C8-linked bulky guanosine DNA adducts: experimental and computational insights into adduct conformational preferences and resulting mutagenicity.

Bulky DNA adducts are formed through the covalent attachment of aryl groups to the DNA nucleobases. Many of these adducts are known to possess conformational heterogeneity, which is responsible for the variety of mutagenic outcomes associated with these lesions. The present contribution reviews several conformational and mutagenic themes that are prevalent among the DNA adducts formed at the C8-site of the guanine nucleobase. The most important conclusions obtained (to date) from experiments are summarized including the anti/syn conformational preference of the adducts, their potential to inflict DNA mutations and mismatch stabilization, and their interactions with DNA polymerases and repair enzymes. Additionally, the unique role that computer calculations can play in understanding the structural properties of these adducts are highlighted.

Fluorescent C-linked C8-aryl-guanine probe for distinguishing syn from anti structures in duplex DNA.

The synthesis and optical properties of the carbon (C)-linked C(8)-(2"-benzo[b]thienyl)-2'-deoxyguanosine ((Bth)dG), which acts as a fluorescent reporter of syn versus anti glycosidic conformations in duplex DNA, are described. In the syn-conformation, the probe stabilizes a G:G mismatch, emits at ∼385 nm (excitation ∼285 nm), and shows an induced circular dichroism (ICD) signal at ∼320 nm. Molecular dynamics (MD) simulations predict a wedge (W)-conformation for the mismatched duplex with the C(8)-benzo[b]thienyl moiety residing in the minor groove. In contrast, the probe destabilizes the duplex when base paired with its normal pyrimidine partner C. With flanking purine bases, a major groove B-type duplex is favored with (Bth)dG present in the anti-conformation emitting at ∼413 nm (excitation ∼326 nm) and no ICD signal. However, with flanking pyrimidine bases, (Bth)dG adopts the syn-conformation when base paired with C, and MD simulations predict a base-displaced stacked (S)-conformation, with the opposing C flipped out of the helix. The different duplex (B-, S-, and W-) conformers formed upon incorporation of (Bth)dG are known to play a critical role in the biological activity of N-linked C8-dG adducts formed by arylamine carcinogens. Bulky environment-sensitive fluorescent C(8)-dG adducts that mimic the duplex structures formed by carcinogens may be useful in luminescence-based DNA polymerase assays.

Fluorescent properties and conformational preferences of C-linked phenolic-DNA adducts.

Phenolic toxins and mutagenic diazoquinones generate C-linked adducts at the C8 site of 2'-deoxyguanosine (dG) through the intermediacy of radical species. We have previously reported the site-specific incorporation of these adducts into oligonucleotides using a postsynthetic palladium-catalyzed cross-coupling strategy [Omumi (2011 ) J. Am. Chem. Soc. 133 , 42 - 50 ]. We report here the structural impact of these lesions within two decanucleotide sequences containing either 5'- and 3'-flanking pyrimidines or purines. In the complementary strands, the base opposite (N) the C-linked adduct was varied to determine the possibility of mismatch stabilization by the modified nucleobases. The resulting adducted duplex structures were characterized using UV thermal denaturation studies, circular dichroism, fluorescence spectroscopy, and molecular dynamics (MD) simulations. The experimental data showed the C-linked adducts to destabilize the duplex when base paired with its normal partner C but to increase duplex stability within a G:G mismatch. The stabilization within the G:G mismatch was sequence dependent, with flanking purine bases playing a key role in the stabilizing influence of the adduct. MD simulations showed no large structural changes to the B form double helix, regardless of the (anti/syn) adduct preference. Consideration of H-bonding and stacking interactions derived from the MD simulations together with the thermal melting data and changes in fluorescent emission of the adducts upon hybridization to the complementary strands implied that the C-linked phenolic adducts preferentially adopt the syn-conformation within both duplexes regardless of the opposite base N. Given that biological outcome in terms of mutagenicity appears to be strongly correlated to the conformational preference of the corresponding N-linked C8-dG adducts, the potential biological implications of phenolic C-linked adducts are discussed.

Conformational flexibility of C8-phenoxylguanine adducts in deoxydinucleoside monophosphates.

M06-2X/6-31G(d,p) is used to calculate the structure of all natural deoxydinucleoside monophosphates with G in the 5' or 3' position, the anti or syn conformation, and each natural (A, C, G, T) base in the corresponding flanking position. When the ortho or para C8-phenoxyl-2'-deoxyguanosine (C8-phenoxyl-dG) adduct replaces G in each model, there is little change in the relative base-base orientation or backbone conformation. However, the orientation of the C8-phenoxyl group can be characterized according to the position (5' versus 3'), conformation (anti versus syn), and isomer (ortho versus para) of damage. Although the degree of coplanarity between the phenoxyl ring and G base in the ortho adduct is highly affected by the sequence since the hydroxyl group can interact with neighboring bases, the para adduct generally does not exhibit discrete interactions with flanking bases. For both adducts, steric clashes between the phenoxyl group and the backbone or flanking base destabilize the anti conformation preferred by the natural nucleotide and thereby result in a clear preference for the syn conformation regardless of the sequence or position. This contrasts the conclusions drawn from smaller (nucleoside, nucleotide) models previously used in the literature, which stresses the importance of using models that address the steric constraints present due to the surrounding environment. Since replication errors for other C8-dG bulky adducts have been linked to a preference for the syn conformation, our findings provide insight into the possible mutagenicity of phenolic adducts.

An indole-linked C8-deoxyguanosine nucleoside acts as a fluorescent reporter of Watson-Crick versus Hoogsteen base pairing.

Pyrrole- and indole-linked C(8)-deoxyguanosine nucleosides act as fluorescent reporters of H-bonding specificity. Their fluorescence is quenched upon Watson-Crick H-bonding to dC, while Hoogsteen H-bonding to G enhances emission intensity. The indole-linked probe is ∼ 10-fold brighter and shows promise as a fluorescent reporter of Hoogsteen base pairing.

Effect of Watson-Crick and Hoogsteen base pairing on the conformational stability of C8-phenoxyl-2'-deoxyguanosine adducts.

Bulky DNA addition products (adducts) formed through attack at the C8 site of guanine can adopt the syn orientation about the glycosidic bond due to changes in conformational stability or hydrogen-bonding preferences directly arising from the bulky group. Indeed, the bulky substituent may improve the stability of (non-native) Hoogsteen pairs. Therefore, such adducts often result in mutations upon DNA replication. This work examines the hydrogen-bonded pairs between the Watson-Crick and Hoogsteen faces of the ortho or para C8-phenoxyl-2'-deoxyguanosine adduct and each natural (undamaged) nucleobase with the goal to clarify the conformational preference of this type of damage, as well as provide insight into the likelihood of subsequent mutation events. B3LYP/6-311+G(2df,p)//B3LYP/6-31G(d) hydrogen-bond strengths were determined using both nucleobase and nucleoside models for adduct pairs, as well as the corresponding complexes involving natural 2'-deoxyguanosine. In addition to the magnitude of the binding strengths, the R(C1'···C1') distances and ∠(N9C1'C1') angles, as well as the degree of propeller-twist and buckle distortions, were carefully compared to the values observed in natural DNA strands. Due to structural changes in the adduct monomer upon inclusion of the sugar moiety, the monomer deformation energy significantly affects the relative hydrogen-bond strengths calculated with the nucleobase and nucleoside models. Therefore, we recommend the use of at least a nucleoside model to accurately evaluate hydrogen-bond strengths of base pairs involving flexible, bulky nucleobase adducts. Our results also emphasize the importance of considering both the magnitude of the hydrogen-bond strength and the structure of the base pair when predicting the preferential binding patterns of nucleobases. Using our best models, we conclude that the Watson-Crick face of the ortho phenoxyl adduct forms significantly more stable complexes than the Hoogsteen face, which implies that the anti orientation of the damaged base will be favored by hydrogen bonding in DNA helices. Additionally, regardless of the hydrogen-bonding face involved, cytosine forms the most stable base pair with the ortho adduct, which implies that misincorporation due to this type of damage is unlikely. Similarly, cytosine is the preferred binding partner for the Watson-Crick face of the para adduct. However, Hoogsteen interactions with the para adduct are stronger than those with natural 2'-deoxyguanosine or the ortho adduct, and this form of damage binds with nearly equal stability to both cytosine and guanine in the Hoogsteen orientation. Therefore, the para adduct may adopt multiple orientations in DNA helices and potentially cause mutations by forming pairs with different natural bases. Models of oligonucleotide duplexes must be used in future work to further evaluate other factors (stacking, major groove contacts) that may influence the conformation and binding preference of these adducts in DNA helices.

Conformational flexibility of c8-phenoxyl-2'-deoxyguanosine nucleotide adducts.

Previous computational work suggests that isolated C8-phenoxyl-2'-deoxyguanosine nucleoside adducts preferentially adopt a syn orientation about the glycosidic bond, which is the first step in the mechanism by which many bulky C8 adducts exert their mutagenic effects. Since it is not clear whether these results can be directly extrapolated to the preferred conformation in DNA helices, approaches that more accurately reflect the physiological environment were used in the present study to understand the anti/syn preference of the ortho and para C8-phenoxyl-2'-deoxyguanosine adducts. Using nucleoside models and methods (B3LYP) similar to those previously implemented, we determine that the syn conformer is less stable than previously predicted when geometries relevant to B-DNA are considered. This indicates that the conformational energy trend is model dependent and stresses the importance of considering models that better mimic the DNA environment when determining the conformational preference of damaged bases. Therefore, we expanded our computational model to include the 5'-monophosphate group. Since the correct anti/syn energy trend for 2'-deoxyguanosine (dG) 5'-monophosphate has only been found using very specific computational models and prior knowledge of the biologically relevant nucleotide conformation, which is unavailable for most damaged systems, we initially benchmark our computational approach by studying the natural nucleotide. Despite the wide use of gas-phase optimizations in the current literature, only through the implementation of solvation-phase optimizations, as well as the use of a counterion model for the phosphate backbone, is the correct anti/syn energy trend predicted. Indeed, this is the first time in the literature that a biologically relevant syn structure is characterized for dG using methods suitable for studying bulky DNA adducts. Subsequently, our newly identified approach for DNA lesions was used to study C8-phenoxyl DNA adducts. In contrast to previously published results, we predict that the ortho and para adducts may adopt both the anti and syn conformations in DNA helices. These results have implications for the base-pairing properties and mutagenicity of these adducts, which must be further considered in future work.

Concerning the hydrolytic stability of 8-aryl-2'-deoxyguanosine nucleoside adducts: implications for abasic site formation at physiological pH.

Direct addition of aryl radical species to the C(8)-site of 2'-deoxyguanosine (dG) affords C(8)-aryl-dG adducts that are produced by carcinogenic arylhydrazines, polycyclic aromatic hydrocarbons (PAHs), and certain phenolic toxins. A common property of C(8)-arylpurine adduction is the accompaniment of abasic site formation. To determine how the C(8)-aryl moiety contributes to sugar loss, UV-vis spectroscopy has been employed to determine N(7) pK(a1) values and hydrolysis kinetics, while density functional theory (DFT) calculations have been utilized to probe the structural features and stability of the C(8)-aryl-dG adducts bearing different para and ortho substituents. In all cases, the C(8)-aryl-dG adducts adopt a syn conformation containing a strong O(5)'-H...N(3) hydrogen bond with the aryl ring twisted with respect to the nucleobase. The adducts undergo N(7)-protonation with ionization constants and calculated N(7) proton affinity (PA) values similar to those measured for dG. The hydrolysis kinetics shows that C(8)-aryl-dG nucleoside adducts are more prone than dG to acid-catalyzed hydrolysis, with those bearing para substituents having k(1) values that are ca. 90- to 200-fold larger than k(1) for dG, while the effects for the ortho adducts are only ca. 9- to 60-fold larger. Changes in the rate of hydrolysis are further explained by calculations showing that glycosidic bond cleavage in the syn orientation of both neutral and N(7)-protonated dG has a lower barrier than the anti orientation, and the bulky (phenyl) group further decreases the barrier. Despite adduct reactivity in acidic media, all adducts are relatively stable at physiological pH with t(1/2) approximately 25 days, suggesting that they are unlikely intermediates leading to abasic site formation at physiological pH. This information has allowed development of a new rationale for the tendency of abasic site formation to accompany C(8)-arylpurine adduction within duplex DNA at neutral pH.

yDNA versus xDNA pyrimidine nucleobases: computational evidence for dependence of duplex stability on spacer location.

The structural and binding properties of the natural and x- and y-pyrimidines were compared using computational methods. Our calculations show that although the x-pyrimidines favor different orientations about the glycosidic bond compared to the natural pyrimidines, which could have implications for the formation and resulting stability of xDNA duplexes and jeopardize the selectivity of expanded nucleobases, y-pyrimidines have rotational profiles more similar to the natural bases. Increasing the pyrimidine size using a benzene spacer leads to relatively minor changes in the hydrogen-bond strength of isolated Watson-Crick base pairs. However, differences in the anomeric carbon distances in pairs composed of x- or y-pyrimidines suggest yDNA may yield a more optimal expanded structure. By stacking two monomers via their centers of mass, we find that the expanded nucleobases stack much stronger than the natural bases. Additionally, although replacing xT by yT changes the stacking energy by less than 5 kJ mol (-1), replacing xC by yC significantly strengthens complexes with the natural nucleobases (by up to 30%). Calculations on larger duplex models composed of four nucleobases reveal that x- and y-pyrimidines can increase duplex stability of natural helices by strengthening both the intra and interstrand stacking interactions. Furthermore, when the total stability (sum of all hydrogen-bonding and (intrastrand and interstrand) stacking interactions) of the larger models is considered, y-pyrimidines lead to more stable complexes than x-pyrimidines for all but three duplex sequences. Thus, through analysis of a variety of properties, our calculations suggest that the location of the benzene spacer affects the properties of expanded nucleobases and the stability of expanded duplexes, and therefore should be carefully considered when designing future expanded analogues.

Computational and experimental evidence for the structural preference of phenolic C-8 purine adducts.

The structural and spectral properties of (ortho and para) C8-aryl-purine adducts formed from carbon attachment by phenolic toxins were investigated through DFT calculations and UV-vis absorbance and emission studies. The global minima of both the deoxyadenosine (dA) and deoxyguanosine (dG) adducts adopted a syn conformation about the glycosidic bond due to the presence of an O5'-H...N3 hydrogen bond, where the anti minima are 20-30 kJ mol-1 higher in energy. While the nucleobase adducts are planar, the presence of the deoxyribose sugar induces a twist about the carbon-carbon bond connecting the phenol and nucleobase rings. ortho-Phenolic adducts are less twisted than the corresponding para adducts due to stabilization provided by an intramolecular O-H...N7 bond. Solvation calculations, in combination with UV-vis studies, demonstrate that the structural preference is solvent dependent, where solvents with hydrogen-bonding abilities disrupt the intramolecular O-H...N7 hydrogen bond such that a greater degree of twist is observed, and less polar solvents stabilize the planar structure. Indeed, the ratio of twisted to planar conformers is estimated to be as large as 50:50 in some aprotic solvents. Thus, the combined experimental and computational approach has provided a greater understanding of the structure of the ortho- and para-dA and dG C-bonded phenoxyl adducts as the first step to understanding the biological consequences of this form of DNA damage.

A kinetic and thermodynamic study of the glycosidic bond cleavage in deoxyuridine.

Density functional theory was used to study the thermodynamics and kinetics for the glycosidic bond cleavage in deoxyuridine. Two reaction pathways were characterized for the unimolecular decomposition in vacuo. However, these processes are associated with large reaction barriers and highly endothermic reaction energies, which is in agreement with experiments that suggest a (water) nucleophile is required for the nonenzymatic glycosidic bond cleavage. Two (S(N)1 and S(N)2) reaction pathways were characterized for direct hydrolysis of the glycosidic bond by a single water molecule; however, both pathways also involve very large barriers. Activation of the water nucleophile via partial proton abstraction steadily decreases the barrier and leads to a more exothermic reaction energy as the proton affinity of the molecule interacting with water increases. Indeed, our data suggests that the barrier heights and reaction energies range from that for hydrolysis by water to that for hydrolysis by the hydroxyl anion, which represents the extreme of (full) water activation (deprotonation). Hydrogen bonds between small molecules (hydrogen fluoride, water, or ammonia) and the nucleobase were found to further decrease the barrier and overall reaction energy but not to the extent that the same hydrogen-bonding interactions increase the acidity of the nucleobase. Our results suggest that the nature of the nucleophile plays a more important role in reducing the barrier to glycosidic bond cleavage than the nature of the small molecule bound, and models with more than one hydrogen fluoride molecule interacting with the nucleobase provide further support for this conclusion. Our results lead to a greater fundamental understanding of the effects of the nucleophile, activation of the nucleophile, and interactions with the nucleobase for this important biological reaction.

Effects of hydrogen-bonding and stacking interactions with amino acids on the acidity of uracil.

The effects of hydrogen-bonding interactions with amino acids on the (N1) acidity of uracil are evaluated using (B3LYP) density functional theory. Many different binding arrangements of each amino acid to three uracil binding sites are considered. The effects on the uracil acidity are found to significantly depend upon the nature of the amino acid and the binding orientation, but weakly depend on the binding site. Our results reveal that in some instances small models for the amino acids can be used, while for other amino acids larger models are required to properly describe the binding to uracil. The gas-phase acidity of uracil is found to increase by up to approximately 60 kJ mol(-1) due to discrete hydrogen-bonding interactions. Although (MP2) stacking interactions with aromatic amino acids decrease the acidity of uracil, unexpected increases in the acidity are found when any of the aromatic amino acids, or the backbone, hydrogen bond to uracil. Consideration of enzymatic and aqueous environments leads to decreases in the effects of the amino acids on the acidity of uracil. However, we find that the magnitude of the decrease varies with the nature of the molecule bound, as well as the (gas-phase) binding orientations and strengths, and therefore solvation effects should be considered on a case-by-case basis in future work. Nevertheless, the effects of amino acid interactions within enzymatic environments are as much as approximately 35 kJ mol(-1). The present study has general implications for understanding the nature of active site amino acids in enzymes, such as DNA repair enzymes, that catalyze reactions involving anionic nucleobase intermediates.