Mutagenic role of Watson-Crick protons in alkylated DNA bases: A theoretical study

Loss of Watson-Crick protons following DNA base alkylation has been proposed as a key event which confers mutation-inducing properties on to alkylated DNA bases. In this theoretical study, the promutagenic O⁶-guanine and O⁴-thymine sites are clearly distinguished from the nonmutagenic N⁷-guanine site on the basis of calculated values of mechanistic indicators for Watson-Crick proton acidity following alkylation at these respective sites. The degree of acidity predicted for these protons for each type of alkylated base accords well with the presence or absence of mutagenicity observed experimentally in each case.     

Mechanism of DNA strand breakage by piperidine at sites of N7-alkylguanines

The volatile, secondary amine piperidine is used in the Maxam-Gilbert chemical method of DNA sequencing to create strand breaks in DNA at sites of damaged bases. As such it is often used in generalized studies of DNA damage to identify 'alkali-labile lesions'. We confirm the mechanism proposed by Maxam and Gilbert (Maxam, A. and Gilbert, W. (1980) Methods Enzymol. 65, 499-560) by which aqueous piperidine creates strand breaks at sites of N7-guanine alkylations: alkaline conditions catalyze rupture of the C8-N9 bond, forming a formamido-pyrimidine structure which is displaced from the ribose moiety by piperidine. In keeping with this mechanism, the tertiary amine, N-methylpiperidine, does not catalyze the formation of strand breaks in alkylated DNA. Our data confirm the prediction that high pH in and of itself will not create strand breaks at sites of N7-alkylguanines.     

Stable isotope labeling-mass spectrometry analysis of methyl- and pyridyloxobutyl-guanine adducts of 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone in p53-derived DNA sequences

The p53 tumor suppressor gene is a primary target in smoking-induced lung cancer. Interestingly, p53 mutations observed in lung tumors of smokers are concentrated at guanine bases within endogenously methylated (Me)CG dinucleotides, e.g., codons 157, 158, 245, 248, and 273 ((Me)C = 5-methylcytosine). One possible mechanism for the increased mutagenesis at these sites involves targeted binding of metabolically activated tobacco carcinogens to (Me)CG sequences. In the present work, a stable isotope labeling HPLC-ESI(+)-MS/MS approach was employed to analyze the formation of guanine lesions induced by the tobacco-specific lung carcinogen 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) within DNA duplexes representing p53 mutational "hot spots" and surrounding sequences. Synthetic DNA duplexes containing p53 codons 153-159, 243-250, and 269-275 were prepared, where (Me)C was incorporated at all physiologically methylated CG sites. In each duplex, one of the guanine bases was replaced with [1,7,NH(2)-(15)N(3)-2-(13)C]-guanine, which served as an isotope "tag" to enable specific quantification of guanine lesions originating from that position. After incubation with NNK diazohydroxides, HPLC-ESI(+)-MS/MS analysis was used to determine the yields of NNK adducts at the isotopically labeled guanine and at unlabeled guanine bases elsewhere in the sequence. We found that N7-methyl-2'-deoxyguanosine and N7-[4-oxo-4-(3-pyridyl)but-1-yl]guanine lesions were overproduced at the 3'-guanine bases within polypurine runs, while the formation of O(6)-methyl-2'-deoxyguanosine and O(6)-[4-oxo-4-(3-pyridyl)but-1-yl]-2'-deoxyguanosine adducts was specifically preferred at the 3'-guanine base of 5'-GG and 5'-GGG sequences. In contrast, the presence of 5'-neighboring (Me)C inhibited O(6)-guanine adduct formation. These results indicate that the N7- and O(6)-guanine adducts of NNK are not overproduced at the endogenously methylated CG dinucleotides within the p53 tumor suppressor gene, suggesting that factors other than NNK adduct formation are responsible for mutagenesis at these sites.     

Endogenous 5-methylcytosine protects neighboring guanines from N7 and O6-methylation and O6-pyridyloxobutylation by the tobacco carcinogen 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone

All CG dinucleotides along exons 5-8 of the p53 tumor suppressor gene contain endogenous 5-methylcytosine (MeC). These same sites (e.g., codons 157, 158, 245, 248, and 273) are mutational hot spots in smoking-induced lung cancer. Several groups used the UvrABC endonuclease incision assay to demonstrate that methylated CG dinucleotides of the p53 gene are the preferred binding sites for the diol epoxides of bay region polycyclic aromatic hydrocarbons (PAH). In contrast, effects of endogenous cytosine methylation on the distribution of DNA lesions induced by tobacco-specific nitrosamines, e.g., 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK), have not been elucidated. In the work presented here, a stable isotope labeling HPLC-ESI-MS/MS approach was employed to analyze the reactivity of the N7 and O6 positions of guanines within hemimethylated and fully methylated CG dinucleotides toward NNK-derived methylating and pyridyloxobutylating species. 15N3-labeled guanine bases were placed within synthetic DNA sequences representing endogenously methylated p53 codons 154, 157, and 248, followed by treatment with acetylated precursors to NNK diazohydroxides. HPLC-ESI-MS/MS analysis was used to determine the relative yields of N7- and O6-guanine adducts at the 15N3-labeled position. In all cases, the presence of MeC inhibited the formation of N7-methylguanine, O6-methylguanine, and O6-pyridyloxobutylguanine at a neighboring G, with the greatest decrease observed in fully methylated dinucleotides and at guanines preceded by MeC. Furthermore, the O6-Me-dG/N7-Me-G molar ratios were decreased in the presence of the 5'-neighboring MeC, suggesting that the observed decline in O6-alkylguanine adduct yields is, at least partially, a result of an altered reactivity pattern in methylated CG dinucleotides. These results indicate that, unlike N2-guanine adducts of PAH diol epoxides, NNK-induced N7- and O6-alkylguanine adducts are not preferentially formed at the endogenously methylated CG sites within the p53 tumor suppressor gene.     

Covalent carcinogenic lesions in DNA: NMR studies of O6-methylguanosine containing oligonucleotide duplexes

We report on proton and phosphorus high resolution NMR investigations of the self-complementary dodecanucleotide d(C1-G2-N3-G4-A5-A6-T7-T8-C9-O6meG10-C11-G12) duplexes (henceforth called O6 meG.N 12-mers), N = C, T, A and G, which contain N3.O6meG10 interactions in the interior of the helix. These sequences containing a single modified O6meG per strand were prepared by phosphoamidite synthesis and provide an excellent model for probing the structural basis for covalent carcinogenic lesions in DNA. Distance dependent nuclear Overhauser effect (NOE) measurements and line widths of imino protons demonstrate that the N3 and O6meG.10 bases stack into the duplex and are flanked by stable Watson-Crick base pairs at low temperature for all four O6meG.N 12-mer duplexes. The imino proton of T3 in the O6meG.T 12-mer and G3 in the O6meG.N 12-mer helix, which are associated with the modification site, resonate at unusually high field (8.5 to 9.0 ppm) compared to imino protons in Watson-Crick base pairs (12.5 to 14.5 ppm). The nonexchangeable base and sugar protons have been assigned from two dimensional correlated (COSY) and nuclear Overhauser effect (NOESY) measurements on the O6meG.N 12-mer helices. The directionality of the distance dependent NOEs establish all O6meG.N duplexes to be right-handed helices in solution. The glycosidic torsion angles are in the anti range at the N3.O6meG10 modification site except for O6meG10 in the O6meG.G 12-mer duplex which adopts a syn configuration. This results in altered NOEs between the G3 (anti).O6meG10 (syn) pair and flanking G2.C11 and G4.C9 base pairs in the O6meG.G 12-mer duplex. We observe pattern reversal for cross peaks in the COSY spectrum linking the sugar H1' protons with the H2',2" protons at the G2 and O6meG10 residues in the O6meG.N 12-mer duplexes with the effect least pronounced for the O6meG.T 12-mer helix. The proton chemical shift and NOE data have been analyzed to identify regions of conformational perturbations associated with N3.O6meG10 modification sites in the O6meG.N 12-mer duplexes. The proton decoupled phosphorus spectrum of O6meG.T 12-mer duplex exhibits an unperturbed phosphodiester backbone in contrast to the phosphorus spectra of the O6meG.C 12-mer, O6meG.G 12-mer and O6meG.A 12-mer duplexes which exhibit phosphorus resonances dispersed over 2 ppm characteristic of altered phosphodiester backbones at the modification site. Tentative proposals are put forward for N3.O6meG10 pairing models based on the available NMR data and serve as a guide for the design of future experiments.     

Role of hoogsteen edge hydrogen bonding at template purines in nucleotide incorporation by human DNA polymerase iota

Human DNA polymerase iota (Pol iota) differs from other DNA polymerases in that it exhibits a marked template specificity, being more efficient and accurate opposite template purines than opposite pyrimidines. The crystal structures of Pol iota with template A and incoming dTTP and with template G and incoming dCTP have revealed that in the Pol iota active site, the templating purine adopts a syn conformation and forms a Hoogsteen base pair with the incoming pyrimidine which remains in the anti conformation. By using 2-aminopurine and purine as the templating residues, which retain the normal N7 position but lack the N(6) of an A or the O(6) of a G, here we provide evidence that whereas hydrogen bonding at N(6) is dispensable for the proficient incorporation of a T opposite template A, hydrogen bonding at O(6) is a prerequisite for C incorporation opposite template G. To further analyze the contributions of O(6) and N7 hydrogen bonding to DNA synthesis by Pol iota, we have examined its proficiency for replicating through the (6)O-methyl guanine and 8-oxoguanine lesions, which affect the O(6) and N7 positions of template G, respectively. We conclude from these studies that for proficient T incorporation opposite template A, only the N7 hydrogen bonding is required, but for proficient C incorporation opposite template G, hydrogen bonding at both the N7 and O(6) is an imperative. The dispensability of N(6) hydrogen bonding for proficient T incorporation opposite template A has important biological implications, as that would endow Pol iota with the ability to replicate through lesions which impair the Watson-Crick hydrogen bonding potential at both the N1 and N(6) positions of templating A.     

Properties of pseudouridine N1 imino protons located in the major groove of an A-form RNA duplex.

The exchangeable N1 imino protons of two pseudouridine (psi) bases located at adjacent internal positions within an undecamer RNA duplex (5'AUAC psi psi ACCUG/3'UAUGAAUGGUC) can report on the environment of the major groove of an A-form double-stranded nucleic acid. The psi N1 imino protons of these residues (which are not involved in interstrand Watson-Crick hydrogen bonding) are protected from chemical exchange with the solvent water and thus are observable in the proton NMR spectrum in H2O (1). These protons will exchange readily at increased pH values or upon thermal denaturation of the duplex. The longitudinal (T1) relaxation times of the psi N1 imino protons in 100 mM NaCl or in 10 mM MgCl2 and 100 mM NaCl are approximately two-fold faster than those of the psi N3 imino protons which are involved in Watson-Crick base pairing. With the addition of spermidine, the psi N1 imino protons become readily exchangeable at a temperature some 20 degrees C below the melting temperature of the duplex.     

B family DNA polymerases asymmetrically recognize pyrimidines and purines

We utilized a series of pyrimidine analogues modified at O(2), N-3, and N(4)/O(4) to determine if two B family DNA polymerases, human DNA polymerase α and herpes simplex virus I DNA polymerase, choose whether to polymerize pyrimidine dNTPs using the same mechanisms they use for purine dNTPs. Removing O(2) of a pyrimidine dNTP vastly decreased the level of incorporation by these enzymes and also compromised fidelity in the case of C analogues, while removing O(2) from the templating base had more modest effects. Removing the Watson-Crick hydrogen bonding groups of N-3 and N(4)/O(4) greatly impaired polymerization, both of the resulting dNTP analogues and of natural dNTPs opposite these pyrimidine analogues when present in the template strand. Thus, the Watson-Crick hydrogen bonding groups of a pyrimidine clearly play an important role in enhancing correct dNTP polymerization but are not essential for preventing misincorporation. These studies also indicate that DNA polymerases recognize bases extremely asymmetrically, both in terms of whether they are a purine or pyrimidine and whether they are in the template or are the incoming dNTP. The mechanistic implications of these results with regard to how polymerases discriminate between right and wrong dNTPs are discussed.     

Simulation of DNA bases in water: Comparison of the Monte Carlo algorithm with molecular mechanics force fields

The interaction between the nucleic acid bases and solvent molecules has an important effect in various biochemical processes. We have calculated total energy and free energy of the solvation of DNA bases in water by Monte Carlo simulation. Adenine, guanine, cytosine, and thymine were first optimized in the gas phase and then placed in a cubic box of water. We have used the TIP3 model for water and OPLS for the nucleic acid bases. The canonical (T, V, N) ensemble at 25°C and Metropolis sampling technique have been used. Good agreement with other available computational data was obtained. Radial distribution functions of water around each site of adenine, guanine, cytosine, and thymine have been computed and the results have shown the ability of the sites for hydrogen bonding and other interactions. The computations have shown that guanine has the highest value of solvation free energy and N7 and N6 in adenine and guanine, N3 in cytosine, and N3 and O4 in thymine have the largest radial distribution function. Monte Carlo simulation has also been performed using the CHARMM program under the same conditions, and the results of two procedures are compared.     

The N2-guanine adduct but not the C8-guanine or N6-adenine adducts formed by 4-nitroquinoline 1-oxide blocks the 3'-5' exonuclease action of T4 DNA polymerase

When O-acetyl-4-(hydroxyamino)quinoline 1-oxide (Ac-4HAQO) reacts with double-stranded DNA at 37 degrees C the major products, N2-guanine, C8-guanine, and N6-adenine adducts, are formed in the proportions of 5:3:2, respectively. When the reaction is carried out with single-stranded DNA at 0 degree C, the products are found in the ratio 1:7:2. Unique 174-bp DNA fragments were modified in these ways and used as substrates for the 3'-5' exonuclease activity of T4 DNA polymerase. The results obtained showed that the exonuclease is blocked by the N2-guanine adduct but not the other two adducts. Interpretation of the cleavage patterns suggested that the enzyme stopped 2 nucleotides before the N2-guanine adduct. The N2-guanine adduct lies in the minor groove of the DNA double helix, while the other two adducts are found in the major groove. Apparently, only the former hinders progression of the enzyme.     

Simulation of DNA bases in water: Comparison of the Monte Carlo algorithm with molecular mechanics force fields

The interaction between the nucleic acid bases and solvent molecules has an important effect in various biochemical processes. We have calculated total energy and free energy of the solvation of DNA bases in water by Monte Carlo simulation. Adenine, guanine, cytosine, and thymine were first optimized in the gas phase and then placed in a cubic box of water. We have used the TIP3 model for water and OPLS for the nucleic acid bases. The canonical (T, V, N) ensemble at 25 degrees C and Metropolis sampling technique have been used. Good agreement with other available computational data was obtained. Radial distribution functions of water around each site of adenine, guanine, cytosine, and thymine have been computed and the results have shown the ability of the sites for hydrogen bonding and other interactions. The computations have shown that guanine has the highest value of solvation free energy and N7 and N6 in adenine and guanine, N3 in cytosine, and N3 and O4 in thymine have the largest radial distribution function. Monte Carlo simulation has also been performed using the CHARMM program under the same conditions, and the results of two procedures are compared.     

The effect of spermine binding on the reactivity of DNA towards carcinogenic alkylating agents

The effect of spermine binding on the electrostatic potential of DNA is evaluated. The calculations are performed for the essential reactive sites, atoms N7 and O6 of guanine, N3 and N7 of adenine, of the nucleic acid and for its surface envelope. An important weakening of the potential is found affecting all the important reactive sites in both grooves and spreading moreover along the polynucleotide chain far away from the site of binding of the ligand. These results are discussed in connection with the experimentally observed inhibitory effect of spermine binding on DNA methylation by carcinogenic agents.     

A laser Raman spectroscopic study of the interaction of calf-thymus DNA with Cu(II) and Pb(II) ions: metal ion binding and DNA conformational changes.

The interaction of calf-thymus DNA with Cu(II) and Pb(II) ions has been investigated in H2O and D2O solutions at physiological pH, using laser Raman spectroscopy. The results confirm the destabilizing effect of Cu2+ ions, which are shown to bind strongly to the guanine and cytidine bases, perturbing the A-T base pairs and disrupting the double-helical structure of DNA, whose conformation is markedly altered by these interactions. Earlier claims that Pb2+ ions destabilize DNA are not supported by the present study. These ions are found to interact only weakly with the nucleic bases, binding to the N7 position of the guanine bases and also interacting with the A-T pairs. Both types of ions are found to interact with the charged phosphate groups of DNA, although these sites are preferred over the nucleic bases by Pb2+ ions.     

O6-ethylguanine carcinogenic lesions in DNA: an NMR study of O6etG.C pairing in dodecanucleotide duplexes

The pairing of O6etG with C located four base pairs in from either end of the self-complementary d(C1-G2-C3-O6etG4-A5-G6-C7-T8-C9-G10-C11-G12) duplex (designated O6etG.C 12-mer) has been investigated from an analysis of proton and phosphorus two-dimensional NMR experiments. The structural consequences of increasing the alkyl group size were elucidated from a comparative study of the pairing of O6meG4 with C9 in a related sequence (designated O6meG.C 12-mer). The NMR parameters for both O6alkG-containing dodecanucleotides are also compared with those of the control sequence containing G4.C9 base pairs (designated G.C 12-mer). The NOE cross-peaks detected in the two-dimensional NOESY spectra of the O6alkG.C 12-mer duplexes in H2O solution establish that the O6etG4/O6meG4 and C9 bases at the lesion site stack into the helix between the flanking C3.G10 and A5.T8 Watson-Crick base pairs. The amino protons of C9 at the O6alkG4-C9 lesion site resonate as an average resonance at 7.78 and 7.63 ppm in the O6etG.C 12-mer and O6meG.C 12-mer duplexes, respectively. The observed NOEs between the amino protons of C9 and the CH3 protons of O6alkG4 establish a syn orientation of the O6-alkyl group with respect to the N1 of alkylated guanine. A wobble alignment of the O6alkG4.C9 base pair stablized by two hydrogen bonds, one between the amino group of C9 and N1 of O6alkG and the other between the amino group of O6alkG and N3 of C9, is tentatively proposed on the basis of the NOEs between the amino protons of C9 at the lesion site and the imino protons of flanking Watson-Crick base pairs. The proton and phosphorus chemical shift differences between the O6etG.C 12-mer and O6meG.C 12-mer duplexes are small compared to the differences between these O6alkG-containing duplexes and the control G.C 12-mer duplex.     

Herpes simplex virus 1 primase employs watson-crick hydrogen bonding to identify cognate nucleoside triphosphates

We utilized NTP analogues containing modified bases to probe the mechanism of NTP selection by the primase activity of the herpes simplex virus 1 helicase-primase complex. Primase readily bound NTP analogues of varying base shape, hydrophobicity, and hydrogen-bonding capacity. Remarkably, primase strongly discriminated against incorporating virtually all of the analogues, even though this enzyme misincorporates natural NTPs at frequencies as high as 1 in 7. This included analogues with bases much more hydrophobic than a natural base (e.g., 4- and 7-trifluoromethylbenzimidazole), a base of similar hydrophobicity as a natural base but with the Watson-Crick hydrogen-bonding groups in unusual positions (7-beta-d-guanine), bases shaped almost identically to the natural bases (4-aminobenzimidazole and 4,6-difluorobenzimidazole), bases shaped very differently than a natural base (e.g., 5- and 6-trifluoromethylbenzimidazole), and bases capable of forming just one Watson-Crick hydrogen bond with the template base (purine and 4-aminobenzimidazole). The only analogues that primase readily polymerized into primers (ITP and 3-deaza-ATP) were those capable of forming Watson-Crick hydrogen bonds with the template base. Thus, herpes primase appears to require the formation of Watson-Crick hydrogen bonds in order to efficiently polymerize a NTP. In contrast to primase's narrow specificity for NTP analogues, the DNA-dependent NTPase activity associated with the herpes primase-helicase complex exhibited very little specificity with respect to NTPs containing unnatural bases. The implications of these results with respect to the mechanism of the helicase-primase and current fidelity models are discussed.     

Base-boronated dinucleotides: synthesis and effect of N7-cyanoborane substitution on the base protons.

Boron-modified nucleic acids comprise a new set of DNA mimics that have potential biological and therapeutic applications. A series of nine dinucleotides containing N7-cyanoborane-2'-deoxyguanosine ((7b)dG) at the 3', 5' or both positions of the phosphodiester linkage have been synthesized using solution phase phosphoramidite chemistry. Fmoc was used as the 5'-protecting group because of incompatibility of the cyanoborane moiety with 5'-DMT cations generated during the deprotection step. The presence of the cyanoborane group was confirmed on the basis of Fab-MS and 1H NMR spectroscopy. The H-8 proton of (7b)dG in the dinucleotides shifted 0.35-0.80 p.p.m. downfield relative to that of unmodified dG. A comparison of the D20 exchange kinetics of the H-8 proton at 60 degrees C showed that H-8 of (7b)dG is very labile relative to unmodified dG, indicating that the N7-cyanoborane modification increases the acidity of the H-8 proton of (7b)dG. These studies illustrate the feasibility of synthesizing boron-containing oligonucleotides which are modified at the N7-guanine to block Hoogsteen pairing in the DNA major groove.     

Mapping the binding site of aflatoxin B1 in DNA: systematic analysis of the reactivity of aflatoxin B1 with guanines in different DNA sequences

The mutagenic and carcinogenic chemical aflatoxin B1 (AFB1) reacts almost exclusively at the N(7)-position of guanine following activation to its reactive form, the 8,9-epoxide (AFB1 oxide). In general N(7)-guanine adducts yield DNA strand breaks when heated in base, a property that serves as the basis for the Maxam-Gilbert DNA sequencing reaction specific for guanine. Using DNA sequencing methods, other workers have shown that AFB1 oxide gives strand breaks at positions of guanines; however, the guanine bands varied in intensity. This phenomenon has been used to infer that AFB1 oxide prefers to react with guanines in some sequence contexts more than in others and has been referred to as "sequence specificity of binding". Herein, data on the reaction of AFB1 oxide with several synthetic DNA polymers with different sequences are presented, and (following hydrolysis) adduct levels are determined by high-pressure liquid chromatography. These results reveal that for AFB1 oxide (1) the N(7)-guanine adduct is the major adduct found in all of the DNA polymers, (2) adduct levels vary in different sequences, and, thus, sequence specificity is also observed by this more direct method, and (3) the intensity of bands in DNA sequencing gels is likely to reflect adduct levels formed at the N(7)-position of guanine. Knowing this, a reinvestigation of the reactivity of guanines in different DNA sequences using DNA sequencing methods was undertaken. The reactivities of 190 guanines were determined quantitatively and considered in a pentanucleotide context, 5'-WXGYZ-3', where the central, underlined G represents the reactive guanine and W, X, Y, and Z can be any of the nucleotide bases. Methods are developed to determine that the X (5'-side) base and the Y (3'-side) base are most influential in determining guanine reactivity. The influence of the bases in the 5'-position (X) is 5'-G (1.0) greater than C (0.8) greater than A (0.3) greater than T (0.2), while the influence of the bases in the 3'-position (Y) is 3'-G (1.0) greater than T (0.8) greater than C (0.4) greater than A (0.3). These rules in conjunction with molecular modeling studies (to be published elsewhere) were used to assess the binding sites that might be utilized by AFB1 oxide in its reaction with DNA.     

Allylic compounds bind directly to DNA: investigation of the binding mechanisms in vitro

The in vitro binding of three allyl compounds, allyl methanesulphonate, allyl bromide and allyl chloride to DNA from salmon sperm was investigated. Kinetic DNA-alkylation measurements revealed significant differences between the strongly alkylating allyl methanesulphonate and allyl bromide with half-life values of 1.5 h and 8.1 h, respectively and the weakly alkylating allyl chloride with a half-life value of 360 h. Five alkylated nucleic bases, 3-allyladenine, N6-allyladenine, N2-allylguanine, 7-allylguanine and O6-allylguanine, could be identified after DNA-alkylation with all three allyl compounds.