Martin Dračínský

Computational Analysis of Solvent Effects in NMR Spectroscopy.

Solvent modeling became a standard part of first principles computations of molecular properties. However, a universal solvent approach is particularly difficult for the nuclear magnetic resonance (NMR) shielding and spin-spin coupling constants that in part result from collective delocalized properties of the solute and the environment. In this work, bulk and specific solvent effects are discussed on experimental and theoretical model systems comprising solvated alanine zwitterion and chloroform molecules. Density functional theory computations performed on larger clusters indicate that standard dielectric continuum solvent models may not be sufficiently accurate. In some cases, more reasonable NMR parameters were obtained by approximation of the solvent with partial atomic charges. Combined cluster/continuum models yielded the most reasonable values of the spectroscopic parameters, provided that they are dynamically averaged. The roles of solvent polarizability, solvent shell structure, and bulk permeability were investigated. NMR shielding values caused by the macroscopic solvent magnetizability exhibited the slowest convergence with respect to the cluster size. For practical computations, however, inclusion of the first solvation sphere provided satisfactory corrections of the vacuum values. The simulations of chloroform chemical shifts and CH J-coupling constants were found to be very sensitive to the molecular dynamics model used to generate the cluster geometries. The results show that computationally efficient solvent modeling is possible and can reveal fine details of molecular structure, solvation, and dynamics.

Explicit versus implicit solvent modeling of Raman optical activity spectra.

Raman and Raman optical activity (ROA) spectra of molecules reflect not only molecular structure and conformation but also the dynamics and interactions with the solvent. For polar, biologically relevant molecules in aqueous environment, this often complicates the band assignment and interpretation of the spectra. In the present study, implicit dielectric and explicit solvent models are compared with respect to the influence of the choice of solvent model on the spectral shape. Lactamide and 2-aminopropanol were selected as model compounds, and the Raman and ROA spectra were measured for both enantiomers. Geometries of explicitly solvated clusters were derived from quantum-mechanical calculations, classical (MD), and Car-Parrinello (CPMD) molecular dynamics. The results indicate that although the dielectric model reasonably well reproduces the main spectral features, more faithful intensity profiles, including the inhomogeneous band broadening, are obtained from the explicit MD and CPMD clusters. Additionally, the CPMD clusters are capable of reproducing most spectral features better than the classical dynamics, provided the simulation time is long enough to allow for a complete sampling of the conformational space. The hydrogen-bonded water molecules of the first hydration shell significantly influence the spectral intensities, whereas the effect of loosely attached or distant solvent molecules is minor. In order to average the signal, however, a relatively large number of MD geometries need to be considered, as was also exemplified by simulations of the ROA spectrum of the achiral molecule glycine. An explicit solvent modeling of sizable systems thus requires extensive computations, which became possible only recently due to the development of efficient analytical computational techniques.

A novel and efficient one-pot synthesis of symmetrical diamide (bis-amidate) prodrugs of acyclic nucleoside phosphonates and evaluation of their biological activities

A novel and efficient method for the one-pot synthesis of diamide (bis-amidate) prodrugs of acyclic nucleoside phosphonates, starting from free phosphonic acids or phosphonate diesters is reported. The approach from phosphonate diesters via their bis(trimethylsilyl) esters is highly convenient, eliminates isolation and tedious purification of the phosphonic acids, and affords the corresponding bis-amidates in excellent yields (83-98%) and purity. The methodology has been applied to the synthesis of the potent anticancer agent GS-9219, and symmetrical bis-amidates of other biologically active phosphonic acids. Anti-HIV, antiproliferative, and immunomodulatory activities of the compounds are discussed including the bis-amidate prodrugs 14 and 17 that exhibited anti-HIV activity at submicromolar concentrations with minimal cytotoxicity.

Mammalian peroxidases activate anticancer drug ellipticine to intermediates forming deoxyguanosine adducts in DNA identical to those found in vivo and generated from 12-hydroxyellipticine and 13-hydroxyellipticine

Ellipticine is a potent antineoplastic agent, whose mode of action is considered to be based mainly on DNA intercalation, inhibition of topoisomerase II and cytochrome P450‐mediated formation of covalent DNA adducts. This is the first report on the molecular mechanism of ellipticine oxidation by peroxidases (human myeloperoxidase, human and ovine cyclooxygenases, bovine lactoperoxidase, horseradish peroxidase) to species forming ellipticine‐DNA adducts. Using NMR spectroscopy, the structures of 2 ellipticine metabolites were identified; the major product is the ellipticine dimer, in which the 2 ellipticine skeletons are connected via N6 of the pyrrole ring of one ellipticine molecule and C9 in the second one. The minor metabolite is ellipticine N2‐oxide. Using 32P‐postlabeling and [3H]‐labeled ellipticine, we showed that ellipticine binds covalently to DNA after its activation by peroxidases. The DNA adduct pattern induced by ellipticine consisted of a cluster of up to 4 adducts. The 2 adducts are indistinguishable from the 2 major adducts generated between deoxyguanosine in DNA and either 13‐hydroxy‐ or 12‐hydroxyellipticine or in rats treated with ellipticine, or if ellipticine was activated with human hepatic and renal microsomes. The results presented here are the first characterization of the peroxidase‐mediated oxidative metabolites of ellipticine and we have proposed species, 2 carbenium ions, ellipticine‐13‐ylium and ellipticine‐12‐ylium, as reactive species generating 2 major DNA adducts seen in vivo in rats treated with ellipticine. The study forms the basis to further predict the susceptibility of human cancers to ellipticine.

Structure of the Alanine Hydration Shell as Probed by NMR Chemical Shifts and Indirect Spin-Spin Coupling

The structure of the alanine hydration shell was modeled by Carr-Parinello molecular dynamics (CPMD) to explain subtle differences in NMR chemical shifts and indirect spin-spin coupling constants of the neutral (zwitterionic), cationic, and anionic forms of this amino acid. In comparison with classical molecular dynamics (MD), the quantum mechanical CPMD approach revealed a more structured solvent and significant differences in the radial and angular distributions of the water molecules around the solute. In particular, the solvent was predicted to be organized around the uncharged COOH and NH(2) residues to a similar degree as that for the charged ones. This was not the case with MD. For snapshot CPMD configurations, the NMR parameters were computed by density functional theory (DFT) and averaged. Obtained values were significantly closer to experimental parameters known for (15)N and (13)C isotopically labeled alanine than those calculated by the conventional implicit dielectric solvent model. The NMR results also quantitatively reflect a superiority of the CPMD over the MD explicit solvent treatment. A further improvement of the computed spin-spin coupling constants could be achieved by taking into account vibrational averaging beyond the harmonic approximation. Differently positioned water molecules in the clusters cause an unexpectedly large scattering of the NMR parameters. About 10-15 dynamics snapshots were required for a satisfactory convergence of the shifts and couplings. The NMR chemical shift was found to be much more sensitive to the molecular hydration than the coupling. The results thus indicate a large potential of the NMR spectroscopy and quantum simulations to probe not only the structure of molecules but also their interactions with the environment.

Relative importance of first and second derivatives of nuclear magnetic resonance chemical shifts and spin-spin coupling constants for vibrational averaging

Relative importance of anharmonic corrections to molecular vibrational energies, nuclear magnetic resonance (NMR) chemical shifts, and J-coupling constants was assessed for a model set of methane derivatives, differently charged alanine forms, and sugar models. Molecular quartic force fields and NMR parameter derivatives were obtained quantum mechanically by a numerical differentiation. In most cases the harmonic vibrational function combined with the property second derivatives provided the largest correction of the equilibrium values, while anharmonic corrections (third and fourth energy derivatives) were found less important. The most computationally expensive off-diagonal quartic energy derivatives involving four different coordinates provided a negligible contribution. The vibrational corrections of NMR shifts were small and yielded a convincing improvement only for very accurate wave function calculations. For the indirect spin-spin coupling constants the averaging significantly improved already the equilibrium values obtained at the density functional theory level. Both first and complete second shielding derivatives were found important for the shift corrections, while for the J-coupling constants the vibrational parts were dominated by the diagonal second derivatives. The vibrational corrections were also applied to some isotopic effects, where the corrected values reasonably well reproduced the experiment, but only if a full second-order expansion of the NMR parameters was included. Contributions of individual vibrational modes for the averaging are discussed. Similar behavior was found for the methane derivatives, and for the larger and polar molecules. The vibrational averaging thus facilitates interpretation of previous experimental results and suggests that it can make future molecular structural studies more reliable. Because of the lengthy numerical differentiation required to compute the NMR parameter derivatives their analytical implementation in future quantum chemistry packages is desirable.

Synthesis of 9-phosphonoalkyl and 9-phosphonoalkoxyalkyl purines: evaluation of their ability to act as inhibitors of Plasmodium falciparum, Plasmodium vivax and human hypoxanthine-guanine-(xanthine) phosphoribosyltransferases.

The purine salvage enzyme, hypoxanthine-guanine-(xanthine) phosphoribosyltransferase [HG(X)PRT], catalyses the synthesis of the purine nucleoside monophosphates, IMP, GMP or XMP essential for DNA/RNA production. In protozoan parasites, such as Plasmodium, this is the only route available for their synthesis as they lack the de novo pathway which is present in human cells. Acyclic nucleoside phosphonates (ANPs), analogs of the purine nucleoside monophosphates, have been found to inhibit Plasmodium falciparum (Pf) HGXPRT and Plasmodium vivax (Pv) HGPRT with K(i) values as low as 100 nM. They arrest parasitemia in cell based assays with IC(50) values of the order of 1-10 μM. ANPs with phosphonoalkyl and phosphonoalkoxyalkyl moieties linking the purine base and phosphonate group were designed and synthesised to evaluate the influence of this linker on the potency and/or selectivity of the ANPs for the human and malarial enzymes. This data shows that variability in the linker, as well as the positioning of the oxygen in this linker, influences binding. The human enzyme binds the ANPs with K(i) values of 0.5 μM when the number of atoms in the linker was 5 or 6 atoms. However, the parasite enzymes have little affinity for such long chains unless oxygen is included in the three-position. In comparison, all three enzymes have little affinity for ANPs where the number of atoms linking the base and the phosphonate group is of the order of 2-3 atoms. The chemical nature of the purine base also effects the K(i) values. This data shows that both the linker and the purine base play an important role in the binding of the ANPs to these three enzymes.

Effects of Quantum Nuclear Delocalisation on NMR Parameters from Path Integral Molecular Dynamics

The influence of nuclear delocalisation on NMR chemical shifts in molecular organic solids is explored using path integral molecular dynamics (PIMD) and density functional theory calculations of shielding tensors. Nuclear quantum effects are shown to explain previously observed systematic deviations in correlations between calculated and experimental chemical shifts, with particularly large PIMD-induced changes (up to 23 ppm) observed for carbon atoms in methyl groups. The PIMD approach also enables isotope substitution effects on chemical shifts and J couplings to be predicted in excellent agreement with experiment for both isolated molecules and molecular crystals. An approach based on convoluting calculated shielding or coupling surfaces with probability distributions of selected bond distances and valence angles obtained from PIMD simulations is used to calculate isotope effects.

Cytochrome b5 and epoxide hydrolase contribute to benzo[a]pyrene-DNA adduct formation catalyzed by cytochrome P450 1A1 under low NADPH:P450 oxidoreductase conditions

In previous studies we had administered benzo[a]pyrene (BaP) to genetically engineered mice (HRN) which do not express NADPH:cytochrome P450 oxidoreductase (POR) in hepatocytes and observed higher DNA adduct levels in livers of these mice than in wild-type mice. To elucidate the reason for this unexpected finding we have used two different settings for in vitro incubations; hepatic microsomes from control and BaP-pretreated HRN mice and reconstituted systems with cytochrome P450 1A1 (CYP1A1), POR, cytochrome b5, and epoxide hydrolase (mEH) in different ratios. In microsomes from BaP-pretreated mice, in which Cyp1a1 was induced, higher levels of BaP metabolites were formed, mainly of BaP-7,8-dihydrodiol. At a low POR:CYP1A1 ratio of 0.05:1 in the reconstituted system, the amounts of BaP diones and BaP-9-ol formed were essentially the same as at an equimolar ratio, but formation of BaP-3-ol was ∼ 1.6-fold higher. Only after addition of mEH were BaP dihydrodiols found. Two BaP-DNA adducts were formed in the presence of mEH, but only one when CYP1A1 and POR were present alone. At a ratio of POR:CYP1A1 of 0.05:1, addition of cytochrome b5 increased CYP1A1-mediated BaP oxidation to most of its metabolites indicating that cytochrome b5 participates in the electron transfer from NADPH to CYP1A1 required for enzyme activity of this CYP. BaP-9-ol was formed even by CYP1A1 reconstituted with cytochrome b5 without POR. Our results suggest that in livers of HRN mice Cyp1a1, cytochrome b5 and mEH can effectively activate BaP to DNA binding species, even in the presence of very low amounts of POR.

Design, synthesis, and biological evaluation of novel coxsackievirus B3 inhibitors

The synthesis and SAR study of a novel class of coxsackievirus B3 (CVB3) inhibitors are reported. These compounds could be considered as the 6-chloropurines substituted at position 9 with variously substituted bicyclic scaffolds (bicyclo[2.2.1]heptane/ene-norbornane or norbornene). The synthesis and biological evaluation of 31 target compounds are described. Several of the analogues inhibited CVB3 in the low micromolar range (0.66-2muM). Minimal or no cytotoxicity was observed.