Improving the performance of SOMFA by use of standard multivariate methods.

Self-Organizing Molecular Field Analysis (SOMFA) comes with a built-in regression methodology, the Self-Organizing Regression (SOR), instead of relying on external methods such as PLS. In this article we present a proof of the equivalence between SOR and SIMPLS with one principal component. Thus, the modest performance of SOMFA on complex datasets can be primarily attributed to the low performance of the SOMFA regression methodology. A multi-component extension of the original SOR methodology (MCSOR) is introduced, and the performances of SOR, MCSOR and SIMPLS are compared using several datasets. The results indicate that in general the performance of SOMFA models is greatly improved if SOR is replaced with a more sophisticated regression method. The results obtained for the Cramer (CBG) dataset further underline the fact that it is a very poor benchmark dataset and should not be used to evaluate the performance of QSAR techniques.

Energetic analysis of binding of progesterone and 5 beta-androstane-3,17-dione to anti-progesterone antibody DB3 using molecular dynamics and free energy calculations.

Molecular dynamics simulations and molecular mechanics-Poisson-Boltzmann surface area (MM-PBSA) free energy calculations were used to study the energetics of the binding of progesterone (PRG) and 5 beta-androstane-3,17-dione (5AD) to anti-PRG antibody DB3. Although the two steroids bind to DB3 in different orientations, their binding affinities are of the same magnitude, 1 nM for PRG and 8 nM for 5AD. The calculated relative binding free energy of the steroids, 8.8 kJ/mol, is in fair agreement with the experimental energy, 5.4 kJ/mol. In addition, computational alanine scanning was applied to study the role of selected amino acid residues of the ligand-binding site on the steroid cross-reactivity. The electrostatic and van der Waals components of the total binding free energies were found to favour more the binding of PRG, whereas solvation energies were more favourable for the binding of 5AD. The differences in the free energy components are due to the binding of the A rings of the steroids to different binding pockets: PRG is bound to a pocket in which electrostatic antibody-steroid interactions are dominating, whereas 5AD is bound to a pocket in which van der Waals and hydrophobic interactions dominate.

Elucidating the nature of enzyme catalysis utilizing a new twist on an old methodology: quantum mechanical-free energy calculations on chemical reactions in enzymes and in aqueous solution.

How do enzymes achieve very large rate enhancements compared to corresponding uncatalyzed reactions in solution? We present a computational approach which combines high-level ab initio quantum mechanical calculations with classical free energy calculations to address this question. Our calculations lead to accurate estimates of DeltaG for both trypsin and catechol O-methyltransferase-catalyzed and reference uncatalyzed reactions and give new insights into the nature of enzyme catalysis. The same methodology applied to steps in the catalytic mechanism of citrate synthase further supports the conclusion that one need not invoke special concepts such as "low-barrier hydrogen bonds" or "pK(a) matching" to explain enzyme catalysis.

Inversion of the roles of the nucleophile and acid/base catalysts in the covalent binding of epoxyalkyl xyloside inhibitor to the catalytic glutamates of endo-1,4-beta-xylanase (XYNII): a molecular dynamics study.

X-Ray crystal structures have revealed that 2, 3-epoxypropyl-beta-D-xyloside reacts with endo-1,4-beta-xylanase (XYNII) by forming a covalent bond with Glu86. In contrast, 3, 4-epoxybutyl-beta-D-xyloside forms a covalent bond with Glu177. In the normal enzyme reaction Glu86 acts as the catalytic nucleophile and Glu177 as the acid/base catalyst. To rationalize the observed reactivity of the two mechanism-based inhibitors, we carried out eight 300 ps molecular dynamics simulations for different enzyme-inhibitor complexes. Simulations were done for both stereo isomers (R and S) of the inhibitors and for enzyme in which the protonation state of the nucleophile and acid/base catalyst was normal (Glu86 charged, Glu177 neutral) and in which the roles of the catalytic residues were reversed (Glu86 neutral, Glu177 charged). The number of reactive conformations found in each simulation was used to predict the reactivity of epoxy inhibitors. The conformation was considered to be a reactive one when at the same time (i) the proton of the catalytic acid was close (<2.9/3.4/3.9 A) to the oxirane oxygen of the inhibitor, (ii) the nucleophile was close to the terminal carbon of the oxirane group (<3.4/3.9/4.4 A) and (iii) the nucleophile approached the terminal carbon from a reactive angle (<30/45/60 degrees from an ideal attack angle). On the basis of the number of reactive conformations, 2,3-epoxypropyl-beta-D-xyloside was predicted to form a covalent bond with Glu86 and 3, 4-epoxybutyl-beta-D-xyloside with Glu177, both in agreement with the experiment. Thus, the MD simulations and the X-ray structures indicate that in the covalent binding of 3, 4-epoxybutyl-beta-D-xyloside the roles of the catalytic glutamates of XYNII are reversed from that of the normal enzyme reaction.

Folding of alpha(r)beta and epsilonbeta reverse turns; a nanosecond molecular dynamics simulation of the hexapeptide MSALNT and the octapeptide NMSALNTL in water.

Folding of the hexapeptide MSALNT and the octapeptide NMSALNTL were investigated using 2.8 ns molecular dynamics (MD) simulations in aqueous solution. In the simulation, the central sequence SALN of the hexapeptide folded rapidly within 200 ps into an alpha(r)beta turn conformation (type VIII conformation) and remained in this conformation for the rest of the trajectory. The sequence SALN of the octapeptide needed 2 ns to fold via epsilonbeta conformations into a similar conformation. The results join the sequences into a growing group of sequences which have a tendency to form secondary structures and thereby to direct protein folding. The structures of the reverse turn conformations were in accordance with the experimental results (Hakalehto et al., Eur J. Biochem. 250, 19-29 (1997)). The main driving force of folding seems to be the hydrophobic interaction between the side chains of Ala and Leu at the i+1 and i+2 positions of the beta-turn.

Functional conformational changes of endo-1,4-xylanase II from Trichoderma reesei: a molecular dynamics study.

Recent crystallographic studies have revealed a range of structural changes in the three-dimensional structure of endo-1,4-xylanase (XYNII) from Trichoderma reesei. The observed conformational changes can be described as snapshots of an open-close movement of the active site of XYNII. These structures were further analyzed in this study. In addition, a total of four 1 ns molecular dynamics (MD) simulations were performed representing different states of the enzyme. A comparison of the global and local changes found in the X-ray structures and the MD runs suggested that the simulations reproduced a similar kind of active site opening and closing as predicted by the crystal structures. The open-close movement was characterized by the use of distance difference matrixes and the Hinge-find program (Wriggers and Schulten, Proteins 29:1-14, 1997) to be a 'hinge-bending' motion involving two large rigidly-moving regions and an extended hinge. This conformational feature is probably inherent to this molecular architecture and probably plays a role in the function of XYNII.

Ab initio models for receptor-ligand interactions in proteins. 4. Model assembly study of the catalytic mechanism of triosephosphate isomerase.

The catalytic mechanism of triosephosphate isomerase (TIM) was investigated with ab initio quantum mechanical calculations. Electrostatic interactions between the quantum mechanical active site and the protein and solvent environment were modeled using the finite difference Poission-Boltzman method. The complexes of TIM with the substrate dihydroxyacetone phosphate (DHAP), five possible intermediates and the product glyceraldehyde-3-phosphate (GAP) were optimized in the active-site model at the 3-21G(*) level and energy profile for the proton abstraction from DHAP by the active-site Glu 167 was calculated at the MP2/3-21G(*)13-21G(*) level. Calculated energetics of the enzyme reaction were found to be in reasonable agreement with the experimental findings. Calculations revealed that an enediol of the substrate is a probable intermediate in the enzyme reaction. It was suggested that the proton abstracted from the substrate by the active-site glutamate goes to the carbonyl oxygen of the substrate producing enediol intermediate either directly or after it is exchanged with solvent.

Model assembly study of the ligand binding by p-hydroxybenzoate hydroxylase: correlation between the calculated binding energies and the experimental dissociation constants.

The energies of binding of seven ligands by p-hydroxybenzoate hydroxylase (PHBH) were calculated theoretically. Direct enzyme-ligand interaction energies were calculated using the ab initio quantum mechanical model assembly of the active site at the 3-21G level. Solvation energies of the ligands needed in the evaluation of the binding energies were calculated with the semiempirical AM1-SM2 method and the long-range electrostatic interaction energies between the ligands and the protein matrix classically using the static charge distributions of the ligands and the protein. Energies for proton-transfer between the ligands' OH or SH substituent at position 4 and the active-site tyrosine within the ab initio model assemblies were calculated and compared to the corresponding pKas in aqueous solution. Excluding 3,4-dihydroxybenzoate, the natural product of PHBH, a linear relationship between the calculated binding energies and the experimental binding free energies was found with a correlation coefficient of 0.90. Contributions of the direct enzyme-ligand interaction energies, solvation energies and the long-range electrostatic interaction energies to the calculated binding energies were analyzed. The proton-transfer energies of the ligands with substituents ortho to the ionized OH were found to be perturbed less in the model calculations than the energies of their meta isomers as deduced from the corresponding pKas.

Quantum mechanical model assembly study on the energetics of binding of arabinose, fucose, and galactose to L-arabinose-binding protein.

Binding energies of L-arabinose, D-fucose, and D-galactose to L-arabinose-binding protein was investigated theoretically. The calculated binding energies were composed of three contributions: 1) direct ligand-active site interaction energies calculated using static ab initio model assemblies; 2) solvation energies of the ligands; and 3) long-range electrostatic interaction energies between the ligands and the protein matrix. The calculated binding energies and the contributions of the energy components were used to analyze the experimental affinities of the ligands.

Quantum chemical study on the interaction of some bisphosphonates and Ca2+: the role of molecular electrostatic potentials in the prediction of binding geometry.

Molecular electrostatic potentials have been used to model the calcium binding properties of some bisphosphonate drugs, which are used to treat various bone diseases. The mechanism of action involves the binding of bisphosphonates to the bone surface, where calcium plays an important role. Electrostatic potential maps derived from ab initio partial charges have been compared with both the crystal structure and the fully optimized ab initio structure of (dichloro)methylene-bisphosphonate-calcium ion complex. Molecular electrostatic potentials can correctly predict the calcium binding geometry of bisphosphonate-type compounds and this type of information can be used in the practical drug design work.