Structure and Stability of Duplex DNA Containing the 3

The data establish a regular right-handed conformation with Watson−Crick base .... Tanya Zaliznyak , Mark Lukin , Francis Johnson and Carlos de los ...
0 downloads 0 Views 209KB Size
Chem. Res. Toxicol. 2006, 19, 745-752

745

Structure and Stability of Duplex DNA Containing the 3-(Deoxyguanosin-N2-yl)-2-acetylaminofluorene (dG(N2)-AAF) Lesion: A Bulky Adduct that Persists in Cellular DNA Tanya Zaliznyak, Radha Bonala, Francis Johnson, and Carlos de los Santos* Department of Pharmacological Sciences, State UniVersity of New York at Stony Brook, Stony Brook, New York 11794-8651 ReceiVed January 4, 2006

The carcinogenic environmental pollutant 2-nitrofluorene produces several DNA adducts including the minor 3-(deoxyguanosin-N2-yl)-2-acetylaminofluorene (dG(N2)-AAF) lesion, which persists for long times in rat tissue DNA after discontinuation of carcinogen administration. Here, we present the solution structure of a dG(N2)-AAF duplex as determined by NMR spectroscopy and restrained molecular dynamics. The data establish a regular right-handed conformation with Watson-Crick base pair alignments throughout the duplex. The AAF moiety resides in the minor grove of the helix with its long axis directed toward the 5′-end of the modified strand. Restrained molecular dynamics shows that the duplex structure adjusts to the AAF lesion, reducing its exposure to water molecules. Analysis of UV melting profiles shows that the presence of dG(N2)-AAF increases the thermal and thermodynamic stability of duplex DNA, an effect that is driven by a favorable entropy. The structure and stability of the dG(N2)-AAF duplex have important implications in understanding the recognition of bulky lesions by the DNA repair system. The aromatic amines 2-aminofluorene (AF) and N-acetyl-2aminofluorene (AAF) are among the most extensively studied chemical carcinogens. Originally developed as insecticides, they were never used for such purposes after early studies revealed the strong carcinogenic activity of AAF (1). Although these compounds are not environmental hazards per se, they have served as models for chemical mutagenesis and as prototypes of DNA lesions formed by food mutagens, such as PhIP and IQ, present in heat-cooked meat, chicken, and fish (2-3). As is the case with other bulky aromatic compounds, AF and AAF do not react directly with cellular DNA, but they require metabolic conversion to reactive hydroxylamine derivatives that, when undergoing solvolysis in the presence of DNA, form covalent adducts with guanine bases, namely, N-(deoxyguanosin-8-yl)-2-aminofluorene (dG(C8)-AF), N-(deoxyguanosin8-yl)-2-acetylaminofluorene (dG(C8)-AAF), and 3-(deoxyguanosin-N2-yl)-2-acetylaminofluorene (dG(N2)-AAF) (4). The environmental contaminant 2-nitrofluorene, a major byproduct of diesel and kerosene combustion (5-6), is activated by cellular reductive metabolism to the same solvolytically unstable hydroxylamine derivative, again producing these guanine lesions (7). Because the synthesis and incorporation of the dG(N2)-AAF lesion into DNA has been difficult to accomplish, knowledge of the biological activity of AF and AFF has been mostly limited to studies of the C8 guanine adducts. The mutagenic activity of dG(C8)-AF and dG(C8)-AAF is quite different in bacteria, where the former causes G•C f T•A transversions, whereas the latter induces predominantly 1 and 2 base deletion mutations (8-9). In contrast, both lesions produce mostly G•C f T•A transversion mutations in mammalian cells (9). In the case of the dG(N2)-AAF, an early primer extension study using the Klenow fragment of DNA polymerase I showed that this lesion * To whom correspondence should be addressed. Phone: (631) 4443649. Fax: (631) 444-3218. E-mail: [email protected].

induces the insertion of dAMP, suggesting that G•C f T•A transversion will occur in vivo (10). More recently, it has been shown that while dG(N2)-AAF blocks almost completely the activity of mammalian DNA pol R, the lesion can be bypassed by pol η and pol κ. Furthermore, mutation analysis of singlestranded shuttle vectors in simian kidney cells established that dG(N2)-AAF causes G•C f T•A transversion mutations (11). Nucleotide excision repair proteins readily recognize and excise dG(C8)-AF and dG(C8)-AAF adducts from duplex DNA (1213). However, processing of dG(N2)-AAF lesions by the NER system remains to be studied. It has been observed that dG(N2)-AAF remains in rat tissue several months after the dietary administration of AAF or 2-nitrofluorene has been discontinued. Thus, this lesion is among the most persistent of the DNA adducts known to date (14-15). Whether or not the persistent dG(N2)-AAF lesions are somehow excluded from repair activity or are not sensed as lesions by the NER system is a question that remains to be answered. Several groups have studied the structure of short DNA duplexes containing dG(C8)-AF and dG(C8)-AAF lesions, using mainly NMR spectroscopy and restrained molecular dynamics (16-28). The presence of the dG(C8)-AF adduct causes conformational heterogeneity at the lesion site without perturbing the global B-form duplex structure or the canonical WatsonCrick alignment on flanking base pairs. The damaged residue can adopt a mixture of two main conformations depending mainly on the sequence context flanking the lesion. The structure of one conformer, known as the external major groove form, has the modified dG residue in the anti orientation around its glycosidic torsion angle forming Watson-Crick hydrogen bonds with its partner dC residue. The AF moiety is external to the base pair stack and resides in the major groove of the helix, where it is largely exposed to the solvent (17-19, 21, 22). The other principal conformer, called the base-displaced intercalated form, has the AF moiety inside the helix where it intercalates between flanking base pairs. In this case, the damaged dG

10.1021/tx060002i CCC: $33.50 © 2006 American Chemical Society Published on Web 05/04/2006

746 Chem. Res. Toxicol., Vol. 19, No. 6, 2006

Figure 1. Structure of the dG(N2)-AAF adduct and numbering scheme of the fluorene moiety (A). Sequence of the duplex employed in this study (B).

residue resides in the major groove of the helix with its glycosidic torsion angle in the syn conformation (23). The bulkier dG(C8)-AAF adduct adopts predominantly a basedisplaced, AAF-intercalated conformation (29, 30). However, in one of these studies, a significant fraction (∼30%) of the damaged duplex does have the AAF-external form (30). In contrast to the C8 adducts of aminofluorene, there is only one computational study that deals with the conformation of duplex DNA having the dG(N2)-AAF lesion (31). We describe here the solution structure of the d(CGTACXCATGC) •d(GCATGCGTACG) duplex (called herein the dG(N2)-AAF duplex), where X is the persistent dG(N2)-AAF lesion, established by high-resolution NMR spectroscopy and restrained molecular dynamics simulations. The chemical structure and numbering of the dG(N2)-AAF lesion are shown in Figure 1.

Experimental Procedures Sample Preparation. The preparation of 3-(deoxyguanosin-N2yl)-2-acetylaminofluorene DMT-phosphoramidite needed for the synthesis of the lesion-containing duplex has been reported elsewhere (32). Unmodified and lesion-containing oligodeoxynucleotides were synthesized using standard solid-phase chemical methods. Samples containing a 5′-dimethoxytrityl group were cleaved from the solid support by overnight treatment with concentrated aqueous ammonia at 55 °C. Oligodeoxynucleotidecontaining supernatants were purified by reversed-phase HPLC on a Luna phenyl-hexyl column (5 µm, 250 × 10 mm; Phenomenex, Torrance, CA) using a flow rate of 4 mL/min. A solvent gradient of 16-40% acetonitrile (30 min) in 0.1 M triethylamonium acetate (pH 6.8) was used to collect the dimethoxytrityl-protected oligomers (DMT-ON conditions). The lesion-containing oligodeoxynucleotide was eluted after 20.9 min as a single peak, whereas the unmodified oligomer appeared at 26 min. The 5′-DMT group was removed by treatment with 80% acetic acid for 30 min, and the samples were rechromatographed using a gradient of 5-25% acetonitrile over 40 min. After HPLC purification, the oligomer composition was confirmed by electrospray ionization mass spectroscopy using a Micromass Quattro LC system. Oligodeoxynucleotides were desalted using a Sephadex G-25 column and converted to the sodium salt by percolation through a Dowex 50W ion-exchange resin. Proper duplex stoichiometry was achieved by following NMR signal intensities during the gradual addition of the complementary to the lesion-containing strand, at 60 °C. Duplex annealing was performed by heating the mixture to 95 °C, followed by slow cooling to room temperature. The NMR sample consisted of 1.1 µmol of duplex dissolved in 0.7 mL of 25 mM phosphate buffer, pH 6.8, containing 50 mM NaCl and 0.5 mM EDTA, using either 99.96% D2O or

Zaliznyak et al. 90% H2O/10% D2O (v/v). Samples were vacuum-degassed inside the NMR tube before data collection. NMR Experiments. One- and two-dimensional experiments were carried out using high-field spectrometers operating at 11.75 and 14.1 T field strength. Proton chemical shifts were referenced relative to sodium 3-(trimethylsilyl)-propionate-2,2,3,3-d4. Phasesensitive (33) NOESY (50, 200, and 300 ms mixing time), TOCSY (70 and 120 ms mixing time), COSY, and DQF-COSY spectra in 100% D2O buffer were collected at 30 °C using a repetition time of 1.5 s, during which the residual water signal was suppressed by saturation. Phase-sensitive proton NOESY (120 and 220 ms mixing time) spectra in 90% H2O buffer were recorded at 5 °C, using a jump and return reading pulse (34). Time-domain data sets consisted of 2048 by 300 complex data points in the t2 and t1 dimensions, respectively. NMR data were processed and analyzed using the program Felix (Accelrys Inc., San Diego, CA) running on Silicon Graphics computers. Time domain data were multiplied by shifted sine-bell window functions prior to Fourier transformation. No baseline correction was applied to the transformed spectra. Structure Determination. Distance calculation and molecular dynamics simulations were performed using XPLOR3.85 on Silicon Graphics workstations (35). Interproton distances were computed by subjecting a canonical B-form duplex structure to potential energy minimization, using only a potential energy function that was proportional to the differences between back-calculated and experimental NOE intensities (36). All NOESY spectra were used for distance calculations. A grid search revealed that 1.20 ns was the best-fit isotropic correlation time, and this value was used for distance calculations. Molecular dynamics simulations were performed in vacuo using a CHARMM-derived all-atom force field (37), reduced phosphate charges, and a dielectric constant value of 4 (38). Topology of the AAF moiety was implemented from published research (31). A canonical B-form DNA structure containing the dG(N2)-AAF lesion was generated in InsightII (Accelrys, San Diego, CA), energy minimized, and used as the starting structure for molecular dynamics. A total of 582 interproton distances were restrained during molecular dynamics using square-well potential energy functions. Reflecting the quality of experimental data, bounds of (0.6 or (0.9 Å were added to the back-calculated distance value. Following the NMR evidence, Watson-Crick hydrogen bonds were enforced using square-well potential energy functions, with equilibrium distances taken from crystallographic studies and bounds of (0.1 Å. Backbone dihedral angles were restrained during molecular dynamics within a range encompassing A- and B-form DNA conformations. Our molecular dynamics protocol consisted of a heating step, where the temperature of the system increased from its initial value to 500 K in 60 ps, followed by a hightemperature step, during which the simulation remained at 500 K for different periods of time (140, 145, 150, and 160 ps). After the high-temperature step, the system was cooled to 300 K over a period of 20 ps and equilibrated at this temperature during an additional 100 ps of molecular dynamics. Experimental distances were softly enforced with a penalty constant that gradually increased its value from 10 to 300 kcal/(mol A2) during the heating step and remained at that value until the end of the simulation. A total of 20 structures were computed by starting molecular dynamics at different initial temperatures (100, 110, 120, and 130 K) and using five different lengths of the high-temperature step. Atomic coordinates at the end of the simulations were energy-minimized, generating the ensemble of distance-refined structures. Since the largest pairwise root-meansquare deviation among these structures was smaller than 0.9 Å, atom positions were averaged and energy-minimized producing the average three-dimensional model presented here. Computation of structural parameters was performed with Curves (39). DNA Melting Experiments. A detailed description of methods used for determination of thermodynamic parameters has been reported elsewhere (40). Briefly, thermal denaturation of DNA was carried out using a CARY100 Bio UV-vis spectrophotometer, with a multicell block temperature regulation unit and a fluid circulation thermal regulation enhancement (Varian, Inc.). Calibration of the

Structure of a dG(N2)-AAF-Containing Duplex temperature controller using a buffer solution with an external thermometer showed that temperature readings were stable and accurate to within 1 °C. Initial temperatures were allowed to equilibrate for at least 10 min at either 1 or 80 °C depending upon the experiment. To achieve complete temperature equilibration before absorbance readings, the rates of temperature increase and decrease were set to 0.3 °C per min. DNA samples consisted of between 0.3 and 1.6 OD260 units of duplex dissolved in 1 mL of 25 mM sodium phosphate buffer solution, pH 6.8, containing 100 mM NaCl and 0.5 mM EDTA. Six different sample concentrations and six independent melting profiles at each concentration composed a set of 36 independent determinations performed on each duplex. Absorption versus temperature data were fit with parameters ∆H° and ∆S° using Meltwin (v3.5, McDowell), generating parameter values for each melting curve (41-42). Reported enthalpy and entropy values are the average of individual determinations, and error estimates represent the standard deviation of the data. The Gibbs free energy (∆G°) of duplex formation was then calculated from these values. Plots of Tm-1 versus ln(Ct/4) yielded straight lines with a slope R/∆H° and an intercept ∆S°/∆H° (where Ct is the total DNA concentration and R the universal gas constant), suggesting a two-state melting process (43). ∆H° values computed from these graphs diverged by less than 15% from those calculated from the fitted melting curves, further supporting the presence of two-state processes (44).

Results Nonexchangeable Proton Spectra. The one-dimensional spectrum of the dG(N2)-AAF duplex dissolved in 100% D2O buffer solution at 30 °C shows exceptionally well-resolved and sharp proton signals (Figure 1S, Supporting Information), facilitating the complete NMR characterization of the sample at this temperature. Assignment of the nonexchangeable proton spectrum follows the analysis of NOESY, COSY, and TOCSY spectra using established procedures (45-46). Figure 2 displays an expanded region of a NOESY spectrum (300 ms mixing time) recorded at 30 °C, showing interactions in the base-H1′ sugar proton region of the spectrum. Characteristic of a right-handed helix, each purine-H8 or pyrimidineH6 proton displays NOE connectivity to the H1′ sugar proton of the same residue and to that of the 5′-flanking deoxyribose. These interactions are readily identified all through both strands of the dG(N2)-AAF duplex, indicating that the presence of the lesion induces only minimal perturbations of the canonical B-form structure. Furthermore, the intensity of all intraresidue base-H1′ NOE peaks is significantly weaker than the cytosine H5-H6 interactions, especially in the 50 ms mixing time NOESY spectrum, where spin-diffusion effects were minimized (data not shown). This observation suggests that all nucleotides have their N-glycosidic angle in an anti conformation. The presence of the fluorene moiety results in the observation of unusual chemical shifts for protons at and near the lesion site. The H1′ sugar proton signals of C7 and G18 shift upfield resonating in the H3′ region of the spectrum, whereas that of C17 moves in the opposite direction appearing at 6.30 ppm, a value characteristic of adenosine nucleotides (Figure 2). Unusual chemical shifts in other regions of the spectrum are mostly absent, and only the H3′ proton of C7 moves upfield, resonating among the H4′ sugar protons (data not shown). The chemical shifts of nonexchangeable protons are listed in Table 1S (Supporting Information). Assignment of AAF signals resulted from the analysis of the base proton region of the COSY and NOESY spectra (Figure 2S, Supporting Information). Three COSY peaks correlate the four proton signals of the nonsubstituted benzene ring (Figure

Chem. Res. Toxicol., Vol. 19, No. 6, 2006 747

Figure 2. Expanded contour plot region of a NOESY (300 ms mixing time) spectrum recorded at 30 °C showing interactions between base (8.83-6.86 ppm) and H1′ sugar (6.40-5.17 ppm) protons. Solid lines depict NOE connectivities on the A4-A8 segment of the lesioncontaining strand, and dashed lines do the same for the T15-T19 segment on the complementary strand. Arrows pointing out and coming into the figure reflect the unusual chemical shift of C7H1′ and G18H1′ protons, which resonate outside the normal H1′ region. Residue-labeled peaks indicate the intraresidue base-H1′ NOE connectivity, and asterisks denote cytosine(H5/H6) interactions. Other labels are assigned as follows: A, AAFH4-G6H1′; B, AAFH4-T19H1′; C, AAFH5-G6H1′; D, AAFH5-T19H1′; and E, AAFH1-A8H4′.

2SA, peaks B, E, F, Supporting Information), identifying the H5, H6, H7, and H8 AAF protons. Of these, AAFH5 is readily recognized because it shows a strong NOE peak to the H4 proton present on the substituted benzene ring (Figure 2SB, peak A, Supporting Information) and AAFH8 by its interaction with the pro-chiral methylene protons (H9(r) and H9(s)) of AAF (data not shown). In addition, AAFH1 shows NOE connectivity to the H9(r)/H9(s) protons and the methyl group of AAF, which distinguishes itself as a sharp signal resonating at a characteristic chemical shift value of 2.05 ppm (Figure 1S, Supporting Information). The chemical shift of AAF protons is listed in Table 1S (Supporting Information). Several NOE interactions between AAF and DNA protons define the location of the fluorene moiety in the duplex. The H1′ protons of G6 and C7 exhibit NOE cross-peaks with the aromatic H4 and H5 protons of AAF (Figure 2, peaks A-D), which are located at one edge of the aromatic AAF moiety. We identify NOE interactions between the AAF protons at the other edge of the flourene ring and sugar protons of the lesioncontaining strand. Specifically, NOE cross-peaks connect AAFH1 with A8H4′ (Figure 2, peak F) and AAFH9(r) with G6H1′, G6H4′, C7H1′, and C7H3′ (Figure 3, peaks A-D). In contrast, the H9(s) proton interacts with T19H1′, G18H1′, and T19H3′ (Figure 3, Peaks E-G) located on the undamaged strand of the duplex. Additionally, the methyl group of AAF shows NOE peaks to the H1′ and H4′ protons of the lesion partner C17 and the A8 residue (Figure 3, peaks H-K). All these NOE interactions are qualitatively satisfied by positioning the fluorene moiety in the minor groove of the helix with the long C3-C7 axis (Figure 1) directed toward the 5′-end of the lesioncontaining strand. Exchangeable Proton Spectra. The one-dimensional spectrum of the dG(N2)-AAF duplex dissolved in 9:1 H2O/D2O buffer solution at 5 °C shows eight partially resolved exchange-

748 Chem. Res. Toxicol., Vol. 19, No. 6, 2006

Zaliznyak et al. Table 1. Thermodynamic Stabilitya

dG(N2)-AAF

duplex control duplex a

Tmb(°C)

∆H0(kcal/mol)

∆S0(cal/mol K)

∆G025°C(kcal/mol)

∆G037°C(kcal/mol)

60.5 54.3

-93 ( 9 -98 ( 11

-262 ( 20 -285 ( 23

-14.9 ( 0.3 -13.1 ( 0.3

-11.8 ( 0.3 -9.6 ( 0.3

Determined by fitting individual melting curves. b Calculated for a 1.0 × 10-4 M DNA concentration. Reported errors are standard deviations.

Figure 3. Expanded contour plot region of a NOESY (300 ms mixing time) spectrum recorded at 30 °C displaying interactions AAF and duplex protons. Labeled peaks are assigned as follows: A, G6H1′AAFH9(r); B, G6H4′-AAFH9(r); C, C7H1′-AAFH9(r); D, C7H3′AAFH9(r); E, T19H1′-AAFH9(s); F, G18H1′-AAFH9(s); G, T19H3′AAFH9(s); H, C17H1′-AAFCH3; I, A8H1′-AAFCH3; J, C17H4′AAFCH3; and K, A8H4′-AAFCH3.

Figure 4. Expanded contour plot regions of a NOESY (220 ms mixing time) spectrum recorded at 5 °C with the corresponding one-dimensional spectrum displayed on top. The figure depicts NOE interactions involving the exchangeable and base protons of the duplex. Assignment of imino protons is written in the figure. Labeled cross-peaks are assigned as follows: A, T3N3H-A20H2 and T9N3H-A14H2; B, T15N3H-A8H2; C, T19N3H-A4H2; D/D′, G6N1H-C17N4Hhb/ G6N1H-C17N4Hnhb; E/E′, G2N1H-C21N4Hhb/G2N1H-C21N4Hnhb; F/F′, G10N1H-C13N4Hhb/G10N1H-C13N4Hnhb; G/G′, G16N1H-C7N4Hhb/ G16N1H-C7N4Hnhb; H/H′, G18N1H-C5N4Hhb/G18N1H-C5N4Hnhb; I, G6N1H-G6N2H; J, G6N1H-AAFNH; K, G6N2H-AAFNH; L, G18N1HAAFH5; M, G6N1H-AAFCH3; N, G16N1H-AAFCH3; O, G6N2HAAFCH3; P, T3N3H-A4H2; Q, T19N3H-A20H2; R, G2N1H-A20H2; S, G10N1H-A14H2 and G16N1H-A4H2; and T, G18N1H-A4H2.

able proton signals between 12.5 and 13.8 ppm and an additional peak at 9.5 ppm (Figure 4 and Figure 4S, Supporting Information). Assignment of the exchangeable protons results from the analysis of NOESY (120 and 220 ms mixing time) spectra recorded at low temperature following standard procedures (45). Strong NOE peaks between TH3 and AH2 protons (Figure 4, peaks A-C) and GH1 and CN4H amino protons (Figure 4, peaks D/D′-H/H′) establish Watson-Crick alignments for all A•T and C•G base pairs of the duplex, including the lesioncontaining G6•C17 pair. NOE interactions between adenineH2 and imino protons of flanking base pairs (Figure 4, peaks P-T) and between adjacent imino protons of the duplex (Figure 3S, Supporting Information) suggest proper base stacking throughout the duplex. Exchangeable proton chemical shifts of the dG(N2)-AAF duplex are listed in Table 1S (Supporting Information).

Figure 5. Stability of DNA containing a dG(N2)-AAF residue. Top panel: typical UV melting curves for the AAF-containing and unmodified duplexes at three different DNA concentrations. Bottom panel: van’t Hoff plots derived from melting curves showing a linear correlation between ln([DNA]/4) and 1/Tm, which indicates of a twostate melting process. In both panels, black-filled and unfilled circles correspond to the dG(N2)-AAF-containing and unmodified duplexes, respectively.

The presence of dG(N2)-AAF results in particular spectral characteristics. The imino proton signal of the lesion-containing nucleotide, G6H1, moves downfield into the thymine imino proton region where its resonance overlaps with T19H3. Because of the presence of the fluorene moiety, rotation around the guanine C2-N2 bond is no longer possible, and a strong NOE cross-peak with G6H1 readily identifies the lesion G6N2H proton, which appears as a sharp signal at 9.52 ppm (Figure 4, peak I). The amide and methyl protons of AAF show NOE interactions with G6H1 and G6N2H protons of the damaged residue (Figure 4, peaks J, M, K, and O), suggesting that both groups are directed toward the interior of the helix. In addition, the AAFCH3-G16H1 and G18H1-AAFH5 NOE cross-peaks (Figure 4, peaks N and L, respectively) further define the orientation of the fluorene lesion in the minor groove, with the long axis of the aromatic moiety directed toward the 5′-side of the damaged G6 residue, while its acetylamino group points in the opposite direction. Duplex Stability. At a preliminary stage during the collection of the NMR spectra, it was evident that the dG(N2)-AAF duplex was particularly stable. The temperature dependence of the exchangeable proton spectra (Figure S4, Supporting Information) shows that the imino proton signals of the duplex five-central base pairs are still visible in the 50-65 °C range, temperatures that cause the complete disappearance of other imino protons due to solvent exchange. This observation is in contrast to the general behavior of lesion-containing DNA where the duplex

Structure of a dG(N2)-AAF-Containing Duplex

Chem. Res. Toxicol., Vol. 19, No. 6, 2006 749

Table 2. dG(N2)-AAF Duplex Structure Deviations from experimental restraintsa RMSD NOE distances (Å) 0.015 NOE energy (kcal/mol Å2) 30 ( 1.1 RMSD bond distances (Å) 0.014 RMSD bond angles (deg) 3.9 RMSD dihedral angles (deg) 32 van der Waals energy (kcal/mol) -330 Relevant structural parametersb minor groove width (Å)c 2.0-4.7 (5.9) minor groove depth (Å)c 5.6-7.2 (4.6) base pair propeller twist (°)d 1-34 (