Chem. Res. Toxicol. 1997, 10, 1123-1132
1123
A Molecular Mechanics and Dynamics Study of the Minor Adduct between DNA and the Carcinogen 2-(Acetylamino)fluorene (dG-N2-AAF) Rosalyn Grad,† Robert Shapiro,‡ Brian E. Hingerty,§ and Suse Broyde*,† Departments of Biology and Chemistry, New York University, New York, New York 10003-5180, and Life Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-8077 Received May 29, 1997X
Experimental studies involving the carcinogenic aromatic amine 2-(acetylamino)fluorene (AAF) have afforded two acetylated DNA adducts, the major one bound to C8 of guanine and a minor adduct bound to N2 of guanine. The minor adduct may be important in carcinogenesis because it persists, while the major adduct is rapidly repaired. Primer extension studies of the minor adduct have indicated that it blocks DNA synthesis, with some bypass and misincorporation of adenine opposite the lesion [Shibutani, S., and Grollman, A. P. (1993) Chem. Res. Toxicol. 6, 819-824]. No experimental structural information is available for this adduct. Extensive minimized potential energy searches involving thousands of trials and molecular dynamics simulations were used to study the conformation of this adduct in three sequences: I, d(C1-G2-C3-[AAF]G4-C5-G6-C7)‚d(G8-C9-G10-C11-G12-C13-G14); II, the sequence of Shibutani and Grollman, d(C1-T2-A3-[AAF]G4-T5-C6-A7)‚d(T8-G9-A10-C11-T12-A13-G14); and III, which is the same as II but with a mismatched adenine in position 11, opposite the lesion. AAF was located in the minor groove in the low-energy structures of all sequences. In the lowest energy form of the C3-[AAF]G4-C5 sequence I, the fluorenyl rings point in the 3′ direction along the modified strand and the acetyl in the 5′ direction. These orientations are reversed in the second lowest energy structure of this sequence, and the energy of this structure is 1.4 kcal/mol higher. Watson-Crick hydrogen bonding is intact in both structures. In the two lowest energy structures of the A3-[AAF]G4-T5 sequence II, the AAF is also located in the minor groove with Watson-Crick hydrogen bonding intact. However, in the lowest energy form, the fluorenyl rings point in the 5′ direction and the acetyl in the 3′ direction. The energy of the structure with opposite orientation is 5.1 kcal/mol higher. In sequence III with adenine mismatched to the modified guanine, the lowest energy form also had the fluorenyl rings oriented 5′ in the minor groove with intact Watson-Crick base pairing. However, the mispaired adenine adopts a syn orientation with Hoogsteen pairing to the modified guanine. These results suggest that the orientation of the AAF in the minor groove may be DNA sequence dependent. Mobile aspects of favored structures derived from molecular dynamics simulations with explicit solvent and salt support the essentially undistorting nature of this lesion, which is in harmony with its persistence in mammalian systems.
Introduction The aromatic amines are a class of environmental chemicals which include numerous mutagens and carcinogens. In a summary of animal testing data which covered 506 chemicals, the aromatic amines were the largest category. Of the 70 aromatic amines tested, 49 were positive for carcinogenicity (1). Aromatic amines are found in automobile exhaust, tobacco smoke, dyes, and barbecued meats and fish. 2-(Acetylamino)fluorene (AAF)1 is an aromatic amine which was originally synthesized as a pesticide, but since it was found to be a potent carcinogen (2) it was never released. Instead it became the prototype for the study of aromatic amines in carcinogenesis. Upon activation AAF reacts with DNA both in vivo and in vitro to form major adducts at guanine C8 and a minor product, * Corresponding author: tel, (212) 998-8231; fax, (212) 995-4015; e-mail,
[email protected]. † Department of Biology, New York University. ‡ Department of Chemistry, New York University. § Oak Ridge National Laboratory. X Abstract published in Advance ACS Abstracts, September 15, 1997. 1 Abbreviations: AAF, 2-(acetylamino)fluorene.
S0893-228x(97)00092-1 CCC: $14.00
3-(deoxyguanosine-N2-yl)(acetylamino)fluorene (dG-N2AAF) in which AAF is bound to N2 of guanine (Figure 1). The extensive literature on AAF adducts has been reviewed in recent years by Beland and Kadlubar (3), Kriek (4), and Heflich and Neft (5). It has long been suspected that the minor dG-N2-AAF adduct may be important in carcinogenesis since it persists in mammals and is repaired only slowly compared to the major adduct (3, 6-10). However, NMR studies of this minor adduct have not been possible to date because synthesis has only been achieved at very low yields (11, 12). When subjected to S1 nuclease, unlike the major adduct, the minor adduct is not digested; this observation led to the suggestion that base pairing is not disrupted in the minor adduct as it is in the major one (8-10). In line with this proposal, early model building efforts by computer (13) and by hand (7, 8) showed that the fluorenyl moiety in the dG-N2-AAF adduct could lie in the minor groove of an essentially unperturbed B-DNA helix. In another early effort, a hand model (14) placed the fluorenyl moiety in the B-DNA major groove, with little distortion, although the modified guanine was in © 1997 American Chemical Society
1124 Chem. Res. Toxicol., Vol. 10, No. 10, 1997
Grad et al. Table 1. Torsion Angles in Starting Conformations at the Modified Guanine and Partnera torsion angle
angle value (deg)
χ guanine χ cytosine (partner) χ adenine (mismatched) R′ β′ γ′ δ′ ′
70 (syn), 250 (anti) 250 (anti) 70 (syn), 250 (anti) 0, 45, 90, 135, 180, 225, 270, 315 0, 45, 90, 135, 180, 225, 270, 315 0, 90, 180, 270 0, 90, 180, 270 60
a
Structures created for all combinations of torsion angles.
containing the G:A mismatch that was most prevalently observed by them. These sequences are designated as C3-[AAF]G4-C5, A3-[AAF]G4-T5, and A3-[AAF]G4-T5 mismatch, respectively, representing the central trimer of each sequence.
Methods
Figure 1. Structure of the dG-N2-AAF adduct. Torsion angle definitions: χ ) O4′-C1′-N9-C4; R′ ) N1-C2-N2-C3; β′ ) C2-N2-C3-C2; γ′ ) C3-C2-N-C; δ′ ) C2-N-C-Cm; ′ ) N-C-Cm-H. C and Cm designate carbonyl and methyl carbons, respectively, in the acetyl group. Bold letters designate AAF atoms. Sequences studied are given in the inset.
the syn conformation. Minimized conformational potential energy calculations for this adduct to d(CpG) with a very limited search (15) combined with hand model building to generate a larger duplex suggested that the fluorenyl moiety could reside at a B-DNA/Z-DNA junction, oriented toward the Z-helix direction, on the 5′ side of the modified strand. More recently, Shibutani et al. (12) have synthesized a decadeoxynucleotide containing a single dG-N2-AAF modification; they have used this template to determine kinetic parameters for deoxynucleotide triphosphate (dNTP) incorporation opposite the modification and the rate of extension from the 3′ terminus of guanines containing the lesion (16). DNA synthesis on the dGN2-AAF-modified template using exo- Klenow fragment was largely blocked one base before and opposite the modification. When extension occurred opposite the modification, dAMP was incorporated. Only a small percentage of starting primer was fully extended, and it overwhelmingly contained dAMP opposite the lesion. Chain extension starting with dA opposite the lesion in the primer was 5 times more efficient than when starting with dC opposite the lesion (16). In the present work we have carried out an extensive series of energy minimization searches to elucidate the structure of the dG-N2-AAF adduct in three sequence contexts: I, d(C1-G2-C3-[AAF]G4-C5-G6-C7)‚d(G8-C9G10-C11-G12-C13-G14); II, d(C1-T2-A3-[AAF]G4-T5-C6A7)‚d(T8-G9-A10-C11-T12-A13-G14); and III, d(C1-T2A3-[AAF]G4-T5-C6-A7)‚d(T8-G9-A10-A11-T12-A13G14). The second is the sequence of Shibutani and Grollman (16), and the third is the same sequence but
Creating Starting Conformations. An AAF moiety was docked to the N2 atom of the central guanine of an energyminimized B-DNA double helix computed with the program DUPLEX (17). Then a torsion driver was used to generate starting structures. This program (written by Dr. Bin Li) keeps the B-DNA structure fixed, rotates the AAF-G torsion angles R′ and β′ and the within AAF torsion angles γ′, δ′, and ′ (Figure 1) to chosen values, and then computes coordinates of the resulting structures. This method creates a large unbiased array of high-energy starting structures for energy minimization. It affords a reasonable opportunity for surveying the potential energy surface, i.e., the conformational possibilities accessible to the adduct, with uniform sampling at 45° intervals for R′ and β′. Both the anti and syn domains of the glycosidic torsion angle were explored [χ ) 250° (anti) and 70° (syn)]; 2048 starting conformers were created for the C3-[AAF]G4-C5 sequence and a like number for the A3-[AAF]G4-T5 sequence. For the mismatch sequence, 4096 starting conformations were constructed in order to model guanine and adenine in both the anti and syn orientations, since both have been observed in crystal structures (18-23). For a summary of the starting torsion angles see Table 1. In addition, a special small search was carried out for the A3-[AAF]G4-T5 sequence in a further effort to locate structures with the AAF intercalated into the helix with displacement of the modified guanine, since these have been observed in other bulky adducts to guanine N2 (24). Starting structures for this search were the observed NMR solution structures of the (+)and (-)-cis-anti-benzo[a]pyrene diol epoxide adducts to N2 of guanine (25, 26) which are base-displaced-intercalated, with carcinogen, sequence and length adjusted. For each of these two conformations, γ′ and δ′ were each started at either 0° or 180°, and ′ at 60°, for a total of eight trials in this set. A second set of eight trials was the same except that the R′ angle was rotated by 180°. Energy Minimization. Energy minimization was carried out with the molecular mechanics program AMBER 4.0 (27). The DNA conformations were minimized using explicit hydrated sodium ions to mimic counterion condensation. These were initially placed 5 Å from the phosphorus atoms along the pendant phosphate oxygen bisector. Hydration of these counterions was modeled by an increase in the Na+ van der Waals radius to 5 Å and a van der Waals energy of 0.1 kcal/mol in the Lennard-Jones potential (28). A sigmoidal distance dependent dielectric function (29) which has been demonstrated to be an appropriate implicit treatment for the dielectric function in computing the electrostatic potential of nucleic acids (30) was used in the Coulombic term of the force field. The energy minimization employed 100 initial cycles of steepest descent minimization followed by conjugate gradient minimization until convergence was reached. A system was deemed converged if the difference in energies for successive
Structure of dG-N2-AAF Adduct
Chem. Res. Toxicol., Vol. 10, No. 10, 1997 1125
Table 2. Parameters Assigned to the AMBER Force Field for the dG-N2-AAF Adduct Bond Parameters bond type
Kr (kcal/mol‚Å)
req (Å)
N-CA CN-CT
481 317
1.480 1.490
Angle Parameters angle type
Kθ (kcal/mol‚rad2)
θ (deg)
N-CA-CA CA-N-H CB-CB-N CB-CN-CT CA-CN-CT CN-CT-HC CN-CT-CN C-N-CA CA-N2-CA CA-CA-N2
70 35 85 70 70 35 63 50 70 70
119.5 120.0 117.4 109.3 132.6 110.0 105.6 121.9 120.0 118.6
Torsional Parameters torsion
Vn/2 (kcal/mol)
γ (deg)
n
CA-CA-N-H X-CN-CT-X C-N-CA-CA X-CA-CA-X X-X-CB-CA
6.8 9.0 6.8 5.6 7.0
180 180 180 180 180
2 2 2 2 2
steps was less than 0.001 kcal/mol or if successive gradient norms were less than 0.001 kcal/mol‚Å. Some parameters had to be added to the AMBER 4.0 force field (31, 32) since the dG-N2-AAF adduct had not previously been parametrized. Atom types assigned to each atom in the adduct are given in Figure 2A. Partial charges for the guanine modified by AAF, consistent with those in the AMBER 4.0 force field, were calculated with the program QUEST (33). Neutrality of the dG-N2-AAF moiety was ensured by normalizing the computed charges so that their sum equaled that of the AMBER set for the unadducted dG. Figure 2B shows the partial charges obtained, which were employed in this work. Bond length, bond angle, and torsional parameters missing from the force field parameter file for the dG-N2-AAF moiety were assigned by comparison to parameters for chemically similar bonds and angles already found in the AMBER force field (31, 32). The chosen parameters added to the force field are consistent with the rest of the AMBER force field parameters and are given on Table 2. Dynamics Simulation. A 270 ps molecular dynamics simulation was carried out, in an environment of salt and solvent using AMBER 4.0, for the lowest energy structures of both the C3-[AAF]G4-C5 sequence and the A3-[AAF]G4-T5 sequence as well as for an unmodified C3-G4-C5 sequence and an unmodified A3-G4-T5 sequence that had been energy minimized in the B-form, as controls. The protocol that was employed is as follows: sodium ions were placed 3.6 Å from the pendant phosphate oxygen bisector so that they were in van der Waals contact with the phosphate residues. The DNA was placed in the center of a rectangular box surrounded by repeating cubes of TIP3P water (34). Water molecules that approached any atom by less than 3.6 Å were removed. The number of water molecules employed and the box size for each simulation are given in Table 3. Each DNA atom was at least 10 Å from any side of the box. A dielectric constant of 1 was used. To equilibrate the waters prior to dynamics, a harmonic restraint of 25 kcal/mol‚residue was placed on the DNA starting structure and potential energy minimization was then carried out for 1000 cycles (20 cycles steepest descent and 980 cycles conjugate gradient) followed by 3 ps of dynamics. The restraints on the DNA were then removed in five steps, reducing the restraint by 5 kcal/mol‚residue in each step and minimizing the entire system for 600 cycles of conjugate gradient minimization. In the fifth step the entire system was minimized without any harmonic restraint.
Figure 2. A. Atom types. B. Partial charges assigned to AMBER 4.0 force field for the dG-N2-AAF adduct. Molecular dynamics simulation then began by assigning a random velocity to each atom in the system so that the velocity distribution corresponded to a Maxwellian distribution at 10 K. The system was heated to 300 K with a temperature coupling constant of 0.2 ps and was allowed to equilibrate for 10 ps. The SHAKE routine, with a tolerance of 0.0005 Å (35), which holds all covalent bond lengths to hydrogens constant, was employed. A 5 kcal/mol‚residue harmonic restraint was
1126 Chem. Res. Toxicol., Vol. 10, No. 10, 1997 Table 3. Number of Water Molecules and Box Sizes in Molecular Dynamics Simulations sequence
no. of waters
box size (Å3)
C3-[AAF]G4-C5 C3-G4-C5 A3-[AAF]G4-T5 A3-G4-T5
2411 2493 2373 2386
44 × 45 × 43 45 × 44 × 44 45 × 45 × 42 43 × 45 × 43
Grad et al. Table 4. Minor Groove Widthsa sequence C3-[AAF]G4-C5 A3-[AAF]G4-T5 A3-[AAF]G4-T5 mismatch
employed for the terminal base pairs of the DNA to prevent fraying. The temperature of the system was then held constant at 300 K with a temperature coupling constant of 0.2 ps. The pressure was allowed to fluctuate around 1 bar with a pressure coupling constant of 0.6 ps. Periodic boundary conditions were used, and a 12 Å uniform cutoff was employed in evaluating Lennard-Jones and electrostatic contributions to the energy. The simulation was carried out with a 1 fs time step, and coordinates for the system were saved every 0.1 ps. Following the simulations, the dynamics trajectories were analyzed using the program Toolchest (36-38), examining structures at 10 ps intervals. Calculations were performed on Cray supercomputers at the Department of Energy’s National Energy Research Supercomputer Center, the National Science Foundation’s San Diego Supercomputer Center, on Silicon Graphics workstations at Sandia National Laboratory in Livermore, CA (courtesy of Dr. Michael Colvin), on Silicon Graphics workstations and a DEC VAX Cluster at the Academic Computing Facility at New York University, and on our own Silicon Graphics workstations.
Results C3-[AAF]G4-C5 Sequence. Residues are numbered 1-7 in the 5′ to 3′ direction of the modified strand and 8-14 in the 5′ to 3′ direction of the unmodified partner, as per Figure 1. As summarized in Table 1, 2048 starting conformers were created and energy-minimized. In these starting structures the DNA itself was an energy minimized B-form. Both the anti and syn domains of the glycosidic torsion angle were explored. In the lowest energy conformation of this sequence (Figure 3A,C), the AAF is situated in the B-DNA minor groove, with all Watson-Crick base pairing intact. The fluorenyl moiety is oriented in the 3′ direction of the modified strand with the acetyl directed 5′. The fluorenyl ring appears well embedded in the groove. On the modified strand side, the distal fluorenyl ring contacts the 5′-CH2 and H-4′ of sugar 6. The proximal ring contacts H-1′ of sugar 5. On the unmodified strand, the distal ring contacts H-4′ and the 5′-CH2 protons of sugar 12. A weak hydrogen bond or favorable electrostatic interaction is present between the amide N-H and N-3 of the modified guanine with a distance of 1.95 Å between H and N and an angle of 128.3° between N and N. Of the 2048 starting conformations, 117 produced conformers similar to the global minimum (torsion angles χ, R′, β′, γ′, δ′, and ′ within 20° of the global minimum), 4 had energies within 5 kcal/mol of the lowest energy conformation, and 64 had energies within 10 kcal/mol of the lowest energy structure. The conformer with the fluorene and acetyl oppositely directed (fluorene 5′ and acetyl 3′) is 1.4 kcal/mol higher in energy (Figure 3B,D). Watson-Crick hydrogen bonds are intact with an additional weak hydrogen bond or favorable electrostatic interaction between the amide N-H to O-2 of the partner cytosine, with a distance of 2.29 Å between H and O and an angle of 127° between N and O. The fluorenyl rings are rather exposed on the modified strand side, with a loose contact of the proximal ring to H-4′ of sugar 5. The distal ring is exposed on that side. The fluorenyl face that is turned toward the
conformer, ∆E (kcal/mol)
width, Å
0.0 1.4 0.0 5.1 0.0
4.67; 3.84; 5.58 5.18; 4.17; 4.88 5.66; 4.26; 4.59 4.94; 4.42; 4.27 5.43; 4.45; 4.49
a Phosphate distances P12 to P4, P11 to P5, and P10 to P6 less 5.8 Å [to account for the phosphate group diameter in the groove (43)], for the phosphate sequence:
unmodified strand contacts one 5′-CH of sugar 13 with its middle ring and H-4′ of sugar 13 with its distal ring. In both structures the acetamide group is turned so that the N-H points into the groove and the carbonyl oxygen points outward. Minor groove widths for the two lowest energy conformations are in unexceptional B-DNA ranges (Table 4) (39, 40). No other types of conformers were found within 10 kcal/mol of the lowest energy form. Table 5 gives a summary of conformations with energies up to 10 kcal/mol above the lowest energy conformation. A3-[AAF]G4-T5 Sequence. In this sequence the orientations are reversed in comparison to the C3-[AAF]G4-C5 sequence. In the lowest energy conformation of the A3-[AAF]G4-T5 sequence, the fluorenyl moiety is directed 5′ along the modified strand and the acetyl is oriented 3′ (Figure 4A,C). Watson-Crick hydrogen bonds are intact. The acetamide N-H is pointed into the groove, and the carbonyl oxygen is pointed outward. The fluorenyl rings on the modified strand side are rather exposed in this structure too, with a loose contact on the proximal ring to H-4′ of sugar 5. On that side, the distal ring is almost completely exposed but makes a loose contact to 5′-CH2 of sugar 5. The fluorene face that is turned toward the unmodified strand contacts one 5′-CH of sugar 13 with the middle ring and H-1′ of sugar 12 with the distal ring. Of the 2048 starting conformations, 101 produced conformers similar to this lowest energy conformation, 20 had energies within 5 kcal/mol of the lowest energy structure, and 57 had energies within 10 kcal/ mol of the lowest energy conformation. The oppositely oriented conformer, with fluorenyl rings oriented 3′ and acetyl oriented 5′, is 5.1 kcal/mol higher in energy than the lowest energy structure (Figure 4B,D). The acetamide group is again pointing with the N-H into the groove and the carbonyl oxygen pointing outward. Watson-Crick hydrogen bonds are intact with an additional weak hydrogen bond or favorable electrostatic interaction between the amide NH and the thymine 12 carbonyl oxygen, with a distance of 2.07 Å between H and O and an angle of 130° between N and O. On the modified strand side of the fluorenyl ring, coverage is fair. The distal ring contacts 5′-CH2 of sugar 6. On the unmodified strand side, the middle ring contacts H-4′ of sugar 12. The edge of the fluorene ring that contains the methylene group projects from the groove. A conformer with a syn-modified guanine and the fluorenyl ring in the major groove was found at 10 kcal/mol above the lowest energy conformer in this sequence. Minor groove widths for the two lowest energy conformations are in unexceptional B-DNA ranges (Table 4) (39, 40). Table 6 gives a summary of conformations with energies up to 10 kcal/mol above the lowest energy form. A3-[AAF]G4-T5 Mismatch Sequence. In this sequence, 4096 starting conformations were generated, to
Structure of dG-N2-AAF Adduct
Chem. Res. Toxicol., Vol. 10, No. 10, 1997 1127
Figure 3. C3-[AAF]G4-C5 sequence. A. Lowest energy conformation (∆E ) 0.0 kcal/mol). B. Second lowest energy conformation (∆E ) 1.4 kcal/mol). Hydrogens have been deleted for clarity. C. Color space-filling view of lowest energy conformation. D. Color space-filling view of second lowest energy conformation. Colors: AAF, magenta; modified strand, cyan; unmodified strand, yellow. Structures are shown in stereo, intended for viewing with a stereoviewer. To view with crossed eyes, left and right images must be interchanged. A viewer that is convenient for image pairs of a wide range of sizes is available from Nu 3D Vu Co., 71 E. 28th Ave., Eugene, OR 97405; tel, (541) 465-4930. Other viewers may require reduction of images.
1128 Chem. Res. Toxicol., Vol. 10, No. 10, 1997
Grad et al.
Figure 4. A3-[AAF]G4-T5 sequence. A. Lowest energy conformation (∆E ) 0.0 kcal/mol). B. Second lowest energy conformation (∆E ) 5.1 kcal/mol). Hydrogens have been deleted for clarity. C. Color space-filling view of lowest energy conformation. D. Color space-filling view of second lowest energy conformation. Colors: AAF, magenta; modified strand, cyan; unmodified strand, yellow. Structures are shown in stereoview.
Structure of dG-N2-AAF Adduct
Chem. Res. Toxicol., Vol. 10, No. 10, 1997 1129
Table 5. Low-Energy Conformations in the C3-[AAF]G4-C5 Sequencea
a
family
χ
R′
β′
γ′
δ′
′
∆E
description
1 2
251 251
176 187
40 203
183 184
184 181
93 72
0.0 1.4
G-anti 3′ orientation minor groove G-anti 5′ orientation minor groove
Torsion angles are in degrees; energies are in kcal/mol. Table 6. Low-Energy Conformations in the A3-[AAF]G4-T5 Sequencea
a
family
χ
R′
β′
γ′
δ′
′
∆E
description
1 2 3
240 249 59
185 169 6
202 46 212
190 197 150
182 178 346
60 50 94
0.0 5.1 10.0
G-anti 5′ orientation minor groove G-anti 3′ orientation minor groove G-syn 5′ orientation major groove
Torsion angles are in degrees; energies are in kcal/mol.
Figure 5. A. Lowest energy conformation in the A3-[AAF]G4-T5 mismatch sequence. ∆E ) 0.0 kcal/mol. Adenine syn and guanine anti. Hydrogens have been deleted for clarity. Structures are shown in stereoview. B. Hoogsteen hydrogen bonds. Table 7. Low-Energy Conformations in the A3-[AAF]G4-T5 Mismatch Sequencea family
χ, A11
χ, G4
R′
β′
γ′
δ′
′
∆E
description
1 2 3 4 5 6 7 8
65 255 63 256 239 70 254 56
239 255 241 255 58 75 264 61
185 184 180 174 333 336 196 19
197 210 37 54 154 153 245 202
201 197 199 159 192 190 327 168
179 178 175 175 182 183 178 179
48 33 53 77 67 72 44 63
0.0 3.9 4.0 6.3 6.7 6.8 7.0 10.0
G-anti A-syn 5′ orientation minor groove G-anti A-anti 5′ orientation minor groove G-anti A-syn 3′ orientation minor groove G-anti A-anti 3′ orientation minor groove G-syn A-anti 3′ orientation major groove G-syn A-syn 3′ orientation major groove G-anti A-anti 5′ orientation minor groove G-syn A-syn 5′ orientation major groove
a
Torsion angles are in degrees; energies are in kcal/mol. A11 is the adenine mismatched to G4.
model guanine both anti and syn and adenine both anti and syn. For a summary of the starting torsion angles see Table 1. The lowest energy conformation found in this series is again a minor groove structure. Overall the structure in this sequence is very similar to the normally paired sequence except that the mispaired adenine is syn (Figure 5A). The fluorene is oriented 5′ with respect to the modified strand, as it is when the normal cytosine partner is present in this sequence. Again the acetamide N-H points into the groove and the carbonyl oxygen points outward. The fluorenyl rings are exposed on the modified strand side, with a loose contact of the proximal ring with H-4′ of sugar 5. The distal ring is almost completely exposed on that side. The proximal ring also contacts the 5′-CH2 of sugar 13, and it is close to H-4′ of sugar 12. Watson-Crick hydrogen bonding is intact for the length of the DNA except at the modification site. At the modification site Hoogsteen hydrogen bonds (Figure 5B) are found, as was proposed by Shibutani and Grollman (16). Minor groove widths are normal (Table 4). Table 7 gives a summary of conformations with energies up to 10 kcal/mol above the lowest energy form. Sequence Dependence. In an effort to understand the origin of the sequence dependence suggested by our
computations, a number of further experiments were undertaken. The unmodified sequences were minimized using AMBER. The energy of the unmodified oligomer for each sequence was subtracted from that of the modified 3′-fluorenyl-oriented conformer in the same sequence. The same was done for the 5′-fluorenyloriented conformers. The absolute values of the energy differences produced by this procedure (termed ∆AAF) have no significance, as they represent a comparison of energy of structures having different numbers of atoms. However, each represents the loss of the same unit, an AAF residue, so they may be compared to one another. The values for the C3-[AAF]G4-C5 heptamer were 3′fluorenyl-oriented conformer, 6.1 kcal/mol; 5′-fluorenyloriented conformer, 7.6 kcal/mol. For the A3-[AAF]G4T5 sequence they were 5′-fluorenyl-oriented conformer, 6.8 kcal/mol; 3′-fluorenyl-oriented conformer, 11.9 kcal/ mol. These results suggested that the last structure was destabilized relative to the others. To obtain further insight, the 3′-fluorenyl-oriented structures in the two sequences were superimposed by computer graphics, overlaying the carbons of the aminofluorene rings so that they merged on the screen. The minor groove environments of the aminofluorene ring in the two sequences differed. In the C3-[AAF]G4-C5
1130 Chem. Res. Toxicol., Vol. 10, No. 10, 1997
Grad et al.
Table 8. Low-Energy Conformations in the T3-[AAF]G4-T5 Sequencea
a
family
χ
R′
β′
γ′
δ′
′
∆E
description
1 2
248 251
185 179
202 38
190 175
184 179
69 85
0.0 1.6
G-anti 5′ orientation minor groove G-anti 3′ orientation minor groove
Torsion angles are in degrees; energies are in kcal/mol.
sequence, the aminofluorene ring appeared to be more deeply embedded in the minor groove. In the A3-[AAF]G4-T5 sequence, the distal ring of AF was more exposed on the side facing the unmodified strand. In seeking a rationale for this difference, we noted that in the latter sequence, the acetyl methyl group of AAF was positioned quite close to the N-3 of the 5′-neighboring adenine on the modified strand. This hindrance presumably prevented the aromatic amine residue from attaining the more deeply embedded conformation observed for the 3′fluorenyl-oriented conformer in the C3-[AAF]G4-C5 sequence. In the latter conformer, the approach of the acetyl methyl to O-2 of the 5′-neighbor cytosine was not as close. By contrast, superimposition of the 5′-fluorenyloriented conformers in these sequences revealed that the environments of the aminofluorene ring were quite similar in the two structures. In accord with these observations, we reasoned that if the A3-[AAF]G4-T5 sequence were modified to a T3[AAF]G4-T5 sequence, keeping the base composition the same but placing a pyrimidine on the 5′ side of the modified G, the orientation preference might be more like that of the C3-[AAF]G4-C5 sequence. To test this idea, we modified the starting structures that had produced the two lowest energy conformers in the A3-[AAF]G4T5 sequence to contain T3-[AAF]G4-T5 as the central trimer and minimized the two structures using the protocols described in the Methods section. An additional series of trials was also conducted to optimize the methyl torsion angle ′ in the T3-[AAF]G4-T5 sequence. While the structure of lowest energy still had the fluorenyl rings oriented 5′ with respect to the modified strand, the energy difference between this conformer and the structure with fluorene oriented 3′ was now only 1.6 kcal/mol, instead of 5.1 kcal/mol. The pyrimidine/purine choice 5′ to the modification thus accounted for 3.5 kcal/mol of the energy difference. The molecular conformations and energies for these two structures are found in Table 8. The origin of the remaining 1.6 kcal/mol that disfavors the 3′-oriented fluorene in the T3-[AAF]G4-T5 sequence compared to the C3-[AAF]G4-C5 sequence is uncertain; perhaps the inherently lesser stability of the T-A pair compared to the G-C pair plays a role here and in the A3-[AAF]G4-T5 case. Dynamics Simulations. The 270 ps dynamics for the lowest energy conformations of both the C3-[AAF]G4-C5 and the A3-[AAF]G4-T5 sequences, compared to those of their unmodified analogs, reveal that the AAF moiety in the minor groove does not provoke substantial distortions to the B-form of DNA. Figures 6 and 7 show snapshots along the dynamics trajectory of the modified sequences. These reveal some mobility of the AAF within the groove, manifested in the 30-50° ranges traversed by the torsion angles χ′, R′, and β′ (Table 9) which govern the AAF orientation at the linkage site. However, base pairing remains intact at all residues throughout the simulations. The torsion angle R′ remains in the vicinity of 180 ( 45°, the value required to maintain a Watson-Crick hydrogen bond at the guanine N2 linkage site (24). In addition the DNA does not bend, kink, or unwind to a significant
Figure 6. Stereoviews of structures from the C3-[AAF]G4-C5 dynamics simulation at (A) 60, (B) 180, and (C) 240 ps, respectively.
degree during the simulation. Average twist angles (Table 10) at and surrounding the lesion site are within ranges observed for B-DNA crystals at the same base pair sequence step (41).
Discussion Our results suggest that the orientation in the minor groove of the dG-N2-AAF adduct can be sequence dependent. In the C3-[AAF]G4-C5 sequence the fluorenyl ring is oriented 3′ with respect to the modified strand, and in the A3-[AAF]G4-T5 sequence it is oriented 5′ with respect to the modified strand, in their respective lowest energy conformations. Opposite orientations of stereoisomeric adducts of activated polycyclic aromatic hydrocarbons have previously been computed (42) and observed by high-resolution NMR in solution (reviewed in ref 24). Studies of (+)- and (-)-trans-anti-benzo[a]pyrene diol epoxide adducts to guanine N2 (24) revealed that opposite orientations in the minor groove are governed by the torsion angle β′ in a normally base-paired duplex in which R′ must be near 180°; β′ differs by about 180° in the 5′ and 3′ orientations, and this is true also in the present work (see Tables 4, 6, and 8). While the PAH metabolites show different orientations for different stereoisomers in the same DNA sequence, our results suggest that the fluorenyl residue in the dGN2-AAF adduct may be prone to different orientation preferences in different sequence contexts. Opposite orientations of the fluorene in different sequences of full duplexes might suggest sequence dependent differential
Structure of dG-N2-AAF Adduct
Chem. Res. Toxicol., Vol. 10, No. 10, 1997 1131
favored major groove structure had an energy of 16.8 kcal/mol. In the A3-[AAF]G4-T5 sequence, of the 2048 structures minimized, 144 led to structures with energies within 20 kcal/mol. Of these, 125 produced minor groove structures and only 19 produced major groove structures. The most favored major groove structure in this sequence had an energy of 10.0 kcal/mol. Consequently, we feel fairly confident of the minor groove position of this adduct. The opposite orientations in the lowest energy conformers of the different sequences were an intriguing finding which suggests that the sequence of the DNA may affect the orientation of the adducted carcinogen. However, since the energy differences in a given sequence between the two orientations are small, we cannot go beyond offering this concept as an interesting possibility. According to the present calculations and in agreement with early model building (7, 8, 13), the minor dG-N2AAF adduct does not cause serious distortions in the DNA structure, such as denaturation, bending or kinking, or unwinding of the helix, but lies quite snugly in the minor groove. This could indeed explain why it is not recognized by repair systems and therefore remains persistently bound to the DNA, as was surmised from the early model building.
Conclusion Figure 7. Stereoviews of structures from the A3-[AAF]G4-T5 dynamics simulation at (A) 60, (B) 180, and (C) 240 ps, respectively. Table 9. Dynamic Ranges of Torsion Angles at Modification Site in Lowest Energy Conformations sequence
torsion angle
range/average (deg)
C3-[AAF]G4-C5
χ R′ β′ χ R′ β′
230-260/245 140-190/166 30-70/47 220-270/249 175-205/192 185-215/201
A3-[AAF]G4-T5
Table 10. Average Twist Angles from Dynamics Trajectories base pair
twist (deg)
C3-G12 [AAF]G4-C11 C5-G10
35.1 36.7 34.3
A3-T12 [AAF]G4-C11 T5-A10
34.4 32.2 33.4
repair. Moreover, opposite orientations, if they existed at the replication fork, might lead to different treatment by replicative enzymes. Perhaps the primer extension blockage caused by this lesion (16) may be related to our observed 3′ directed orientation of the acetyl group found in the lowest energy structure of this sequence, if a similar orientation were to occur at a single-stranddouble-strand junction. Our extensive searches located only minor groove structures up to 20 kcal/mol in all the sequences studied. Even the special search for base-displaced-intercalated structures (see Methods) located such structures only at energies of 30 kcal/mol or higher than the lowest energy form. Moreover, of the 2048 structures minimized in the C3-[AAF]G4-C5 sequence, 209 produced structures with energies within 20 kcal/mol of the lowest energy conformation. Of these, 197 produced minor groove structures and only 12 led to major groove structures. The most
In this investigation extensive potential energy minimization searches for dG-N2-AAF, the minor adduct of activated AAF in DNA duplexes, place the fluorenyl moiety in an essentially undistorted B-DNA minor groove. However, an interesting effect of base sequence is suggested. In the C3-[AAF]G4-C5 sequence the fluorene is oriented in the 3′ direction of the modified strand and the acetyl is 5′ directed in the lowest energy conformation, while the opposite orientation is favored for the A3-[AAF]G4-T5 sequence. Moreover, molecular dynamics simulations with explicit solvent as well as counterions reveal some mobility of the AAF in the minor groove but no large kinetic disturbances such as flexible kinks, bends, unwinding, or denaturation. Perhaps the persistence of this adduct may indeed be related to its unperturbing orientation in the B-DNA minor groove, as was suggested long ago (7, 8, 13). Coordinates. Coordinates of all structures are available by e-mail from Suse Broyde (
[email protected]).
Acknowledgment. This research is supported by NIH Grants CA 75449, CA 28038, and RR-06458 and DOE Grant DE-FG02-90ER60931 to S. Broyde and R. Shapiro and DOE Contract DE-AC05-96OR22464 with Lockheed Martin Energy Research to B. E. Hingerty. We thank Dr. Michael Colvin, Sandia National Laboratory, Livermore, CA, for computing resources on Silicon Graphics workstations at Sandia.
References (1) Nesnow, S., Argus, M., Bergman, H., Chu, K., Frith, C., Helmes, T., McGaughy, R., Ray, V., Slaga, T. J., Tennant, R., and Weisburger, E. (1988) Chemical carcinogens. A review and analysis of the literature of selected chemicals and the establishment of the Gene-Tox Carcinogen Data Base. The U.S. Environmental Protection Agency Gene-Tox Program. Mutat. Res. 185, 1-195. (2) Wilson, R. H., DeEds, F., and Cox, A. J. (1941) The toxicity and carcinogenic activity of 2-acetaminofluorene. Cancer Res. 1, 595608. (3) Beland, F. A., and Kadlubar. (1990) Metabolic activation and DNA adducts of aromatic amines and nitroaromatic hydrocarbons. In Handbook of Experimental Pharmacology (Cooper, C. S., and
1132 Chem. Res. Toxicol., Vol. 10, No. 10, 1997
(4) (5) (6) (7) (8)
(9)
(10)
(11)
(12)
(13)
(14) (15) (16) (17) (18)
(19) (20)
(21) (22)
(23)
(24)
Grover, P. L., Eds.) Vol. 94/I: Chemical Carcinogenesis and Mutagenesis, pp 267-325, Springer Verlag, Heidelberg. Kriek, E. (1992) Fifty years of research on N-acetyl-2-aminofluorene, one of the most versatile compounds in experimental cancer research. J. Cancer Res. Clin. Oncol. 118, 481-489. Heflich, R. H., and Neft, R. E. (1994) Genetic toxicity of 2-acetylaminofluorene, 2-aminofluorene and some of their metabolites and model metabolites. Mutat. Res. 318, 73-174. Kriek, E. (1972) Persistent binding of a new reaction product of the carcinogen N-hydroxy-N-2-acetylaminofluorene with guanine in rat liver DNA in vivo. Cancer Res. 32, 2042-2048. Kriek, E. (1974) Carcinogenesis by aromatic amines. Biochim. Biophys. Acta 355, 177-203. Grunberger, D., and Weinstein, I. B. (1979) Conformational changes in nucleic acids modified by chemical carcinogens. In Chemical Carcinogens and DNA (Grover, P. L., Ed.) pp 59-93, CRC Press, Boca Raton, FL. Howard, P. C., Casciano, D. A., Beland, F. A., and Shaddock, J. G. (1981) The binding of N-hydroxy-2-acetylaminofluorene to DNA and repair of the adducts in primary rat hepatocyte cultures. Carcinogenesis 2, 97. Beland, F. A., Dooley, K. L., and Jackson, C. D. (1982) Persistence of DNA adducts in rat liver and kidney after multiple doses of the carcinogen N-hydroxy-2-acetylaminofluorene. Cancer Res. 42, 1348-1354. Westra, J. G., Kriek, E., and Hittenhausen, H. (1976) Identification of the persistently bound form of the carcinogen N-acetyl-2aminofluorene to the rat liver DNA in vivo. Chem.-Biol. Interact. 15, 149-164. Shibutani, S., Gentles, R., Johnson, F., and Grollman, A. P. (1991) Isolation and characterization of oligonucleotides containing dGN2-AAF and oxidation products of dG-C8-AF. Carcinogenesis 12, 813-818. Beland, F. A. (1978) Computer generated graphic models of the N2-substituted deoxyguanosine adducts of 2-acetylaminofluorene and benzo[a]pyrene and the O6-substituted deoxyguanosine adduct of 1-naphthylamine in the DNA double helix. Chem.-Biol. Interact. 22, 329-339. Kadlubar, F. F. (1980) A transversion mutation hypothesis for chemical carcinogenesis by N2-substitution of guanine in DNA. Chem.-Biol. Interact. 31, 253-263. Hingerty, B. E., and Broyde, S. (1983) AAF linked to the guanine amino group: a B-Z junction. Nucleic Acids Res. 11, 3241-3254. Shibutani, S., and Grollman, A. P. (1993) Nucleotide misincorporation on DNA templates containing N-(deoxyguanosin-N2-yl)2-(acetylamino)fluorene. Chem. Res. Toxicol. 6, 819-824. Hingerty, B. E., Figueroa, S., and Hayden, T. L. (1989) Prediction of DNA structure from sequence: a build-up technique. Biopolymers 28, 1195-1222. Brown, T., Hunter, W. N., Kneale, G., and Kennard, O. (1986) Molecular structure of the G-A base pair in DNA and its implications for the mechanism of transversion mutation. Proc. Natl. Acad. Sci. U.S.A. 83, 2402-2406. Brown, T., Leonard, G. A., Booth, E. D., and Chambers, J. (1989) Crystal structure and stability of a DNA duplex containing A(anti)-G(syn) base pairs. J. Mol. Biol. 207, 455-457. Hunter, W. N., Brown, T., and Kennard, O. (1986) Structural features and hydration of d(C-G-C-G-A-A-T-T-A-G-C-G); a double helix containing two G-A mispairs. J. Biomol. Struct. Dyn. 4, 173191. Prive, G. G., Heinemann, U., Chandrasegaran, S., Kan, L. S., Kopka, M. L., and Dickerson, R. E. (1987) Helix geometry, hydration and G-A mismatch. Science 238, 498-504. Prive, G. G., Heinemann, U., Chandrasegaran, S., Kan, L. S., Kopka, M. L., and Dickerson, R. E. (1989) In Structure and Expression (Olson, W. K., Sarma, R. H., and Sundaralingam, M., Eds.) Vol. 2, pp 27-47, Adenine Press, New York. Webster, G., Sanderson, M., Shelly, J., Neidle, S., Swann, P., Li, B., and Tickle, I. (1990) Crystal structure and sequence-dependent conformation of the A-G mispair oligonucleotide d(CGCAGCTGGCG). Proc. Natl. Acad. Sci. U.S.A. 87, 6693-6697. Geacintov, N. E., Cosman, M., Hingerty, B. E., Amin, S., Broyde, S., and Patel, D. J. (1997) NMR solution structures of stereoiso-
Grad et al.
(25)
(26)
(27) (28)
(29) (30) (31)
(32) (33) (34)
(35) (36)
(37)
(38) (39) (40)
(41) (42)
(43)
meric polycyclic aromatic carcinogen-DNA adducts: principles, patterns, and diversity. Chem. Res. Toxicol. 10, 111-146. Cosman, M., Hingerty, B. E., Luneva, N., Amin, S., Geacintov, N. E., Broyde, S., and Patel, D. J. (1993) Solution conformation of the (+)-cis-anti-benzo[a]pyrenyl-dG adduct opposite dC in a DNA duplex: intercalation of the covalently modified deoxyguanosine into the major groove. Biochemistry 32, 4145-4155. Cosman, M., Hingerty, B. E., Luneva, N., Amin, S., Geacintov, N. E., Broyde, S., and Patel, D. J. (1996) Solution conformation of the (-)-cis-anti-benzo[a]pyrenyl-dG adduct opposite dC in a DNA duplex: intercalation of the covalently modified deoxyguanosine into the major groove. Biochemistry 35, 9850-9863. Pearlman, D., Case, D., Caldwell, J., Seibel, G., Singh, U. C., Weiner, P., and Kollman, P. A. (1986) AMBER 4.0 Documentation, University of California, CA. Singh, U. C., Weiner, S. J., and Kollman, P. (1985) Molecular dynamics simulation of d(C-G-C-G-A)‚d(T-C-G-C-G) with and without “hydrated” counterions. Proc. Natl. Acad. Sci. U.S.A. 85, 755. Hingerty, B. E., Ritchie, R. H., Ferrell, T. L., and Turner, J. E. (1985) Dielectric effects in biopolymers: the theory of ionic saturation revisited. Biopolymers 24, 427-439. Friedman, R. A., and Honig, B. (1992) The electrostatic contribution to DNA base stacking interactions. Biopolymers 32, 145159. Weiner, S. J., Kollman, P. A., Case, D. A., Singh, U. C., Ghio, C., Alagona, G., Profeta, S., and Weiner, P. (1984) A new force field for molecular mechanics simulations of nucleic acids and proteins. J. Am. Chem. Soc. 106, 765-784. Weiner, S. J., Kollman, P. A., Nguyen, D. T., and Case, D. A. (1986) An all atom force field for simulations of proteins and nucleic acids. J. Comput. Chem. 7, 230-252. Singh, U. C., and Kollman, P. A. (1984) An approach to computing electrostatic charges for molecules. J. Comput. Chem. 5, 129145. Jorgenson, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W., and Klein, M. L. (1983) Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 79, 926935. Van Gunsteren, W. F., and Berendsen, H. J. C. (1977) Algorithms for macromolecular dynamics and constraint dynamics. Mol. Phys. 34, 1311-1327. Ravishanker, G., Swaminathan, S., Beveridge, D. L., Lavery, R., and Sklenar, H. (1989) Conformational and helicoidal analysis of 30 ps of molecular dynamics on the d(CGCGAATTCGCG) double helix: “curves”, dials and windows. J. Biomol. Struct. Dyn. 6, 669-699. Srinivasan, J., Withka, J. M., and Beveridge, D. L. (1990) Molecular dynamics of an in vacuo model of duplex d(CGCGAATTCGCG) in the B-form based on the AMBER 3.0 force field. Biophys. J. 58, 533-547. Ravishanker, G., and Beveridge, D. L. (1991) Molecular Dynamics Analysis Toolchest, Wesleyan University, Middletown, CT. Saenger, W. (1983) Principles of Nucleic Acid Structure, SpringerVerlag, New York. Berman, H. M., Olson, W. K., Beveridge, D. L., Westbrook, J., Gelbin, A., Demey, T., Hsieh, S. H., Srinivasan, A. R., and Schneider, B. (1992) The nucleic acid database: a comprehensive relational database of three-dimensional structures of nucleic acids. Biophys. J. 63, 751-759. Gorin, A. A., Zhurkin, V. B., and Olson, W. K. (1995) B-DNA twisting correlates with base pair morphology. J. Mol. Biol. 247, 34-48. Singh, S. B., Hingerty, B. E., Singh, U. C., Greenberg, J. P., Geacintov, N. E., and Broyde, S. (1991) Structures of the (+)- and (-)-trans-7,8-dihydroxy-anti-9,10-epoxy-7,8,9,10-tetrahydrobenzo(a)pyrene adducts to guanine-N2 in a duplex dodecamer. Cancer Res. 51, 3482-3492. Fratini, A., Kopka, M. L., Drew, H. R., and Dickerson, R. E. (1982) Reversible bending and helix geometry in a B-DNA dodecamer: CGCGAATTBrCGCG. J. Biol. Chem. 257, 14686-14707.
TX970092E
