The vascular endothelial growth factor (VEGF) seems to be the most important regulator of physiological and pathological angiogenesis, being, for this reason, a favorite target for therapies against angiogenesis-related diseases. VEGF is a homodimer in which the monomers are formed by beta-strands interconnected on the poles by three loops. A recent work showed that an intimate relationship between loops-1 and -3 is required for high affinity binding to the receptors (Kiba et al., J Biol Chem 2003;278:13453-13461). In this work, we report the results of a 10-ns molecular dynamics simulation of VEGF. We analyzed the dynamical behavior of the protein (using a dynamical cross-correlation map) and found that it is governed by a high degree of correlation between the motions of the loops. We also performed a principal component analysis and found an overall motion in which the opposite poles are projected against each other, just like the movement of the wings of a butterfly. From the biological point of view, it is likely that this motion would facilitate receptor binding since VEGF must enter a restricted cavity formed by the two subunits of the receptor.

Simulações de Monte Carlo foram realizadas para soluções aquosas concentradas de uréia nas concentrações de 4, 5, 6, 7 e 8 mol L -1 no ensemble NpT a 298 K e 1 atm. As superfícies de energia potencial foram representadas pela soma dos potenciais clássicos de Coulomb e de Lennard-Jones, nos quais a aditividade de pares foi considerada. Nas simulações a água foi descrita pelo modelo TIP4P e a uréia por dois modelos: o modelo OPLS planar e um modelo no qual a molécula de uréia não é totalmente planar. Os resultados das simulações para os dois modelos mostraram que a uréia não causa perturbações significativas na estrutura da água, na faixa de concentrações estudadas. No entanto, foi observada diminuição do número médio de ligações de hidrogênio entre as moléculas de água. Para as soluções mais concentradas os resultados obtidos com o modelo OPLS sugerem a formação de dímeros de uréia. Observou-se similaridade entre alguns dos resultados das simulações obtidos com os modelos: OPLS planar e não-planar da uréia , nas mesmas concentrações e condições.Monte Carlo simulations were carried out for concentrated aqueous urea solutions at 4, 5, 6, 7 and 8 mol L -1 in the NpT ensemble at 298 K and 1 atm. The potential energy surfaces were represented by the sum of Coulomb and Lennard-Jones potentials in which pair-wise additivity was considered. Water was described by the TIP4P model, while urea was described by the OPLS planar model and also by a proposed non-planar model. The simulation results for both urea models showed that urea does not induce any significant changes in the structure of water over all the concentrations studied. However, it was noted that the number of hydrogen bonds between the water molecules decrease as the concentration increases. The simulation results of the OPLS planar model suggest the formation of urea dimers in the more concentrated solutions. The simulations with the non-planar model showed several features similar to those of the OPLS planar model.

A series of 74 dihydroalkoxybenzyloxopyrimidines (DABOs), a class of highly potent non-nucleoside reverse transcriptase inhibitors (NNRTIs), was retrieved from the literature and studied by comparative molecular field analysis (CoMFA) in order to derive three-dimensional quantitative structure-activity relationship (3D-QSAR) models. The CoMFA study has been performed with a training set of 59 compounds, testing three alignments and four charge schemes (DFT, HF, AM1, and PM3) and using defaults probe atom (Csp (3), +1 charge), cutoffs (30 kcal.mol (-1) for both steric and electrostatic fields), and grid distance (2.0 A). The best model ( N = 59), derived from Alignment 1 and PM3 charges, shows q (2) = 0.691, SE cv = 0.475, optimum number of components = 6, r (2) = 0.930, SEE = 0.226, and F-value = 115.544. The steric and electrostatic contributions for the best model were 43.2% and 56.8%, respectively. The external predictive ability (r (2) pred = 0.918) of the resultant best model was evaluated using a test set of 15 compounds. In order to design more potent DABO analogues as anti-HIV/AIDS agents, attention should be taken in order to select a substituent for the 4-oxopyrimidine ring, since, as revealed by the best CoMFA model, there are a steric restriction at the C2-position, a electron-rich group restriction at the C6-position ( para-substituent of the 6-benzyl group), and a steric allowed region at the C5-position.

No abstract

No abstract

Recebido em 3/1/01; aceito em 12/9/01 STUDIES OF UREA GEOMETRY BY MEANS OF AB INITIO METHODS AND COMPUTER SIMULATIONS OF LIQUIDS. A study was carried out on the urea geometries using ab initio calculation and Monte Carlo computational simulation of liquids. The ab initio calculated results showed that urea has a non-planar conformation in the gas phase in which the hydrogen atoms are out of the plane formed by the heavy atoms. Free energies associated to the rotation of the amino groups of urea in water were obtained using the Monte Carlo method in which the thermodynamic perturbation theory is implemented. The magnitude of the free energy obtained from this simulation did not permit us to conclude that urea is non-planar in water.

Recebido em 9/12/03; aceito em 1/9/04; publicado na web em 12/11/04 STRUCTURE AND PROPERTIES OF ELLIPTICINES. The ellipticines constitute a broad class of molecules with antitumor activity. In the present work we analyzed the structure and properties of a series of ellipticine derivatives in the gas phase and in solution using quantum mechanical and Monte Carlo methods. The results showed a good correlation between the solvation energies in water obtained with the continuum model and the Monte Carlo simulation. Molecular descriptors were considered in the development of QSAR models using the DNA association constant (log K app ) as biological data. The results showed that the DNA binding is dominated by electronic parameters, with small contributions from the molecular volume and area.Keywords: ellipticine; ab initio calculation; QSAR. INTRODUÇÃOElipticinas (Figura 1) constituem uma ampla classe de compostos com atividade antitumoral 1 . Além da alta citotoxicidade em célu-las cancerosas, estas moléculas são especialmente interessantes do ponto de vista clínico devido aos efeitos colaterais reduzidos 2 . O modo de ação das elipticinas tem sido amplamente estudado, entretanto, conclusões definitivas sobre os aspectos moleculares relevantes para a ação biológica não foram ainda relatadas 3 . O núcleo molecular básico das elipticinas é constituído de um grupamento carbazol ligado a um anel piridina, propiciando a deslocalização eletrônica sobre toda a molécula. Esta estrutura relativamente simples tem permitido a síntese de mais de 70 análogos distintos 3 , com as principais modificações sendo a inclusão de um grupo hidroxila nas posições C 9 e C 7 , e alquilação das posições N 2 , N 6 e C 1 (ver Figura 1). Mais recentemente 4 , novas elipticinas substituídas em C 9 foram sintetizadas visando obter derivados com citotoxicidade seletiva.No presente trabalho, a estrutura e as propriedades moleculares de algumas elipticinas foram determinadas através de cálculos ab initio em fase gasosa. As energias de solvatação foram calculadas utilizando modelos contínuos de solvatação e simulação de Monte Carlo. A estrutura do solvente ao redor de alguns sítios das molécu-las foi analisada visando identificar características do processo de solvatação, específicas para os diferentes análogos. Na parte final do trabalho, modelos quantitativos de relação estrutura-atividade (QSAR) foram construídos e analisados. METODOLOGIA DE CÁLCULOO conjunto de moléculas estudadas no presente trabalho é apresentado na Tabela 1. Estes análogos foram selecionados de acordo com a disponibilidade de parâmetros biológicos quantitativos 3 e características estruturais dissimilares dentro desta classe de moléculas. Para cada composto descrito na Tabela 1, a geometria foi completamente otimizada no nível HF/6-31G(d) e caracterizada como mínimo verdadeiro na Superfície de Energia Potencial (PES) através do cálcu-lo de freqüências harmônicas. Cargas atômicas ChelpG (Q), momento de dipolo elétrico (µ) e energias dos orbitais moleculares de fronteira (HOMO e ...

We recently described the dynamical behavior of the vascular endothelial growth factor (VEGF) by the application of molecular dynamics (MD) simulations. 1 The collective motions were extracted by the use of two methods: (i) the analysis of the dynamical cross-correlation map (DCCM) obtained by plotting the elements of the N 3 N dynamical cross-correlation matrix; and (ii) the analysis of essential motional modes obtained from the diagonalization of the 3N 3 3N covariance matrix (i.e. referred herein as PCA: principal component analysis). Both analyses suggested that there exists a type of motion in which the residues that form the same pole of the protein move with a high degree of correlation, but residues belonging to different poles move with an anticorrelation.The DCCM analysis is based on the interpretation of maps that indicates whether two residues show correlations or anticorrelations between their fluctuations. The map is a two dimensional plot of the elements of the dynamical cross-correlation matrix, which corresponds to the normalized covariance matrix of atomic displacements from the average structure. 2,3In Cartesian coordinates, the overall translational and rotational motions must be removed. Otherwise, spurious long-range correlations and anticorrelations will appear since these overall motions refer to the motion of the whole molecule (i.e. motion of a rigid body), in which all atoms are strongly coupled. The removal of overall motions is usually performed by least square fitting superimposition. The a-carbon atoms are often used in the least square fitting. Hünenberger et al. have shown that spurious correlations may arise if the most flexible residues are used in the fitting procedure, 2 thus it is preferred to use only the most rigid residues (i.e. those with lowest B-factor values).In Ref. 1 only the most rigid residues (18-33, 46-58, 68-78 and 90-100) were employed for the fitting procedure for DCCM calculations. 1 An ensemble of PDB structure was generated using the ''trjconv'' module of the GROMACS simulation package 4,5 and the analysis performed using a code developed by our group. The results were then interpreted and provided a good explanation for the observed experimental data. 1,6,7 Erroneously, only the overall translational motion was removed, the overall rotational motion was not removed. Here we show the corrected DCCM (i.e. calculated with the same trajectory of the original paper) for the 10 ns simulation and its interpretation. However, we stress that the rest of the analysis presented (figure 4 and figures 7 and 8) were not affected. Figure 1 shows the corrected DCCM calculated by averaging 5 blocks of 1 ns length over the last 5 ns subpart of a 10 ns trajectory (exactly as was performed in the original paper). By comparing figure 1 (new DCCM) with figure 5 of Ref. 1 (old DCCM), it is clear that the old DCCM shows much more correlations and anticorrelations than the new DCCM. The presence of more correlated regions in the old DCCM is due to the presence of overall rotation...