Mutagenesis by O(6)-methyl-, O(6)-ethyl-, and O(6)-benzylguanine and O(4)-methylthymine in human cells: effects of O(6)-alkylguanine-DNA alkyltransferase and mismatch repair.

Double-stranded and gapped shuttle vectors were used to study mutagenesis in human cells by O(6)-methyl (m(6)G)-, O(6)-ethyl (e(6)G)-, and O(6)-benzylguanine (b(6)G), and O(4)-methylthymine (m(4)T) when these bases were incorporated site-specifically in the ATG initiation codon of a lacZ' gene. Vectors were transfected into either human kidney cells (293) or colon tumor cells (SO) or into mismatch repair defective human colon tumor cells (H6 and LoVo). Cellular O(6)-alkylguanine-DNA alkyltransferase (alkyltransferase) was optionally inactivated by treating cells with O(6)-benzylguanine prior to transfection. In alkyltransferase competent cells, the mutagenicity of all the modified bases was substantially higher in gapped plasmids than in double-stranded plasmids. Alkyltransferase inactivation increased mutagenesis by the three O(6)-substituted guanines in both double-stranded and gapped plasmids but did not affect m(4)T mutagenesis. In the absence of alkyltransferase, mutagenesis by m(6)G and to a lesser extent e(6)G in double-stranded vectors was higher in the mismatch repair defective H6 and LoVo cells than in SO or 293 cells indicating that e(6)G as well as m(6)G were subject to mismatch repair processing in these cells. The level of mutagenesis by m(4)T and b(6)G was not affected by mismatch repair status. When incorporated in gapped plasmids and in the absence of alkyltransferase, the order of mutagenicity for the modified bases was m(4)T > e(6)G congruent with m(6)G > b(6)G. The O(6)-substituted guanines primarily produced G-->A transitions while m(4)T primarily produced T-->C transitions. However, m(4)T also produced a significant number of T-->A transversion mutations in addition to T-->C transitions in mismatch repair deficient LoVo cells.

Inactivation of human O(6)-alkylguanine-DNA alkyltransferase by modified oligodeoxyribonucleotides containing O(6)-benzylguanine.

Inactivation of the DNA repair protein O(6)-alkylguanine-DNA alkyltransferase (AGT) enhances tumor cell killing by therapeutic alkylating agents. O(6)-Benzylguanine (b(6)G) can inactivate AGT and is currently in clinical trials to enhance therapy. Short oligodeoxyribonucleotides containing b(6)G are much more effective inactivators, but their use for therapeutic purposes is likely to be compromised by metabolic instability. We have therefore examined the ability to inactivate AGT of an 11-mer oligodeoxyribonucleotide containing b(6)G (11-mpBG) when modified with terminal methylphosphonate linkages to protect it from nucleases. This modification did not reduce the ability to serve as a substrate/inactivator for AGT, and 11-mpBG had an ED(50) value of 1.3 nM, more than 300-fold lower than that for b(6)G. A similar oligodeoxyribonucleotide containing O(6)-methylguanine (m(6)G) was also found to be a good substrate (ED(50) value of 10 nM), but the benzylated form was repaired more rapidly and preferentially. When added to HT29 cell cultures, 5 microM 11-mpBG was able to cause a prolonged inactivation of cellular AGT for at least 72 h and to greatly sensitize the cells to killing by 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU). The 11-mpMG was ineffective at up to 20 microM, suggesting that the benzyl group allows better uptake into the cell. However, even with 11-mpBG, the 1000-fold decrease in potency toward AGT in HT29 cells compared to that toward the protein in vitro suggests that uptake may be a limiting factor. These results suggest that oligodeoxyribonucleotides such as 11-mpBG may prove to be useful drugs for potentiation of alkylating agent chemotherapy if uptake can be improved.

Reaction of O6-benzylguanine-resistant mutants of human O6-alkylguanine-DNA alkyltransferase with O6-benzylguanine in oligodeoxyribonucleotides.

Inactivation of the human DNA repair protein, O6-alkylguanine-DNA alkyltransferase (AGT), by O6-benzylguanine renders tumor cells susceptible to killing by alkylating agents. AGT mutants resistant to O6-benzylguanine can be made by converting Pro140 to an alanine (P140A) or Gly156 to an alanine (G156A). These mutations had a much smaller effect on the reaction with O6-benzylguanine when it was incorporated into a short single-stranded oligodeoxyribonucleotide. Such oligodeoxyribonucleotides could form the basis for the design of improved AGT inhibitors. AGT and mutants P140A and G156A preferentially reacted with O6-benzylguanine when incubated with a mixture of two 16-mer oligodeoxyribonucleotides, one containing O6-benzylguanine and the other, O6-methylguanine. When the 6 amino acids located in positions 159-164 in AGT were replaced by the equivalent sequence from the Escherichia coli Ada-C protein (mutant AGT/6ada) the preference for benzyl repair was eliminated. Further mutation incorporating the P140A change into AGT/6ada giving mutant P140A/6ada led to a protein that resembled Ada-C in preference for the repair of methyl groups, but P140A/6ada did not differ from P140A in reaction with the free base O6-benzylguanine. Changes in the AGT active site pocket can therefore affect the preference for repair of O6-benzyl or -methyl groups when present in an oligodeoxyribonucleotide without altering the reaction with free O6-benzylguanine.

Comparison of mutagenesis by O6-methyl- and O6-ethylguanine and O4-methylthymine in Escherichia coli using double-stranded and gapped plasmids.

To compare mutagenesis by O6-methylguanine (m6G), O4-methylthymine (m4T) and O6-ethylguanine (e6G), and assess their genotoxicity in Escherichia coli, double-stranded and gapped plasmids were constructed containing a single m6G, e6G or m4T in the initiation codon (ATG) of a lacZ' gene. Modified base induced mutations were scored by the loss of lacZ' activity on X-gal-containing media resulting in formation of white or sectored (mutant) rather than blue (non-mutant) colonies. Genotoxicity experiments with gapped plasmids containing the modified bases indicated that m4T produced a greater number of bacterial colonies than m6G or e6G. m4T was more mutagenic (45% mutant colonies) than m6G (6%) or e6G (11%) in repair competent (w.t.) E. coli when incorporated in double-stranded plasmids. In gapped plasmids, m4T produced 99% mutant colonies (as was observed previously for e6G) in both w.t. E. coli or E. coli deficient in both O6-alkylguanine-DNA alkyltransferases as well as methylation-directed mismatch repair (ada(-)-ogt(-)-mutS[-]). m6G in gapped plasmids produced 62% mutant colonies in w.t. E. coli, but this percentage increased to 94% in the ada(-)-ogt(-)-mutS(-) strain. In double-stranded plasmids both m4T and m6G produced very similar distributions of mutant and non-mutant colonies in the ada(-)-ogt(-)-mutS(-) strain. These observations led to the conclusion that differences in the mutagenicity of m6G and m4T in w.t. E. coli were a result of preferential repair of m6G compared to m4T by alkyltransferase and mismatch repair mechanisms, and did not reflect differences in their respective coding efficiency or their inherent obstructiveness to DNA synthesis as was observed with e6G. The combination of alkyltransferase and mismatch repair was concluded to be primarily responsible for the apparent genotoxicity of m6G compared to m4T in double-stranded plasmids.

Repair of O6-benzylguanine by the Escherichia coli Ada and Ogt and the human O6-alkylguanine-DNA alkyltransferases.

O6-Methylguanine is removed from DNA via the transfer of the methyl group to a cysteine acceptor site present in the DNA repair protein O6-alkylguanine-DNA alkyltransferase. The human alkyltransferase is inactivated by the free base O6-benzylguanine, raising the possibility that substantially larger alkyl groups could also be accepted as substrates. However, the Escherichia coli alkyltransferase, Ada-C, is not inactivated by O6-benzylguanine. The Ada-C protein was rendered capable of reaction by the incorporation of two site-directed mutations converting Ala316 to a proline (A316P) and Trp336 to alanine (W336A) or glycine (W336G). These changes increase the space at the active site of the protein where Cys321 is buried and thus permit access of the O6-benzylguanine inhibitor. Reaction of the mutant A316P/W336A-Ada-C with O6-benzylguanine was greatly stimulated by the presence of DNA, providing strong support for the concept that binding of DNA to the Ada-C protein activates the protein. The Ada-C protein was able to repair O6-benzylguanine in a 16-mer oligodeoxyribonucleotide. However, the rate of repair was very slow, whereas the E. coli Ogt, the human alkyltransferase, and the mutant A316P/W336A-Ada-C alkyltransferases reacted very rapidly with this 16-mer substrate and preferentially repaired it when incubated with a mixture of the methylated and benzylated 16-mers. These results show that benzyl groups are better substrates than methyl groups for alkyltransferases provided that steric factors do not prevent binding of the substrate in the correct orientation for alkyl group transfer.

O6-ethylguanine and O6-benzylguanine incorporated site-specifically in codon 12 of the rat H-ras gene induce semi-targeted as well as targeted mutations in Rat4 cells.

To examine the miscoding properties of modified guanine residues bearing increasingly bulky O6-substituents, Rat4 cells, grown in the presence of O6-benzylguanine to deplete the DNA repair protein O6-alkylguanine-DNA alkyltransferase, were transfected with plasmids carrying H-ras genes in which O6-methyl, O6-ethyl- and O6-benzylguanine were substituted for the first, second or both the first and second guanine residues of codon 12 (GGA). DNA from isolated transformed colonies was amplified by PCR and directly sequenced by high-temperature manual and automated methods. The results show that O6-ethylguanine and O6-benzylguanine induced semi-targeted as well as targeted mutations, in contrast to O6-methylguanine, which induced only targeted mutations. When incorporated in place of the first guanine of H-ras codon 12, the targeted mutations induced by all these modified guanines were exclusively G-->A transitions. When incorporated at the second position of codon 12, O6-benzylguanine induced G-->A, G-->T and G-->C mutations. O6-Ethylguanine at the second position induced chiefly G-->A transitions, and O6-methylguanine induced G-->A transitions exclusively. Semi-targeted mutations were strictly G-->A at the base 3' to a position 1 adduct or 5' to a position 2 adduct. The mechanism for induction of targeted mutations probably involves decreasing preference to thymidine incorporation opposite an O6-modified guanine as the size of the O6-substituent increases, while the mechanism for non-targeted mutations may be related to abasic site formation or to translesion synthesis which might be made error-prone by obstructive DNA lesions in this context.

Mutagenesis in Escherichia coli by three O6-substituted guanines in double-stranded or gapped plasmids.

Plasmids were constructed with guanine (G) or O6-methyl- (m6G), O6-ethyl-(e6G), or O6-benzyl- (b6G) guanine in the initiation codon (ATG) of the lacZ' gene. Four deoxyuridine residues were incorporated near the modified guanine in the complementary strand. The deoxyuridine-containing plasmids exhibited similarly high transformation efficiencies in ung- Escherichia coli, although the frequency of mutations induced by m6G, e6G, and b6G residues was relatively low. Treatment of the plasmids with uracil-DNA glycosylase (UDG), to remove the uracil residues, or UDG and exonuclease III, to create a gap in the deoxyuridine-containing strand, reduced transformation efficiency for adduct-containing plasmids but did not affect transformation efficiency for control plasmids. However, the same treatments dramatically enhanced mutagenesis by m6G, e6G, and b6G. These results were consistent with blockage of replication by the modified guanines in double-stranded plasmids resulting in preferential replication of the complementary strand. Replication past the modified guanines was forced in the gapped plasmids. The frequency of modified guanine-induced mutations in gapped vectors was similar in strains of E. coli that were proficient in DNA polymerase III but deficient in either DNA polymerase I or II or both polymerase I and II suggesting either that polymerase III was primarily responsible for adduct bypass in all strains or that the probability of base misinsertion during bypass by either polymerase I or II was similar to that for polymerase III. Repair studies with gapped plasmids indicated that m6G was subject to repair by Ada methyltransferase and to postreplication processing by methylation-directed mismatch repair. Neither e6G nor b6G were similarly repaired.(ABSTRACT TRUNCATED AT 250 WORDS)

Response of repair-competent and repair-deficient Escherichia coli to three O6-substituted guanines and involvement of methyl-directed mismatch repair in the processing of O6-methylguanine residues.

Plasmids containing a site-specifically incorporated O6-methyl- (m6G), O6-ethyl- (e6G), or O6-benzylguanine (b6G) within the ATG initiation codon of the lacZ' gene were used to transform Escherichia coli that were repair proficient or deficient in one or both of the E. coli O6-alkylguanine-DNA alkyltransferases, the uvr(ABC) excision repair system, the recA-mediated recombination system, or the methylation-directed mismatch repair system. Colonies were scored phenotypically for adduct-induced mutations. With plasmids containing either e6G or b6G, the frequency of adduct-induced mutation was low and independent of the repair proficiency of the strain transformed. Plasmids containing an m6G residue elicited similar responses in all but the mismatch repair-deficient strain. The generally low mutagenicity of all the O6-substituted guanines was interpreted as reflecting an adduct-induced arrest of replication of the modified strand while the unmodified complementary strand was replicated normally. Studies of the involvement of mismatch repair in m6G mutagenesis showed that m6G:T base pairs were more readily processed than m6G:C base pairs, indicating that mismatch repair involving m6G residues occurs after replication. These data support a model in which the E. coli methylation-directed mismatch repair system diverts plasmids containing promutagenic m6G:T base pairs into replication-arrested complexes providing another line of defense against O6-methylguanine mutagenicity in addition to O6-alkylguanine-DNA alkyltransferase repair and excision repair mechanisms.

The role of O6-alkylguanine-DNA alkyltransferase in protecting Rat4 cells against the mutagenic effects of O6-substituted guanine residues incorporated in codon 12 of the H-ras gene.

The role of rat O6-alkylguanine-DNA alkyltransferase (AGT) in modulating mutagenesis by O6-substituted guanines in the rat H-ras gene was examined. Rat4 cells were transfected with vectors carrying O6-methyl-, O6-ethyl- or O6-benzylguanine residues in place of the normal guanines at either the first, second, or both the first and second positions in codon 12 (GGA) of the H-ras coding sequence. The percentage of transformed colonies was determined for cells grown in normal medium or in medium containing O6-benzylguanine to completely deplete AGT. In parallel experiments with O6-methylguanine-containing vectors, the percentage of cellular DNA harboring codon 12 mutations was determined for normal cells and cells lacking AGT. A reasonable correspondence was observed between the percentage of mutated DNA and the percentage of transformed colonies produced in both types of cells. The results indicate that the contribution of AGT to the repair of O6-substituted guanine damage decreases as the O6 substituent is changed from methyl > ethyl > benzyl. Additionally, cellular AGT appears to repair an O6-methylguanine more readily at the first position of codon 12 than the second position. However, other repair mechanisms in these mammalian cells appear to play a major role in correcting low levels of O6-substituted guanine damage including O6-methylguanine damage.

A sectored colony assay for monitoring mutagenesis by specific carcinogen-DNA adducts in Escherichia coli.

To study the mutagenicity of various carcinogen-DNA adducts in Escherichia coli, a cassette plasmid was developed that permits positioning of specific carcinogen-modified bases within the ATG initiation codon of the lacZ' alpha-complementation gene. Adduct-induced mutations inactivate the gene and lead to formation of blue and white sectored colonies when transformants from an alpha-complementing version of E. coli strain AB1157 are grown on media containing 5-bromo-4-chloro-3-indolyl beta-D-galactoside. In the absence of mutation, blue colonies are produced. This system has been used to measure the mutagenicity of O6-methyl-, O6-ethyl-, and O6-benzyl-2'-deoxyguanosine residues incorporated in place of the normal 2'-deoxyguanosine of the ATG initiation codon. Although a low percentage of sectored colonies was produced in this repair-proficient strain, pretreatment of the bacteria with N-methyl-N'-nitro-N-nitrosoguanidine to disable DNA repair led to a dose-dependent increase in the percentage of sectored colonies. This percentage increased as a function of modified guanine in the order O6-benzyl- less than O6-methyl- less than O6-ethyl-2'-deoxy-guanosine. The only mutations detected at the site of incorporation of these O6-substituted guanines were G-to-A transitions. This sectored colony assay system permits convenient screening of large numbers of colonies and simplifies quantification of modified-base-induced mutations whether they be single-base changes, frameshifts, insertions, or deletions.

Molecular analysis of O6-substituted guanine-induced mutagenesis of ras oncogenes.

We have designed an Ha-ras/thymidine kinase (TK) cassette that permits the incorporation of chemically synthesized adducts within specific domains of the rat Ha-ras protooncogene. This cassette has been used to evaluate the mutagenicity of O6-substituted guanine residues, including O6-methylguanine and O6-benzylguanine, incorporated within the 12th codon of this locus. Mutations were monitored by the ability of these modified Ha-ras DNAs to transform Rat4 TK-cells. Our results indicate that both types of O6-substituted guanines are substantially mutagenic, although the methyl substituent induced a 2-fold higher percentage of transformed Rat4 TK+ colonies than its bulkier benzyl analogue. Interestingly, the mutagenicity of both O6-substituted guanines was found to be independent of their relative position within codon 12, therefore suggesting that the specific activation of Ha-ras oncogenes by GGA----GAA mutations in tumors induced by methylating carcinogens might be due to differences in the accessibility of these guanine residues to the carcinogen rather than to a differential rate of repair. Molecular analysis of the mutations induced by these O6-substituted guanines indicated that O6-methylguanine exclusively induced G----A transitions. In contrast, O6-benzylguanine produced G----C and G----T transversions in addition to G----A transitions. These results suggest that O6-methylguanine and its bulkier analogue O6-benzylguanine may induce mutagenesis by different mechanisms.

Synthesis and properties of H-ras DNA sequences containing O6-substituted 2'-deoxyguanosine residues at the first, second, or both positions of codon 12.

Nine 16-base oligodeoxyribonucleotides having the sequence of codons 9 through the first base of codon 14 of the rodent H-ras gene, i.e., 5'-d(GTGGGCGCTG*G*AGGCG)-3', have been synthesized containing either an O6-methyl- (G* = m6G), O6-ethyl- (G* = e6G), or the newly described O6-benzyl-2'-deoxyguanosine residue (G* = b6G) at position 10 and/or 11 from the 5'-end. The conversion of the protected O6-substituted 2'-deoxyguanosine derivatives to the corresponding 3'-[O-(2-cyanoethyl) diisopropylphosphoramidites] and their incorporation into oligodeoxyribonucleotides were conveniently accomplished by using an "in situ" activation approach and automated phosphite triester synthetic methods. These oligomers were characterized by enzymatic digestion to their component nucleosides and were shown to be free of detectable contamination by known nucleoside impurities that can be produced during these syntheses. The melting behavior and circular dichroism spectra are described for duplexes of the nine O6-substituted 2'-deoxyguanosine containing oligomers paired with the complementary strand 5'-d(CGCCTCCAGCGCCCAC)-3', and these data have been compared with those for the "wild-type" unsubstituted duplex.

Enzymatic sequence-specific spin labeling of a DNA fragment containing the recognition sequence of EcoRI endonuclease.

Deoxyuridine analogs spin labeled in position 5 have been enzymatically incorporated sequence specifically into an oligodeoxyribonucleotide to form a spin-labeled 26-mer. The 26-mer contains the EcoRI-binding site and two labels which are located symmetrically close to the binding site. The labels are separated from one another far beyond the Heisenberg spin-exchange distance. The local base motion as determined by ESR spectroscopy is of the order of 4 ns in the oligonucleotide duplex. This is the same value as reported earlier for local T motions in polynucleotide duplexes, thereby providing direct experimental evidence that the ESR line shape of spin levels covalently attached to nucleic acids depends primarily on the local dynamics of the nucleic acid building blocks.

Base dynamics of nitroxide-labeled thymidine analogues incorporated into (dA-dT)n by DNA polymerase I from Escherichia coli.

Nitroxide-labeled thymidine substrates (dL) for Escherichia coli DNA polymerase I (pol I) were used to synthesize spin-labeled alternating double-stranded copolymers with (dA-dT)n as a template. All dL substrates use an alkane or alkene tether substituted into the 5-position of the pyrimidine ring to link a five- or six-membered ring nitroxide to the pyrimidine base. The kinetics of dL incorporation show some tether dependence with respect to tether length and tether geometry. The electron spin resonance (ESR) spectra of (dA-dT,dL)n duplexes directly formed by polymerization with pol I are compared with the ESR spectra of (dA)n(dT,dL)n duplexes, which are obtained after annealing of nitroxide-labeled single strands with complementary unlabeled single strands. The ESR spectra indicate that nitroxide-labeled analogues with tethers short enough to let the nitroxide ring reside in the major groove are excellent reporter groups for monitoring hybridization. A small difference between the ESR line shapes of the alternating duplexes (dA-dT,dL)n and the homopolymer duplexes (dA)n(dT,dL)n containing the same dL is detectable, suggesting the presence of subtle differences in the base dynamics between both systems. Computer simulation of the ESR spectra of the (dA-dT,dL)n duplexes was successful with the same motional model reported earlier [Kao, S.-C., & Bobst, A.M. (1985) Biochemistry 24, 5465-5469]. The thymidine motion arising from tilting and torsion of base pairs and base twisting in (dA-dT)n is similar to that in (dA)n(dT)n and is of the order of 4 ns.

Redox-active daunomycin-spin-labeled nucleic acid complexes.

Interaction studies between daunomycin (DM) and enzymatically spin-labeled nucleic acid duplexes reveal two modes of binding by electron spin resonance (ESR) spectroscopy. At a low drug/nucleotide (D/N) ratio, the drug binds in the intercalative mode with only a slight reduction in base mobility. Saturation in the intercalative mode is achieved at a lower D/N ratio for B' DNA than for B DNA. After full intercalation, further addition of DM seems to destabilize the helix and to allow the formation of redox-active DM stacks complexed to the nucleic acid lattice. These stacks will irreversibly oxidize all the nitroxides covalently bound to the 4- or 5-position of the pyrimidine base. Interactions between DM and spin-labeled single-stranded nucleic acids lead directly to the formation of redox-active complexes, while mixing of the drug with spin-labeled nucleic acid building blocks not incorporated in a nucleic acid lattice causes no ESR signal change. Complete disappearance of the ESR signal of spin-labeled nucleic acids extrapolates to a D/N value which is a constant for a particular lattice system and is independent of spin-labeling content.

Molecular dynamics in protein-single stranded DNA complexes. Two distinct nucleoside mobilities in poly(deoxythymidylic acid)-gene 5 protein complexes.

A set of differently spin labeled (dT)n is used to evaluate thymidine dynamics and some of the structural features in a (dT)n-gene 5 protein complex. ESR evidence is presented that only one of the four thymidine residues bound in the DNA binding channel shows strong immobilization, whereas the other three display significant mobility of the order of nanoseconds. It is hypothesized that the accessability of such mobile bases could be critical to the recognition of the (dT)n-gene 5 protein complex in auxiliary interactions with other proteins and competitive DNAs.

Molecular dynamics in protein-single stranded DNA complexes. Two distinct nucleoside mobilities, in poly(deoxythymidylic acid)-poly-L-lysine complexes.

Stoichiometric amounts of poly-L-lysine were added to site-specifically spin labeled single stranded nucleic acids and the resulting complexes analyzed by electron spin resonance spectroscopy (ESR). The nucleic acids were spin labeled to different extents and with labels of varying tether length. The ESR data are used to determine nucleoside dynamics and some structural features in these complexes. It is concluded that two distinct base mobilities exist in the complexes; one set is characterized by a mean correlation time tau -R = 2 ns, and the other one by a tau -R greater than or equal to 50 ns. A model is proposed which suggests that a poly-L-lys single stranded nucleic acid complex consists of low mobility segments flanked by more mobile bases. An interesting feature of the proposed model is its applicability to explain ESR data of single strand binding protein-spin labeled nucleic acid complexes, which can also be interpreted in terms of two distinct nucleoside mobility states. It is hypothesized that this phenomenon could be of biological significance for the release of protein ligands from a protein-nucleic acid complex.