[show abstract][hide abstract] ABSTRACT: Bisphosphonates are a class of molecules in widespread use in treating bone resorption diseases and are also of interest as immunomodulators and anti-infectives. They function by inhibiting the enzyme farnesyl diphosphate synthase (FPPS), but the details of how these molecules bind are not fully understood. Here, we report the results of a solid-state (13)C, (15)N, and (31)P magic-angle sample spinning (MAS) NMR and quantum chemical investigation of several bisphosphonates, both as pure compounds and when bound to FPPS, to provide information about side-chain and phosphonate backbone protonation states when bound to the enzyme. We then used computational docking methods (with the charges assigned by NMR) to predict how several bisphosphonates bind to FPPS. Finally, we used X-ray crystallography to determine the structures of two potent bisphosphonate inhibitors, finding good agreement with the computational results, opening up the possibility of using the combination of NMR, quantum chemistry and molecular docking to facilitate the design of other, novel prenytransferase inhibitors.
Journal of the American Chemical Society 12/2006; 128(45):14485-97. · 10.68 Impact Factor
[show abstract][hide abstract] ABSTRACT: Farnesyl diphosphate synthase (FPPS) catalyses the formation of farnesyl diphosphate from dimethylallyl diphosphate and isopentenyl diphosphate and is an RNAi-validated drug target in Trypanosoma brucei, the causative agent of African sleeping sickness. A T. brucei FPPS (390 amino acids) has been expressed in Escherichia coli and the recombinant protein has been crystallized in the absence and presence of the bisphosphonate inhibitor minodronate. Diffraction data were collected at 100 K using synchrotron radiation from both crystal types. Crystals obtained in the absence of minodronate belong to space group I222, with unit-cell parameters a = 61.43, b = 118.12, c = 120.04 A, while crystals grown in the presence of minodronate belong to space group C2, with unit-cell parameters a = 131.98, b = 118.10, c = 63.25 A, beta = 112.48 degrees. An initial model of the drug-free protein has been built using a homology model with the molecular-replacement method and refined to 3.3 A resolution. It shows mostly helical structure and resembles the structure of avian farnesyl diphosphate synthase, but with the addition of two loop regions.
[show abstract][hide abstract] ABSTRACT: We report the activities of 62 bisphosphonates as inhibitors of the Leishmania major mevalonate/isoprene biosynthesis pathway enzyme, farnesyl pyrophosphate synthase. The compounds investigated exhibit activities (IC(50) values) ranging from approximately 100 nM to approximately 80 microM (corresponding to K(i) values as low as 10 nM). The most active compounds were found to be zoledronate (whose single-crystal X-ray structure is reported), pyridinyl-ethane-1-hydroxy-1,1-bisphosphonates or picolyl aminomethylene bisphosphonates. However, N-alicyclic aminomethylene bisphosphonates, such as incadronate (N-cycloheptyl aminomethylene bisphosphonate), as well as aliphatic aminomethylene bisphosphonates containing short (n = 4, 5) alkyl chains, were also active, with IC(50) values in the 200-1700 nM range (corresponding to K(i) values of approximately 20-170 nM). Bisphosphonates containing longer or multiple (N,N-) alkyl substitutions were inactive, as were aromatic species lacking an o- or m-nitrogen atom in the ring, or possessing multiple halogen substitutions or a p-amino group. To put these observations on a more quantitative structural basis, we used three-dimensional quantitative structure-activity relationship techniques: comparative molecular field analysis (CoMFA) and comparative molecular similarity index analysis (CoMSIA), to investigate which structural features correlated with high activity. Training set results (N = 62 compounds) yielded good correlations with each technique (R(2) = 0.87 and 0.88, respectively), and were further validated by using a training/test set approach. Test set results (N = 24 compounds) indicated that IC(50) values could be predicted within factors of 2.9 and 2.7 for the CoMFA and CoMSIA methods, respectively. The CoMSIA fields indicated that a positive charge in the bisphosphonate side chain and a hydrophobic feature contributed significantly to activity. Overall, these results are of general interest since they represent the first detailed quantitative structure-activity relationship study of the inhibition of an expressed farnesyl pyrophosphate synthase enzyme by bisphosphonate inhibitors and that the activity of these inhibitors can be predicted within about a factor of 3 by using 3D-QSAR techniques.
Journal of Medicinal Chemistry 12/2003; 46(24):5171-83. · 5.61 Impact Factor
[show abstract][hide abstract] ABSTRACT: We report the results of a series of density functional theory (DFT) calculations aimed at predicting the (57)Fe Mössbauer electric field gradient (EFG) tensors (quadrupole splittings and asymmetry parameters) and their orientations in S = 0, (1)/(2), 1, (3)/(2), 2, and (5)/(2) metalloproteins and/or model systems. Excellent results were found by using a Wachter's all electron basis set for iron, 6-311G for other heavy atoms, and 6-31G for hydrogen atoms, BPW91 and B3LYP exchange-correlation functionals, and spin-unrestricted methods for the paramagnetic systems. For the theory versus experiment correlation, we found R(2) = 0.975, slope = 0.99, intercept = -0.08 mm sec(-)(1), rmsd = 0.30 mm sec(-)(1) (N = 23 points) covering a DeltaE(Q) range of 5.63 mm s(-)(1) when using the BPW91 functional and R(2) = 0.978, slope = 1.12, intercept = -0.26 mm sec(-)(1), rmsd = 0.31 mm sec(-)(1) when using the B3LYP functional. DeltaE(Q) values in the following systems were successfully predicted: (1) ferric low-spin (S = (1)/(2)) systems, including one iron porphyrin with the usual (d(xy))(2)(d(xz)d(yz))(3) electronic configuration and two iron porphyrins with the more unusual (d(xz)d(yz))(4)(d(xy))(1) electronic configuration; (2) ferrous NO-heme model compounds (S = (1)/(2)); (3) ferrous intermediate spin (S = 1) tetraphenylporphinato iron(II); (4) a ferric intermediate spin (S = (3)/(2)) iron porphyrin; (5) ferrous high-spin (S = 2) deoxymyoglobin and deoxyhemoglobin; and (6) ferric high spin (S = (5)/(2)) metmyoglobin plus two five-coordinate and one six-coordinate iron porphyrins. In addition, seven diamagnetic (S = 0, d(6) and d(8)) systems studied previously were reinvestigated using the same functionals and basis set scheme as used for the paramagnetic systems. All computed asymmetry parameters were found to be in good agreement with the available experimental data as were the electric field gradient tensor orientations. In addition, we investigated the electronic structures of several systems, including the (d(xy))(2)(d(xz),d(yz))(3) and (d(xz),d(yz))(4)(d(xy))(1) [Fe(III)/porphyrinate](+) cations as well as the NO adduct of Fe(II)(octaethylporphinate), where interesting information on the spin density distributions can be readily obtained from the computed wave functions.
Journal of the American Chemical Society 12/2002; 124(46):13921-30. · 10.68 Impact Factor
[show abstract][hide abstract] ABSTRACT: We report the first detailed investigation of the (1)H, (13)C, (15)N, and (19)F nuclear magnetic resonance (NMR) spectroscopic shifts in paramagnetic metalloprotein and metalloporphyrin systems. The >3500 ppm range in experimentally observed hyperfine shifts can be well predicted by using density functional theory (DFT) methods. Using spin-unrestricted methods together with large, locally dense basis sets, we obtain very good correlations between experimental and theoretical results: R(2) = 0.941 (N = 37, p < 0.0001) when using the pure BPW91 functional and R(2) = 0.981 (N = 37, p < 0.0001) when using the hybrid functional, B3LYP. The correlations are even better for C(alpha) and C(beta) shifts alone: C(alpha), R(2) = 0.996 (N = 8, p < 0.0001, B3LYP); C(beta), R(2) = 0.995 (N = 8, p < 0.0001, B3LYP), but are worse for C(meso), in part because of the small range in C(meso) shifts. The results of these theoretical calculations also lead to a revision of previous heme and proximal histidine residue (13)C NMR assignments in deoxymyoglobin which are confirmed by new quantitative NMR measurements. Molecular orbital (MO) analyses of the resulting wave functions provide a graphical representation of the spin density distribution in the [Fe(TPP)(CN)(2)](-) (TPP = 5,10,15,20-tetraphenylporphyrinato) system (S = (1)/(2)), where the spin density is shown to be localized primarily in the d(xz) (or d(yz)) orbital, together with an analysis of the frontier MOs in Fe(TPP)Cl (S = (5)/(2)), Mn(TPP)Cl (S = 2), and a deoxymyoglobin model (S = 2). The ability to now begin to predict essentially all heavy atom NMR hyperfine shifts in paramagnetic metalloporphyrins and metalloproteins using quantum chemical methods should open up new areas of research aimed at structure prediction and refinement in paramagnetic systems in much the same way that DFT methods have been used successfully in the past to predict/refine elements of diamagnetic heme protein structures.
Journal of the American Chemical Society 11/2002; 124(46):13911-20. · 10.68 Impact Factor
[show abstract][hide abstract] ABSTRACT: We report the results of density functional theory (DFT) calculations of the (57)Fe Mössbauer isomer shifts (delta(Fe)) for a series of 24 inorganic, organometallic, and metalloprotein/metalloporphyrin model systems in S = 0, (1)/(2), 1, (3)/(2), 2, and (5)/(2) spin states. We find an excellent correlation between calculation and experiment over the entire 2.34 mm s(-1) range of isomer shifts: a 0.07-0.08 mm s(-1) rms deviation between calculation and experiment (corresponding to 3-4% of the total delta(Fe) range, depending on the functionals used) with R(2) values of 0.973 and 0.981 (p < 0.0001). The best results are obtained by using the hybrid exchange-correlation functional B3LYP, used previously for (57)Fe Mössbauer quadrupole splittings and (57)Fe NMR chemical shifts and chemical shielding anisotropies. The relativistically corrected value of alpha, alpha(rel), converges with the large basis set used in this work, but the exact values vary somewhat with the methods used: -0.253 a(0)(3) mm s(-1) (Hartree-Fock; HF); -0.316 a(0)(3) mm s(-1) (hybrid HF-DFT; B3LYP), or -0.367 a(0)(3) mm s(-1) (pure DFT; BPW91). Both normal and intermediate spin state isomer shifts are well reproduced by the calculations, as is the broad range of delta(Fe) values: from [Fe(VI)O(4)](2-) (-0.90 mm s(-1) expt; -1.01 mm s(-1) calc) to KFe(II)F(3) (1.44 mm s(-1) expt; 1.46 mm s(-1) calc). Molecular orbital analyses of all inorganic solids as well as all organometallic and metalloporphyrin systems studied reveal that there are three major core MO contributions to rho(tot)(0), the total charge density at the iron nucleus (and hence delta(Fe)), that do not vary with changes in chemistry, while the valence MO contributions are highly correlated with delta(Fe) (R(2) = 0.915-0.938, depending on the functionals used), and the correlation between the valence MO contributions and the total MO contribution is even better (R(2) = 0.965-0.976, depending on the functionals used). These results are of general interest since they demonstrate that DFT methods now enable the accurate prediction of delta(Fe) values in inorganic, organometallic, and metalloporphyrin systems in all spin states and over a very wide range of delta(Fe) values with a very small rms error.
Journal of the American Chemical Society 07/2002; 124(26):7829-39. · 10.68 Impact Factor
[show abstract][hide abstract] ABSTRACT: By using density functional theory it is demonstrated that the long-range 19F−19F J-couplings (>3J) seen in small organic molecules can be calculated with good accuracy using small molecule fragments and in some cases complete molecules. The results reproduce the exponential distance dependence of J seen experimentally, demonstrate the dominance of the Fermi contact interaction, and rule out any significant covalent or through-bond contributions to J in these systems. The calculations also verify an experimentally observed 19F−19F J-coupling seen between two [6-F]Trp residues in the protein dihydrofolate reductase (for d = 2.98 Å), where there is clearly no covalent bonding between the two 19F sites. The results also clarify the abnormally small J-couplings seen previously in phenanthrenes and cyclohexenes, which are shown by ab initio and molecular mechanics geometry optimizations to be due to conversion of the supposedly planar structures to more distorted but less sterically hindered structures. These distortions increase the F−F distance and thereby reduce JFF. The lack of any appreciable covalent bonding between the 19F atoms in both the protein and the model systems, but the presence of significant J-couplings, emphasizes that all that is required is Fermi contact, and the close spatial proximity of atoms. This result is of considerable current interest in the context of (long range/through-space) hydrogen bond J-couplings in macromolecules.
Journal of The American Chemical Society - J AM CHEM SOC. 01/2000; 122(49):12164-12168.