-yl)-2-acetylaminofluorene adduct (dG-C8-AAF) and its deacetylated derivative, the N-(2'-deoxyguanosin-8-yl)-2-aminofluorene adduct (dG-C8-AF). It is well established that the AAF adduct is a very strong block for DNA synthesis in vitro while the AF adduct is more easily bypassed. In an effort to understand the molecular mechanism of this phenomenon, the structure of the complex of an exonuclease-deficient Escherichia coli DNA polymerase I (Klenow fragment) bound to primer-templates containing either an AF or AAF adduct in or near the active site was probed by nuclease and protease digestion analyses. The results of these experiments suggest that positioning the AAF adduct in the polymerase active site strongly inhibits the conformational change that is required for the insertion of a nucleotide. Similar experiments with AF-modified primer-templates shows a much less pronounced effect. The inhibition of the conformational change by either adduct is not detected if they are positioned in the single-stranded part of the template just one nucleotide before the active site. These findings may explain the different abilities of these lesions to block DNA synthesis.     

Observing translesion synthesis of an aromatic amine DNA adduct by a high-fidelity DNA polymerase

Aromatic amines have been studied for more than a half-century as model carcinogens representing a class of chemicals that form bulky adducts to the C8 position of guanine in DNA. Among these guanine adducts, the N-(2'-deoxyguanosin-8-yl)-aminofluorene (G-AF) and N-2-(2'-deoxyguanosin-8-yl)-acetylaminofluorene (G-AAF) derivatives are the best studied. Although G-AF and G-AAF differ by only an acetyl group, they exert different effects on DNA replication by replicative and high-fidelity DNA polymerases. Translesion synthesis of G-AF is achieved with high-fidelity polymerases, whereas replication of G-AAF requires specialized bypass polymerases. Here we have presented structures of G-AF as it undergoes one round of accurate replication by a high-fidelity DNA polymerase. Nucleotide incorporation opposite G-AF is achieved in solution and in the crystal, revealing how the polymerase accommodates and replicates past G-AF, but not G-AAF. Like an unmodified guanine, G-AF adopts a conformation that allows it to form Watson-Crick hydrogen bonds with an opposing cytosine that results in protrusion of the bulky fluorene moiety into the major groove. Although incorporation opposite G-AF is observed, the C:G-AF base pair induces distortions to the polymerase active site that slow translesion synthesis.     

Mutagenesis by site-specific arylamine adducts in plasmid DNA: enhancing replication of the adducted strand alters mutation frequency

Site specifically modified plasmids were used to determine the mutagenic effects of single arylamine adducts in bacterial cells. A synthetic heptadecamer bearing a single N-(guanin-8-yl)-2-aminofluorene (AF) or N-(guanin-8-yl)-2-(acetylamino)fluorene (AAF) adduct was used to introduce the adducts into a specific site in plasmid DNA that contained a 17-base single-stranded region complementary to the modified oligonucleotide. Following transformation of bacterial cells with the adduct-bearing DNA, putative mutants were detected by colony hybridization techniques that allowed unbiased detection of all mutations at or near the site of the adduct. The site-specific AF or AAF adducts were also placed into plasmid DNA that contained uracil residues on the strand opposite that bearing the lesions. The presence of uracil in one strand of the DNA decreases the ability of the bacterial replication system to use the uracil-containing strand, thereby favoring the use of the strand bearing the adducts. In a comparison of the results obtained with site specifically modified DNA, either with or without uracil, the presence of the uracil increased the mutation frequencies of the AF adduct by greater than 7-fold to 2.9% and of the AAF adduct by greater than 12-fold to 0.75%. The mutation frequency of the AF adduct was greatly reduced in a uvrA- strain while no mutations occurred with the AAF adduct in this strain. The sequence changes resulting from these treatments were dependent on adduct structure and the presence or absence of uracil on the strand opposite the adducts.(ABSTRACT TRUNCATED AT 250 WORDS)     

Evidence for in vitro translesion DNA synthesis past a site-specific aminofluorene adduct

The ability of Escherichia coli DNA polymerase I and T7 DNA polymerase to bypass bulky C-8 guanyl-2-aminofluorene adducts in DNA was studied by in vitro DNA synthesis reactions on a site-specific aminofluorene-modified M13mp9 template. This site-specifically modified DNA was prepared by ligating an oligonucleotide containing a single aminofluorene adduct into a gapped heteroduplex of M13mp9 DNA (Johnson, D. L., Reid, T. M., Lee, M.-S., King, C. M., and Romano, L. J. (1986) Biochemistry 25, 449-456). The resulting covalently closed duplex DNA molecule was then cleaved with a restriction endonuclease, denatured, and annealed to a primer on the 3' side of the adduct to form a template specifically designed to study bypass. In this system, any synthesis that was not blocked by the bulky aminofluorene adduct would proceed to the 5' terminus of the single-stranded template, while synthesis interrupted by the adduct would terminate at or near the adduct location. We have measured DNA synthesis on this template and find that the amount of radiolabeled nucleotide incorporated by either E. coli DNA polymerase I (large fragment) or T7 DNA polymerase was much greater than would be predicted if the aminofluorene adduct were an absolute block to DNA synthesis. Furthermore, the products of similar reactions electrophoresed on polyacrylamide gels showed conclusively that the majority of the DNA synthesized by either the T7 DNA polymerase or E. coli DNA polymerase I bypassed the aminofluorene lesion. Substitution of Mn2+ for Mg2+ as the divalent cation resulted in even higher levels of translesion synthesis.     

Mutagenic properties of 3-(deoxyguanosin-N2-yl)-2-acetylaminofluorene, a persistent acetylaminofluorene-derived DNA adduct in mammalian cells

The carcinogen 2-acetylaminofluorene is metabolically activated in cells and reacts with DNA to form N-(deoxyguanosin-8-yl)-2-acetylaminofluorene (dG-C8-AAF), N-(deoxyguanosin-8-yl)-2-aminofluorene (dG-C8-AF), and 3-(deoxyguanosin-N(2)()-yl)-2-acetylaminofluorene (dG-N(2)-AAF) DNA adducts. The dG-N(2)-AAF adduct is the least abundant of the three isomers, but it persists in the tissues of animals treated with this carcinogen. The miscoding and mutagenic properties of dG-C8-AAF and dG-C8-AF have been established; these adducts are readily excised by DNA repair enzymes engaged in nucleotide excision repair. In the present study, oligodeoxynucleotides modified site-specifically with dG-N(2)-AAF were used as DNA templates in primer extension reactions catalyzed by mammalian DNA polymerases. Reactions catalyzed by pol alpha were strongly blocked at a position one base before dG-N(2)-AAF and also opposite this lesion. In contrast, during translesion synthesis catalyzed by pol eta or pol kappa nucleotides were incorporated opposite the lesion. Both pol eta and pol kappa incorporated dCMP, the correct base, opposite dG-N(2)-AAF. In reactions catalyzed by pol eta, small amounts of dAMP misincorporation and one-base deletions were detected at the lesion site. With pol kappa, significant dTMP misincorporation was observed opposite the lesion. Steady-state kinetic analysis confirmed the results obtained from primer extension studies. Single-stranded shuttle vectors containing (5)(')TCCTCCTCXCCTCTC (X = dG-N(2)-AAF, dG-C8-AAF, or dG) were used to establish the frequency and specificity of dG-N(2)-AAF-induced mutations in simian kidney (COS-7) cells. Both lesions promote G --> T transversions overall, with dG-N(2)-AAF being less mutagenic than dG-C8-AAF (3.4% vs 12.5%). We conclude from this study that dG-N(2)-AAF, by virtue of its persistence in tissues, contributes significantly to the mutational spectra observed in AAF-induced mutagenesis and that pol eta, but not pol kappa, may play a role in this process.     

Ring-opened 7-methylguanine residues in DNA are a block to in vitro DNA synthesis.

Single-stranded M13mp18 phage DNA was methylated with dimethylsulfate (DMS), and further treated with alkali to ring-open N7-methylguanine residues and yield 2-6-diamino-4-hydroxy-5N-methylformamidopyrimidine (Fapy) residues. Nucleotide incorporation during in vitro DNA synthesis on methylated template using E. coli DNA polymerase Klenow fragment (Kf polymerase) was reduced compared to the unmethylated template. Additional treatment of the methylated template with NaOH to generate Fapy residues, further reduced in vitro DNA synthesis compared to the synthesis on methylated templates, which suggested that Fapy residues were a block to in vitro DNA synthesis. Analysis of the termination products on sequencing gels, assuming that synthesis stops one base before a blocking lesion, indicated that arrest of DNA synthesis upon direct alkylation of single-stranded DNA occurred 1 base 3' to template adenine residues in the case of Kf polymerase and 1 base 3' to adenine and cystosine residues for T4 polymerase. When the alkylated templates were treated with NaOH to produce a template which converted all the N7-methylguanine residues to Fapy residues, the blocks to DNA synthesis were still observed one base before adenine residues. In addition to the stops previously observed for the methylated templates, however, new stops occurred one base 3' to template guanine residues for synthesis using both Kf polymerase and T4 polymerase. Fapy residues, therefore, represent a potential lethal lesion which may also arrest in vivo DNA synthesis if not repaired.     

In situ detection of acetylaminofluorene-DNA adducts in human cells using monoclonal antibodies

The present study was performed to generate monoclonal antibodies capable of detecting N-acetoxy-2-acetylaminofluorene (NA-AAF)-derived DNA adducts in human cells in situ. As an immunogen, we employed NA-AAF-modified single-stranded DNA coupled electrostatically to methylated protein and we produced five different monoclonal antibodies. All of them showed strong binding to NA-AAF-modified DNA, but had undetectable or minimal binding to undamaged DNA. Competitive inhibition experiments revealed that the epitope recognized by these antibodies is N-(deoxyguanosin-8-yl)-2-acetylaminofluorene (dG-C8-AAF) in DNA, although deacetylated N-(deoxyguanosin-8-yl)-2-aminofluorene in DNA is also recognized with slightly less efficiency. In contrast, these antibodies did not bind to 3-(deoxyguanosin-N(2)-yl)-2-acetylaminofluorene in DNA or to UV-induced lesions in DNA. Interestingly, they showed only minimal binding to small AAF-nucleoside adducts (dG-C8-AAF), indicating that DNA regions flanking a DNA-bound adduct, in addition to the adduct itself, are essential for the stable binding of the antibodies. Using an enzyme-linked immunosorbent assay with the most promising antibody (AAF-1), we detected the concentration-dependent induction of NA-AAF-modified adducts in DNA from repair deficient xeroderma pigmentosum (XP) cells treated with physiological concentrations of NA-AAF. Moreover, the assay enabled to confirm that normal human cells efficiently repaired NA-AAF-induced DNA adducts but not XP-A cells. Most importantly, the formation of NA-AAF-induced DNA adducts in individual nuclei of XP cells could be clearly visualized using indirect immunofluorescence. Thus, we succeeded in establishing novel monoclonal antibodies capable of the in situ detection of NA-AAF-induced DNA adducts in human cells.     

Mutational spectrum and recombinogenic effects induced by aminofluorene adducts in bacteriophage M13

Double-stranded replicative form (RFI) DNA of bacteriophage M13 strain M13mp10 which carries partial lacZ gene has been modified in vitro to various extents with N-hydroxy-2-amino-fluorene (N-OH-AF) and then transfected into E. coli cells. High-performance liquid chromatography (HPLC) analysis results demonstrate that the sole adduct (95%) formed in modified DNA is N-(deoxyguanosine-8-yl)-2-aminofluorene (dG-C8-AF). Approximately 20 adducts per RFI molecule constitute 1 lethal event when plaque-forming ability is assayed on E. coli cells which have received no prior SOS induction. The mutagenicity of dG-C8-AF adducts was assayed by measuring loss of beta-galactosidase activity as a function of adducts per molecule. A dose-dependent increase in Lac- mutants was observed, with a 4-fold increase in mutants per survivor at 30 adducts/molecule. The mutations produced, characterized by DNA sequencing, occur predominantly at either G or C positions different from those observed in the spontaneous mutant spectrum. Restriction-mapping results show that in our assay system, dG-C8-AF adducts induce a previously unreported recombinogenic activity.     

Translesional synthesis past acetylaminofluorene-derived DNA adducts catalyzed by human DNA polymerase kappa and Escherichia coli DNA polymerase IV.

Human DNA polymerase kappa (pol kappa) has a sequence significantly homologous with that of Escherichia coli DNA polymerase IV (pol IV). We used a truncated form of human pol kappa (pol kappaDeltaC) and full-length pol IV to explore the miscoding properties of these enzymes. Oligodeoxynucleotides, modified site-specifically with N-(deoxyguanosin-8-yl)-2-acetylaminofluorene (dG-AAF) and N-(deoxyguanosin-8-yl)-2-aminofluorene (dG-AF), were used as DNA templates in primer extension reactions that included all four dNTPs. Reactions catalyzed by pol kappaDeltaC were partially blocked one base prior to dG-AAF or dG-AF, and also opposite both lesions. At higher enzyme concentrations, a significant fraction of primer was extended. Analysis of the fully extended reaction product revealed incorporation of dTMP opposite dG-AAF, accompanied by much smaller amounts of dCMP, dAMP, and dGMP and some one- and two-base deletions. The product terminating 3' to the adduct site contained AMP misincorporated opposite dC. On templates containing dG-AF, dAMP, dTMP, and dCMP were incorporated opposite the lesion in approximately equal amounts, together with some one-base and two-base deletions. Steady-state kinetics analysis confirmed the results obtained from primer extension reactions catalyzed by pol kappa. In contract, primer extension reactions catalyzed by pol IV were blocked effectively by dG-AAF and dG-AF. At high concentrations of pol IV, full-length products were formed containing primarily one- or two-base deletions with dCMP, the correct base, incorporated opposite dG-AF. The miscoding properties of pol kappa observed in this study are consistent with mutational spectra observed when plasmid vectors containing dG-AAF or dG-AF are introduced into simian kidney cells [Shibutani, S., et al. (2001) Biochemistry 40, 3717-3722], supporting a model in which pol kappa plays a role in translesion synthesis past acetylaminofluorene-derived lesions in mammalian cells.     

The role of specific amino acid residues in the active site of Escherichia coli DNA polymerase I on translesion DNA synthesis across from and past an N-2-aminofluorene adduct

Understanding how carcinogenic DNA adducts compromise accurate DNA replication is an important goal in cancer research. A central part of these studies is to determine the molecular mechanism that allows a DNA polymerase to incorporate a nucleotide across from and past a bulky adduct in a DNA template. To address the importance of polymerase architecture on replication across from this type of bulky DNA adduct, three active-site mutants of Escherichia coli DNA polymerase I (Klenow fragment) were used to study DNA synthesis on DNA modified with the carcinogen N-2-aminofluorene (AF). Running-start synthesis studies showed that full-length synthesis past the AF adduct was inhibited for all of the mutants, but that this inhibition was substantially less for the F762A mutant. Single nucleotide extension and steady-state kinetic experiments showed that the Y766S mutant displayed higher rates of insertion of each incorrect nucleotide relative to WT across from the dG-AF adduct. This effect was not observed for F762A or E710A mutants. Similar experiments that measured synthesis one nucleotide past the dG-AF adduct revealed an enhanced preference by the F762A mutant for dG opposite the T at this position. Finally, synthesis at the +1 and +2 positions was inhibited to a greater extent for the Y766S and E710A mutants compared with both the WT and F762A mutants. Taken together, this work is consistent with the model that polymerase geometry plays a crucial role in both the insertion and extension steps during replication across from bulky DNA lesions.     

DNA polymerase action blocked by adenine adducts induced by 5-hydroxymethylchrysene sulfate.

Modification of M13mp10 single-stranded DNA with 5-hydroxymethylchrysene (5HCR) sulfate, the ultimate carcinogenic metabolite of 5-methylchrysene, resulted in formation of N6[(chrysen-5-yl)methyl]adenine and N2[(chrysen-5-yl)methyl]-guanine at the ratio of 2.7:1. Measurement of DNA synthesis using this modified template and E.coli DNA polymerase I (Klenow fragment) demonstrated that increasing levels of adducts caused a progressive decline in replication. Analysis of reaction products on DNA-sequence gels revealed DNA elongation to be arrested exclusively at adenine adducts in -AAAGGA- and -AACA- sequences.     

DNA sequence analysis of mutations induced by N-2-acetylamino-7-iodofluorene in plasmid pBR322 in Escherichia coli

The spectrum of mutations induced by N-2-acetylamino-7-iodofluorene (AAIF) was analyzed in a forward mutation system based on mutagenesis directed to a small restriction fragment in the tetracycline resistance gene of plasmid pBR322. AAIF was found to induce frameshift mutations and base-pair substitutions at approximately equal frequencies. The frameshift mutations were mostly deletions of single base-pairs, but -2 frameshifts and +1 frameshifts were also detected. With one exception, the substitutions were transversions initiated at a G.C base-pair. Both frameshift mutations and transversions occurred preferentially at sites of repetitive guanine residues. Although AAIF and the related aromatic amines N-2-acetylaminofluorene (AAF) and N-2-aminofluorene (AF) all bind to the C-8 position of guanine, they have different effects on DNA conformation, and these differences are reflected in their mutation spectra. Previous studies have provided evidence that AAF adducts can trigger a B to Z conformational change in alternating GC sequences or displacement of the guanine by the fluorene ring in other sequences; the principal result is two classes of frameshift mutations. AF, whose DNA interaction involves outside binding rather than insertion and denaturation, primarily induces base-pair substitutions. AAIF adducts are chemically similar to AAF adducts, but the iodo group apparently hinders insertion of the fluorene ring into DNA. Consistent with this model, the mutation spectrum of AAIF combines properties of the mutation spectra of both AAF and AF.     

Mutagenic events in Escherichia coli and mammalian cells generated in response to acetylaminofluorene-derived DNA adducts positioned in the Nar I restriction enzyme site

Comparative mutagenesis studies of N-(2'-deoxyguanosin-8-yl)-2-acetylaminofluorene (dG-AAF) and N-(2'-deoxyguanosin-8-yl)-2-aminofluorene (dG-AF) adducts positioned in the Nar I restriction enzyme site were performed using Escherichia coli (E. coli) and simian kidney (COS-7) cells. Oligodeoxynucleotides ((5)(')TCCTCG(1)G(2)CG(3)CCTCTC) containing a recognition sequence for the Nar I restriction enzyme were modified site-specifically with dG-AAF or dG-AF. Modified and unmodified oligomers inserted into single-stranded phagemid shuttle vectors were used to transform E. coli or to transfect COS-7 cells. Following replication in host cells, progeny plasmids were recovered and analyzed for mutations. In SOS-induced E. coli, dG-AAF primarily induced one- and two-base deletions. The mutational frequency varied, depending on the position modified in the Nar I site; 91% two-base deletions were observed at G(3), while 8.4% and 2.8% deletions were detected at G(2) and G(1), respectively. In contrast, dG-AF at any position in the Nar I site failed to produce deletions, generating primarily G --> T transversions (mutational frequency, 7.6-8.4%). In COS-7 cells, both dG-AAF and dG-AF primarily induced G --> T transversions. Mutation frequencies for dG-AAF were 9.4-24%, the highest values being at G(1) and G(3). Mutation frequencies for dG-AF were 9.3-21%, the higher value at G(2). We conclude from this study that the mutation potential of dG-AAF and dG-AF depends on the structure of the adduct, the sequence context of the lesion, and the host cell used for the experiment.