Benchmark Theoretical and Experimental Study on 15N NMR Shifts of

Jan 4, 2016 - DFT design of polyguanidine – a unique two-dimensional material with high-energy density. Sergey V. Bondarchuk , Boris F. Minaev. Mole...
2 downloads 0 Views 2MB Size
Article pubs.acs.org/JPCB

Benchmark Theoretical and Experimental Study on 15N NMR Shifts of Oxidatively Damaged Guanine Martin Dračínský,† Michal Šála,† Blanka Klepetárǒ vá,† Jakub Šebera,†,‡ Jiří Fukal,† Veronika Holečková,† Yoshiyuki Tanaka,§ Radim Nencka,† and Vladimír Sychrovský*,† †

Institute of Organic Chemistry and Biochemistry, Academy of Sciences of the Czech Republic v.v.i., Flemingovo náměstí 2, 16610 Praha, Czech Republic ‡ Institute of Physics, Academy of Sciences of the Czech Republic, v.v.i, Na Slovance 2, CZ-182 21 Prague 8, Czech Republic § Faculty of Pharmaceutical Sciences, Tokushima Bunri University, 180 Nishihama-Boji, Yamashirocho, Tokushima, Tokushima 980-8578, Japan S Supporting Information *

ABSTRACT: The 15N NMR shifts of 9-ethyl-8-oxoguanine (OG) were calculated and measured in liquid DMSO and in crystal. The OG molecule is a model for oxidatively damaged 2′-deoxyguanosine that occurs owing to oxidative stress in cell. The DNA lesion is repaired with human 8-oxoguanine glycosylase 1 (hOGG1) base-excision repair enzyme, however, the exact mechanism of excision of damaged nucleobase with hOGG1 is currently unknown. This benchmark study on 15N NMR shifts of OG aims their accurate structural interpretation and calibration of the calculation protocol utilizable in future studies on mechanism of hOGG1 enzyme. The effects of NMR reference, DFT functional, basis set, solvent, structure, and dynamics on calculated 15N NMR shifts were first evaluated for OG in crystal to calibrate the best performing calculation method. The effect of large-amplitude motions on 15N NMR shifts of OG in liquid was calculated employing molecular dynamics. The B3LYP method with Iglo-III basis used for B3LYP optimized geometry with 6-311++G(d,p) basis and including effects of solvent and molecular dynamic was the calculation protocol used for calculation of 15N NMR shifts of OG. The NMR shift of N9 nitrogen of OG was particularly studied because the atom is involved in an N-glycosidic bond that is cleaved with hOGG1. The change of N9 NMR shift owing to oxidation of 9ethylguanine (G) measured in liquid was −27.1 ppm. The calculated N9 NMR shift of OG deviated from experiment in crystal and in liquid by 0.45 and 0.65 ppm, respectively. The calculated change of N9 NMR shift owing to notable N9-pyramidalization of OG in one previously found polymorph was 20.53 ppm. We therefore assume that the pyramidal geometry of N9 nitrogen that could occur for damaged DNA within hOGG1 catalytic site might be detectable with 15N NMR spectroscopy. The calculation protocol can be used for accurate structural interpretation of 15N NMR shifts of oxidatively damaged guanine DNA residue.



technique is usable for in situ probing of nucleic acids.16−19 The structure and conformation of damaged 8-oxo-2′-deoxyguanosine was studied with 13C and 15N NMR spectroscopy.20−22 Structural interpretation of the NMR spectroscopic parameters acquired for nucleic acids commonly employs empirical rules that need not be valid for damaged nucleosides. The theoretical calculation methods can describe accurately effects of structure, local conformation, base-pairing and specific solvation on NMR parameters of nucleic acids.23−27 We therefore anticipated that the structural interpretation of NMR experiment involving damaged 8-oxoguanine can also employ theoretical methods; however, their performance needs to be benchmarked.

INTRODUCTION

The oxidatively damaged guanine base can be recognized during DNA replication as mispairing lesion with mutagenic potential.1 One of the repair mechanisms targeted on this DNA lesion in cell involves the hOGG1 base-excision repair enzyme. However, the exact chemical mechanism of repair including the mechanism of excision of 8-oxoguanine base is currently unknown.2−4 Several excision mechanisms with hOGG1 enzyme were also proposed based on the theoretical modeling.5−15 The glycosidic nitrogen of 8-oxo-2′-deoxyguanosine is a “hot spot” of the base excision reaction operated by the hOGG1 enzyme with key importance in all currently assumed mechanisms. The hOGG1 catalytic pathway could be thus validated when structural information on activated substrate within catalytic site is available. The 15N NMR spectroscopy could be employed for this purpose since the spectroscopy © XXXX American Chemical Society

Received: November 23, 2015 Revised: December 22, 2015

A

DOI: 10.1021/acs.jpcb.5b11428 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

The δN1(OG), δN9(OG), δN1(G), and δN9(G) are NMR shifts of the N1 and N9 nitrogens of OG and G with respect to NMR reference R. The ΔδN9 and ΔδN9(N1) are called the NMR shift of N9 nitrogen of OG in the text. The effects of basis set, optimized geometry and solvent on calculated ΔδN9 NMR shift are reported in the Supporting Information. The molecular dynamics (MD) calculations employed GAFF force field40 and the ESP atomic charges obtained at the Hartree−Fock level with 6-31G(d,p) basis. The G and OG were solvated by TIP3P water41 or explicit DMSO42 within initial cubic box with the edge of 50 Å. The equilibration of system including energy minimization followed by 30 ps free dynamics was followed by 20 ns MD simulation at standard laboratory conditions; temperature 298.15 K, pressure 1 atm. The MD snapshot geometries were used for calculation of averaged ΔδN9 NMR shift as was described in the Supporting Information. 4000 MD snapshots acquired for sampling step of 5 ps within 20 ns MD run were used for calculating probability distributions of the χ and κ′ torsion angles depicted in the Figure 1. 200 snapshots acquired for sampling step of 100 ps within 20 ns MD run were optimized preserving the MD calculated orientation of the 9-ethyl group and N9-pyramidalization. Two methods were employed for the dynamical averaging of the ΔδN9 NMR shift. Briefly, the adiabatic averaging employed adiabatic dependence of σN9 shielding on χ torsion that was averaged employing χ-probability distribution. The MD averaging employed statistical average of σN9 shielding calculated for MD snapshots that were geometry optimized or original. The adiabatic dependence of σN9 shielding on χ torsion was parametrized by employing the general function

The utilization of 15N NMR shifts acquired for nucleic acids as source of structural information is less common owing to obstacles in their measurement such as low natural abundance and sensitivity.28 Also the calculation of 15N NMR shifts in organic compounds is nontrivial.29 The theoretical 15N NMR shifts are known to depend on calculation method more than 13 C and 1H NMR shifts.30 Furthermore, the effects of solvent and molecular dynamics affect calculated 15N NMR shifts.31−33 This study therefore aims also to demonstrate usability of 15N NMR spectroscopy in structural studies of nucleic acids and particularly the usefulness of theoretical methods for detailed interpretation of measured NMR parameters. The OG was chosen for the benchmark study since the basic chemical structure of damaged DNA residue was kept and accuracy of 15 N NMR benchmark calculations can be pushed toward the limit owing to moderate size of studied molecule. The effect of rotation of 9-ethyl group on 15N NMR shifts was studied in relation with rotation flexibility of N-glycosidic bond in DNA nucleosides. The effect of pyramidal geometry of N9 nitrogen on N9 NMR shift was particularly studied in relation to the recently proposed hOGG1 base-excision scheme.12 The OG and G molecules were synthesized and the 15N NMR spectra were acquired in crystal and in liquid. The original chemical synthesis of molecules and X-ray and 15N NMR experiments enabled reliable benchmarking of calculation protocol for accurate interpretation of 15N NMR shifts of oxidatively damaged guanine.



METHODS AND MATERIALS Theoretical Calculations. The geometries of G and OG molecules (Figure 1) were optimized with the B3LYP

σN9(χ ) = a cos(χ )5 + b cos(χ )4 + c cos(χ )3 + d cos(χ )2 + e cos(χ ) + f

as was described in the Supporting Information. Here, eq 3 was utilized in analysis of dynamical contribution to the ΔδN9 NMR shift. The atomic coordinates of the FURGAA03 crystal structure including two crystallographically equivalent OG molecules in the unit cells were acquired from the Cambridge Crystallographic Database.43 Calculations that utilized the crystal structure employed the plane-wave DFT with the generalized gradient approximation of Perdew, Burke, and Ernzerhof44 as implemented within the GIPAW approach.45,46 Calculations that utilized molecular cluster derived from the crystal employed the molecular-orbital DFT approach and the B3LYP method. The plane-wave cutoff energy of 610 eV and Monkhorst−Pack grid47 of a minimum k-point sampling of 0.04 Å−1 in integral evaluation over the Brillouin zone were used in optimization of atomic positions and in calculations of 15 N NMR shifts. The dynamical averaging of ΔδN9(N1) NMR shift in the crystal employed less tight criteria; the plane-wave cutoff energy of 300 eV, the Monkhorst−Pack grid 0.1 Å−1, NVT ensemble at constant temperature of 300 K, Langevin thermostat, 0.5 fs integration time step, and ultrasoft pseudopotentials.48 The lattice parameters were fixed to the experimental values. The MD averaged ΔδN9(N1) NMR shift was calculated employing 182 geometries since 91 geometries were acquired at 1.0, 1.1, 1.2, ..., 10.0 ps of each MD run for the two crystallographically equivalent molecules (Z′ = 2). All the calculations in relation with NMR experiment in liquid and the calculations that utilized cluster model derived

Figure 1. 9-Ethylguanine (G) and 9-ethyl-8-oxoguanine (OG) molecules. The orientation of 9-ethyl group with respect to base was described with torsion angle χ = C4−N9−C(H2)−C(H3) and pyramidal geometry of N9 nitrogen (N9-pyramidalization) was described with torsion angle κ′ = C4−N9−C(H2)−C8 − 180°.

method,34,35 6-311++G(d,p) atomic basis36,37 and by including the implicit polarizable continuum model (PCM)38 of the DMSO solvent. The NMR shielding of the N9 nitrogen of G and OG (σN9(G) and σN9(OG)) was calculated employing the B3LYP method, Iglo-III basis,39 and PCM DMSO solvent. To compare calculated and measured N9 NMR shifts of OG, we introduced a special NMR reference as follows. For liquid state calculations and experiment, the N9 NMR shift of OG was related to the N9 NMR shift of G: Δδ N9 = δ N9(OG) − δ N9(G) = σN(R) − σN9(OG) − (σN(R) − σN9(G)) = σN9(G) − σN9(OG)

(1)

For solid-state calculations and experiment, the N9 NMR shift was related to the N1 NMR shift of OG: Δδ N9(N1) = σN1(OG) − σN9(OG) = σN(R) − σN9(OG) − (σN(R) − σN1(OG)) = δ N9(OG) − δ N1(OG)

(3)

(2) B

DOI: 10.1021/acs.jpcb.5b11428 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B Scheme 1. Description of the Synthesis of the G Molecule and Preparation of the OG Molecule

repositioned geometrically and then refined with riding constraints. The resulting structure was almost identical to that of FURGAA03. Minor differences in lattice parameters ( 2σ(I) and 137 parameters. CCDC 1435942. An ORTP58 view of OG molecule can be seen in the Figure S9. NMR Spectroscopy Measurements. High-resolution 15N solid-state NMR spectra at natural isotope abundance were obtained using a Bruker Avance II spectrometer operating at 499.9 MHz for 1H and 50.7 MHz for 15N. Samples were packed into 3.2 mm magic angle spinning rotors and measurements taken using a MAS rate of 12 kHz using cross-polarization (CP) with ramped amplitude shape pulse. The typical CP conditions were used: recycle delay 5 s, contact time 4 ms, acquisition time 35. The nitrogen chemical shifts were referenced to crystalline α-glycine as a secondary reference (δR = 34.1 ppm). For a partial signal assignment, short CP experiments with contact time of 400 μs were used leading to a suppression of signals of N3 and N9 nitrogen atoms without any directly attached hydrogen. All signals were assigned by a comparison with liquid-state and computational data. The liquid state NMR spectra were recorded in DMSO-d6 at 25 °C with a Bruker Avance III spectrometer equipped with a 5 mm diameter cryoprobe operating at 500.0 MHz for 1H and 50.7 MHz for 15N. The nitrogen spectra were referenced to nitromethane used as an external standard (δR = 381.7 ppm). 2D correlation experiments (1H−15N HSQC and 1H−15N HMBC) were used for nitrogen signal detection and assignment (Figure S8).

from FURGAA03 crystal structure were carried out with the Gaussian 09, revision D01 program.49 All the plane-wave calculations including MD averaging of 15N shifts in the crystal were carried out with the CASTEP program.50 The MD calculations in liquid were carried out with the Amber 10 program package.51 Synthesis and Crystallization of 9-Ethyl-8-oxoguanine. Guanine derivative 3 and 8-oxoguanine derivative 6 were synthesized by simple built-up strategy52 starting from 2-amino4,6-dichloro-5-formamidopyrimidine (1). 9-Ethyl derivative 2 was prepared in high yield (74%) and final 9-ethylguanine compound 3 was obtained by acidic hydrolysis with trifluoroacetic acid (90%). The characteristics of 2 matched to those previously reported.53 A three-step methodology was used for introduction of two oxo groups into the structure of derivative 6. First, the chlorine atom at positon 6 was exchanged with OBn group by nucleophilic displacement with benzylalcohol under basic conditions (76%, DABCO, K2CO3). Subsequent bromination of the purine scaffold with N‑bromosuccinimide (NBS) afforded compound 5. Finally, the bromine atom and the benzyl group were removed by acidic hydrolysis, which yielded the desired 9-ethyl-8-oxoguanine 6. The experimental details of synthesis depicted in the Scheme 1 are reported in the Supporting Information. Compound 6 was subjected to crystallization experiments (1, water; 2, water:ethanol:dimethylformanide 6:4:1 with a catalytic amount of guanosine54). Both crystallization experiments provided the same polymorph according to solid-state NMR spectroscopy measurement in the crystal. The crystal structure was determined by X-ray diffraction. X-ray Diffraction Measurement. The diffraction data of OG were collected on Xcalibur PX diffractometer by monochromatized Cu (Kα) radiation (λ = 1.54180 Å) at 180 K. CrysAlisProCCD55 software was used for data collection, cell refinement, and data reduction. The structure was solved by direct methods with SIR9256 and refined by full-matrix leastsquares on F with CRYSTALS.57 The positional and anisotropic thermal parameters of all non-hydrogen atoms were refined. All hydrogen atoms were located in a Fourier difference map, but those attached to carbon atoms were C

DOI: 10.1021/acs.jpcb.5b11428 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B Table 1. Calculated and X-ray Determined Geometrical Parameters of G and OG Moleculesa G methodb QM GAFF MD X-rayc FURGAA03 X-rayc JICLAI

OG

χ

κ′

χ

κ′

84.1 85.2 177.6 (53.5; 326.8) − −

−2.6 0.3 0.4 (−24.9; 29.8) − −

86.2 85.0 188.8 (45.2; 302.2) 89.5 103.5

−1.2 0.1 −0.2 (−21.5; 20.6) −0.3 28.7

The χ = C4−N9−C(H2)−C(H3) and κ′ = C4−N9−C(H2)−C8 − 180° torsion angles in degrees. bQM: optimization employing B3LYP method, 6-311++G(d,p) basis, PCM DMSO solvent. GAFF: optimization employing GAFF force field neglecting solvent. MD: statistical average employing 4000 snapshots calculated with GAFF force field. The maximal and minimal amplitudes of torsion angles are given in parentheses. cThe FURGAA0359 (X-ray in this work) and JICLAI54 crystal structure. a



RESULTS The Structures of G and OG. The calculated energy minimum structures of G and OG and the structure of OG in FURGAA0359 crystal were very similar (Table 1). The orientation of the 9-ethyl group with respect to base was described with χ torsion angle and the degree of pyramidal geometry of N9 nitrogen was described with κ′ torsion angle (Figure 1). The X-ray resolved χ and κ′ of OG in FURGAA03 differed only minutely from the geometry optimized parameters. The orientation of 9-ethyl group with respect to base was almost perpendicular and geometry of N9 nitrogen was nearly planar. Interestingly, a notable N9-pyramidalization of OG (κ′ = 28.7°) was reported by Doi in a different polymorph of OG captured in the JICLAI crystal.54 Similar magnitudes of N9pyramidalizations found in the crystal of DNA quadruplex containing normal 2′-deoxyguanosine were enforced by specifically coordinated molecules of solvent.60 We therefore assume that notable N9-pyramidalization of OG in JICLAI was induced also by crystal packing and molecules surrounding OG since their kind and positioning in JICLAI and in FURGAA03 are different.54 The energy barrier on adiabatic energy surface describing rotation of 9-ethyl group for OG was higher by 3.4 kcal.mol−1 than the barrier for G (Figure 2). The oxidative damage to G thus resulted in more confined rotation flexibility of 9-ethyl group. The energy barriers for G and OG corresponded to MD calculated frequencies of the rotation conformers having χ ≈ 180° (Figure 3). The peaks on χ-distributions centered near χ ≈ 90° and χ ≈ 270° coincided with χ values calculated for energy minimum of G and OG. The χ-distributions for OG decreased to zero for rotation of 9-ethyl farther than ca. 40° from minimum while the χ-distributions for G indicated occurrence of rotation conformers even within the barrier (Figure 3). The N9-pyramidalizations of some MD snapshots was notable, however, the overall N9-pyramidalization of both G and OG was negligible (Table 1). The MD calculated κ′ ranged from −24.9° to +29.8° and from −21.5° to +20.6° for G and OG, respectively. The κ′-distributions were centered at 0°, and the overall MD geometry of N9 nitrogen was planar since the out-of-plane fluctuations of N9 nitrogen were symmetric (Figure 4). The N9-pyramidalization can be thus only enforced by some external factors since N9 nitrogen of G and OG is inherently planar. Such enforcement can occur in normal nucleic acids as was demonstrated by κ′ ranging from −21° to 24° in the ultra high-resolution crystal structures of DNA and RNA molecules.61 The external factors can be modeled explicitly, however, the geometric constraints employed in

Figure 2. Dependences of energy E relative to energy minimum on χ torsion angle calculated employing the B3LYP method, 6-311+ +G(d,p) basis and PCM DMSO solvent. The adiabatic potential energy surfaces calculated employing χ-constraint for G (black filled circle) and OG (red filled square). The correlation of relative energies with χ-torsion calculated for G (black open circle) and OG (red open square) optimized MD snapshots.

Figure 3. MD calculated probability distribution of χ torsion angle for G (black full circle) and OG (red full square) molecules.

adiabatic geometry optimization also enforce N9-pyramidalization since they represent external forces.60 This explains notable N9-pyramidalizations calculated employing adiabatic χconstraint for G and OG (Figure S3). The difference of D

DOI: 10.1021/acs.jpcb.5b11428 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

Figure 4. MD calculated probability distribution of κ′ torsion angle for G (black full circle) and OG (red full square) molecules.

Figure 5. Experimental 15N solid-state NMR spectra of OG acquired with standard cross-polarization (black spectrum) and with short cross-polarization to suppress signals of nitrogen atoms without directly attached hydrogen atoms (red spectrum). The 15N NMR shifts were referenced to solid α-glycine (δR = 34.1 ppm).

energies calculated for MD snapshots and energies within adiabatic potential surface correspond to energy increase owing to adiabatically conserved N9-pyramidalization (Figure 2). The parabolic-like potential may be anticipated from the correlation of MD snapshot energies with κ′ torsion (Figure S4). The 15N NMR Shifts of OG Molecule. I. The 15N NMR Reference. The choice of NMR reference in theoretical calculations may become critical. Accuracy of the calculated NMR shift is often affected by the error that occurs owing to inappropriate theoretical description of chemically different atoms employed for calculations of NMR shielding in studied molecule and in NMR reference molecule. The aim is to employ NMR reference that is chemically similar to the studied molecule since such referencing usually ensures that similar systematic errors in calculated NMR shieldings cancel out when NMR shift is calculated. The error can be also avoided or suppressed by employing relevant secondary NMR reference as was suggested for 31P NMR by van Wüllen.62 The measurement of δN9 NMR shift of OG molecule in crystal and in liquid provided unique opportunity to demonstrate effect of inappropriately chosen NMR reference on calculated δN9 NMR shift. The FURGAA03 crystal was Xray resolved and high-accuracy structure of OG molecule including surrounding molecules was obtained (Figure S5 and S9). All the 15N NMR resonances of OG molecule in liquid and in crystal were acquired (Figure 5 and Figure S8). The δN9 NMR shift of OG measured in liquid DMSO and in crystal was referenced to nitrogen of nitromethane and α-glycine, respectively. Particularly the nitrogen of nitromethane is different as compared to N9 nitrogen of OG. We may therefore assume substantial error in calculated δN9 NMR shift referenced to nitromethane. The calculated σN9(OG) NMR shielding including solvent effects and neglecting effect of molecular dynamics was 82.47 ppm (QM calculation, Table 4). The NMR shielding of nitrogen of nitromethane calculated employing the same method was −178.80 ppm. The calculated and measured δN9 NMR shifts of OG referenced to nitromethane were −261.27 and −243.0 ppm, respectively. The difference between calculated and measured δN9 NMR shifts was much larger than calculated changes of σN9(OG) NMR shielding owing to effects of local geometry, molecular dynamics and solvent reported below. By employing the N9 nitrogen of G molecule

(eq 1) or N1 nitrogen of OG molecule (eq 2) as NMR reference in calculations of N9 NMR shift we avoided the systematic error employing nitromethane reference. The deviations of calculated ΔδN9 and ΔδN9(N1) NMR shifts from experiment reported below are smaller by one order in magnitude as compared to difference of measured and calculated δN9 NMR shift referenced to nitromethane. The calculations of ΔδN9 and ΔδN9(N1) NMR shifts allowed unbiased benchmarking of calculation methods. II. The N9 NMR Shift of the OG Molecule in Crystal. The structure of OG in the FURGAA03 crystal and the structure of optimized OG were very close. The X-ray diffraction structures could be however affected by molecular motions and geometry optimization prior to NMR calculation is therefore desirable.63 The effect of geometry optimization on ΔδN9(N1) NMR shift (eq 2) was therefore calculated. The plane-wave (PW) NMR calculation employing PBE functional describes infinite periodic arrangement of molecules in crystal. The PBE calculated ΔδN9(N1) NMR shift in original FURGAA03 structure where only added hydrogen atoms were PW geometry optimized was 5.04 ppm and when FURGAA03 was fully optimized ΔδN9(N1) was 3.85 ppm. The geometry optimization improved agreement of calculated ΔδN9(N1) NMR shift with experiment. The measured ΔδN9(N1) NMR shift was −0.90 ppm. Next, we calculated the effect of molecular dynamics on ΔδN9(N1) NMR shift. (The PW-ST and PW-MD calculations in the Table 2.) The ΔδN9(N1) NMR shift decreased by 1.08 ppm owing to dynamical averaging since the σN9(OG) and σN1(OG) shielding increased by 5.79 and 4.71 ppm, respectively. Next, the performance of B3LYP method was tested. The B3LYP calculations of ΔδN9(N1) NMR shifts employed Iglo-III basis and structural model adopted from FURGAA03 that included OG surrounded by 10 OG molecules and four water molecules (Figure S5). The B3LYP calculated ΔδN9(N1) NMR shift employing the cluster model with PW PBE optimized hydrogen atoms was 3.55 ppm. The B3LYP calculated ΔδN9(N1) NMR shift employing fully PW PBE optimized cluster model where only the central OG was B3LYP optimized was 0.63 ppm. The utilization of B3LYP method in both geometry optimization and NMR calculation thus further improved agreement of calculated ΔδN9(N1) NMR shift with experiment. The ΔδN9(N1) NMR shift calculated with B3LYP and PBE method neglecting surrounding molecules (“gas phase” E

DOI: 10.1021/acs.jpcb.5b11428 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B Table 2. Calculated and Measured NMR and Geometric Parameters of the OG Moleculea calculations methodb

σN1

σN9

ΔδN9(N1)

χ

PW1 PW2 PW3 PW-ST PW-MD QM1 QM2 QM3 QM4

82.08 79.37 76.63 75.96 80.67 85.06 83.36 79.47 83.62

77.04 75.52 56.51 71.76 77.55 81.51 82.73 79.14 84.49

5.04 3.85 20.13 4.20 3.12 3.55 0.63 0.33 −0.87

89.5 90.0 103.5 86.9 86.3 89.5 88.6 84.6 87.2

FURGAA03

κ′ −0.3 0.0 28.7 −2.7 −1.0 −0.3 −1.5 −1.7 −1.2 experiment

rN9−C(H2)

rN9−C4

rN9−C8

1.459 1.456 1.581 1.456 1.461 1.459 1.459 1.459 1.460

1.373 1.376 1.393 1.377 1.381 1.373 1.376 1.379 1.375

1.390 1.399 1.403 1.397 1.403 1.390 1.396 1.420 1.412

δN1

δN9

ΔδN9(N1)

χ

κ′

rN9−C(H2)

rN9−C4

rN9−C8

144.30

143.40

−0.90

89.5

−0.3

1.459

1.373

1.390

The NMR shielding σN9, the ΔδN9(N1) NMR shift (eq 2), and the measured δN1 and δN9 NMR shifts referenced to solid α-glycine (δR = 34.1 ppm) are given in ppm. The torsion angles χ and κ′ are given in degrees and bond lengths in Angstroms (Figure 1). bPW: the plane-wave-PBE calculations employing cutoff energy 610 eV and grid 0.04 Å−1. PW-MD: the plane-wave-PBE dynamical averaging employing cutoff energy 300 eV and grid 0.1 Å−1. QM: the molecular-orbital DFT calculations employing PBE or B3LYP functional and Iglo-III basis. PW1: only H atoms in FURGAA03 optimized, PW2: fully optimized FURGAA03, PW3: only H atoms in JICLAI optimized, PW-ST: fully optimized (static) FURGAA03. PW-MD: dynamically averaged FURGAA03. QM1: NMR B3LYP, OG including nearest surrounding in FURGAA03 from PW1 (see Figure S5), QM2: NMR B3LYP, OG including FURGAA03 surrounding from PW2 (see Figure S5), only central OG molecule optimized with B3LYP/6-311++G(d,p), QM3: NMR PBE, isolated OG molecule optimized with PBE/6-311++G(d,p), QM4: NMR B3LYP, isolated OG molecule optimized with B3LYP/6-311+ +G(d,p). Experiment: the solid state NMR experiment in the FURGAA03. a

Table 3. Deviations of Mutual ΔΔδN NMR Shifts Calculated for OG Molecule from Experimenta atom N7

N3

N2 (NH2)

N1

atom

PW1

PW2

QM1

QM2

PW1

PW2

QM1

QM2

PW1

PW2

QM1

QM2

PW1

PW2

QM1

QM2

N9 N7 N3 N2

−0.04 − − −

1.20 − − −

−1.23 − − −

−1.79 − − −

−4.88 −4.84 − −

−1.00 −2.20 − −

−2.40 −1.17 − −

−1.12 0.67 − −

−0.98 −0.94 3.90 −

0.62 −0.58 1.62 −

1.07 2.30 3.47 −

−0.85 0.94 0.27 −

−5.94 −5.90 −1.06 −4.96

−4.75 −5.95 −3.75 −5.37

−4.45 −3.22 −2.05 −5.52

−1.53 0.26 −0.41 −0.68

The experimental mutual NMR shift ΔΔδN = δNj − δNi minus theoretical mutual NMR shift ΔΔδN = σNi − σNj calculated for nitrogen atoms of OG molecule; Ni, Nj = N1, N2 (NH2), N3, N7, and N9, employing the PW1, PW2, QM1, and QM2 methods (Table 2) in ppm.

a

ΔδN9 employed optimized geometries. The adiabatically averaged ΔδN9 was calculated as averaged adiabatic σN9(χ) dependence with MD calculated χ-probability distribution. The MD averaged ΔδN9 was calculated as statistical average of ΔδN9 shifts calculated for explicit optimized or original MD snapshots. The methods for dynamical averaging of ΔδN9 shift are described in the Supporting Information. The ΔδN9 NMR shifts calculated in liquid varied from −27.75 to −43.47 ppm and the measured ΔδN9 NMR shift was −27.1 ppm (Table 4). The Effect of Solvent. The “static” σN9(OG) calculated neglecting solvent, in FURGAA03 crystal, and including implicit PCM DMSO and PCM water solvent, was 84.49, 82.73, 82.47, and 81.39 ppm, respectively. The “static” σN9(G) shielding calculated neglecting solvent and including PCM DMSO and PCM water was 59.90, 54.36, and 53.40 ppm, respectively. The effects of explicit surrounding of OG in crystal and implicit PCM solvent on σN9(OG) were thus very similar. The inclusion of solvent resulted in decrease of calculated σN9(G) and σN9(OG). The “static” ΔδN9 NMR shift calculated neglecting solvent, including PCM DMSO and PCM water was −24.59, −28.11, and −27.98 ppm, respectively. The inclusion of solvent resulted in better agreement of calculated ΔδN9 NMR shift with experiment.

calculation) was −0.87 and 0.33 ppm, respectively. As compared to PBE calculated ΔδN9(N1) NMR shift, the “net” effect of hybrid B3LYP functional on ΔδN9(N1) NMR shift was −1.20 ppm. The effect of explicit surrounding of OG on ΔδN9(N1) NMR shift was +1.50 ppm since the B3LYP calculated σN1(OG) and σN9(OG) increased by 0.26 and 1.76 ppm, respectively. Considering the dynamical effect on ΔδN9(N1) NMR shift −1.08 ppm (PW-ST and PW-MD calculations, Table 2) for the superior approach (QM2 calculation, Table 2) the resulting ΔδN9(N1) NMR shift was 0.63−1.08 = −0.45 ppm. The other 15N NMR shifts of OG were calculated employing the PW1, PW2, QM1, and QM2 methods described in Table 2 to find out whether the superior performance of B3LYP method is systematic. The calculated mutual ΔΔδN NMR shifts for N1, N2 (NH2), N3, N7, and N9 atoms of the OG molecule were compared with experiment (Table 3). The B3LYP calculated ΔΔδN NMR shifts differed from experiment less than PBE ΔΔδN NMR shifts. The B3LYP geometry optimization with a 6-311++G(d,p) basis followed by a B3LYP NMR calculation employing the Iglo-III basis can be thus regarded as a superior, well-performing method for calculating 15N NMR shifts of OG. III. The N9 NMR Shift of the OG Molecule in Liquid. The dynamical averaging of ΔδN9 NMR shift (eq 1) included gradually effects of rotation of 9-ethyl group, N9-pyramidalization, and all vibrations/rotations. The “static” calculation of F

DOI: 10.1021/acs.jpcb.5b11428 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

effect of adiabatic averaging on ΔδN9 was almost independent of atomic basis because the curvatures σN9(χ) dependences calculated with Iglo-III and cc-pVTZ basis near minima were very similar. The optimal calculation strategy thus may employ large atomic basis for “static” NMR calculation and relatively moderate atomic basis for adiabatic averaging. The effect of atomic basis on curvature of σN9(χ) dependence can be numerically tested only for area on σN9(χ) that is actually “probed” with χ-distribution. The numerical stability of χ-distribution with respect to number of MD snapshots was demonstrated with adiabatically averaged σN9(OG) employing χ-distributions calculated for 200, 500, 1000, 2000, and 4000 MD snapshots. The shape of χdistributions seemed to be stable already for 500 snapshots (Figure S6). The respective averaged σN9(OG) varied from 82.80 to 82.85 ppm. The χ-distribution was also stable with regard to solvent employed in MD calculation since the adiabatic σN9(OG) calculated with explicit TIP3P water and explicit DMSO differed only by 0.01 ppm. The MD Averaged ΔδN9 NMR Shift. The averaged ΔδN9 shift calculated with optimized snapshots included effect of rotation of 9-ethyl group and N9-pyramidalization. The averaged ΔδN9 shift calculated with original MD snapshots included effect of all vibration motions. The averaged σN9(OG) was particularly affected by geometry optimization. The σN9(G) and σN9(OG) calculated with relaxed and original snapshots differed by 0.42 and 15.43 ppm, respectively. The ΔδN9 shift averaged with optimized and original snapshot was −28.46 and −43.47 ppm, respectively. The σN9(OG) averaged with original snapshots was affected primarily by local geometry since the “static” σN9(OG) calculated for OG geometry optimized with GAFF was 102.05 ppm. The “static” σN9(OG) calculated for OG geometry optimized with ff9964,65 force field was 97.64 ppm. A closer look on the bonds involving N9 nitrogen atom of OG unveiled that N9−C8 distance optimized with GAFF (1.343 Å) and ff99 (1.360 Å) was notably shorter than N9−C8 optimized with B3LYP (1.412 Å, Table 2) or resolved with Xray diffraction (1.390 Å, Table 2). The MD averaged ΔδN9 shift calculated employing original MD snapshots thus deviated from experiment because the force field failed to describe correctly geometry of the five-membered ring of OG. The geometry optimization of MD snapshots improved convergences of ΔδN9 NMR shifts (Figures 7 and 8). The standard deviations of the mean (SM) calculated for σN9(G) and σN9(OG) with optimized and original snapshots was smaller

Table 4. Calculated and Measured NMR Shifts of N9 Nitrogen in G and OG Moleculea calculations method

b

static QM adiabatic-MD MD-relaxed MD-original

experiment

σN9(G) 54.36 55.08 55.95 ± 0.12 56.37 ± 0.53

σN9(OG) 82.47 82.83 84.41 ± 0.09 99.84 ± 0.45 experiment

ΔδN9 −28.11 −27.75 −28.46 ± 0.21 −43.47 ± 0.98

δN9(G)

δN9(OG)

ΔδN9

165.8

138.7

−27.1

The calculated NMR shielding σN9, relative NMR shift ΔδN9 (eq 1), and measured δN9 NMR shift in DMSO with respect to nitromethane (δR = 381.7 ppm) in ppm. bThe calculations employing B3LYP method, Iglo-III basis, and PCM DMSO solvent for geometries optimized with B3LYP method, 6-311++G(d,p) basis and PCM DMSO. Static QM: NMR calculation for energy minimum, AdiabaticMD: adiabatic σN9(χ) dependence averaged employing χ-distribution, MD-relaxed/original: statistical average of σN9 calculated for 200 geometry optimized/original MD snapshots. The errors were calculated as standard deviation of the mean. a

The Adiabatically Averaged ΔδN9 NMR Shift. The adiabatically averaged σN9(G) and σN9(OG) increased by 0.72 and 0.36 ppm as compared to “static” σN9 values. The dynamical increase of σN9 can be assumed since minima on σN9(χ) dependences (Figure 6) and maxima on χ-distributions (Figure 3) coincided. The dynamical averaging thus inevitably included larger σN9 values as compared to minima on σN9(χ) dependence and this was why averaged σN9 increased. The sharper χ-distribution was, the larger was statistical weight of minima on calculated σN9(χ) dependence. The adiabatic σN9(G) increased more than σN9(OG) because χ-distribution calculated for G was relatively wider as compared to χdistribution for OG (Figure 3). The decrease of adiabatic ΔδN9 shift by 0.36 ppm relative to “static” ΔδN9 was thus caused by different rotation flexibility of 9-ethyl group in G and OG. The σN9(χ) dependence was calculated with Iglo-III basis. The basis set effect on “static” ΔδN9 shift was calculated (Supporting Information). For adiabatic averaging of ΔδN9 we compared performances of Iglo-III and cc-pVTZ bases. The adiabatic Iglo-III ΔδN9 shift was larger than cc-pVTZ ΔδN9 shift by 0.26 ppm. The “static” Iglo-III ΔδN9 shift was larger than ccpVTZ ΔδN9 shift by 0.32 ppm (Table S1). Interestingly, the

Figure 6. Dependence of σN9 NMR shielding on χ torsion angle calculated for adiabatically optimized geometries of G and OG molecules (full symbols) and correlation of σN9 NMR shielding with χ torsion angle calculated for geometry-optimized MD snapshots (open symbols). G

DOI: 10.1021/acs.jpcb.5b11428 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

Figure 7. Statistical averaging of σN9(G) NMR shielding calculated for MD snapshots. (A) Values of σN9(G) shielding for optimized snapshots (open circle) and the cumulative average of σN9(G) (full circle). (B) Cumulative standard deviation of the mean sM for σN9(G) employing optimized snapshots. (C) Cumulative average of σN9(G) employing optimized (full circle) and original (open circle) snapshots. (D) Cumulative sM for σN9(G) employing optimized (full circle) and original (open circle) snapshots.

shielding was 0.87 and 1.58 ppm, respectively. The MD averaged ΔδN9 NMR shift calculated with optimized snapshots therefore decreased as compared to adiabatically averaged ΔδN9 NMR shift by 0.36 ppm owing to N9-pyramidalization although the overall N9-pyramidalization of G and OG was zero.

than 0.12 and 0.53 ppm, respectively. The cumulative average of σN9 converged faster with optimized snapshots although the σN9 values calculated for snapshots varied notably (Figures 7 and 8). The utilization of original snapshots resulted in variation of cumulative average of σN9 even for last of 200 snapshots and SM error was therefore relatively large. The erroneous deviation of MD averaged σN9(OG) owing to incorrectly described local geometry was evident (Figure 8C). The averaged ΔδN9 calculated with optimized snapshots included dynamical effects of rotation of 9-ethyl group and N9pyramidalization. The larger was amplitude of rotation of 9ethyl, the larger was dynamical increase of σN9 as was evidenced for adiabatically averaged σN9. On the contrary, the larger N9pyramidalization of OG resulted in relative decrease of σN9(OG) as was evidenced by PW2 and PW3 calculations (FURGAA03 and JICALI, Table 2). The calculated σN9(OG) in JICLAI was smaller by 15.09 ppm than σN9(OG) in FURGAA03. The κ′ torsion of OG in FURGAA03 and JICLAI was −0.3° and 28.7°, respectively. To find out how much the two structural effects compensate each other the magnitude of MD averaged σN9 was decomposed. The contribution to σN9 owing to N9-pramidalization was calculated as σN9 for snapshot minus σN9 calculated with the eq 3 for snapshot geometry (Figure S7). The correlation of contribution to σN9 owing to N9-pyramidalization on κ′ torsion was rather indistinct probably because overall N9-pramidalization was zero. Nevertheless, the eq 3 is accurate. The sum of contributions in Figure S7 for σN9(G) and σN9(OG) was 0.90 and 1.55 ppm, respectively. The increase of MD averaged σN9(G) and σN9(OG) shielding as compared to adiabatic σN9 NMR



DISCUSSION The calculated N9 NMR shift of OG molecule was affected by choice of NMR reference, solvent, and structure and dynamics of the molecule. The choice of relevant NMR reference was essential for unbiased calculation of 15N NMR shifts of OG. The calculated and measured δN9 NMR shifts of OG referenced to nitromethane were −261.27 and −243.0 ppm, respectively. The calculated and measured ΔδN9(N1) NMR shifts of OG referenced to N1 nitrogen of OG were +0.63 and −0.90 ppm, respectively. The ΔδN9(N1) NMR shift of OG calculated in crystal including dynamical correction was −0.45 ppm. The B3LYP method with Iglo-III basis used in NMR calculations for OG optimized with B3LYP method and 6-311++G(d,p) basis was superior method for calculating 15N NMR shifts of the OG molecule in the FURGA03 crystal. The NMR calculations employing the FURGAA03 crystal structure enabled reliable comparison of B3LYP and PBE functional and quantification of effects of solvent, local geometry, and molecular dynamics. The B3LYP 15N NMR shifts deviated from experiment less than the PBE NMR shifts. The B3LYP geometry optimization improved accuracy of calculated 15N NMR shifts as compared to NMR shifts H

DOI: 10.1021/acs.jpcb.5b11428 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

Figure 8. Statistical averaging of σN9(OG) NMR shielding calculated for MD snapshots. (A) Values of σN9(OG) shielding for optimized snapshots (open square) and the cumulative average of σN9(OG) (full square). (B) Cumulative standard deviation of the mean sM for σN9(OG) employing optimized snapshots. (C) Cumulative average of σN9(OG) employing optimized (full square) and original (open square) snapshots. (D) Cumulative sM for σN9(OG) employing optimized (full square) and original (open square) snapshots.

16.37 ppm because the geometry of the OG molecule described with GAFF was incorrect. The oxidative damage to G resulted in a measured ΔδN9 NMR shift of −27.1 ppm in liquid. The oxidative damage to 2′deoxyguannosine resulted in a measured ΔδN9 NMR shift of −30.4 ppm.20 The method for calculation of the ΔδN9 NMR shift of OG should be therefore employable also for structural interpretation of 15N NMR shifts of damaged DNA. The calculation method that was benchmarked in this work should therefore guarantee reliable interpretation of 15N NMR shifts of damaged DNA substrate deposited within the hOGG1 catalytic site.

calculated for OG geometry optimized with PBE or original Xray geometry. The ΔδN9 NMR shifts of OG calculated with B3LYP method and atomic bases of Pople’s, Kutzelnigg’s Iglo, and Dunning’s cc-pV and aug-cc-pV series differed by up to 3.22 ppm. The ΔδN9(N1) NMR shifts of OG calculated with the B3LYP or PBE method changed by 1.20 ppm. The effect of explicit surrounding of the OG molecule in crystal on the ΔδN9(N1) NMR shift was +1.50 ppm. The effect of molecular dynamics on the ΔδN9(N1) NMR shift in crystal was −1.08 ppm. The dynamical averaging of the ΔδN9 NMR shift in liquid applied particularly to rotation of the 9-ethyl group since this rotation motion was largest of all the vibration/rotation motions in the OG molecule. The ΔδN9 NMR shift calculated neglecting dynamical effects was smaller than the measured ΔδN9 NMR shift by 1.01 ppm. The adiabatically averaged ΔδN9 NMR shift including effect of 9-ethyl rotation was smaller than the measured ΔδN9 NMR shift by 0.65 ppm. The dynamical increase of the adiabatic ΔδN9 NMR shift by 0.36 ppm was caused by different flexibility of the 9-ethyl group in G and OG. The MD averaged ΔδN9 NMR shift including effects of 9-ethyl rotation and N9-pyramidalization was smaller than the measured ΔδN9 NMR shift by 1.36 ppm. The calculated ΔδN9 of OG decreased by 20.53 ppm when the geometry of N9 nitrogen changed from planar (FURGAA03, κ′ = −0.3°) to pyramidal (JICLAI, κ′ = 28.7°). The experimental detection of pyramidal N9 nitrogen with 15N NMR spectroscopy thus might be possible providing that the pyramidal geometry of inherently planar N9 nitrogen is stabilized by some external factors. The MD averaged ΔδN9 NMR shift calculated with original snapshots was smaller than the measured ΔδN9 NMR shift by



CONCLUSIONS The effects of NMR reference, DFT functional, basis set, solvent, structure, and dynamics on calculated 15N NMR shifts of the OG molecule were evaluated, and the calculated 15N NMR shifts were compared with the NMR shifts of OG molecule measured in crystal and in liquid. The calculated N9 NMR shift of OG molecule deviated from experiment in crystal and in liquid by 0.45 and 0.65 ppm, respectively. The pyramidal geometry of N9 nitrogen in oxidatively damaged DNA nucleoside should be therefore detectable with 15N NMR spectroscopy since the sizable N9-pyramidalization of OG molecule captured in JICLAI crystal caused change of the calculated N9 NMR shift by 20.53 ppm. Experimental detection of pyramidal N9 nitrogen might be possible only when surrounding it during the NMR experiment stabilizes the pyramidal structure of the DNA residue. I

DOI: 10.1021/acs.jpcb.5b11428 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B



deoxyguanosine: Insights from First Principles Calculations. Theor. Chem. Acc. 2011, 128, 619−626. (11) Kellie, J. L.; Wetmore, S. D. Mechanistic and Conformational Flexibility of the Covalent Linkage Formed During Beta-Lyase Activity on an AP-Site: Application to hOgg1. J. Phys. Chem. B 2012, 116, 10786−10797. (12) Š ebera, J.; Trantírek, L.; Tanaka, Y.; Sychrovský, V. Pyramidalization of the Glycosidic Nitrogen Provides the Way for Efficient Cleavage of the N-Glycosidic Bond of 8-OxoG with the hOGG1 DNA Repair Protein. J. Phys. Chem. B 2012, 116, 12535− 12544. (13) Sadeghian, K.; Ochsenfeld, C. Unraveling the Base Excision Repair Mechanism of Human DNA Glycosylase. J. Am. Chem. Soc. 2015, 137, 9824−9831. (14) Kellie, J. L.; Wilson, K. A.; Wetmore, S. D. An ONIOM and MD Investigation of Possible Monofunctional Activity of Human 8Oxoguanine-DNA Glycosylase (hOgg1). J. Phys. Chem. B 2015, 119, 8013−8023. (15) Šebera, J.; Trantírek, L.; Tanaka, Y.; Nencka, R.; Fukal, J.; Sychrovský, V. The Activation of N-glycosidic Bond Cleavage Performed by Base-excision Repair Enzyme hOGG1; Theoretical Study of the Role of Lys 249 Residue in Activation of G, OxoG and FapyG. RSC Adv. 2014, 4, 44043−44051. (16) Lam, S. L.; Chi, L. M. Use of Chemical Shifts for Structural Studies of Nucleic Acids. Prog. Nucl. Magn. Reson. Spectrosc. 2010, 56, 289−310. (17) Lukin, M.; de los Santos, C. NMR Structures of Damaged DNA. Chem. Rev. 2006, 106, 607−686. (18) Dračínský, M.; Pohl, R. NMR Studies of Purines. Annu. Rep. NMR Spectrosc. 2014, 82, 59−113. (19) Dračínský, M.; Hodgkinson, P. Solid-state NMR Studies of Nucleic Acid Components. RSC Adv. 2015, 5, 12300−12310. (20) Cho, B. P.; Kadlubar, F. F.; Culp, S. J.; Evans, F. E. 15N NuclearMagnetic-Resonance Studies on the Tautomerism of 8-hydroxy-2′Deoxyguanosine, 8-Hydroxyguanosine, and Other C8-Substituted Guanine Nucleosides. Chem. Res. Toxicol. 1990, 3, 445−452. (21) Culp, S. J.; Cho, B. P.; Kadlubar, F. F.; Evans, F. E. Structural and Conformational Analyses of 8-Hydroxy-2′-Deoxyguanosine. Chem. Res. Toxicol. 1989, 2, 416−422. (22) Uesugi, S.; Ikehara, M. Carbon-13 Magnetic Resonance Spectra of 8-Substituted Purine Nucleosides − Characteristic Shifts for the Syn Conformation. J. Am. Chem. Soc. 1977, 99, 3250−3253. (23) Sychrovský, V.; Šponer, J.; Hobza, P. Theoretical Calculation of the NMR spin-spin Coupling Constants and the NMR Shifts Allow Distinguishability Between the Specific Direct and the Water-mediated Binding of a Divalent Metal Cation to Guanine. J. Am. Chem. Soc. 2004, 126, 663−672. (24) Sychrovský, V.; Müller, N.; Schneider, B.; Smrecki, V.; Špirko, V.; Šponer, J.; Trantírek, L. Sugar Pucker Modulates the Crosscorrelated Relaxation Rates Across the Glycosidic Bond in DNA. J. Am. Chem. Soc. 2005, 127, 14663−14667. (25) Brumovská, E.; Sychrovský, V.; Vokácǒ vá, Z.; Šponer, J.; Schneider, B.; Trantírek, L. Effect of Local Sugar and Base Geometry on C-13 and N-15 Magnetic Shielding Anisotropy in DNA Nucleosides. J. Biomol. NMR 2008, 42, 209−223. (26) Vokácǒ vá, Z.; Šponer, J.; Šponer, J. E.; Sychrovský, V. Theoretical Study of the Scalar Coupling Constants Across the Noncovalent Contacts in RNA Base Pairs: The cis- and trans-WatsonCrick/sugar Edge Base Pair Family. J. Phys. Chem. B 2007, 111, 10813−10824. (27) Fonville, J. M.; Swart, M.; Vokácǒ vá, Z.; Sychrovský, V.; Šponer, J. E.; Šponer, J.; Hilbers, C. W.; Bickelhaupt, F. M.; Wijmenga, S. S. Chemical Shifts in Nucleic Acids Studied by Density Functional Theory Calculations and Comparison with Experiment. Chem. - Eur. J. 2012, 18, 12372−12387. (28) Dairaku, T.; Furuita, K.; Sato, H.; Šebera, J.; Yamanaka, D.; Otaki, H.; Kikkawa, S.; Kondo, Y.; Katahira, R.; Bickelhaupt, F. M.; Guerra, C. F.; et al. Direct Detection of the Mercury-nitrogen Bond in

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcb.5b11428. Description of chemical synthesis of 9-ethyl-8-oxoguanine and 9-ethylguanine, computational details including basis set effect and effect of solvent on NMR shift, description of methods for dynamical averaging of NMR shift and procedure of numerical fitting of eq 3, experimental 15N liquid-state NMR spectra of OG molecule, an ORTEP view of the OG molecule, and complete refs 28, 49, and 51 (PDF)



AUTHOR INFORMATION

Corresponding Author

*(V.S.) ́ Telephone: 00420 220183234. E-mail: vladimir. [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the Czech Science Foundation GA Č R; by Grant Number 13-27676S to V.S., and Grant Number 15-11223S to M.D. Y.T. and V.S. acknowledge the Young Investigator’s Grant of the Human Frontier Science Program (HFSP) RGY0082/2008. Computational resources were provided by the MetaCentrum under Program LM2010005 and the CERIT-SC under Program Centre CERIT Scientific Cloud, part of the Operational Program Research and Development for Innovations, Reg. No. CZ.1.05/3.2.00/ 08.0144.



REFERENCES

(1) Kuchino, Y.; Mori, F.; Kasai, H.; Inoue, H.; Iwai, S.; Miura, K.; Ohtsuka, E.; Nishimura, S. Misreading of DNA Templates Containing 8-hydroxydeoxyguanosine at the Modified Base and at Adjancent Residues. Nature 1987, 327, 77−79. (2) Kow, Y. W.; Wallace, S. S. Mechanism of Action of EscherichiaColi Endonuclease-III. Biochemistry 1987, 26, 8200−8206. (3) Nash, H. M.; Bruner, S. D.; Scharer, O. D.; Kawate, T.; Addona, T. A.; Spooner, E.; Lane, W. S.; Verdine, G. L. Cloning of a Yeast 8oxoguanine DNA Glycosylase Reveals the Existence of a Base-excision DNA-repair Protein Superfamily. Curr. Biol. 1996, 6, 968−980. (4) Norman, D. P. G.; Chung, S. J.; Verdine, G. L. Structural and Biochemical Exploration of a Critical Amino Acid in Human 8oxoguanine Glycosylase. Biochemistry 2003, 42, 1564−1572. (5) Osakabe, T.; Fujii, Y.; Hata, M.; Tsuda, M.; Neya, S.; Hoshino, T. Quantum Chemical Study on Base Excision Mechanism of 8Oxoguanine DNA Glycosylase. Chem-Bio Inf. J. 2004, 4, 73−92. (6) Schyman, P.; Danielsson, J.; Pinak, M.; Laaksonen, A. Theoretical Study of the Human DNA Repair Protein HOGG1 Activity. J. Phys. Chem. A 2005, 109, 1713−1719. (7) McCann, J. A. B.; Berti, P. J. Transition-state Analysis of the DNA Repair Enzyme MutY. J. Am. Chem. Soc. 2008, 130, 5789−5797. (8) Millen, A. L.; Wetmore, S. D. Glycosidic Bond Cleavage in Deoxynucleotides - A Density Functional Study. Can. J. Chem. 2009, 87, 850−863. (9) Shim, E. J.; Przybylski, J. L.; Wetmore, S. D. Effects of Nucleophile, Oxidative Damage, and Nucleobase Orientation on the Glycosidic Bond Cleavage in Deoxyguanosine. J. Phys. Chem. B 2010, 114, 2319−2326. (10) Rios-Font, R.; Bertran, J.; Sodupe, M.; Rodriguez-Santiago, L. On the Mechanism of the N-glycosydic Bond Hydrolysis of 2′J

DOI: 10.1021/acs.jpcb.5b11428 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

W.; Merz, K. M.; et al. AMBER 10: University of California: San Francisco, 2008. (52) Dejmek, M.; Kovačková, S.; Zborníková, E.; Hřebabecký, H.; Šála, M.; Dračínský, M.; Nencka, R. One-pot Build-up Procedure for the Synthesis of Variously Substituted Purine Derivatives. RSC Adv. 2012, 2, 6970−6980. (53) Č esnek, M.; Holý, A.; Masojídková, M.; Zídek, Z. Synthesis of 9alkyl and 9-Heteroalkyl Substituted 2-amino-6-guanidinopurines and their Influence on the NO-production in Macrophages. Bioorg. Med. Chem. 2005, 13, 2917−2926. (54) Doi, M.; In, Y.; Kikuchi, M.; Ishida, T.; Inoue, M. Multiple BasePairing Mode of 9-Ethyl-8-Hydroxyguanine in 3 Different Crystal Phases. J. Chem. Soc., Perkin Trans. 1 1991, 1, 55−61. (55) CrysAlisPro, O. D., 2002. (56) Altomare, A.; Cascarano, G.; Giacovazzo, G.; Guagliardi, A.; Burla, M. C.; Polidori, G.; Camalli, M. SIR92 - a Program for Automatic Solution of Crystal Structures by Direct Methods. J. Appl. Crystallogr. 1994, 27, 435−435. (57) Betteridge, P. W.; Carruthers, J. R.; Cooper, R. I.; Prout, K.; Watkin, D. J. CRYSTALS Version 12: Software for Guided Crystal Structure Analysis. J. Appl. Crystallogr. 2003, 36, 1487−1487. (58) Farrugia, L. J. WinGX and ORTEP for Windows: an Update. J. Appl. Crystallogr. 2012, 45, 849−854. (59) Marsh, R. E. P1 or P(1)over-bar? Or something else? Acta Crystallogr., Sect. B: Struct. Sci. 1999, 55, 931−936. (60) Sychrovský, V.; Vokácǒ vá, Z. S.; Trantírek, L. Guanine Bases in DNA G-Quadruplex Adopt Nonplanar Geometries Owing to Solvation and Base Pairing. J. Phys. Chem. A 2012, 116, 4144−4151. (61) Sychrovský, V.; Foldýnová-Trantírková, S.; Špačková, N.; Robeyns, K.; Van Meervelt, L.; Blankenfeldt, W.; Vokácǒ vá, Z.; Šponer, J.; Trantírek, L. Revisiting the Planarity of Nucleic Acid Bases: Pyramidilization at Glycosidic Nitrogen in Purine Bases is Modulated by Orientation of Glycosidic Torsion. Nucleic Acids Res. 2009, 37, 7321−7331. (62) van Wullen, C. A Comparison of Density Functional Methods for the Calculation of Phosphorus-31 NMR Chemical Shifts. Phys. Chem. Chem. Phys. 2000, 2, 2137−2144. (63) Dračínský, M.; Hodgkinson, P. A Molecular Dynamics Study of the Effects of Fast Molecular Motions on Solid-state NMR Parameters. CrystEngComm 2013, 15, 8705−8712. (64) Hornak, V.; Abel, R.; Okur, A.; Strockbine, B.; Roitberg, A.; Simmerling, C. Comparison of Multiple Amber Force Fields and Development of Improved Protein Backbone Parameters. Proteins: Struct., Funct., Genet. 2006, 65, 712−725. (65) Perez, A.; Marchan, I.; Svozil, D.; Šponer, J.; Cheatham, T. E.; Laughton, C. A.; Orozco, M. Refinenement of the AMBER Force Field for Nucleic Acids: Improving the Description of Alpha/Gamma Conformers. Biophys. J. 2007, 92, 3817−3829.

the Thymine-Hg-II-thymine Base-pair with Hg-199 NMR Spectroscopy. Chem. Commun. 2015, 51, 8488−8491. (29) Olsson, L.; Cremer, D. Prediction of Nitrogen and Oxygen NMR Chemical Shifts in Organic Compounds by Density Functional Theory. J. Phys. Chem. 1996, 100, 16881−16891. (30) Procházková, E.; Šála, M.; Nencka, R.; Dračínský, M. C6substituted Purine Derivatives: an Experimental and Theoretical 1H, 13 C and 15N NMR Study. Magn. Reson. Chem. 2012, 50, 181−186. (31) Dračínský, M.; Bouř, P. Computational Analysis of Solvent Effects in NMR Spectroscopy. J. Chem. Theory Comput. 2010, 6, 288− 299. (32) Dračínský, M.; Kaminský, J.; Bouř, P. Structure of the Alanine Hydration Shell as Probed by NMR Chemical Shifts and Indirect SpinSpin Coupling. J. Phys. Chem. B 2009, 113, 14698−14707. (33) Dračínský, M.; Möller, H. M.; Exner, T. E. Conformational Sampling by Ab Initio Molecular Dynamics Simulations Improves NMR Chemical Shift Predictions. J. Chem. Theory Comput. 2013, 9, 3806−3815. (34) Becke, A. D. Density-Functional Thermochemistry. 3. The Role of Exact Exchange. J. Chem. Phys. 1993, 98, 5648−5652. (35) Lee, C. T.; Yang, W. T.; Parr, R. G. Development of the ColleSalvetti Correlation-Energy Formula into a Functional of the ElectronDensity. Phys. Rev. B: Condens. Matter Mater. Phys. 1988, 37, 785−789. (36) Hehre, W. J.; Ditchfie, R.; Pople, J. A. Self-Consisitent Molecular-Orbital Methods 0.12. Further Extensions of GaussianType Basis Sets for Use in Molecular-Orbital Studies of OrganicMolecules. J. Chem. Phys. 1972, 56, 2257. (37) Hariharan, P. C.; Pople, J. A. Influence of Polarization Functions on Molecular-Orbital Hydrogenation Energies. Theor. Chim. Acta 1973, 28, 213−222. (38) Tomasi, J.; Mennucci, B.; Cammi, R. Quantum Mechanical Continuum Solvation Models. Chem. Rev. 2005, 105, 2999−3093. (39) Kutzelnigg, W. F.; Fleischer, U.; Schindler, M. NMR Basic Princ. Progr. 1990, 23, 165−262. (40) Wang, J. M.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A. Development and Testing of a General Amber Force Field (vol 25, pg 1157, 2004). J. Comput. Chem. 2005, 26, 114−114. (41) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. Comparison of Simple Potential Functions For Simulating Liquid Water. J. Chem. Phys. 1983, 79, 926−935. (42) Fox, T.; Kollman, P. A. Application of the RESP Methodology in the Parametrization of Organic Solvents. J. Phys. Chem. B 1998, 102, 8070−8079. (43) Yao, J. W.; Cole, J. C.; Pidcock, E.; Allen, F. H.; Howard, J. A. K.; Motherwell, W. D. S. CSD Symmetry: the Definitive Database of Point-group and Space-group Symmetry Relationships in Smallmolecule Crystal Structures. Acta Crystallogr., Sect. B: Struct. Sci. 2002, 58, 640−646. (44) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868. (45) Pickard, C. J.; Mauri, F. All-electron Magnetic Response with Pseudopotentials: NMR Chemical Shifts. Phys. Rev. B: Condens. Matter Mater. Phys. 2001, 63, 245101. (46) Yates, J. R.; Pickard, C. J.; Mauri, F. Calculation of NMR Chemical Shifts for Extended Systems Using Ultrasoft Pseudopotentials. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 76, 024401. (47) Monkhorst, H. J.; Pack, J. D. Special Points for Brillouin-Zone Integrations. Phys. Rev. B 1976, 13, 5188−5192. (48) Vanderbilt, D. Soft Self-Consistent Pseudopotentials in a Generalized Eigenvalue Formalism. Phys. Rev. B: Condens. Matter Mater. Phys. 1990, 41, 7892−7895. (49) Frisch, M. J.; et al. Gaussian 09, revision D.01; Gaussian, Inc.: Wallingford, CT, 2013. (50) Clark, S. J.; Segall, M. D.; Pickard, C. J.; Hasnip, P. J.; Probert, M. J.; Refson, K.; Payne, M. C. First Principles Methods Using CASTEP. Z. Kristallogr. - Cryst. Mater. 2005, 220, 567−570. (51) Case, D. A.; Cheatham, T. E., III; Darden, T. A.; Simmerling, C. L.; Wang, J.; Duke, R. E.; Luo, R.; Crowley, M.; Walker, R. C.; Zhang, K

DOI: 10.1021/acs.jpcb.5b11428 J. Phys. Chem. B XXXX, XXX, XXX−XXX