Chem. Res. Toxicol. 2005, 18, 1347-1363
1347
Articles Molecular Dynamics of a Food Carcinogen-DNA Adduct in a Replicative DNA Polymerase Suggest Hindered Nucleotide Incorporation and Extension Ling Zhang,† Robert Shapiro,† and Suse Broyde*,‡ Departments of Chemistry and Biology, New York University, New York, New York 10003 Received May 18, 2005
2-Amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) is the most abundant of the carcinogenic heterocyclic aromatic amines in the human diet, and the major mutagenic effect of dietary PhIP is GfT transversions. The major PhIP-derived DNA adduct is to C8 of guanine. We have investigated this adduct in a PhIP-induced mutational hotspot 5′-GGGA-3′ of the Apc tumor suppressor gene, frequently mutated in mammalian colon tumors. We have carried out a molecular dynamics study to elucidate on a structural level nucleotide incorporation and extension opposite this major adduct during replication. The PhIP adduct was modeled into the ternary complex closed conformation of DNA polymerase RB69, at incorporation and extension positions, with normal cytosine or mismatched partner adenine. RB69 polymerase is a member of the B family as are most replicative eukaryotic DNA polymerases such as DNA polymerase R. These systems were subjected to molecular dynamics simulations with AMBER. Our results show that the adduct can reside on the major groove side of the modified DNA template opposite an incoming dCTP or dATP. In the case of the normal partner, disturbance to the active site is observed at the incorporation step, but there is less perturbance in the extension simulation. In the case of the mismatched partner, a less disturbed active site is observed during the incorporation step, but extension appears to be more difficult. Disturbances include adverse impacts on Watson-Crick hydrogen bonding in the nascent base pair, on the distance between the R-phosphate of the incoming dNTP and the primer terminus 3′-OH, and on critical protein interactions with the dNTP. However, in all of these cases, a near reaction ready distance (within 3.5 Å) between the 3′-terminal oxygen of the primer and the PR of the incoming nucleotide triphosphate is sampled occasionally (0.4-23.5% of the time). Thus, error-free bypass or the induction of a GfT transversion mutation could occur at times and contribute to an extent to the mutagenic effect of PhIP. Polymerase stalling would be the more common outcome and in vivo could lead to switch to an error-prone bypass polymerase.
Introduction Heterocyclic aromatic amines (HAAs)1 are environmental carcinogens formed during high temperature cooking of proteinaceous foods, such as meat and fish (14). After metabolic activation, the reactive derivatives of these chemicals can form covalent adducts with DNA, and the carcinogen-damaged DNA may cause mutations during DNA replication (5-10). If these mutations are * To whom correspondence should be addressed. Tel: 212-998-8231. Fax: 212-995-4015. E-mail:
[email protected]. † Department of Chemistry. ‡ Department of Biology. 1 Abbreviations: HAA, heterocyclic aromatic amine; PhIP, 2-amino1-methyl-6-phenylimidazo[4,5-b]pyridine; dG-C8-PhIP, N2-(2′-deoxyguanosin-8-yl)-PhIP; dG, 2′-deoxyguanosine; Apc, adenomatous polyposis coli; dNTP, 2′-deoxynucleoside triphosphate; dCTP, 2′-deoxycytidine triphosphate; dATP, 2′-deoxyadenosine triphosphate; pol R, DNA polymerase R; MD, molecular dynamics; SD, steepest descent; CG, conjugate gradient; AF, aminofluorene; dG-C8-AF, N2-(2′-deoxyguanosin-8-yl)-2-aminofluorene; BF, Bacillus DNA polymerase I fragment; AAF, acetylaminofluorene; dG-C8-AAF, N2-(2′-deoxyguanosin-8-yl)-2acetylaminofluorene; pol β, DNA polymerase β.
in genes such as oncogenes or tumor suppressor genes, which play key roles in cell cycle control, the process of carcinogenesis can be initiated (11-13). The most abundant of the HAAs in the human diet is PhIP, 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (4), a substance known to cause mammary and colon cancers in mammals (14-16). PhIP is metabolically activated into reactive intermediates via two steps (1719); these involve N-hydroxylation, catalyzed primarily by cytochrome P450 in humans, to produce the proximate carcinogen N2-OH-PhIP, followed by acetylation via trans-acetylase enzymes to yield N2-acetoxy-PhIP. This ultimate mutagenic metabolite is reactive toward DNA and forms a major adduct to C8 of guanine, dG-C8-PhIP, [N2-(2′-deoxyguanosin-8-yl)-PhIP] (Figure 1a) (20, 21). In addition to the dG-C8-PhIP adduct, a number of other products including N2-dG-PhIP are formed in the reaction (22, 23). Most of these minor adducts of PhIP-dG have not yet been fully characterized (23, 24).
10.1021/tx050132b CCC: $30.25 © 2005 American Chemical Society Published on Web 08/31/2005
1348
Chem. Res. Toxicol., Vol. 18, No. 9, 2005
Zhang et al.
Figure 2. Possible mutagenesis mechanisms. Although not illustrated, there is also the possibility of slippage in pol R.
Figure 1. (a) Structure and numbering scheme of the dG-C8PhIP adduct. Torsion angle definitions are as follows: χ, O4′C1′-N9-C4; R′, N9-C8-[PhIP]N2-[PhIP]C2; β′, C8-[PhIP]N2-[PhIP]C2-[PhIP]N1; γ′, [PhIP]C5-[PhIP]C6-[PhIP]C1′[PhIP]C2′. (b) dG-C8-PhIP modified DNA sequence used in the G3*‚dCTP incorporation system. (c) dG-C8-PhIP modified DNA sequence used in the G4*‚C32 extension system. (d) dG-C8-PhIP modified DNA sequence used in the G3*‚syn-dATP incorporation system. (e) dG-C8-PhIP modified DNA sequence used in the G4*‚syn-A32 extension system.
A high resolution NMR solution structure of the dGC8-PhIP adduct in an 11-mer duplex has been published and reveals that the adduct adopts predominantly a conformation termed base-displaced intercalation (25). In this conformation, the modified guanine is syn and displaced into the major groove, while the PhIP moiety intercalates into the DNA double helix. The imidazopyridine ring of the PhIP stacks with the adjacent base pairs. The N-methyl group and phenyl ring protrude into the minor groove, and the phenyl ring is twisted with respect to the imidazopyridine ring, with a torsion angle γ′ (Figure 1a) of 26°. A minor, uncharacterized conformation was also noted, and it was suggested to be located outside the helix in a minimally perturbed B-DNA duplex and exposed to the solvent. However, nothing is known about the adduct structure within a DNA polymerase. Such structures are needed in order to understand how the carcinogen-damaged DNA may affect polymerase function. GfT transversion mutations are the major base substitution caused by dietary PhIP in the rat and mouse
and in hamster, simian, and human cells treated with PhIP (26-31). GfT transversion (5′-GTGGGAT-3′f5′GTTGGAT-3′) has been detected in the adenomatous polyposis coli (Apc) gene of PhIP-treated rat colon (32). The Apc gene is an 8.5 kb tumor suppressor gene, frequently mutated in sporadic human colon cancers (33, 34). In addition, metabolically activated PhIP can cause frameshift mutations (29-31, 35, 36). A -1 frameshift hotspot was also observed in a 5′-GGGA-3′ sequence of the Apc gene of PhIP-induced rat colon tumors (36) and in the mammary gland of PhIP-treated rats harboring the λ lacI mutational reporter transgene (31). This mutation involves a deletion of a G in a sequence of three Gs during replication. However, the experiments did not reveal which G was deleted or modified. At this stage, we consider the major PhIP-guanine adduct (Figure 1a), although other adducts may be involved in PhIP-induced mutagenesis, as well. Site specific mutagenesis experiments for this adduct have been performed in simian kidney (COS-7) cells (30). In all sequences, normal bypass was the most common result, but a significant number of GfT transversions (8-24%) were also observed, with the amount dependent on the 5′-neighbor. Frameshift mutations were observed in a GG* and TG* context. Current evidence suggests that the predominant processing of bulky carcinogen-DNA adducts by DNA polymerases involves a polymerase switch (37-42). Such lesions on DNA often stall or block the replicative high fidelity DNA polymerases, which frequently leads to their release. An error-prone Y family bypass polymerase may then be called in to transit the lesion (43-45). Possible mechanisms for induction of a GfT transversion or a -1 deletion by PhIP are shown in Figure 2 (30, 46-52). In the first step, the incoming nucleotide is in the active site of a replicative polymerase, with the primer in position to react with the correct incoming
PhIP-DNA Adduct/Polymerase Molecular Dynamics
nucleotide 2′-deoxycytidine triphosphate (dCTP), or a mismatched 2′-deoxyadenosine triphosphate (dATP), opposite the PhIP-damaged guanine. The polymerase may stall at this step, and incorporation of the incoming nucleotide may be inhibited because of structural disturbance to the active site by the bulky adduct. The replicative polymerase may then be released from the replication site and replaced by a bypass polymerase. Slippage may occur directly via a 2′-deoxynucleoside triphosphate (dNTP)-stabilized misalignment (46-48) or by insertion opposite the adduct followed by rearrangement (30, 49-52) in a bypass polymerase. The possibility of slippage within a replicative polymerase also exists. Alternatively, the incoming nucleotide may be incorporated and further extension of the primer may take place. In the present work, we examined the behavior of the major dG-C8-PhIP adduct in a ternary complex (primer, template, and incoming nucleoside triphosphate) of a replicative polymerase. No suitable unmodified crystal structure of such a complex is currently available from a mammalian polymerase. One has been reported, however, from bacteriophage RB69, which has been classified as a member of the B family, similar to mammalian DNA polymerase R (pol R) (53). This classification was made on the basis of a structural comparison to other pol R family polymerases (54-56). The pol R’s are a replicative polymerase family (57), which have associated DNA primase activity, and are believed to be involved in both the initiation of DNA replication and the repeated priming events necessary for replication of the lagging strand (57, 58). Although the 903 amino acids of RB69 show only 15% sequence homology with human pol R, RB69 shows a high degree of structural equivalence to other pol R family polymerases, which suggests that it is a good structural model for them (53). Details of sequence comparison between RB69 and human pol R and their possible relation to functional domains are given in Rodriguez et al. (55) and Wang et al. (59). Key amino acid residues important in maintaining the polymerase active site for efficient and faithful nucleotide incorporation are conserved in the pol R family. They have a shape like a right hand, containing palm, fingers, and thumb domains, undergo a conformational transition from an open to a closed conformation upon binding of the incoming nucleotide (53, 59, 60), and employ a two metal ion catalytic mechanism for the phosphoryl transfer reaction (61). The conformational transition brings critical residues into position for the nucleophilic attack of the 3′-ΟΗ of the primer on the R-phosphate of the incoming nucleotide. Our goal in the present work is to investigate the effect of the major dG-C8-PhIP adduct on the RB69 ternary complex structure with dCTP or mismatched dATP as the incoming triphosphate. We studied the processes of incorporation opposite the adduct (Figure 1b,d) and extension beyond it (Figure 1c,e). Molecular modeling and dynamics simulations have been carried out to obtain an ensemble of structures for detailed structural analyses. Our simulations suggest the possibility of infrequent nucleotide incorporation and extension for all cases. However, active site disturbances indicate that polymerase stalling would be more prevalent. In vivo, switch to a bypass polymerase would then be likely.
Chem. Res. Toxicol., Vol. 18, No. 9, 2005 1349
Materials and Methods Molecular Modeling. The crystal structure of the RB69 DNA polymerase ternary complex was used to obtain the starting models, using coordinates obtained from the Protein Data Bank (PDB ID: 1IG9) (62). Hydrogen atoms that were not present in the crystal structure were added by the AMBER 6.0 molecular dynamics (MD) software package (63). A hydroxyl group was added to the 3′-end of the primer, since in the crystallographic structure a dideoxynucleotide was used in this position to prevent reaction within the crystal. The active site was then remodeled. First, because the 3′-end of the primer was a dideoxynucleotide, the R-phosphate (PR) of the incoming nucleotide was somewhat distant from the modeled 3′-OH of the primer. Therefore, small adjustments in the DNA backbone torsion angles R and (64) in the last four nucleotides before the 3′-terminus of the primer were made to reduce the distance from 5.3 to 3.4 Å (Table S1 gives the remodeled values). Second, the two metal ions in the active site were Ca2+ in the original crystallographic structure. We replaced them with the more biologically relevant Mg2+ ions. Third, an extra Ca2+ ion, which has no physiological role, was trapped in the crystal. Because it is not near the active site, we deleted it. However electrostatic neutrality was maintained through added Na+ ions (see below). Finally, the two Mg2+ ions were slightly repositioned to achieve proper octahedral coordination with protein residues and a water molecule. Figure S1 shows the geometry of the octahedral coordination. The DNA sequences were remodeled to match a part of the Apc gene around mutational hotspot codon 635, which contains three Gs (36). The guanine selected for modification with adenine opposite the damaged base corresponds to a sequence context supporting G to T transversion mutations (32). With cytosine opposite the modified base, we selected modification of the third G in the frameshift hotspot, to allow for the possibility of deletion slippage during dynamics. We retained the number of DNA residues on each strand as in the RB69 crystal structure (53). In the incorporation complexes, the PhIP moiety was linked to the guanine in the template opposite incoming nucleotide, dCTP (G3*‚dCTP incorporation complex, Figure 1b), or dATP (G3*‚syn-dATP incorporation complex, Figure 1d); in the extension complexes, it was linked to the guanine at the 5′-end of the templating base opposite incoming nucleotide, dCTP (G4*‚C32 extension complex, Figure 1c) or dATP (G4*‚syn-A32 extension complex, Figure 1e). The remodeled DNA sequences are shown in Figure 1b-e. As a control for all of these modified systems, we employed unmodified anti guanine with Watson-Crick partners, the normal substrate for the enzyme. However, we also simulated a G3‚syn-dATP as an additional control for the G3*‚syn-dATP modified system. The torsion angles R′, β′ at the linkage site of the dG-C8-PhIP adduct (defined in Figure 1a) were rotated at 10° intervals, in combination to find locations within the polymerase without steric clashes (see Results). The torsion angles in the starting models for MD simulation are summarized in Table S2. Force Field Parametrization. To carry out the MD simulations, new force field parameters had to be developed for the modified residues that are not present in the AMBER force field (65). These must be consistent with the rest of the AMBER force field, which only includes standard protein and nucleic acid residues. The modified residues in our system included the dGC8-PhIP adduct and the incoming nucleotides dCTP and dATP. For the partial charge calculations, we constructed a nucleoside containing the PhIP adduct. Hartree-Fock calculations with basis set 6-31G* in GAUSSIAN 98 (66) were used to quantum mechanically geometry optimize the structure and calculate the electrostatic potential. The least squares chargefitting algorithm, RESP, provided with AMBER, was then used to fit the charge to each atomic center (67). Two structures of the dG-C8-PhIP adduct, corresponding to family 1 and family 2 conformations (see Results), were used, and final charges were averaged. Finally, the charges were normalized to maintain a charge of -1 on the PhIP-modified nucleotide to avoid charge
1350
Chem. Res. Toxicol., Vol. 18, No. 9, 2005
imbalance (68). In the case of dCTP or syn-dATP, a similar protocol was used. Tables S3-5 show the AMBER atom type, connection type, and partial charge assignments for the modified residues. Partial charges for anti-dATP were computed in previous work (69). Equilibrium bond lengths and angles for the PhIP moiety were obtained from the quantum mechanical geometry optimization. Force constants for the carcinogen bond lengths and bond angles were assigned by employing values in the force field database for chemically similar bonds and angles. The dihedral parameters for the dG-C8-PhIP linkage torsion angles R′ and β′ (Figure 1a), which include the rotation barrier, the dihedral angle, and the periodicity, were previously obtained (68). Table S6 shows the parameters added to the AMBER force field. MD Simulation Protocol. Simulations were carried out using the SANDER module of AMBER 6.0, the Cornell et al. force field (65), and the parm99.dat parameter set (66). The electrostatic interactions were approximated by the particle mesh Ewald method (70), and a 12 Å cutoff was applied to Lennard-Jones interactions. All bonds involving hydrogen atoms were constrained by the SHAKE algorithm (71) with a tolerance of 0.0005 Å, and a 2 fs time step was used in the dynamics simulation. Periodic boundary conditions were applied, and all MD simulations were carried out under constant pressure and constant temperature. The translational motion of the center of mass was removed every 1 ps. No obvious overall rotation of the system was observed during the simulation; thus, energy leakage from internal motion to global rotation through the “flying ice cube effect” did not happen here (72). The initial model of the G3*‚dCTP incorporation complex (Figure 1b) was further prepared for MD by the LEaP module of AMBER 6.0. First, the solute was neutralized with counterions. Thirty-eight Na+ ions were placed at the positions of minimum electrostatic potential of the system. Then, 356 crystallographic water molecules, whose hydrogen atoms were added by the AMBER suite, were added to the neutralized system. Next, hydrogen atoms, Na+, and crystallographic waters were minimized by AMBER 6.0 for 600 steps of steepest descent (SD) followed by 400 steps of conjugate gradient (CG) while holding the solute fixed. Then, the whole system was solvated with a truncated octahedral box of TIP3P water molecules (73), using a 9 Å buffer around the solute, to create a periodic box. TIP3P water molecules (23832 and 23933) were added respectively to the G3*‚dCTP incorporation complex system and the unmodified control system, to yield a box with a total of 88307 and 88583 atoms, respectively. The following minimization and equilibration protocol was carried out for all of the systems. First, Na+ ions and water molecules were minimized by AMBER 6.0 for 50 steps of SD followed by 3950 steps of CG, while holding the solute fixed with harmonic restraints of 55 kcal/mol Å. Then, the Na+ ions and water molecules were relaxed with 30 ps of MD conducted at 10 K, while holding the solute fixed with harmonic restraints of 20 kcal/mol Å. The system was then heated to 300 K over 80 ps, using the Berendsen coupling algorithm (74) with a coupling parameter of 0.2 ps, and then held at 300 K for 20 ps while restraining the solute with 20 kcal/mol Å. The rest of the equilibration was carried out at 300 K. At this stage, the restraints on the solute were released slowly over 130 ps of MD by running 30 ps of MD with 10 kcal/ mol Å restraints, 40 ps of MD with 1 kcal/mol Å restraints, and 60 ps of unrestrained MD. Unrestrained production MD of 2.5 ns was begun after the equilibration, at 300 K with a temperature coupling parameter of 4.0 ps and at atmospheric pressure with a 1.0 ps coupling parameter. Trajectory Analysis. MD simulations of 2.5 ns were carried out for each modified system and its respective control. Trajectories were collected for analysis of the structural characteristics of the modified systems in comparison with the respective unmodified controls. The PTRAJ and CARNAL models of the AMBER package were employed to analyze the trajectories (63). Stability of the MD Simulation. The RMSDs (root-meansquare deviations) of the G3*‚dCTP incorporation, G4*‚C32
Zhang et al. extension, G3*‚syn-dATP incorporation, and G4*‚syn-A32 extension systems calculated relative to the starting structures are shown in Figure S2. The RMSDs of the active site, including every residue within 5 Å of the nascent base pair relative to the starting structures, are also given. We found that the incorporation systems (G3*‚dCTP and G3*‚syn-dATP) are stable from 0.5 to 2.5 ns and the extension systems (G4*‚C32 and G4*‚ syn-A32) from 1.0 to 2.5 ns (Figure S2). The rest of our analyses are based on these stable ranges. Stereoviews of unmodified and modified systems after 2.5 ns of unrestrained MD simulations are shown in Figure S3a-d, and the active site of the modified systems is shown in Figure 3a-d. All molecular modeling was carried out with Insight II 97.0 (Accelrys,Inc., a subsidiary of Pharmacopeia, Inc.). Structural figures were prepared with PyMOL (75).
Results Initial Models. 1. G3*‚dCTP Normal Pair Incorporation Complex. We first built initial models for MD simulations with the PhIP-modified guanine in the active site of the RB69 DNA polymerase closed ternary complex. The PhIP moiety contains the heterocyclic ring system, the bulky N-methyl group, and the distal phenyl ring. Our goal in building the starting models was to locate structures with minimal crowding between the PhIP ring system and the polymerase. For the G3*‚dCTP incorporation complex (Figure 1b), flexibility in torsion angles R′, β′, and γ′ (Figure 1a) was considered. On the basis of structural studies of biphenyl, γ′ twist angles are in the range of about (20-40°) (76, 77). In the NMR solution structure of the PhIP major adduct, γ′ ) 26° (25). To simplify the search, we utilized this value for γ′ in the starting models, but the rotation of the phenyl ring is flexible during the MD simulations, and indeed, the range of -38 to 38° was observed, as described below. We investigated the anti conformation of the PhIPmodified guanine, initially retaining the glycosidic torsion χ as in the unmodified crystal structure (53), so that Watson-Crick base pairing with the incoming dCTP was maintained in the G3*‚dCTP incorporation complex, and the PhIP ring system was on the major groove side of the template. We then rotated R′ and β′ at 10° intervals, in combination, to find structures without collisions. Two such families were found. In the first family, the PhIPN-methyl group is oriented toward the 5′-end of the modified template, and in the second family, it is oriented toward the 3′-end. There is no stacking between the PhIP ring system, which is 3′-directed along the modified strand, and any DNA bases. These families differ by a ∼180° rotation about β′. In family 1, the R′, β′ ranges are, respectively, 125 ((25°) and 120° ((20°), and in family 2, these values are, respectively, 125 ((25°) and 305° ((20°). Figure S4 shows these models. We selected family 1 for our first simulation (G3*‚dCTP incorporation complex), because its β′ was in the same range as in the NMR solution structure (25). As shown below, early in the simulation, this structure rearranged to the family 2 model. We also attempted to find a model with the PhIPmodified guanine syn, as in the NMR solution structure, but all R′, β′ combinations with syn guanine produced structures in which the PhIP collided severely with the amino acid residues that are in close contact with the minor groove side of the nascent base pair; these residues constitute a minor groove scanning track, which is critical for fidelity in replicative polymerases (53, 61, 78-82) (Figure S5). Such structures would be expected to block
PhIP-DNA Adduct/Polymerase Molecular Dynamics
Chem. Res. Toxicol., Vol. 18, No. 9, 2005 1351
Figure 3. Stereoviews of the active sites in the (a) G3*‚dCTP incorporation complex, (b) G4*‚C32 extension complex, (c) G3*‚syndATP incorporation complex, and (d) G4*‚syn-A32 extension complex. The RB69 DNA polymerase, shown in ribbon representation, is colored as follows: the N-terminal domain is in violet, the exonuclease domain is in cyan, the palm is in green, the fingers are in yellow, and the thumb is in orange. The DNA ternary complex is shown in stick representation. The primer and template DNA strands and the normally paired incoming nucleotide are colored in gray, while the syn-dATP or syn-A32 is colored in blue, and the PhIP-guanine adduct is colored in red. Mg2+ ions A and B, shown in spheres representation, are colored in purple. All stereo figures were made for viewing with a stereoviewer.
the polymerase, as has been observed for other polymerase/adduct systems (83, 84). 2. G3*‚syn-dATP Mismatched Pair Incorporation Complex. For the G3*‚dATP incorporation complex (Figure 1d), a similar approach was employed to obtain the initial models, and the same family 1 and family 2 anti conformations can be adopted by the dG-C8-PhIP
adduct. To accommodate two purines in the active site of the replicative polymerase, either the templating base or the incoming nucleotide must adopt the syn conformation. Because the template G3* can only adopt the anti conformation to avoid PhIP’s clashing with minor groove protein residues, the incoming dATP is syn in the initial model. With χ ) 12°, dATP pairs with G3* through a
1352
Chem. Res. Toxicol., Vol. 18, No. 9, 2005
Zhang et al.
Figure 4. Stereoviews of snapshots of the PhIP moiety of dG-C8-PhIP (G3*) and Lys 279 at (a) 200 and (b) 500 ps in the G3*‚dCTP incorporation complex system; see text.
hydrogen bond between N7 on the base of syn-dATP and N1 on the base of G3*. In this case of the G3*‚syn-dATP incorporation complex (Figure 1d), both family 1 and family 2 complexes were investigated. A syn incoming dATP has been observed in the active site of another DNA polymerase (85). Conformational Change of dG-C8-PhIP. 1. G3*‚ dCTP Normal Pair Incorporation Complex. In the G3*‚dCTP incorporation complex (Figure 1b), the conformation of the dG-C8-PhIP adduct in the starting model rotated around β′ by ∼150° (from an average of 157 ( 15.6° to an average value of 310 ( 15.6°) in the 250-480 ps time frame, a transition from family 1 to family 2, and remained stably in this state for the duration of the simulation. The β′-rotation leads to flip of the PhIP rings. At the same time, Lys 279 is rotated around the torsion angle between Cβ and Cγ by ∼180°. As a result of PhIP’s flip and Lys 279’s rotation, the negatively charged atoms N3 and N4 on the imidazopyridine ring of the PhIP move closer (about 3 Å) to the positively charged three H atoms (H1,2,3) of Lys 279. Electrostatic interactions between these atoms on the PhIP and Lys 279 help to stabilize the PhIP in the new position (Figure 4). Also, hydrogen bonds are formed between the N3 and N4 of PhIP with H1,2,3 of Lys 279, with occupancies of, respectively, 10.7 and 43.3%. Figure S6 shows the time dependence of the torsion angles, χ, R′, β′, and γ′. The glycosidic torsion angle of the PhIP-modified guanine, χ, has an average value of 200.6 ( 16.8°, keeping the guanine in the anti conformation opposite the incoming nucleotide dCTP. The torsion angle governing the orientation of the PhIP, R′, has an average value of 145.3 ( 22.8°, in the range of the starting model. The phenyl ring orientation is flexible during the MD simulation as shown in Figure S6; the γ′ torsion angle samples the -38 to 38° range, with an average value of 0.9 ( 12.7°.
2. G3*‚syn-dATP Mismatched Pair Incorporation Complex. In the G3*‚syn-dATP incorporation system (Figure 1d), no major conformational change of the dGC8-PhIP adduct was observed during the MD simulation in either the family 1 or the family 2 case. During the simulation, the adduct that adopted the family 1 conformation in the initial model remained in family 1, due to formation of a hydrogen bond between N2 of the PhIP and O4′ on the deoxyribose of the modified guanine. Similarly, the initial family 2 model remained in family 2, due to electrostatic interactions between the PhIP and the Lys 279, as discussed previously for the normally paired G3*‚dCTP incorporation complex (Figure 1b). In both families 1 and 2, the PhIP maintained its position on the major groove side of the modified template, and the PhIP-modified guanine actually achieved one full and one bifurcated hydrogen bond with the syn-dATP (Figure 5). We present the simulation results for family 2 here, while those for family 1 are given in the Supporting Information (Table S7 and Figures S7-S9). The main difference between these two family models is the orientation of the adduct (rotated ∼180° around its long axis), while the other structural characteristics of the active sites relevant for reaction are quite similar. However, the distance between the 3′-terminal oxygen of the primer and the PR of the syn-dATP is more frequently near the reaction ready range of ∼3.1-3.5 Å in family 2, as described below. Hydrogen Bonds in the Nascent Base Pair. 1. G3*‚ dCTP Normal Pair Incorporation Complex. Hydrogen bonds between the dCTP and the PhIP-damaged guanine (G3*) in the G3*‚dCTP incorporation system (Figure 1b) are less stable than in the unmodified control. As shown in Table 1, two of these Watson-Crick hydrogen bonds have occupancies reduced by more than 10% in the modified system.
PhIP-DNA Adduct/Polymerase Molecular Dynamics
Chem. Res. Toxicol., Vol. 18, No. 9, 2005 1353
Figure 5. Hydrogen bonds between the nascent base pair G3*‚ syn-dATP in the G3*‚syn-dATP incorporation system are shown in solid cyan lines. Table 1. Hydrogen Bond Occupancies Involving the Nascent Base Pair in the G3*‚dCTP Incorporation and the G3*‚syn-dATP Incorporation Systemsa
donor
acceptor
identified in undamaged damaged RB69 crystal system (%) system (%) structure (53) G3*‚dCTP 96.5 97.8 89.6 70.5 44.2 58.6 99.3 56.2 36.8 43.2
G3* H21 G3* H1 dCTP HN42 Ile362 H Asn572 Hγ22 Tyr416 H Arg482 Hη22 Arg482 Hη12 Lys486 Hζ1,2,3 Lys560 Hζ1,2,3
dCTP O2 dCTP N3 G3* O6 G3* O2P G3* O3′ dCTP O3′ dCTP O3γb dCTP O3γb dCTP O3γ dCTP O3R
G3* H21 G3* H1 syn-dATP H62 Ile362 H Asn572 Hγ22 Tyr416 H Arg482 Hη22 Arg482 Hη12 Lys486 Hζ1,2,3 Lys560 Hζ1,2,3
G3*‚syn-dATP syn-dATP N7 N/Ac syn-dATP N7 N/A G3* O6 N/A G3* O2P 88.6 G3* O3′ 73.6 dNTPd O3′ 97.5 dNTP O2γb 99.9 dNTP O2γb 54.2 dNTP O2γ 19.4 dNTP O3R 48.3
99.5 87.7 75.4 88.3 50.9 60.8 98.7 51.5 36.5 0
yes yes yes yes yes yes yes no no yes
93.4 74.0 72.1 93.1 56.7 3.0 99.8 49.1 0 51.8
yes yes yes yes yes yes yes no no yes
a % occupied indicates the percentage of the time (time range, 0.5-2.5 ns) that the hydrogen bond was intact. Hydrogen bonding criteria were 3.3 Å between heavy atoms and a hydrogen bonding angle of 135°. b All Oγs are nonbridging oxygens and are equivalent. c Not applicable, since the unmodified system does not have a syn-dATP. d dNTP: dCTP in the unmodified system and syndATP in the modified system. Results for the G3‚syn-dATP unmodified control simulation are given in Table S8.
2. G3*‚syn-dATP Mismatched Pair Incorporation Complex. In the G3*‚syn-dATP incorporation system (Figure 1d), one full and one bifurcated hydrogen bond is present between the templating base G3* and the mismatched incoming syn-dATP during the MD simulation (Figure 5); a similar hydrogen bonding pattern was observed in the G3‚syn-dATP unmodified control (Table S8 gives full results for this simulation). The occupancies of these hydrogen bonds are greater than 72% in the G3*‚ syn-dATP incorporation system (Table 1); thus, these hydrogen bonds act to stabilize the mismatched nascent base pair. Achievement of these hydrogen bonds between the G3* and the syn-dATP occurs at the expense of optimal stacking between the base of the syn-dATP and the base of the 3′-end primer nucleotide C32. The N1 acceptor on the base of the syn-dATP forms a hydrogen bond with water molecules during 83% of the simulation.
Figure 6. Rearrangement of dCTP in the G3*‚dCTP incorporation complex. dCTP in the starting structure (the triphosphate is colored in gray and wheat) and the simulated G3*‚dCTP incorporation complex (the triphosphate is colored in purple and red) are superimposed. The specific amino acid residues are colored according to the different domains as in Figure 3, as are Mg2+ ions A and B. The lost interactions (between Hζ of Lys 560 and O3R of dCTP, between O2β of dCTP and Mg2+ B) due to the conformational change of the triphosphate caused by the adduct in the G3*‚dCTP incorporation system are indicated by the lines colored in cyan.
Hydrogen Bonds Involving Key Amino Acid Residues and the Nascent Base Pair. 1. G3*‚dCTP Normal Pair Incorporation Complex. In the G3*‚ dCTP incorporation system (Figure 1b), one of the hydrogen bonds between key amino acid residues and the triphosphate of the incoming nucleotide has a much lower occupancy than in the unmodified control, as shown in Table 1. These interactions are needed to maintain the proper geometry of the incoming nucleotide. Specifically, the positions of the residues Arg 482, Lys 486, and Lys 560 in the long R-helixes of the fingers domain are important for catalysis of the phosphoryl transfer step, which adds the incoming nucleotide to the primer end (53). The fingers residues Arg 482 and Lys 486 interact with the γ-phosphate of dCTP and separately form hydrogen bonds with one of the nonbridging oxygens (O3γ) of the γ-phosphate. These hydrogen bonds have similar occupancies in both the G3*‚dCTP incorporation complex and the unmodified system (Table 1). In the crystal structure, Lys 560 forms a hydrogen bond with the bridging oxygen (O3R) between the R-phosphate and the β-phosphate of dCTP. The Lys 560 hydrogen bond with the bridging O3R is particularly important as it may serve to stabilize the negative charge on the PR oxygens during catalysis. In the G3*‚dCTP incorporation complex (Figure 1b), however, this interaction is totally lost during the MD simulation, as a result of a conformational change of the triphosphate of dCTP. As shown in Figure 6, the triphosphate of the incoming nucleotide dCTP is rearranged in the G3*‚dCTP incorporation system as compared with the unmodified control. On the other hand, one of the hydrogen bonds between the PhIPmodified guanine (G3*) and the protein residues has a higher hydrogen bond occupancy than in the unmodified control (Table 1). Ile 362 makes a hydrogen bond to the phosphodiester of G3*, which has a higher occupancy in the modified system. The presence of the PhIP moiety
1354
Chem. Res. Toxicol., Vol. 18, No. 9, 2005
Zhang et al.
Figure 7. Nucleotide binding pocket in the G3*‚syn-dATP incorporation system (a,b) and its respective unmodified control system (c,d). Parts a and c show the minor groove side wall (defined by G568, Y567, Y416, and L415) and the length of the pocket (measured by the distance between C1′ of G3 and C1′ of dNTP). Parts b and d show the floor (defined by the flat face of G4‚C32) and roof (defined by L561 and K560) of the pocket and the height of the pocket [measured by the distance between K560 CR and carbon 6 of C32 (primer), as well as the distance between L561 CR and carbon 6 of G4 (template)].
on the major groove side of the modified guanine causes the backbone to move slightly and facilitates the higher occupancy of hydrogen bonding with Ile 362. 2. G3*‚syn-dATP Mismatched Pair Incorporation Complex. In the G3*‚syn-dATP incorporation system (Figure 1d), two of the hydrogen bonds formed between the conserved protein residues and the incoming nucleotide syn-dATP have lesser occupancies than in the unmodified control, as shown in Table 1, while the other hydrogen bonds have similar occupancies as in the unmodified system. Tyr 416 interacts with syn-dATP through a hydrogen bond between the amide N of Tyr 416 and O3′ of the syn-dATP. In the modified system, this hydrogen bond has a very small occupancy as compared to the unmodified control. The hydrogen bond between Nζ of Lys 486 and one of the nonbridging oxygen (O2γ) of the γ-phosphate of syn-dATP is lost in the modified system. Nucleotide Binding Pocket of the Nascent Base Pair. 1. G3*‚dCTP Normal Pair Incorporation Complex. The nucleotide binding pocket checks the geometric fit of the nascent base pair. We measured the length of the pocket by the distance between C1′ of G3 and C1′ of dNTP. In the G3*‚dCTP incorporation system (Figure 1b), this distance has an average value of 10.6 ( 0.2 Å and a value of 10.8 ( 0.2 Å in the unmodified control. The minor groove side wall of the pocket is defined by fingers domain residues Gly 568 and Tyr 567 and palm domain residues Tyr 416 and Leu 415 (Figure 7a,c). The proper positions of these residues are preserved in the modified system. The floor of the pocket is defined by the flat face of the +1 position base pair (A4‚T32), and the ceiling is defined by the protein residues Leu 561 and Lys 560 (Figure 7b,d). Thus, the height of the pocket is
measured by the distance between Lys 560 CR and carbon 6 of T32, as well as the distance between Leu 561 CR and carbon 6 of A4. Respectively, the average values of these two distances are 10.5 ( 0.5 and 12.2 ( 0.4 Å in the G3*‚dCTP incorporation complex system and 10.1 ( 0.5 and 12.2 ( 0.4 Å in the unmodified control. 2. G3*‚syn-dATP Mismatched Pair Incorporation Complex. Despite the mismatch between the templating base G3* and the incoming nucleotide syn-dATP, the shape and the size of the nucleotide binding pocket are retained in the G3*‚syn-dATP incorporation complex (Figure 1d) as well as in the unmodified control (Figure 7). The length of the pocket has an average value of 11.0 ( 0.2 Å in the G3*‚syn-dATP incorporation complex and a value of 10.7 ( 0.2 Å in the unmodified control. As seen from Figure 7a, the proper positions of the minor groove side wall residues are preserved in the modified system. Respectively, the height of the pocket is 10.6 ( 0.4 and 11.8 ( 0.3 Å in the G3*‚syn-dATP incorporation complex system and 10.8 ( 0.4 and 11.9 ( 0.3 Å in the unmodified control (Figure 7b,d). Octahedral Coordination of the Mg2+ Ions. 1. G3*‚ dCTP Normal Pair Incorporation Complex. In the G3*‚dCTP incorporation system (Figure 1b), the octahedral coordination of the Mg2+ ions is less ideal than the unmodified control. There are two metal ions in the active site of the RB69 DNA polymerase. Mg2+ A is essential for catalyzing nucleotide incorporation by activating the 3′-hydroxyl group of the primer for attack on the R-phosphate of the dCTP. One of the octahedral interactions of the Mg2+ A, to Oγ of Ser 624, is lost in the G3*‚dCTP incorporation complex during the MD simulation (Figure 6). The distance between Oγ of Ser 624 and Mg2+ A in the G3*‚dCTP incorporation system is 5.0 ( 0.3 Å but
PhIP-DNA Adduct/Polymerase Molecular Dynamics
2.3 ( 0.1 Å in the unmodified control. The other metal ion, Mg2+ B, is responsible for stabilizing the negative charge that builds up on the leaving oxygen and chelating the β- and γ-phosphates of the dCTP. In the G3*‚dCTP incorporation system, the interaction between Oγ of Asp 623 and Mg2+ B is lost during the MD simulation. The distance between Oγ of Asp 623 and Mg2+ B in the modified system is 4.2 ( 0.3 Å but 1.9 ( 0.1 Å in the unmodified system. In addition, in the modified system, the interaction between the β-phosphate of the dCTP and the Mg2+ B no longer exists, due to the conformational change of the triphosphate of the dCTP (Figure 6). The distance between one of the nonbridging oxygens (O2β) of the β-phosphate of the dCTP and Mg2+ B in the G3*‚ dCTP incorporation system has an average value of 5.6 ( 0.1 Å but has a value of 1.9 ( 0.1 Å in the unmodified control. 2. G3*‚syn-dATP Mismatched Pair Incorporation Complex. The octahedral coordination of the two Mg2+ ions is maintained as well in the G3*‚syn-dATP incorporation system (Figure 1d) as in the unmodified control. Except in the unmodified system, the octahedral interaction between Mg2+ A and Oγ of Ser 624 is replaced by interaction with a water molecule. Distance between the 3′-Primer Oxygen and Pr of the dNTP. 1. G3*‚dCTP Normal Pair Incorporation Complex. The close distance between the 3′terminal oxygen atom of the primer and the R-phosphate (PR) of the incoming dCTP, needed for nucleotidyl transfer, is achieved less in the G3*‚dCTP incorporation complex (Figure 1b) than in the respective unmodified system. In this unmodified system, a distance near the reaction-ready range of ∼3.1-3.5 Å is achieved with occupancy of 49.1% (average value 3.6 ( 0.4 Å), but in the G3*‚dCTP incorporation system, this occupancy is only 0.4% (average value 3.9 ( 0.2 Å) (Figure S10a,b). 2. G3*‚syn-dATP Mismatched Pair Incorporation Complex. This distance in the range of ∼3.1-3.5 Å is also achieved less in the G3*‚syn-dATP incorporation complex (Figure 1d) than in the respective unmodified system. The occupancy of this distance is 23.5 (average value 3.6 ( 0.2 Å) and 51.4% (average value 3.4 ( 0.1 Å) (Figure S11a,b) in the modified and unmodified cases, respectively. Achievement of this distance suggests the structural possibility of successful incorporation but with reduced efficiency in the modified systems. Initial Models of G4*‚C32 and G4*‚syn-A32 Extension Complexes. We built initial models to study the feasibility of further extension of the dG-C8-PhIP adduct once incorporation of the normal or mismatched partner has taken place. The adduct is now in the +1 position, facing a normally paired cytosine (Figure 1c) or a mismatched syn-adenine (Figure 1e). With the PhIP on the major groove side, protein residues are out of range for making contacts with it, equally permitting PhIP orientations in family 1 or 2 (Figure S12). For the G4*‚ C32 extension complex (Figure 1c), we selected an initial model that employs average values obtained from the previous incorporation step (χ ) 200°, R′ ) 143°, and β′ ) 306°, Family 2). γ′ has a starting value of 26°, as in the NMR solution structure (25). For the G4*‚syn-A32 extension complex (Figure 1e), average values from the previous incorporation step had to be modified slightly to obtain a structure without collisions, selected from family 1 (χ ) 213°, R′ ) 118°, β′ ) 85°, and γ′ ) 26°).
Chem. Res. Toxicol., Vol. 18, No. 9, 2005 1355 Table 2. Hydrogen Bond Occupancies Involving the Nascent Base Pair and +1 Base Pair in the G4*‚C32 Extension and G4*‚syn-A32 Extension Systemsa identified in undamaged damaged RB69 crystal system (%) system (%) structure (53)
donor
acceptor
G3 H21 G3 H1 dCTP HN42 Ile362 H Asn572 Hγ22 Tyr416 H Arg482 Hη22 Arg482 Hη12 Lys486 Hζ1,2,3 Lys560 Hζ1,2,3 G4* H21 G4* H1 C32 HN42 Tyr391 Hη Asn572 Hγ22 Tyr708 Hη
dCTP O2 dCTP N3 G3 O6 G3 O2P G3 O3′ dCTP O3′ dCTP O2γ dCTP O2γ dCTP O3γ dCTP O3R C32 O2 C32 N3 G4* O6 G4* O3′ G4* O1P C32 O1P
T3 H3 dATP H61 Ile362 H Asn572 Hγ22 Tyr416 H Arg482 Hη22 Arg482 Hη12 Lys560 Hζ1,2,3 Lys560 Hζ1,2,3 G4* H21 G4* H1 syn-A32 H62 Tyr391 Hη Asn572 Hγ22 Tyr708 Hη
G4*‚syn-A32 dATP N1 96.7 T3 O4 93.1 T3 O2P 91.8 T3 O3′ 39.3 dATP O3′ 9.0 dATP O2γ 99.9 dATP O2γ 83.9 dATP O3γ 69.9 dATP O3R 56.4 syn-A32 N7 N/Ab syn-A32 N7 N/A G4* O6 N/A G4* O3′ 9.2 G4* O1P 87.7 N32c O1P 97.9
G4*‚C32 99.9 97.7 84.4 93.3 65.5 85.0 99.8 48.3 36.6 50.2 93.9 98.1 94.5 65.8 77.9 100
99.3 98.6 91.9 86.9 72.1 75.7 95.3 45.6 40.5 24.7 97.5 98.9 95.1 24.9 62.5 99.9
yes yes yes yes yes yes yes no no yes yes yes yes no yes yes
92.8 92.1 86.7 51.1 4.1 96.3 5.3 22.5 57.2 66.1 88.6 95.4 0.1 88.5 99.7
yes yes yes yes yes yes no no yes yes yes yes no yes yes
a % occupied indicates the percentage of the time (time range, 1-2.5 ns) that the hydrogen bond was intact. Hydrogen bonding criteria were 3.3 Å between heavy atoms and a hydrogen bonding angle of 135°. b Not applicable, since the unmodified system does not have a syn-A32. c N32: C32 in the unmodified system and synA32 in the modified system.
Conformation of dG-C8-PhIP in G4*‚C32 and G4*‚ syn-A32 Extension Complexes. The conformation of the dG-C8-PhIP adduct undergoes only slight changes during the MD simulation, and the PhIP moiety remains on the major groove side of the DNA duplex with no contact with the protein residues. During the MD simulations of the G4*‚C32 (Figure 1c) and G4*‚syn-A32 (Figure 1e) extension complexes, the glycosidic torsion angle χ, governing the orientation of the base, has values of 204.8 ( 7.2 and 205.7 ( 12.6°, respectively, stabilizing the guanine G4* in the anti conformation; hydrogen bonding with C32 at the 3′-end of the primer in the G4*‚ C32 extension complex (Figure 1c) or with syn-A32 in the G4*‚syn-A32 extension complex (Figure 1e) is maintained. The torsion angles involving the PhIP, namely, R′, β′, and γ′, all remain in stable regions throughout the MD simulation, as shown in Figure S6. Hydrogen bonds formed between the N2 of the PhIP and O5′ or O4′ of the deoxyribose in the dG-C8-PhIP adduct contribute to the stability in both complexes. Hydrogen Bonds in the Nascent Base Pair. 1. G4*‚ C32 Normal Pair Extension Complex. The WatsonCrick hydrogen bonds within the nascent base pair, G3‚ dCTP (Figure 1c), are as well maintained in the G4*‚ C32 extension system as in the unmodified control. From Table 2, we can see that these hydrogen bonds have similar occupancies in the modified and the unmodified systems.
1356
Chem. Res. Toxicol., Vol. 18, No. 9, 2005
2. G4*‚syn-A32 Mismatched Pair Incorporation Complex. In the G4*‚syn-A32 extension system (Figure 1e), the Watson-Crick hydrogen bonds within the normal nascent base pair, T3‚dATP are as well-preserved as in the unmodified control. Hydrogen Bonds Involving Key Amino Acids and the Nascent Base Pair. 1. G4*‚C32 Normal Pair Extension Complex. In the G4*‚C32 extension system (Figure 1c), most of the hydrogen bonds between key amino acid residues and the nascent base pair of dCTP and G3 are preserved and have less than 10% difference in hydrogen bond occupancies as compared to the unmodified control, as shown in Table 2. Specifically, hydrogen bonds formed between Arg 482, Lys 486 and Lys 560 with the triphosphate of dCTP generally have similar occupancies (differences are less than 5%) in the modified system as in the unmodified (Table 2). However, the hydrogen bond formed between the Lys 560 and the bridging oxygen (O3R) between the R- and the β-phosphate of dCTP has a 25% lower hydrogen bond occupancy in the modified system than the unmodified one. The two amino acid residues Ile 362 and Asn 572, which interact with the phosphodiester oxygen O2P and O3′ of G3 through H and Ηγ22, respectively, also have similar hydrogen bond occupancies in both the modified and the unmodified systems. 2. G4*‚syn-A32 Mismatched Pair Incorporation Complex. In the G4*‚syn-A32 extension system (Figure 1e), two of the hydrogen bonds formed between key protein residues and the triphosphate of the incoming dATP have lower occupancies than in the unmodified control. Specifically, the hydrogen bonds between Arg 482 and Lys 560 and the nonbridging oxygen (O2γ, O3γ ) of the γ-phosphate of dATP have less than 45% occupancies, while the other hydrogen bonds have similar occupancies in both the modified system and the unmodified control. Nucleotide Binding Pocket of the Nascent Base Pair. 1. G4*‚C32 Normal Pair Extension Complex. The geometry of the nucleotide binding pocket is as wellpreserved in the G4*‚C32 extension complex (Figure 1c) as in the respective unmodified control system. The length of the pocket is 10.5 ( 0.5 Å in the modified system and 10.7 ( 0.4 Å in the unmodified control. The height (Figure 7) is 11.5 ( 0.3 and 10.8 ( 0.1 Å in the modified system and 11.9 ( 0.4 and 10.7 ( 0.2 Å in the unmodified control. 2. G4*‚syn-A32 Mismatched Pair Extension Complex. The geometry of the nucleotide binding pocket is as well-preserved in the G4*‚syn-A32 extension complex (Figure 1e) as in the respective unmodified control system. The length of the pocket is 10.4 ( 0.4 Å in the modified system and 10.3 ( 0.3 Å in the unmodified control. The height is 11.7 ( 0.3 and 10.9 ( 0.2 Å in the modified system and 12.0 ( 0.3 and 10.8 ( 0.2 Å in the unmodified control. Hydrogen Bonds in the +1 Base Pair. 1. G4*‚C32 Normal Pair Extension Complex. The Watson-Crick hydrogen bonds within the +1 base pair G4*‚C32 (Figure 1c) are as well-maintained in the G4*‚C32 extension system as in the unmodified control. From Table 2, we can see that these hydrogen bonds have similar occupancies in the modified and the unmodified systems. 2. G4*‚syn-A32 Mismatched Pair Extension Complex. In the mismatched +1 position base pair, G4*‚synA32 (Figure 1e), we observed the same one full and one bifurcated hydrogen bond throughout the MD simulation
Zhang et al.
Figure 8. Minor groove contacts in the G4*‚C32 extension complex involving the +1 and +2 position base pairs (Figure 1c). Hydrogen bonds are shown in solid cyan lines.
as in the G3*‚syn-dATP incorporation results described previously (Figure 5). Hydrogen Bonds Involving Amino Acids and the +1 Base Pair. 1. G4*‚C32 Normal Pair Extension Complex. In the G4*‚C32 extension system (Figure 1c), two of the three hydrogen bonds involving the +1 base pair are weakened. The hydrogen bond between Tyr 708, from the conserved KKRY sequence, and the 3′-end primer nucleotide C32 is well-preserved in the G4*‚C32 extension system and has a very similar hydrogen bond occupancy to the unmodified system (Table 2). However, the hydrogen bonds between the backbone of the PhIPmodified template base G4* and Asn 572 and Tyr 391 are weakened, as compared to the unmodified guanine in the control, as reflected in lower hydrogen bond occupancies (Table 2). 2. G4*‚syn-A32 Mismatched Pair Extension Complex. In the G4*‚syn-A32 extension system (Figure 1e), only one of the above three hydrogen bonds has lesser occupancy than in the unmodified control. Specifically, the hydrogen bond between Tyr 391 and O3′ of G4* has a very small occupancy, while the hydrogen bonds between Tyr 708, Asn 574 and G4* have similar occupancies as in the unmodified control. Minor Groove Contacts Involving +1 Base Pair. 1. G4*‚C32 Normal Pair Extension Complex. In the G4*‚C32 extension system (Figure 1c), Lys 706, from the conserved sequence KKRY, forms a hydrogen bond with O2 on the base of the primer nucleotide T31 at the +2 position on one side. On the other side, Lys 706 hydrogen bonds with Asp 621, as in the crystal structure (Figure 8). These interactions are preserved in both the modified and the unmodified system. Tyr 567, one of the highly conserved fingers residues, hydrogen bonds via a water molecule with N3 on the base of G4* at the +1 position and with O4′ on the deoxyribose and N3 on the base of A5 at the +2 position (Figure 8) (53, 86). The interactions between Tyr 567 and G4* at the +1 position are somewhat stronger in the modified system than the unmodified system. The two acceptors on the A5, namely, O4′ and N3, take turns interacting with Tyr 567 through the same water molecule. The overall water-mediated hydrogen bond interactions between Tyr 567 and A5 are similar in the modified system and unmodified system. Hydrogen bond occupancies are given in Table 3.
PhIP-DNA Adduct/Polymerase Molecular Dynamics
Chem. Res. Toxicol., Vol. 18, No. 9, 2005 1357
Table 3. Hydrogen Bond Occupancies of the Minor Groove Contacts in the G4*‚C32 Extension and G4*‚syn-A32 Extension Systemsa
donor
acceptor
identified in undamaged damaged RB69 crystal system (%) system (%) structure (53)
Lys706 Hζ1,2,3 Lys706 Hζ1,2,3 water H1, H2 water H1, H2 water H1, H2 water H1, H2
G4*‚C32 Asp621 Oγ1,2 100 T31 O2 85.6 Tyr567 Oη 63.1 G4* N3 1.2 A5 O4′ 18.5 A5 N3 28.7
100 78.2 70.0 15.1 38.3 1.9
yes yes yes yes yes no
Lys706 Hζ1,2,3 Lys706 Hζ1,2,3 water H1, H2 water H1, H2 water H1, H2 water H1, H2
G4*‚syn-A32 Asp621 Oγ1,2 100 C31 O2 98.1 Tyr567 Oη 52.8 G4* N3 3.5 G5 O4′ 50.0 G5 N3 0.2
100 93.3 20.5 0.3 59.6 0
yes yes yes yes yes no
a % occupied indicates the percentage of the time (time range, 1-2.5 ns) that the hydrogen bond was intact. Hydrogen bonding criteria were 3.3 Å between heavy atoms and a hydrogen bonding angle of 135°.
2. G4*‚syn-A32 Mismatched Pair Extension Complex. In the G4*‚syn-A32 extension system (Figure 1e), the hydrogen bonds connecting Lys 706, Asp 621 and C31 at the +2 position are preserved and have similar occupancies as in the unmodified control (Table 3). The hydrogen bond between Tyr 567 and the water molecule, which mediates hydrogen bonding to the two template bases, has occupancy of 21% in the modified system, as compared to 53% in the unmodified control, as shown in Table 3. As a result, the interactions between Tyr 567 and G4* at the +1 position and G5 at the +2 position are weaker in the modified system than the unmodified control. Octahedral Coordination of the Mg2+ Ions. 1. G4*‚ C32 Normal Pair Extension Complex. The octahedral coordination of the two Mg2+ ions located in the active site of the RB69 DNA polymerase, as shown in Figure 9, is maintained as well in the G4*‚C32 extension system (Figure 1c) as in the unmodified control. Unlike in the G3*‚dCTP incorporation system (Figure 1b), the overall conformation of the triphosphate of the incoming nucleotide dCTP in the G4*‚C32 extension system is very similar to the unmodified control. Therefore, the interaction between the β-phosphate of the dCTP and Mg2+ B is not lost, as in the G3*‚dCTP incorporation system. However, in both the G4*‚C32 extension system and the respective unmodified control, the octahedral interaction between Mg2+ A and Oγ of Ser 624 is lost after Ser 624 forms a hydrogen bond with Phe 410. The needed coordination is replaced by interaction with a water molecule. 2. G4*‚syn-A32 Mismatched Pair Extension Complex. The octahedral coordination of the two Mg2+ ions is maintained as well in the G4*‚syn-A32 extension system (Figure 1d) as in the respective unmodified control. Except in the unmodified system, the octahedral interaction between Mg2+ A and Oγ of Ser 624 is replaced by interaction with a water molecule. Distance between the 3′-Primer Oxygen and Pr of the dNTP. 1. G4*‚C32 Normal Pair Extension Complex. The distance of 3.1-3.5 Å between the 3′terminal oxygen atom of primer and PR of the incoming dCTP, needed for approaching nucleotidyl transfer, is sampled during the MD simulation in both the G4*‚C32 extension complex (Figure 1c) (19.2%, average value of
Figure 9. Active site of the G4*‚C32 extension complex. Octahedral coordination of the two Mg2+ ions A and B is shown in solid blue lines. Coordination of Mg2+ A is to the 3′-terminal oxygen, oxygens from two water molecules, Oγ2 of Asp 411, Oγ2 of Asp 623, and O2R of dCTP. Coordination of Mg2+ B is to O2R of dCTP, O2β of dCTP, O1γ of dCTP, Oγ1 of Asp 411, Oγ1 of Asp 623, and O of Leu 412. The color code is the same as in Figure 3, including Mg2+ ions A and B.
3.5 ( 0.2 Å) and the respective unmodified control (40.6%, average value of 3.5 ( 0.1 Å) as shown in Figure S10c,d, although to a reduced extent in the modified system. 2. G4*‚syn-A32 Mismatched Pair Extension Complex. This distance in a range of 3.1-3.5 Å is also achieved less in the G4*‚syn-A32 extension system (Figure 1e) than in the respective unmodified one. While in the unmodified control system this distance is achieved during 70.7% (average value of 3.4 ( 0.2 Å) of the MD simulation, it is only 0.7% (average value of 3.8 ( 0.1 Å) in the G4*‚syn-A32 system (Figure S11c,d), suggesting much less efficient nucleotidyl incorporation for mismatch than normal partner in the +1 position during extension. Base Sequence Context Effect in Incorporation and Extension of the Unmodified Controls. During the simulations of the two unmodified control systems for the dG-C8-PhIP with the correct partner (Figure 1b,c), an interesting difference concerning one of the octahedral interactions of the catalytic Mg2+ A was noticed. These two control systems are identical except for the DNA sequence. The 3′-end primer nucleotide is thymine in the control system of the incorporation step (Figure 1b), and a twist movement of the backbone toward the minor groove, caused by the presence of the bulky methyl group on the base of the thymine, is observed; the interaction between 3′-primer oxygen and Mg2+ A is disrupted by this movement (Figure 10a). On the other hand, the 3′-end of the primer nucleotide is cytosine in the control system of the extension step (Figure 1c), and no movement of the backbone is observed; the interaction between 3′primer oxygen and Mg2+ A is preserved (Figure 10b). This sequence difference in the two systems also causes greater flexibility when the primer terminus contains thymine as compared to the case of cytosine (Figure S10a,c). These results indicate a sequence effect of the base at the 3′-end of the primer on the octahedral
1358
Chem. Res. Toxicol., Vol. 18, No. 9, 2005
Zhang et al.
Figure 10. Stereoviews of the average structures of dCTP and T32 in the unmodified control system of the G3‚dCTP incorporation complex (a) and dCTP and C32 in the unmodified control system of the G4‚C32 extension complex (b). The O3′ H groups at the primer termini are circled.
coordination of the catalytic Mg2+ A and primer terminus flexibility, which may produce a sequence effect on nucleotide insertion efficiency. Kinetic studies have revealed such a base sequence effect in unmodified DNA (87, 88).
Discussion Active Site Disturbances Suggest Impeded Incorporation with dATP Less Disturbed than dCTP. Our results, summarized in Table 4, suggest that the incorporation of a nucleotide opposite a PhIP-damaged guanine in RB69 DNA polymerase is likely to be impeded due to the distortion of the active site caused by the presence of the PhIP residue. A PhIP-modified antiguanine can form Watson-Crick hydrogen bonds with an incoming dCTP, or it can partner with an incoming syn-dATP using alternative hydrogen bonds. In the two incorporation systems, stable structures are formed after ∼500 ps of unrestrained MD simulation. In the normally paired G3*‚dCTP incorporation system (Figure 1b), the PhIP is stabilized through favorable electrostatic and hydrogen bonding interactions with protein residue Lys 279, while in the mismatched G3*‚syn-dATP incorporation system (Figure 1d), the PhIP is stabilized through hydrogen bonding with O4′ on the deoxyribose of G3*. As the PhIP residue is positioned on the major groove side of the DNA helix in both modified systems, the minor groove interactions needed for faithful nucleotide incor-
Table 4. Composite Evaluation of Structural Distortion in G3*‚dCTP and G3*‚syn-dATP Incorporation Complexes as Compared to Undamaged System undamaged G3*‚dCTP undamaged G3*‚syn-dATP system incorporation system incorporation PR-O3′a HB geometryb pocket geometryc key amino acidsDNA interactiond Mg2+ ions octahedral coordinatione
x x x x
×× ×× x ×
x x x x
× × x ××
x
×××
x
x
composite scoref
0
-8
0
-4
a
One cross denotes a PR-O3′ distance occupancy of 3.1-3.5 Å between 1 and 25% of the simulation time; two crosses denote this PR-O3′ distance occupancy
